diff --git a/sql/upgrades/174.sql b/sql/upgrades/174.sql new file mode 100644 index 00000000..e4a6cda3 --- /dev/null +++ b/sql/upgrades/174.sql @@ -0,0 +1,12 @@ +-- SWH DB schema upgrade +-- from_version: 173 +-- to_version: 174 +-- description: remove authority and fetcher metadata + +insert into dbversion(version, release, description) + values(174, now(), 'Work In Progress'); + +alter table metadata_authority + drop column metadata; +alter table metadata_fetcher + drop column metadata; diff --git a/swh/storage/backfill.py b/swh/storage/backfill.py index d1bd9deb..3df10b2c 100644 --- a/swh/storage/backfill.py +++ b/swh/storage/backfill.py @@ -1,646 +1,646 @@ # Copyright (C) 2017-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information """Storage backfiller. The backfiller goal is to produce back part or all of the objects from a storage to the journal topics Current implementation consists in the JournalBackfiller class. It simply reads the objects from the storage and sends every object identifier back to the journal. """ import logging from typing import Any, Callable, Dict, Optional from swh.core.db import BaseDb from swh.model.identifiers import ExtendedObjectType from swh.model.model import ( BaseModel, Directory, DirectoryEntry, ExtID, RawExtrinsicMetadata, Release, Revision, Snapshot, SnapshotBranch, TargetType, ) from swh.storage.postgresql.converters import ( db_to_extid, db_to_raw_extrinsic_metadata, db_to_release, db_to_revision, ) from swh.storage.replay import object_converter_fn from swh.storage.writer import JournalWriter logger = logging.getLogger(__name__) PARTITION_KEY = { "content": "sha1", "skipped_content": "sha1", "directory": "id", "extid": "target", "metadata_authority": "type, url", "metadata_fetcher": "name, version", "raw_extrinsic_metadata": "target", "revision": "revision.id", "release": "release.id", "snapshot": "id", "origin": "id", "origin_visit": "origin_visit.origin", "origin_visit_status": "origin_visit_status.origin", } COLUMNS = { "content": [ "sha1", "sha1_git", "sha256", "blake2s256", "length", "status", "ctime", ], "skipped_content": [ "sha1", "sha1_git", "sha256", "blake2s256", "length", "ctime", "status", "reason", ], "directory": ["id", "dir_entries", "file_entries", "rev_entries"], "extid": ["extid_type", "extid", "target_type", "target"], - "metadata_authority": ["type", "url", "metadata",], - "metadata_fetcher": ["name", "version", "metadata",], + "metadata_authority": ["type", "url"], + "metadata_fetcher": ["name", "version"], "origin": ["url"], "origin_visit": ["visit", "type", ("origin.url", "origin"), "date",], "origin_visit_status": [ ("origin_visit_status.visit", "visit"), ("origin.url", "origin"), ("origin_visit_status.date", "date"), "type", "snapshot", "status", "metadata", ], "raw_extrinsic_metadata": [ "raw_extrinsic_metadata.type", "raw_extrinsic_metadata.target", "metadata_authority.type", "metadata_authority.url", "metadata_fetcher.name", "metadata_fetcher.version", "discovery_date", "format", "raw_extrinsic_metadata.metadata", "origin", "visit", "snapshot", "release", "revision", "path", "directory", ], "revision": [ ("revision.id", "id"), "date", "date_offset", "date_neg_utc_offset", "committer_date", "committer_date_offset", "committer_date_neg_utc_offset", "type", "directory", "message", "synthetic", "metadata", "extra_headers", ( "array(select parent_id::bytea from revision_history rh " "where rh.id = revision.id order by rh.parent_rank asc)", "parents", ), ("a.id", "author_id"), ("a.name", "author_name"), ("a.email", "author_email"), ("a.fullname", "author_fullname"), ("c.id", "committer_id"), ("c.name", "committer_name"), ("c.email", "committer_email"), ("c.fullname", "committer_fullname"), ], "release": [ ("release.id", "id"), "date", "date_offset", "date_neg_utc_offset", "comment", ("release.name", "name"), "synthetic", "target", "target_type", ("a.id", "author_id"), ("a.name", "author_name"), ("a.email", "author_email"), ("a.fullname", "author_fullname"), ], "snapshot": ["id", "object_id"], } JOINS = { "release": ["person a on release.author=a.id"], "revision": [ "person a on revision.author=a.id", "person c on revision.committer=c.id", ], "origin_visit": ["origin on origin_visit.origin=origin.id"], "origin_visit_status": ["origin on origin_visit_status.origin=origin.id",], "raw_extrinsic_metadata": [ "metadata_authority on " "raw_extrinsic_metadata.authority_id=metadata_authority.id", "metadata_fetcher on raw_extrinsic_metadata.fetcher_id=metadata_fetcher.id", ], } def directory_converter(db: BaseDb, directory_d: Dict[str, Any]) -> Directory: """Convert directory from the flat representation to swh model compatible objects. """ columns = ["target", "name", "perms"] query_template = """ select %(columns)s from directory_entry_%(type)s where id in %%s """ types = ["file", "dir", "rev"] entries = [] with db.cursor() as cur: for type in types: ids = directory_d.pop("%s_entries" % type) if not ids: continue query = query_template % { "columns": ",".join(columns), "type": type, } cur.execute(query, (tuple(ids),)) for row in cur: entry_d = dict(zip(columns, row)) entry = DirectoryEntry( name=entry_d["name"], type=type, target=entry_d["target"], perms=entry_d["perms"], ) entries.append(entry) return Directory(id=directory_d["id"], entries=tuple(entries),) def raw_extrinsic_metadata_converter( db: BaseDb, metadata: Dict[str, Any] ) -> RawExtrinsicMetadata: """Convert a raw extrinsic metadata from the flat representation to swh model compatible objects. """ return db_to_raw_extrinsic_metadata(metadata) def extid_converter(db: BaseDb, extid: Dict[str, Any]) -> ExtID: """Convert an extid from the flat representation to swh model compatible objects. """ return db_to_extid(extid) def revision_converter(db: BaseDb, revision_d: Dict[str, Any]) -> Revision: """Convert revision from the flat representation to swh model compatible objects. """ revision = db_to_revision(revision_d) assert revision is not None, revision_d["id"] return revision def release_converter(db: BaseDb, release_d: Dict[str, Any]) -> Release: """Convert release from the flat representation to swh model compatible objects. """ release = db_to_release(release_d) assert release is not None, release_d["id"] return release def snapshot_converter(db: BaseDb, snapshot_d: Dict[str, Any]) -> Snapshot: """Convert snapshot from the flat representation to swh model compatible objects. """ columns = ["name", "target", "target_type"] query = """ select %s from snapshot_branches sbs inner join snapshot_branch sb on sb.object_id=sbs.branch_id where sbs.snapshot_id=%%s """ % ", ".join( columns ) with db.cursor() as cur: cur.execute(query, (snapshot_d["object_id"],)) branches = {} for name, *row in cur: branch_d = dict(zip(columns[1:], row)) if branch_d["target"] is not None and branch_d["target_type"] is not None: branch: Optional[SnapshotBranch] = SnapshotBranch( target=branch_d["target"], target_type=TargetType(branch_d["target_type"]), ) else: branch = None branches[name] = branch return Snapshot(id=snapshot_d["id"], branches=branches,) CONVERTERS: Dict[str, Callable[[BaseDb, Dict[str, Any]], BaseModel]] = { "directory": directory_converter, "extid": extid_converter, "raw_extrinsic_metadata": raw_extrinsic_metadata_converter, "revision": revision_converter, "release": release_converter, "snapshot": snapshot_converter, } def object_to_offset(object_id, numbits): """Compute the index of the range containing object id, when dividing space into 2^numbits. Args: object_id (str): The hex representation of object_id numbits (int): Number of bits in which we divide input space Returns: The index of the range containing object id """ q, r = divmod(numbits, 8) length = q + (r != 0) shift_bits = 8 - r if r else 0 truncated_id = object_id[: length * 2] if len(truncated_id) < length * 2: truncated_id += "0" * (length * 2 - len(truncated_id)) truncated_id_bytes = bytes.fromhex(truncated_id) return int.from_bytes(truncated_id_bytes, byteorder="big") >> shift_bits def byte_ranges(numbits, start_object=None, end_object=None): """Generate start/end pairs of bytes spanning numbits bits and constrained by optional start_object and end_object. Args: numbits (int): Number of bits in which we divide input space start_object (str): Hex object id contained in the first range returned end_object (str): Hex object id contained in the last range returned Yields: 2^numbits pairs of bytes """ q, r = divmod(numbits, 8) length = q + (r != 0) shift_bits = 8 - r if r else 0 def to_bytes(i): return int.to_bytes(i << shift_bits, length=length, byteorder="big") start_offset = 0 end_offset = 1 << numbits if start_object is not None: start_offset = object_to_offset(start_object, numbits) if end_object is not None: end_offset = object_to_offset(end_object, numbits) + 1 for start in range(start_offset, end_offset): end = start + 1 if start == 0: yield None, to_bytes(end) elif end == 1 << numbits: yield to_bytes(start), None else: yield to_bytes(start), to_bytes(end) def raw_extrinsic_metadata_target_ranges(start_object=None, end_object=None): """Generate ranges of values for the `target` attribute of `raw_extrinsic_metadata` objects. This generates one range for all values before the first SWHID (which would correspond to raw origin URLs), then a number of hex-based ranges for each known type of SWHID (2**12 ranges for directories, 2**8 ranges for all other types). Finally, it generates one extra range for values above all possible SWHIDs. """ if start_object is None: start_object = "" swhid_target_types = sorted(type.value for type in ExtendedObjectType) first_swhid = f"swh:1:{swhid_target_types[0]}:" # Generate a range for url targets, if the starting object is before SWHIDs if start_object < first_swhid: yield start_object, ( first_swhid if end_object is None or end_object >= first_swhid else end_object ) if end_object is not None and end_object <= first_swhid: return # Prime the following loop, which uses the upper bound of the previous range # as lower bound, to account for potential targets between two valid types # of SWHIDs (even though they would eventually be rejected by the # RawExtrinsicMetadata parser, they /might/ exist...) end_swhid = first_swhid # Generate ranges for swhid targets for target_type in swhid_target_types: finished = False base_swhid = f"swh:1:{target_type}:" last_swhid = base_swhid + ("f" * 40) if start_object > last_swhid: continue # Generate 2**8 or 2**12 ranges for _, end in byte_ranges(12 if target_type == "dir" else 8): # Reuse previous uppper bound start_swhid = end_swhid # Use last_swhid for this object type if on the last byte range end_swhid = (base_swhid + end.hex()) if end is not None else last_swhid # Ignore out of bounds ranges if start_object >= end_swhid: continue # Potentially clamp start of range to the first object requested start_swhid = max(start_swhid, start_object) # Handle ending the loop early if the last requested object id is in # the current range if end_object is not None and end_swhid >= end_object: end_swhid = end_object finished = True yield start_swhid, end_swhid if finished: return # Generate one final range for potential raw origin URLs after the last # valid SWHID start_swhid = max(start_object, end_swhid) yield start_swhid, end_object def integer_ranges(start, end, block_size=1000): for start in range(start, end, block_size): if start == 0: yield None, block_size elif start + block_size > end: yield start, end else: yield start, start + block_size RANGE_GENERATORS = { "content": lambda start, end: byte_ranges(24, start, end), "skipped_content": lambda start, end: [(None, None)], "directory": lambda start, end: byte_ranges(24, start, end), "extid": lambda start, end: byte_ranges(24, start, end), "revision": lambda start, end: byte_ranges(24, start, end), "release": lambda start, end: byte_ranges(16, start, end), "raw_extrinsic_metadata": raw_extrinsic_metadata_target_ranges, "snapshot": lambda start, end: byte_ranges(16, start, end), "origin": integer_ranges, "origin_visit": integer_ranges, "origin_visit_status": integer_ranges, } def compute_query(obj_type, start, end): columns = COLUMNS.get(obj_type) join_specs = JOINS.get(obj_type, []) join_clause = "\n".join("left join %s" % clause for clause in join_specs) where = [] where_args = [] if start: where.append("%(keys)s >= %%s") where_args.append(start) if end: where.append("%(keys)s < %%s") where_args.append(end) where_clause = "" if where: where_clause = ("where " + " and ".join(where)) % { "keys": "(%s)" % PARTITION_KEY[obj_type] } column_specs = [] column_aliases = [] for column in columns: if isinstance(column, str): column_specs.append(column) column_aliases.append(column) else: column_specs.append("%s as %s" % column) column_aliases.append(column[1]) query = """ select %(columns)s from %(table)s %(join)s %(where)s """ % { "columns": ",".join(column_specs), "table": obj_type, "join": join_clause, "where": where_clause, } return query, where_args, column_aliases def fetch(db, obj_type, start, end): """Fetch all obj_type's identifiers from db. This opens one connection, stream objects and when done, close the connection. Args: db (BaseDb): Db connection object obj_type (str): Object type start (Union[bytes|Tuple]): Range start identifier end (Union[bytes|Tuple]): Range end identifier Raises: ValueError if obj_type is not supported Yields: Objects in the given range """ query, where_args, column_aliases = compute_query(obj_type, start, end) converter = CONVERTERS.get(obj_type) with db.cursor() as cursor: logger.debug("Fetching data for table %s", obj_type) logger.debug("query: %s %s", query, where_args) cursor.execute(query, where_args) for row in cursor: record = dict(zip(column_aliases, row)) if converter: record = converter(db, record) else: record = object_converter_fn[obj_type](record) logger.debug("record: %s", record) yield record def _format_range_bound(bound): if isinstance(bound, bytes): return bound.hex() else: return str(bound) MANDATORY_KEYS = ["storage", "journal_writer"] class JournalBackfiller: """Class in charge of reading the storage's objects and sends those back to the journal's topics. This is designed to be run periodically. """ def __init__(self, config=None): self.config = config self.check_config(config) def check_config(self, config): missing_keys = [] for key in MANDATORY_KEYS: if not config.get(key): missing_keys.append(key) if missing_keys: raise ValueError( "Configuration error: The following keys must be" " provided: %s" % (",".join(missing_keys),) ) if "cls" not in config["storage"] or config["storage"]["cls"] != "local": raise ValueError( "swh storage backfiller must be configured to use a local" " (PostgreSQL) storage" ) def parse_arguments(self, object_type, start_object, end_object): """Parse arguments Raises: ValueError for unsupported object type ValueError if object ids are not parseable Returns: Parsed start and end object ids """ if object_type not in COLUMNS: raise ValueError( "Object type %s is not supported. " "The only possible values are %s" % (object_type, ", ".join(sorted(COLUMNS.keys()))) ) if object_type in ["origin", "origin_visit", "origin_visit_status"]: if start_object: start_object = int(start_object) else: start_object = 0 if end_object: end_object = int(end_object) else: end_object = 100 * 1000 * 1000 # hard-coded limit return start_object, end_object def run(self, object_type, start_object, end_object, dry_run=False): """Reads storage's subscribed object types and send them to the journal's reading topic. """ start_object, end_object = self.parse_arguments( object_type, start_object, end_object ) db = BaseDb.connect(self.config["storage"]["db"]) writer = JournalWriter({"cls": "kafka", **self.config["journal_writer"]}) assert writer.journal is not None for range_start, range_end in RANGE_GENERATORS[object_type]( start_object, end_object ): logger.info( "Processing %s range %s to %s", object_type, _format_range_bound(range_start), _format_range_bound(range_end), ) objects = fetch(db, object_type, start=range_start, end=range_end) if not dry_run: writer.write_additions(object_type, objects) else: # only consume the objects iterator to check for any potential # decoding/encoding errors for obj in objects: pass if __name__ == "__main__": print('Please use the "swh-journal backfiller run" command') diff --git a/swh/storage/cassandra/model.py b/swh/storage/cassandra/model.py index 00eb1057..e3f8d569 100644 --- a/swh/storage/cassandra/model.py +++ b/swh/storage/cassandra/model.py @@ -1,323 +1,321 @@ # Copyright (C) 2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information """Classes representing tables in the Cassandra database. They are very close to classes found in swh.model.model, but most of them are subtly different: * Large objects are split into other classes (eg. RevisionRow has no 'parents' field, because parents are stored in a different table, represented by RevisionParentRow) * They have a "cols" field, which returns the list of column names of the table * They only use types that map directly to Cassandra's schema (ie. no enums) Therefore, this model doesn't reuse swh.model.model, except for types that can be mapped to UDTs (Person and TimestampWithTimezone). """ import dataclasses import datetime from typing import Any, ClassVar, Dict, List, Optional, Tuple, Type, TypeVar from swh.model.model import Person, TimestampWithTimezone MAGIC_NULL_PK = b"" """ NULLs (or all-empty blobs) are not allowed in primary keys; instead we use a special value that can't possibly be a valid hash. """ T = TypeVar("T", bound="BaseRow") def content_index_table_name(algo: str, skipped_content: bool) -> str: """Given an algorithm name, returns the name of one of the 'content_by_*' and 'skipped_content_by_*' tables that serve as index for the 'content' and 'skipped_content' tables based on this algorithm's hashes. For now it is a simple substitution, but future versions may append a version number to it, if needed for schema updates.""" if skipped_content: return f"skipped_content_by_{algo}" else: return f"content_by_{algo}" class BaseRow: TABLE: ClassVar[str] PARTITION_KEY: ClassVar[Tuple[str, ...]] CLUSTERING_KEY: ClassVar[Tuple[str, ...]] = () @classmethod def from_dict(cls: Type[T], d: Dict[str, Any]) -> T: return cls(**d) # type: ignore @classmethod def cols(cls) -> List[str]: return [field.name for field in dataclasses.fields(cls)] def to_dict(self) -> Dict[str, Any]: return dataclasses.asdict(self) @dataclasses.dataclass class ContentRow(BaseRow): TABLE = "content" PARTITION_KEY: ClassVar[Tuple[str, ...]] = ( "sha1", "sha1_git", "sha256", "blake2s256", ) sha1: bytes sha1_git: bytes sha256: bytes blake2s256: bytes length: int ctime: datetime.datetime status: str @dataclasses.dataclass class SkippedContentRow(BaseRow): TABLE = "skipped_content" PARTITION_KEY = ("sha1", "sha1_git", "sha256", "blake2s256") sha1: Optional[bytes] sha1_git: Optional[bytes] sha256: Optional[bytes] blake2s256: Optional[bytes] length: Optional[int] ctime: Optional[datetime.datetime] status: str reason: str origin: str @classmethod def from_dict(cls, d: Dict[str, Any]) -> "SkippedContentRow": d = d.copy() for k in ("sha1", "sha1_git", "sha256", "blake2s256"): if d[k] == MAGIC_NULL_PK: d[k] = None return super().from_dict(d) @dataclasses.dataclass class DirectoryRow(BaseRow): TABLE = "directory" PARTITION_KEY = ("id",) id: bytes @dataclasses.dataclass class DirectoryEntryRow(BaseRow): TABLE = "directory_entry" PARTITION_KEY = ("directory_id",) CLUSTERING_KEY = ("name",) directory_id: bytes name: bytes target: bytes perms: int type: str @dataclasses.dataclass class RevisionRow(BaseRow): TABLE = "revision" PARTITION_KEY = ("id",) id: bytes date: Optional[TimestampWithTimezone] committer_date: Optional[TimestampWithTimezone] type: str directory: bytes message: bytes author: Person committer: Person synthetic: bool metadata: str extra_headers: dict @dataclasses.dataclass class RevisionParentRow(BaseRow): TABLE = "revision_parent" PARTITION_KEY = ("id",) CLUSTERING_KEY = ("parent_rank",) id: bytes parent_rank: int parent_id: bytes @dataclasses.dataclass class ReleaseRow(BaseRow): TABLE = "release" PARTITION_KEY = ("id",) id: bytes target_type: str target: bytes date: TimestampWithTimezone name: bytes message: bytes author: Person synthetic: bool @dataclasses.dataclass class SnapshotRow(BaseRow): TABLE = "snapshot" PARTITION_KEY = ("id",) id: bytes @dataclasses.dataclass class SnapshotBranchRow(BaseRow): TABLE = "snapshot_branch" PARTITION_KEY = ("snapshot_id",) CLUSTERING_KEY = ("name",) snapshot_id: bytes name: bytes target_type: Optional[str] target: Optional[bytes] @dataclasses.dataclass class OriginVisitRow(BaseRow): TABLE = "origin_visit" PARTITION_KEY = ("origin",) CLUSTERING_KEY = ("visit",) origin: str visit: int date: datetime.datetime type: str @dataclasses.dataclass class OriginVisitStatusRow(BaseRow): TABLE = "origin_visit_status" PARTITION_KEY = ("origin",) CLUSTERING_KEY = ("visit", "date") origin: str visit: int date: datetime.datetime type: str status: str metadata: str snapshot: bytes @classmethod def from_dict(cls: Type[T], d: Dict[str, Any]) -> T: return cls(**d) # type: ignore @dataclasses.dataclass class OriginRow(BaseRow): TABLE = "origin" PARTITION_KEY = ("sha1",) sha1: bytes url: str next_visit_id: int @dataclasses.dataclass class MetadataAuthorityRow(BaseRow): TABLE = "metadata_authority" PARTITION_KEY = ("url",) CLUSTERING_KEY = ("type",) url: str type: str - metadata: str @dataclasses.dataclass class MetadataFetcherRow(BaseRow): TABLE = "metadata_fetcher" PARTITION_KEY = ("name",) CLUSTERING_KEY = ("version",) name: str version: str - metadata: str @dataclasses.dataclass class RawExtrinsicMetadataRow(BaseRow): TABLE = "raw_extrinsic_metadata" PARTITION_KEY = ("target",) CLUSTERING_KEY = ( "authority_type", "authority_url", "discovery_date", "id", ) id: bytes type: str target: str authority_type: str authority_url: str discovery_date: datetime.datetime fetcher_name: str fetcher_version: str format: str metadata: bytes origin: Optional[str] visit: Optional[int] snapshot: Optional[str] release: Optional[str] revision: Optional[str] path: Optional[bytes] directory: Optional[str] @dataclasses.dataclass class ObjectCountRow(BaseRow): TABLE = "object_count" PARTITION_KEY = ("partition_key",) CLUSTERING_KEY = ("object_type",) partition_key: int object_type: str count: int @dataclasses.dataclass class ExtIDRow(BaseRow): TABLE = "extid" PARTITION_KEY = ("target", "target_type", "extid", "extid_type") extid_type: str extid: bytes target_type: str target: bytes @dataclasses.dataclass class ExtIDByTargetRow(BaseRow): TABLE = "extid_by_target" PARTITION_KEY = ("target_type", "target") CLUSTERING_KEY = ("target_token",) target_type: str target: bytes target_token: int diff --git a/swh/storage/cassandra/schema.py b/swh/storage/cassandra/schema.py index 12ff1820..19e62007 100644 --- a/swh/storage/cassandra/schema.py +++ b/swh/storage/cassandra/schema.py @@ -1,317 +1,315 @@ # Copyright (C) 2019-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information CREATE_TABLES_QUERIES = [ """ CREATE OR REPLACE FUNCTION ascii_bins_count_sfunc ( state tuple>, -- (nb_none, map) bin_name ascii ) CALLED ON NULL INPUT RETURNS tuple> LANGUAGE java AS $$ if (bin_name == null) { state.setInt(0, state.getInt(0) + 1); } else { Map counters = state.getMap( 1, String.class, Integer.class); Integer nb = counters.get(bin_name); if (nb == null) { nb = 0; } counters.put(bin_name, nb + 1); state.setMap(1, counters, String.class, Integer.class); } return state; $$ ;""", """ CREATE OR REPLACE AGGREGATE ascii_bins_count ( ascii ) SFUNC ascii_bins_count_sfunc STYPE tuple> INITCOND (0, {}) ;""", """ CREATE TYPE IF NOT EXISTS microtimestamp ( seconds bigint, microseconds int );""", """ CREATE TYPE IF NOT EXISTS microtimestamp_with_timezone ( timestamp frozen, offset smallint, negative_utc boolean );""", """ CREATE TYPE IF NOT EXISTS person ( fullname blob, name blob, email blob );""", """ CREATE TABLE IF NOT EXISTS content ( sha1 blob, sha1_git blob, sha256 blob, blake2s256 blob, length bigint, ctime timestamp, -- creation time, i.e. time of (first) injection into the storage status ascii, PRIMARY KEY ((sha1, sha1_git, sha256, blake2s256)) );""", """ CREATE TABLE IF NOT EXISTS skipped_content ( sha1 blob, sha1_git blob, sha256 blob, blake2s256 blob, length bigint, ctime timestamp, -- creation time, i.e. time of (first) injection into the storage status ascii, reason text, origin text, PRIMARY KEY ((sha1, sha1_git, sha256, blake2s256)) );""", """ CREATE TABLE IF NOT EXISTS revision ( id blob PRIMARY KEY, date microtimestamp_with_timezone, committer_date microtimestamp_with_timezone, type ascii, directory blob, -- source code "root" directory message blob, author person, committer person, synthetic boolean, -- true iff revision has been created by Software Heritage metadata text, -- extra metadata as JSON(tarball checksums, etc...) extra_headers frozen> > -- extra commit information as (tuple(key, value), ...) );""", """ CREATE TABLE IF NOT EXISTS revision_parent ( id blob, parent_rank int, -- parent position in merge commits, 0-based parent_id blob, PRIMARY KEY ((id), parent_rank) );""", """ CREATE TABLE IF NOT EXISTS release ( id blob PRIMARY KEY, target_type ascii, target blob, date microtimestamp_with_timezone, name blob, message blob, author person, synthetic boolean, -- true iff release has been created by Software Heritage );""", """ CREATE TABLE IF NOT EXISTS directory ( id blob PRIMARY KEY, );""", """ CREATE TABLE IF NOT EXISTS directory_entry ( directory_id blob, name blob, -- path name, relative to containing dir target blob, perms int, -- unix-like permissions type ascii, -- target type PRIMARY KEY ((directory_id), name) );""", """ CREATE TABLE IF NOT EXISTS snapshot ( id blob PRIMARY KEY, );""", """ -- For a given snapshot_id, branches are sorted by their name, -- allowing easy pagination. CREATE TABLE IF NOT EXISTS snapshot_branch ( snapshot_id blob, name blob, target_type ascii, target blob, PRIMARY KEY ((snapshot_id), name) );""", """ CREATE TABLE IF NOT EXISTS origin_visit ( origin text, visit bigint, date timestamp, type text, PRIMARY KEY ((origin), visit) );""", """ CREATE TABLE IF NOT EXISTS origin_visit_status ( origin text, visit bigint, date timestamp, type text, status ascii, metadata text, snapshot blob, PRIMARY KEY ((origin), visit, date) );""", """ CREATE TABLE IF NOT EXISTS origin ( sha1 blob PRIMARY KEY, url text, next_visit_id int, -- We need integer visit ids for compatibility with the pgsql -- storage, so we're using lightweight transactions with this trick: -- https://stackoverflow.com/a/29391877/539465 );""", """ CREATE TABLE IF NOT EXISTS metadata_authority ( url text, type ascii, - metadata text, PRIMARY KEY ((url), type) );""", """ CREATE TABLE IF NOT EXISTS metadata_fetcher ( name ascii, version ascii, - metadata text, PRIMARY KEY ((name), version) );""", """ CREATE TABLE IF NOT EXISTS raw_extrinsic_metadata ( id blob, type text, target text, -- metadata source authority_type text, authority_url text, discovery_date timestamp, fetcher_name ascii, fetcher_version ascii, -- metadata itself format ascii, metadata blob, -- context origin text, visit bigint, snapshot text, release text, revision text, path blob, directory text, PRIMARY KEY ((target), authority_type, authority_url, discovery_date, id) -- An explanation is in order for this primary key: -- -- Intuitively, the primary key should only be 'id', because two metadata -- entries are the same iff the id is the same; and 'id' is used for -- deduplication. -- -- However, we also want to query by -- (target, authority_type, authority_url, discovery_date) -- The naive solution to this would be an extra table, to use as index; -- but it means 1. extra code to keep them in sync 2. overhead when writing -- 3. overhead + random reads (instead of linear) when reading. -- -- Therefore, we use a single table for both, by adding the column -- we want to query with before the id. -- It solves both a) the query/order issues and b) the uniqueness issue because: -- -- a) adding the id at the end of the primary key does not change the rows' order: -- for two different rows, id1 != id2, so -- (target1, ..., date1) < (target2, ..., date2) -- <=> (target1, ..., date1, id1) < (target2, ..., date2, id2) -- -- b) the id is a hash of all the columns, so: -- rows are the same -- <=> id1 == id2 -- <=> (target1, ..., date1, id1) == (target2, ..., date2, id2) );""", """ CREATE TABLE IF NOT EXISTS object_count ( partition_key smallint, -- Constant, must always be 0 object_type ascii, count counter, PRIMARY KEY ((partition_key), object_type) );""", """ CREATE TABLE IF NOT EXISTS extid ( extid_type ascii, extid blob, target_type ascii, target blob, PRIMARY KEY ((extid_type, extid), target_type, target) );""", """ CREATE TABLE IF NOT EXISTS extid_by_target ( target_type ascii, target blob, target_token bigint, -- value of token(pk) on the "primary" table PRIMARY KEY ((target_type, target), target_token) );""", ] CONTENT_INDEX_TEMPLATE = """ -- Secondary table, used for looking up "content" from a single hash CREATE TABLE IF NOT EXISTS content_by_{main_algo} ( {main_algo} blob, target_token bigint, -- value of token(pk) on the "primary" table PRIMARY KEY (({main_algo}), target_token) ); CREATE TABLE IF NOT EXISTS skipped_content_by_{main_algo} ( {main_algo} blob, target_token bigint, -- value of token(pk) on the "primary" table PRIMARY KEY (({main_algo}), target_token) ); """ TABLES = [ "skipped_content", "content", "revision", "revision_parent", "release", "directory", "directory_entry", "snapshot", "snapshot_branch", "origin_visit", "origin", "raw_extrinsic_metadata", "object_count", "origin_visit_status", "metadata_authority", "metadata_fetcher", "extid", "extid_by_target", ] HASH_ALGORITHMS = ["sha1", "sha1_git", "sha256", "blake2s256"] for main_algo in HASH_ALGORITHMS: CREATE_TABLES_QUERIES.extend( CONTENT_INDEX_TEMPLATE.format( main_algo=main_algo, other_algos=", ".join( [algo for algo in HASH_ALGORITHMS if algo != main_algo] ), ).split("\n\n") ) TABLES.append("content_by_%s" % main_algo) TABLES.append("skipped_content_by_%s" % main_algo) diff --git a/swh/storage/cassandra/storage.py b/swh/storage/cassandra/storage.py index 2cb1aa5c..816e86f6 100644 --- a/swh/storage/cassandra/storage.py +++ b/swh/storage/cassandra/storage.py @@ -1,1483 +1,1468 @@ # Copyright (C) 2019-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import base64 import datetime import itertools -import json import random import re from typing import ( Any, Callable, Counter, Dict, Iterable, List, Optional, Sequence, Set, Tuple, Union, ) import attr from swh.core.api.classes import stream_results from swh.core.api.serializers import msgpack_dumps, msgpack_loads from swh.model.hashutil import DEFAULT_ALGORITHMS from swh.model.identifiers import CoreSWHID, ExtendedObjectType, ExtendedSWHID from swh.model.identifiers import ObjectType as SwhidObjectType from swh.model.model import ( Content, Directory, DirectoryEntry, ExtID, MetadataAuthority, MetadataAuthorityType, MetadataFetcher, Origin, OriginVisit, OriginVisitStatus, RawExtrinsicMetadata, Release, Revision, Sha1Git, SkippedContent, Snapshot, SnapshotBranch, TargetType, ) from swh.storage.interface import ( VISIT_STATUSES, ListOrder, PagedResult, PartialBranches, Sha1, ) from swh.storage.objstorage import ObjStorage from swh.storage.utils import map_optional, now from swh.storage.writer import JournalWriter from . import converters from ..exc import HashCollision, StorageArgumentException from ..utils import remove_keys from .common import TOKEN_BEGIN, TOKEN_END, hash_url from .cql import CqlRunner from .model import ( ContentRow, DirectoryEntryRow, DirectoryRow, ExtIDByTargetRow, ExtIDRow, MetadataAuthorityRow, MetadataFetcherRow, OriginRow, OriginVisitRow, OriginVisitStatusRow, RawExtrinsicMetadataRow, RevisionParentRow, SkippedContentRow, SnapshotBranchRow, SnapshotRow, ) from .schema import HASH_ALGORITHMS # Max block size of contents to return BULK_BLOCK_CONTENT_LEN_MAX = 10000 class CassandraStorage: def __init__( self, hosts, keyspace, objstorage, port=9042, journal_writer=None, allow_overwrite=False, ): """ A backend of swh-storage backed by Cassandra Args: hosts: Seed Cassandra nodes, to start connecting to the cluster keyspace: Name of the Cassandra database to use objstorage: Passed as argument to :class:`ObjStorage` port: Cassandra port journal_writer: Passed as argument to :class:`JournalWriter` allow_overwrite: Whether ``*_add`` functions will check if an object already exists in the database before sending it in an INSERT. ``False`` is the default as it is more efficient when there is a moderately high probability the object is already known, but ``True`` can be useful to overwrite existing objects (eg. when applying a schema update), or when the database is known to be mostly empty. Note that a ``False`` value does not guarantee there won't be any overwrite. """ self._hosts = hosts self._keyspace = keyspace self._port = port self._set_cql_runner() self.journal_writer: JournalWriter = JournalWriter(journal_writer) self.objstorage: ObjStorage = ObjStorage(objstorage) self._allow_overwrite = allow_overwrite def _set_cql_runner(self): """Used by tests when they need to reset the CqlRunner""" self._cql_runner: CqlRunner = CqlRunner(self._hosts, self._keyspace, self._port) def check_config(self, *, check_write: bool) -> bool: self._cql_runner.check_read() return True def _content_get_from_hash(self, algo, hash_) -> Iterable: """From the name of a hash algorithm and a value of that hash, looks up the "hash -> token" secondary table (content_by_{algo}) to get tokens. Then, looks up the main table (content) to get all contents with that token, and filters out contents whose hash doesn't match.""" found_tokens = self._cql_runner.content_get_tokens_from_single_hash(algo, hash_) for token in found_tokens: assert isinstance(token, int), found_tokens # Query the main table ('content'). res = self._cql_runner.content_get_from_token(token) for row in res: # re-check the the hash (in case of murmur3 collision) if getattr(row, algo) == hash_: yield row def _content_add(self, contents: List[Content], with_data: bool) -> Dict[str, int]: # Filter-out content already in the database. if not self._allow_overwrite: contents = [ c for c in contents if not self._cql_runner.content_get_from_pk(c.to_dict()) ] if with_data: # First insert to the objstorage, if the endpoint is # `content_add` (as opposed to `content_add_metadata`). # Must add to the objstorage before the DB and journal. Otherwise: # 1. in case of a crash the DB may "believe" we have the content, but # we didn't have time to write to the objstorage before the crash # 2. the objstorage mirroring, which reads from the journal, may attempt to # read from the objstorage before we finished writing it summary = self.objstorage.content_add( c for c in contents if c.status != "absent" ) content_add_bytes = summary["content:add:bytes"] self.journal_writer.content_add(contents) content_add = 0 for content in contents: content_add += 1 # Check for sha1 or sha1_git collisions. This test is not atomic # with the insertion, so it won't detect a collision if both # contents are inserted at the same time, but it's good enough. # # The proper way to do it would probably be a BATCH, but this # would be inefficient because of the number of partitions we # need to affect (len(HASH_ALGORITHMS)+1, which is currently 5) if not self._allow_overwrite: for algo in {"sha1", "sha1_git"}: collisions = [] # Get tokens of 'content' rows with the same value for # sha1/sha1_git rows = self._content_get_from_hash(algo, content.get_hash(algo)) for row in rows: if getattr(row, algo) != content.get_hash(algo): # collision of token(partition key), ignore this # row continue for other_algo in HASH_ALGORITHMS: if getattr(row, other_algo) != content.get_hash(other_algo): # This hash didn't match; discard the row. collisions.append( {k: getattr(row, k) for k in HASH_ALGORITHMS} ) if collisions: collisions.append(content.hashes()) raise HashCollision(algo, content.get_hash(algo), collisions) (token, insertion_finalizer) = self._cql_runner.content_add_prepare( ContentRow(**remove_keys(content.to_dict(), ("data",))) ) # Then add to index tables for algo in HASH_ALGORITHMS: self._cql_runner.content_index_add_one(algo, content, token) # Then to the main table insertion_finalizer() summary = { "content:add": content_add, } if with_data: summary["content:add:bytes"] = content_add_bytes return summary def content_add(self, content: List[Content]) -> Dict[str, int]: to_add = { (c.sha1, c.sha1_git, c.sha256, c.blake2s256): c for c in content }.values() contents = [attr.evolve(c, ctime=now()) for c in to_add] return self._content_add(list(contents), with_data=True) def content_update( self, contents: List[Dict[str, Any]], keys: List[str] = [] ) -> None: raise NotImplementedError( "content_update is not supported by the Cassandra backend" ) def content_add_metadata(self, content: List[Content]) -> Dict[str, int]: return self._content_add(content, with_data=False) def content_get_data(self, content: Sha1) -> Optional[bytes]: # FIXME: Make this method support slicing the `data` return self.objstorage.content_get(content) def content_get_partition( self, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, ) -> PagedResult[Content]: if limit is None: raise StorageArgumentException("limit should not be None") # Compute start and end of the range of tokens covered by the # requested partition partition_size = (TOKEN_END - TOKEN_BEGIN) // nb_partitions range_start = TOKEN_BEGIN + partition_id * partition_size range_end = TOKEN_BEGIN + (partition_id + 1) * partition_size # offset the range start according to the `page_token`. if page_token is not None: if not (range_start <= int(page_token) <= range_end): raise StorageArgumentException("Invalid page_token.") range_start = int(page_token) next_page_token: Optional[str] = None rows = self._cql_runner.content_get_token_range( range_start, range_end, limit + 1 ) contents = [] for counter, (tok, row) in enumerate(rows): if row.status == "absent": continue row_d = row.to_dict() if counter >= limit: next_page_token = str(tok) break row_d.pop("ctime") contents.append(Content(**row_d)) assert len(contents) <= limit return PagedResult(results=contents, next_page_token=next_page_token) def content_get(self, contents: List[Sha1]) -> List[Optional[Content]]: contents_by_sha1: Dict[Sha1, Optional[Content]] = {} for sha1 in contents: # Get all (sha1, sha1_git, sha256, blake2s256) whose sha1 # matches the argument, from the index table ('content_by_sha1') for row in self._content_get_from_hash("sha1", sha1): row_d = row.to_dict() row_d.pop("ctime") content = Content(**row_d) contents_by_sha1[content.sha1] = content return [contents_by_sha1.get(sha1) for sha1 in contents] def content_find(self, content: Dict[str, Any]) -> List[Content]: # Find an algorithm that is common to all the requested contents. # It will be used to do an initial filtering efficiently. filter_algos = list(set(content).intersection(HASH_ALGORITHMS)) if not filter_algos: raise StorageArgumentException( "content keys must contain at least one " f"of: {', '.join(sorted(HASH_ALGORITHMS))}" ) common_algo = filter_algos[0] results = [] rows = self._content_get_from_hash(common_algo, content[common_algo]) for row in rows: # Re-check all the hashes, in case of collisions (either of the # hash of the partition key, or the hashes in it) for algo in HASH_ALGORITHMS: if content.get(algo) and getattr(row, algo) != content[algo]: # This hash didn't match; discard the row. break else: # All hashes match, keep this row. row_d = row.to_dict() row_d["ctime"] = row.ctime.replace(tzinfo=datetime.timezone.utc) results.append(Content(**row_d)) return results def content_missing( self, contents: List[Dict[str, Any]], key_hash: str = "sha1" ) -> Iterable[bytes]: if key_hash not in DEFAULT_ALGORITHMS: raise StorageArgumentException( "key_hash should be one of {','.join(DEFAULT_ALGORITHMS)}" ) for content in contents: res = self.content_find(content) if not res: yield content[key_hash] def content_missing_per_sha1(self, contents: List[bytes]) -> Iterable[bytes]: return self.content_missing([{"sha1": c} for c in contents]) def content_missing_per_sha1_git( self, contents: List[Sha1Git] ) -> Iterable[Sha1Git]: return self.content_missing( [{"sha1_git": c} for c in contents], key_hash="sha1_git" ) def content_get_random(self) -> Sha1Git: content = self._cql_runner.content_get_random() assert content, "Could not find any content" return content.sha1_git def _skipped_content_add(self, contents: List[SkippedContent]) -> Dict[str, int]: # Filter-out content already in the database. if not self._allow_overwrite: contents = [ c for c in contents if not self._cql_runner.skipped_content_get_from_pk(c.to_dict()) ] self.journal_writer.skipped_content_add(contents) for content in contents: # Compute token of the row in the main table (token, insertion_finalizer) = self._cql_runner.skipped_content_add_prepare( SkippedContentRow.from_dict({"origin": None, **content.to_dict()}) ) # Then add to index tables for algo in HASH_ALGORITHMS: self._cql_runner.skipped_content_index_add_one(algo, content, token) # Then to the main table insertion_finalizer() return {"skipped_content:add": len(contents)} def skipped_content_add(self, content: List[SkippedContent]) -> Dict[str, int]: contents = [attr.evolve(c, ctime=now()) for c in content] return self._skipped_content_add(contents) def skipped_content_missing( self, contents: List[Dict[str, Any]] ) -> Iterable[Dict[str, Any]]: for content in contents: if not self._cql_runner.skipped_content_get_from_pk(content): yield {algo: content[algo] for algo in DEFAULT_ALGORITHMS} def directory_add(self, directories: List[Directory]) -> Dict[str, int]: to_add = {d.id: d for d in directories}.values() if not self._allow_overwrite: # Filter out directories that are already inserted. missing = self.directory_missing([dir_.id for dir_ in to_add]) directories = [dir_ for dir_ in directories if dir_.id in missing] self.journal_writer.directory_add(directories) for directory in directories: # Add directory entries to the 'directory_entry' table for entry in directory.entries: self._cql_runner.directory_entry_add_one( DirectoryEntryRow(directory_id=directory.id, **entry.to_dict()) ) # Add the directory *after* adding all the entries, so someone # calling snapshot_get_branch in the meantime won't end up # with half the entries. self._cql_runner.directory_add_one(DirectoryRow(id=directory.id)) return {"directory:add": len(directories)} def directory_missing(self, directories: List[Sha1Git]) -> Iterable[Sha1Git]: return self._cql_runner.directory_missing(directories) def _join_dentry_to_content(self, dentry: DirectoryEntry) -> Dict[str, Any]: contents: Union[List[Content], List[SkippedContentRow]] keys = ( "status", "sha1", "sha1_git", "sha256", "length", ) ret = dict.fromkeys(keys) ret.update(dentry.to_dict()) if ret["type"] == "file": contents = self.content_find({"sha1_git": ret["target"]}) if not contents: tokens = list( self._cql_runner.skipped_content_get_tokens_from_single_hash( "sha1_git", ret["target"] ) ) if tokens: contents = list( self._cql_runner.skipped_content_get_from_token(tokens[0]) ) if contents: content = contents[0] for key in keys: ret[key] = getattr(content, key) return ret def _directory_ls( self, directory_id: Sha1Git, recursive: bool, prefix: bytes = b"" ) -> Iterable[Dict[str, Any]]: if self.directory_missing([directory_id]): return rows = list(self._cql_runner.directory_entry_get([directory_id])) for row in rows: entry_d = row.to_dict() # Build and yield the directory entry dict del entry_d["directory_id"] entry = DirectoryEntry.from_dict(entry_d) ret = self._join_dentry_to_content(entry) ret["name"] = prefix + ret["name"] ret["dir_id"] = directory_id yield ret if recursive and ret["type"] == "dir": yield from self._directory_ls( ret["target"], True, prefix + ret["name"] + b"/" ) def directory_entry_get_by_path( self, directory: Sha1Git, paths: List[bytes] ) -> Optional[Dict[str, Any]]: return self._directory_entry_get_by_path(directory, paths, b"") def _directory_entry_get_by_path( self, directory: Sha1Git, paths: List[bytes], prefix: bytes ) -> Optional[Dict[str, Any]]: if not paths: return None contents = list(self.directory_ls(directory)) if not contents: return None def _get_entry(entries, name): """Finds the entry with the requested name, prepends the prefix (to get its full path), and returns it. If no entry has that name, returns None.""" for entry in entries: if entry["name"] == name: entry = entry.copy() entry["name"] = prefix + entry["name"] return entry first_item = _get_entry(contents, paths[0]) if len(paths) == 1: return first_item if not first_item or first_item["type"] != "dir": return None return self._directory_entry_get_by_path( first_item["target"], paths[1:], prefix + paths[0] + b"/" ) def directory_ls( self, directory: Sha1Git, recursive: bool = False ) -> Iterable[Dict[str, Any]]: yield from self._directory_ls(directory, recursive) def directory_get_random(self) -> Sha1Git: directory = self._cql_runner.directory_get_random() assert directory, "Could not find any directory" return directory.id def revision_add(self, revisions: List[Revision]) -> Dict[str, int]: # Filter-out revisions already in the database if not self._allow_overwrite: to_add = {r.id: r for r in revisions}.values() missing = self.revision_missing([rev.id for rev in to_add]) revisions = [rev for rev in revisions if rev.id in missing] self.journal_writer.revision_add(revisions) for revision in revisions: revobject = converters.revision_to_db(revision) if revobject: # Add parents first for (rank, parent) in enumerate(revision.parents): self._cql_runner.revision_parent_add_one( RevisionParentRow( id=revobject.id, parent_rank=rank, parent_id=parent ) ) # Then write the main revision row. # Writing this after all parents were written ensures that # read endpoints don't return a partial view while writing # the parents self._cql_runner.revision_add_one(revobject) return {"revision:add": len(revisions)} def revision_missing(self, revisions: List[Sha1Git]) -> Iterable[Sha1Git]: return self._cql_runner.revision_missing(revisions) def revision_get(self, revision_ids: List[Sha1Git]) -> List[Optional[Revision]]: rows = self._cql_runner.revision_get(revision_ids) revisions: Dict[Sha1Git, Revision] = {} for row in rows: # TODO: use a single query to get all parents? # (it might have lower latency, but requires more code and more # bandwidth, because revision id would be part of each returned # row) parents = tuple(self._cql_runner.revision_parent_get(row.id)) # parent_rank is the clustering key, so results are already # sorted by rank. rev = converters.revision_from_db(row, parents=parents) revisions[rev.id] = rev return [revisions.get(rev_id) for rev_id in revision_ids] def _get_parent_revs( self, rev_ids: Iterable[Sha1Git], seen: Set[Sha1Git], limit: Optional[int], short: bool, ) -> Union[ Iterable[Dict[str, Any]], Iterable[Tuple[Sha1Git, Tuple[Sha1Git, ...]]], ]: if limit and len(seen) >= limit: return rev_ids = [id_ for id_ in rev_ids if id_ not in seen] if not rev_ids: return seen |= set(rev_ids) # We need this query, even if short=True, to return consistent # results (ie. not return only a subset of a revision's parents # if it is being written) if short: ids = self._cql_runner.revision_get_ids(rev_ids) for id_ in ids: # TODO: use a single query to get all parents? # (it might have less latency, but requires less code and more # bandwidth (because revision id would be part of each returned # row) parents = tuple(self._cql_runner.revision_parent_get(id_)) # parent_rank is the clustering key, so results are already # sorted by rank. yield (id_, parents) yield from self._get_parent_revs(parents, seen, limit, short) else: rows = self._cql_runner.revision_get(rev_ids) for row in rows: # TODO: use a single query to get all parents? # (it might have less latency, but requires less code and more # bandwidth (because revision id would be part of each returned # row) parents = tuple(self._cql_runner.revision_parent_get(row.id)) # parent_rank is the clustering key, so results are already # sorted by rank. rev = converters.revision_from_db(row, parents=parents) yield rev.to_dict() yield from self._get_parent_revs(parents, seen, limit, short) def revision_log( self, revisions: List[Sha1Git], limit: Optional[int] = None ) -> Iterable[Optional[Dict[str, Any]]]: seen: Set[Sha1Git] = set() yield from self._get_parent_revs(revisions, seen, limit, False) def revision_shortlog( self, revisions: List[Sha1Git], limit: Optional[int] = None ) -> Iterable[Optional[Tuple[Sha1Git, Tuple[Sha1Git, ...]]]]: seen: Set[Sha1Git] = set() yield from self._get_parent_revs(revisions, seen, limit, True) def revision_get_random(self) -> Sha1Git: revision = self._cql_runner.revision_get_random() assert revision, "Could not find any revision" return revision.id def release_add(self, releases: List[Release]) -> Dict[str, int]: if not self._allow_overwrite: to_add = {r.id: r for r in releases}.values() missing = set(self.release_missing([rel.id for rel in to_add])) releases = [rel for rel in to_add if rel.id in missing] self.journal_writer.release_add(releases) for release in releases: if release: self._cql_runner.release_add_one(converters.release_to_db(release)) return {"release:add": len(releases)} def release_missing(self, releases: List[Sha1Git]) -> Iterable[Sha1Git]: return self._cql_runner.release_missing(releases) def release_get(self, releases: List[Sha1Git]) -> List[Optional[Release]]: rows = self._cql_runner.release_get(releases) rels: Dict[Sha1Git, Release] = {} for row in rows: release = converters.release_from_db(row) rels[row.id] = release return [rels.get(rel_id) for rel_id in releases] def release_get_random(self) -> Sha1Git: release = self._cql_runner.release_get_random() assert release, "Could not find any release" return release.id def snapshot_add(self, snapshots: List[Snapshot]) -> Dict[str, int]: if not self._allow_overwrite: to_add = {s.id: s for s in snapshots}.values() missing = self._cql_runner.snapshot_missing([snp.id for snp in to_add]) snapshots = [snp for snp in snapshots if snp.id in missing] for snapshot in snapshots: self.journal_writer.snapshot_add([snapshot]) # Add branches for (branch_name, branch) in snapshot.branches.items(): if branch is None: target_type: Optional[str] = None target: Optional[bytes] = None else: target_type = branch.target_type.value target = branch.target self._cql_runner.snapshot_branch_add_one( SnapshotBranchRow( snapshot_id=snapshot.id, name=branch_name, target_type=target_type, target=target, ) ) # Add the snapshot *after* adding all the branches, so someone # calling snapshot_get_branch in the meantime won't end up # with half the branches. self._cql_runner.snapshot_add_one(SnapshotRow(id=snapshot.id)) return {"snapshot:add": len(snapshots)} def snapshot_missing(self, snapshots: List[Sha1Git]) -> Iterable[Sha1Git]: return self._cql_runner.snapshot_missing(snapshots) def snapshot_get(self, snapshot_id: Sha1Git) -> Optional[Dict[str, Any]]: d = self.snapshot_get_branches(snapshot_id) if d is None: return None return { "id": d["id"], "branches": { name: branch.to_dict() if branch else None for (name, branch) in d["branches"].items() }, "next_branch": d["next_branch"], } def snapshot_count_branches( self, snapshot_id: Sha1Git, branch_name_exclude_prefix: Optional[bytes] = None, ) -> Optional[Dict[Optional[str], int]]: if self._cql_runner.snapshot_missing([snapshot_id]): # Makes sure we don't fetch branches for a snapshot that is # being added. return None return self._cql_runner.snapshot_count_branches( snapshot_id, branch_name_exclude_prefix ) def snapshot_get_branches( self, snapshot_id: Sha1Git, branches_from: bytes = b"", branches_count: int = 1000, target_types: Optional[List[str]] = None, branch_name_include_substring: Optional[bytes] = None, branch_name_exclude_prefix: Optional[bytes] = None, ) -> Optional[PartialBranches]: if self._cql_runner.snapshot_missing([snapshot_id]): # Makes sure we don't fetch branches for a snapshot that is # being added. return None branches: List = [] while len(branches) < branches_count + 1: new_branches = list( self._cql_runner.snapshot_branch_get( snapshot_id, branches_from, branches_count + 1, branch_name_exclude_prefix, ) ) if not new_branches: break branches_from = new_branches[-1].name new_branches_filtered = new_branches # Filter by target_type if target_types: new_branches_filtered = [ branch for branch in new_branches_filtered if branch.target is not None and branch.target_type in target_types ] # Filter by branches_name_pattern if branch_name_include_substring: new_branches_filtered = [ branch for branch in new_branches_filtered if branch.name is not None and ( branch_name_include_substring is None or branch_name_include_substring in branch.name ) ] branches.extend(new_branches_filtered) if len(new_branches) < branches_count + 1: break if len(branches) > branches_count: last_branch = branches.pop(-1).name else: last_branch = None return PartialBranches( id=snapshot_id, branches={ branch.name: None if branch.target is None else SnapshotBranch( target=branch.target, target_type=TargetType(branch.target_type) ) for branch in branches }, next_branch=last_branch, ) def snapshot_get_random(self) -> Sha1Git: snapshot = self._cql_runner.snapshot_get_random() assert snapshot, "Could not find any snapshot" return snapshot.id def object_find_by_sha1_git(self, ids: List[Sha1Git]) -> Dict[Sha1Git, List[Dict]]: results: Dict[Sha1Git, List[Dict]] = {id_: [] for id_ in ids} missing_ids = set(ids) # Mind the order, revision is the most likely one for a given ID, # so we check revisions first. queries: List[Tuple[str, Callable[[List[Sha1Git]], List[Sha1Git]]]] = [ ("revision", self._cql_runner.revision_missing), ("release", self._cql_runner.release_missing), ("content", self._cql_runner.content_missing_by_sha1_git), ("directory", self._cql_runner.directory_missing), ] for (object_type, query_fn) in queries: found_ids = missing_ids - set(query_fn(list(missing_ids))) for sha1_git in found_ids: results[sha1_git].append( {"sha1_git": sha1_git, "type": object_type,} ) missing_ids.remove(sha1_git) if not missing_ids: # We found everything, skipping the next queries. break return results def origin_get(self, origins: List[str]) -> Iterable[Optional[Origin]]: return [self.origin_get_one(origin) for origin in origins] def origin_get_one(self, origin_url: str) -> Optional[Origin]: """Given an origin url, return the origin if it exists, None otherwise """ rows = list(self._cql_runner.origin_get_by_url(origin_url)) if rows: assert len(rows) == 1 return Origin(url=rows[0].url) else: return None def origin_get_by_sha1(self, sha1s: List[bytes]) -> List[Optional[Dict[str, Any]]]: results = [] for sha1 in sha1s: rows = list(self._cql_runner.origin_get_by_sha1(sha1)) origin = {"url": rows[0].url} if rows else None results.append(origin) return results def origin_list( self, page_token: Optional[str] = None, limit: int = 100 ) -> PagedResult[Origin]: # Compute what token to begin the listing from start_token = TOKEN_BEGIN if page_token: start_token = int(page_token) if not (TOKEN_BEGIN <= start_token <= TOKEN_END): raise StorageArgumentException("Invalid page_token.") next_page_token = None origins = [] # Take one more origin so we can reuse it as the next page token if any for (tok, row) in self._cql_runner.origin_list(start_token, limit + 1): origins.append(Origin(url=row.url)) # keep reference of the last id for pagination purposes last_id = tok if len(origins) > limit: # last origin id is the next page token next_page_token = str(last_id) # excluding that origin from the result to respect the limit size origins = origins[:limit] assert len(origins) <= limit return PagedResult(results=origins, next_page_token=next_page_token) def origin_search( self, url_pattern: str, page_token: Optional[str] = None, limit: int = 50, regexp: bool = False, with_visit: bool = False, visit_types: Optional[List[str]] = None, ) -> PagedResult[Origin]: # TODO: remove this endpoint, swh-search should be used instead. next_page_token = None offset = int(page_token) if page_token else 0 origin_rows = [row for row in self._cql_runner.origin_iter_all()] if regexp: pat = re.compile(url_pattern) origin_rows = [row for row in origin_rows if pat.search(row.url)] else: origin_rows = [row for row in origin_rows if url_pattern in row.url] if with_visit: origin_rows = [row for row in origin_rows if row.next_visit_id > 1] if visit_types: def _has_visit_types(origin, visit_types): for origin_visit in stream_results(self.origin_visit_get, origin): if origin_visit.type in visit_types: return True return False origin_rows = [ row for row in origin_rows if _has_visit_types(row.url, visit_types) ] origins = [Origin(url=row.url) for row in origin_rows] origins = origins[offset : offset + limit + 1] if len(origins) > limit: # next offset next_page_token = str(offset + limit) # excluding that origin from the result to respect the limit size origins = origins[:limit] assert len(origins) <= limit return PagedResult(results=origins, next_page_token=next_page_token) def origin_count( self, url_pattern: str, regexp: bool = False, with_visit: bool = False ) -> int: raise NotImplementedError( "The Cassandra backend does not implement origin_count" ) def origin_add(self, origins: List[Origin]) -> Dict[str, int]: if not self._allow_overwrite: to_add = {o.url: o for o in origins}.values() origins = [ori for ori in to_add if self.origin_get_one(ori.url) is None] self.journal_writer.origin_add(origins) for origin in origins: self._cql_runner.origin_add_one( OriginRow(sha1=hash_url(origin.url), url=origin.url, next_visit_id=1) ) return {"origin:add": len(origins)} def origin_visit_add(self, visits: List[OriginVisit]) -> Iterable[OriginVisit]: for visit in visits: origin = self.origin_get_one(visit.origin) if not origin: # Cannot add a visit without an origin raise StorageArgumentException("Unknown origin %s", visit.origin) all_visits = [] nb_visits = 0 for visit in visits: nb_visits += 1 if not visit.visit: visit_id = self._cql_runner.origin_generate_unique_visit_id( visit.origin ) visit = attr.evolve(visit, visit=visit_id) self.journal_writer.origin_visit_add([visit]) self._cql_runner.origin_visit_add_one(OriginVisitRow(**visit.to_dict())) assert visit.visit is not None all_visits.append(visit) self._origin_visit_status_add( OriginVisitStatus( origin=visit.origin, visit=visit.visit, date=visit.date, type=visit.type, status="created", snapshot=None, ) ) return all_visits def _origin_visit_status_add(self, visit_status: OriginVisitStatus) -> None: """Add an origin visit status""" if visit_status.type is None: visit_row = self._cql_runner.origin_visit_get_one( visit_status.origin, visit_status.visit ) if visit_row is None: raise StorageArgumentException( f"Unknown origin visit {visit_status.visit} " f"of origin {visit_status.origin}" ) visit_status = attr.evolve(visit_status, type=visit_row.type) self.journal_writer.origin_visit_status_add([visit_status]) self._cql_runner.origin_visit_status_add_one( converters.visit_status_to_row(visit_status) ) def origin_visit_status_add( self, visit_statuses: List[OriginVisitStatus] ) -> Dict[str, int]: # First round to check existence (fail early if any is ko) for visit_status in visit_statuses: origin_url = self.origin_get_one(visit_status.origin) if not origin_url: raise StorageArgumentException(f"Unknown origin {visit_status.origin}") for visit_status in visit_statuses: self._origin_visit_status_add(visit_status) return {"origin_visit_status:add": len(visit_statuses)} def _origin_visit_apply_status( self, visit: Dict[str, Any], visit_status: OriginVisitStatusRow ) -> Dict[str, Any]: """Retrieve the latest visit status information for the origin visit. Then merge it with the visit and return it. """ return { # default to the values in visit **visit, # override with the last update **visit_status.to_dict(), # visit['origin'] is the URL (via a join), while # visit_status['origin'] is only an id. "origin": visit["origin"], # but keep the date of the creation of the origin visit "date": visit["date"], # We use the visit type from origin visit # if it's not present on the origin visit status "type": visit_status.type or visit["type"], } def _origin_visit_get_latest_status(self, visit: OriginVisit) -> OriginVisitStatus: """Retrieve the latest visit status information for the origin visit object. """ assert visit.visit row = self._cql_runner.origin_visit_status_get_latest(visit.origin, visit.visit) assert row is not None visit_status = converters.row_to_visit_status(row) return attr.evolve(visit_status, origin=visit.origin) @staticmethod def _format_origin_visit_row(visit): return { **visit.to_dict(), "origin": visit.origin, "date": visit.date.replace(tzinfo=datetime.timezone.utc), } def origin_visit_get( self, origin: str, page_token: Optional[str] = None, order: ListOrder = ListOrder.ASC, limit: int = 10, ) -> PagedResult[OriginVisit]: if not isinstance(order, ListOrder): raise StorageArgumentException("order must be a ListOrder value") if page_token and not isinstance(page_token, str): raise StorageArgumentException("page_token must be a string.") next_page_token = None visit_from = None if page_token is None else int(page_token) visits: List[OriginVisit] = [] extra_limit = limit + 1 rows = self._cql_runner.origin_visit_get(origin, visit_from, extra_limit, order) for row in rows: visits.append(converters.row_to_visit(row)) assert len(visits) <= extra_limit if len(visits) == extra_limit: visits = visits[:limit] next_page_token = str(visits[-1].visit) return PagedResult(results=visits, next_page_token=next_page_token) def origin_visit_status_get( self, origin: str, visit: int, page_token: Optional[str] = None, order: ListOrder = ListOrder.ASC, limit: int = 10, ) -> PagedResult[OriginVisitStatus]: next_page_token = None date_from = None if page_token is not None: date_from = datetime.datetime.fromisoformat(page_token) # Take one more visit status so we can reuse it as the next page token if any rows = self._cql_runner.origin_visit_status_get_range( origin, visit, date_from, limit + 1, order ) visit_statuses = [converters.row_to_visit_status(row) for row in rows] if len(visit_statuses) > limit: # last visit status date is the next page token next_page_token = str(visit_statuses[-1].date) # excluding that visit status from the result to respect the limit size visit_statuses = visit_statuses[:limit] return PagedResult(results=visit_statuses, next_page_token=next_page_token) def origin_visit_find_by_date( self, origin: str, visit_date: datetime.datetime ) -> Optional[OriginVisit]: # Iterator over all the visits of the origin # This should be ok for now, as there aren't too many visits # per origin. rows = list(self._cql_runner.origin_visit_get_all(origin)) def key(visit): dt = visit.date.replace(tzinfo=datetime.timezone.utc) - visit_date return (abs(dt), -visit.visit) if rows: return converters.row_to_visit(min(rows, key=key)) return None def origin_visit_get_by(self, origin: str, visit: int) -> Optional[OriginVisit]: row = self._cql_runner.origin_visit_get_one(origin, visit) if row: return converters.row_to_visit(row) return None def origin_visit_get_latest( self, origin: str, type: Optional[str] = None, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, ) -> Optional[OriginVisit]: if allowed_statuses and not set(allowed_statuses).intersection(VISIT_STATUSES): raise StorageArgumentException( f"Unknown allowed statuses {','.join(allowed_statuses)}, only " f"{','.join(VISIT_STATUSES)} authorized" ) # TODO: Do not fetch all visits rows = self._cql_runner.origin_visit_get_all(origin) latest_visit = None for row in rows: visit = self._format_origin_visit_row(row) for status_row in self._cql_runner.origin_visit_status_get( origin, visit["visit"] ): updated_visit = self._origin_visit_apply_status(visit, status_row) if type is not None and updated_visit["type"] != type: continue if allowed_statuses and updated_visit["status"] not in allowed_statuses: continue if require_snapshot and updated_visit["snapshot"] is None: continue # updated_visit is a candidate if latest_visit is not None: if updated_visit["date"] < latest_visit["date"]: continue if updated_visit["visit"] < latest_visit["visit"]: continue latest_visit = updated_visit if latest_visit is None: return None return OriginVisit( origin=latest_visit["origin"], visit=latest_visit["visit"], date=latest_visit["date"], type=latest_visit["type"], ) def origin_visit_status_get_latest( self, origin_url: str, visit: int, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, ) -> Optional[OriginVisitStatus]: if allowed_statuses and not set(allowed_statuses).intersection(VISIT_STATUSES): raise StorageArgumentException( f"Unknown allowed statuses {','.join(allowed_statuses)}, only " f"{','.join(VISIT_STATUSES)} authorized" ) rows = list(self._cql_runner.origin_visit_status_get(origin_url, visit)) # filtering is done python side as we cannot do it server side if allowed_statuses: rows = [row for row in rows if row.status in allowed_statuses] if require_snapshot: rows = [row for row in rows if row.snapshot is not None] if not rows: return None return converters.row_to_visit_status(rows[0]) def origin_visit_status_get_random(self, type: str) -> Optional[OriginVisitStatus]: back_in_the_day = now() - datetime.timedelta(weeks=12) # 3 months back # Random position to start iteration at start_token = random.randint(TOKEN_BEGIN, TOKEN_END) # Iterator over all visits, ordered by token(origins) then visit_id rows = self._cql_runner.origin_visit_iter(start_token) for row in rows: visit = converters.row_to_visit(row) visit_status = self._origin_visit_get_latest_status(visit) if visit.date > back_in_the_day and visit_status.status == "full": return visit_status return None def stat_counters(self): rows = self._cql_runner.stat_counters() keys = ( "content", "directory", "origin", "origin_visit", "release", "revision", "skipped_content", "snapshot", ) stats = {key: 0 for key in keys} stats.update({row.object_type: row.count for row in rows}) return stats def refresh_stat_counters(self): pass def raw_extrinsic_metadata_add( self, metadata: List[RawExtrinsicMetadata] ) -> Dict[str, int]: self.journal_writer.raw_extrinsic_metadata_add(metadata) counter = Counter[ExtendedObjectType]() for metadata_entry in metadata: if not self._cql_runner.metadata_authority_get( metadata_entry.authority.type.value, metadata_entry.authority.url ): raise StorageArgumentException( f"Unknown authority {metadata_entry.authority}" ) if not self._cql_runner.metadata_fetcher_get( metadata_entry.fetcher.name, metadata_entry.fetcher.version ): raise StorageArgumentException( f"Unknown fetcher {metadata_entry.fetcher}" ) try: row = RawExtrinsicMetadataRow( id=metadata_entry.id, type=metadata_entry.target.object_type.name.lower(), target=str(metadata_entry.target), authority_type=metadata_entry.authority.type.value, authority_url=metadata_entry.authority.url, discovery_date=metadata_entry.discovery_date, fetcher_name=metadata_entry.fetcher.name, fetcher_version=metadata_entry.fetcher.version, format=metadata_entry.format, metadata=metadata_entry.metadata, origin=metadata_entry.origin, visit=metadata_entry.visit, snapshot=map_optional(str, metadata_entry.snapshot), release=map_optional(str, metadata_entry.release), revision=map_optional(str, metadata_entry.revision), path=metadata_entry.path, directory=map_optional(str, metadata_entry.directory), ) self._cql_runner.raw_extrinsic_metadata_add(row) counter[metadata_entry.target.object_type] += 1 except TypeError as e: raise StorageArgumentException(*e.args) return { f"{type.value}_metadata:add": count for (type, count) in counter.items() } def raw_extrinsic_metadata_get( self, target: ExtendedSWHID, authority: MetadataAuthority, after: Optional[datetime.datetime] = None, page_token: Optional[bytes] = None, limit: int = 1000, ) -> PagedResult[RawExtrinsicMetadata]: if page_token is not None: (after_date, id_) = msgpack_loads(base64.b64decode(page_token)) if after and after_date < after: raise StorageArgumentException( "page_token is inconsistent with the value of 'after'." ) entries = self._cql_runner.raw_extrinsic_metadata_get_after_date_and_id( str(target), authority.type.value, authority.url, after_date, id_, ) elif after is not None: entries = self._cql_runner.raw_extrinsic_metadata_get_after_date( str(target), authority.type.value, authority.url, after ) else: entries = self._cql_runner.raw_extrinsic_metadata_get( str(target), authority.type.value, authority.url ) if limit: entries = itertools.islice(entries, 0, limit + 1) results = [] for entry in entries: discovery_date = entry.discovery_date.replace(tzinfo=datetime.timezone.utc) assert str(target) == entry.target result = RawExtrinsicMetadata( target=target, authority=MetadataAuthority( type=MetadataAuthorityType(entry.authority_type), url=entry.authority_url, ), fetcher=MetadataFetcher( name=entry.fetcher_name, version=entry.fetcher_version, ), discovery_date=discovery_date, format=entry.format, metadata=entry.metadata, origin=entry.origin, visit=entry.visit, snapshot=map_optional(CoreSWHID.from_string, entry.snapshot), release=map_optional(CoreSWHID.from_string, entry.release), revision=map_optional(CoreSWHID.from_string, entry.revision), path=entry.path, directory=map_optional(CoreSWHID.from_string, entry.directory), ) results.append(result) if len(results) > limit: results.pop() assert len(results) == limit last_result = results[-1] next_page_token: Optional[str] = base64.b64encode( msgpack_dumps((last_result.discovery_date, last_result.id,)) ).decode() else: next_page_token = None return PagedResult(next_page_token=next_page_token, results=results,) def metadata_fetcher_add(self, fetchers: List[MetadataFetcher]) -> Dict[str, int]: self.journal_writer.metadata_fetcher_add(fetchers) for fetcher in fetchers: self._cql_runner.metadata_fetcher_add( - MetadataFetcherRow( - name=fetcher.name, - version=fetcher.version, - metadata=json.dumps(map_optional(dict, fetcher.metadata)), - ) + MetadataFetcherRow(name=fetcher.name, version=fetcher.version,) ) return {"metadata_fetcher:add": len(fetchers)} def metadata_fetcher_get( self, name: str, version: str ) -> Optional[MetadataFetcher]: fetcher = self._cql_runner.metadata_fetcher_get(name, version) if fetcher: - return MetadataFetcher( - name=fetcher.name, - version=fetcher.version, - metadata=json.loads(fetcher.metadata), - ) + return MetadataFetcher(name=fetcher.name, version=fetcher.version,) else: return None def metadata_authority_add( self, authorities: List[MetadataAuthority] ) -> Dict[str, int]: self.journal_writer.metadata_authority_add(authorities) for authority in authorities: self._cql_runner.metadata_authority_add( - MetadataAuthorityRow( - url=authority.url, - type=authority.type.value, - metadata=json.dumps(map_optional(dict, authority.metadata)), - ) + MetadataAuthorityRow(url=authority.url, type=authority.type.value,) ) return {"metadata_authority:add": len(authorities)} def metadata_authority_get( self, type: MetadataAuthorityType, url: str ) -> Optional[MetadataAuthority]: authority = self._cql_runner.metadata_authority_get(type.value, url) if authority: return MetadataAuthority( - type=MetadataAuthorityType(authority.type), - url=authority.url, - metadata=json.loads(authority.metadata), + type=MetadataAuthorityType(authority.type), url=authority.url, ) else: return None # ExtID tables def extid_add(self, ids: List[ExtID]) -> Dict[str, int]: if not self._allow_overwrite: extids = [ extid for extid in ids if not self._cql_runner.extid_get_from_pk( extid_type=extid.extid_type, extid=extid.extid, target=extid.target, ) ] else: extids = list(ids) self.journal_writer.extid_add(extids) inserted = 0 for extid in extids: target_type = extid.target.object_type.value target = extid.target.object_id extidrow = ExtIDRow( extid_type=extid.extid_type, extid=extid.extid, target_type=target_type, target=target, ) (token, insertion_finalizer) = self._cql_runner.extid_add_prepare(extidrow) indexrow = ExtIDByTargetRow( target_type=target_type, target=target, target_token=token, ) self._cql_runner.extid_index_add_one(indexrow) insertion_finalizer() inserted += 1 return {"extid:add": inserted} def extid_get_from_extid(self, id_type: str, ids: List[bytes]) -> List[ExtID]: result: List[ExtID] = [] for extid in ids: extidrows = list(self._cql_runner.extid_get_from_extid(id_type, extid)) result.extend( ExtID( extid_type=extidrow.extid_type, extid=extidrow.extid, target=CoreSWHID( object_type=extidrow.target_type, object_id=extidrow.target, ), ) for extidrow in extidrows ) return result def extid_get_from_target( self, target_type: SwhidObjectType, ids: List[Sha1Git] ) -> List[ExtID]: result: List[ExtID] = [] for target in ids: extidrows = list( self._cql_runner.extid_get_from_target(target_type.value, target) ) result.extend( ExtID( extid_type=extidrow.extid_type, extid=extidrow.extid, target=CoreSWHID( object_type=SwhidObjectType(extidrow.target_type), object_id=extidrow.target, ), ) for extidrow in extidrows ) return result # Misc def clear_buffers(self, object_types: Sequence[str] = ()) -> None: """Do nothing """ return None def flush(self, object_types: Sequence[str] = ()) -> Dict[str, int]: return {} diff --git a/swh/storage/postgresql/db.py b/swh/storage/postgresql/db.py index 60dd8e48..71aa13fa 100644 --- a/swh/storage/postgresql/db.py +++ b/swh/storage/postgresql/db.py @@ -1,1442 +1,1438 @@ # Copyright (C) 2015-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime import logging import random import select from typing import Any, Dict, Iterable, List, Optional, Tuple from swh.core.db import BaseDb from swh.core.db.db_utils import execute_values_generator from swh.core.db.db_utils import jsonize as _jsonize from swh.core.db.db_utils import stored_procedure from swh.model.identifiers import ObjectType from swh.model.model import SHA1_SIZE, OriginVisit, OriginVisitStatus from swh.storage.interface import ListOrder logger = logging.getLogger(__name__) def jsonize(d): return _jsonize(dict(d) if d is not None else None) class Db(BaseDb): """Proxy to the SWH DB, with wrappers around stored procedures """ - current_version = 173 + current_version = 174 def mktemp_dir_entry(self, entry_type, cur=None): self._cursor(cur).execute( "SELECT swh_mktemp_dir_entry(%s)", (("directory_entry_%s" % entry_type),) ) @stored_procedure("swh_mktemp_revision") def mktemp_revision(self, cur=None): pass @stored_procedure("swh_mktemp_release") def mktemp_release(self, cur=None): pass @stored_procedure("swh_mktemp_snapshot_branch") def mktemp_snapshot_branch(self, cur=None): pass def register_listener(self, notify_queue, cur=None): """Register a listener for NOTIFY queue `notify_queue`""" self._cursor(cur).execute("LISTEN %s" % notify_queue) def listen_notifies(self, timeout): """Listen to notifications for `timeout` seconds""" if select.select([self.conn], [], [], timeout) == ([], [], []): return else: self.conn.poll() while self.conn.notifies: yield self.conn.notifies.pop(0) @stored_procedure("swh_content_add") def content_add_from_temp(self, cur=None): pass @stored_procedure("swh_directory_add") def directory_add_from_temp(self, cur=None): pass @stored_procedure("swh_skipped_content_add") def skipped_content_add_from_temp(self, cur=None): pass @stored_procedure("swh_revision_add") def revision_add_from_temp(self, cur=None): pass @stored_procedure("swh_extid_add") def extid_add_from_temp(self, cur=None): pass @stored_procedure("swh_release_add") def release_add_from_temp(self, cur=None): pass def content_update_from_temp(self, keys_to_update, cur=None): cur = self._cursor(cur) cur.execute( """select swh_content_update(ARRAY[%s] :: text[])""" % keys_to_update ) content_get_metadata_keys = [ "sha1", "sha1_git", "sha256", "blake2s256", "length", "status", ] content_add_keys = content_get_metadata_keys + ["ctime"] skipped_content_keys = [ "sha1", "sha1_git", "sha256", "blake2s256", "length", "reason", "status", "origin", ] def content_get_metadata_from_sha1s(self, sha1s, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ select t.sha1, %s from (values %%s) as t (sha1) inner join content using (sha1) """ % ", ".join(self.content_get_metadata_keys[1:]), ((sha1,) for sha1 in sha1s), ) def content_get_range(self, start, end, limit=None, cur=None): """Retrieve contents within range [start, end]. """ cur = self._cursor(cur) query = """select %s from content where %%s <= sha1 and sha1 <= %%s order by sha1 limit %%s""" % ", ".join( self.content_get_metadata_keys ) cur.execute(query, (start, end, limit)) yield from cur content_hash_keys = ["sha1", "sha1_git", "sha256", "blake2s256"] def content_missing_from_list(self, contents, cur=None): cur = self._cursor(cur) keys = ", ".join(self.content_hash_keys) equality = " AND ".join( ("t.%s = c.%s" % (key, key)) for key in self.content_hash_keys ) yield from execute_values_generator( cur, """ SELECT %s FROM (VALUES %%s) as t(%s) WHERE NOT EXISTS ( SELECT 1 FROM content c WHERE %s ) """ % (keys, keys, equality), (tuple(c[key] for key in self.content_hash_keys) for c in contents), ) def content_missing_per_sha1(self, sha1s, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT t.sha1 FROM (VALUES %s) AS t(sha1) WHERE NOT EXISTS ( SELECT 1 FROM content c WHERE c.sha1 = t.sha1 )""", ((sha1,) for sha1 in sha1s), ) def content_missing_per_sha1_git(self, contents, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT t.sha1_git FROM (VALUES %s) AS t(sha1_git) WHERE NOT EXISTS ( SELECT 1 FROM content c WHERE c.sha1_git = t.sha1_git )""", ((sha1,) for sha1 in contents), ) def skipped_content_missing(self, contents, cur=None): if not contents: return [] cur = self._cursor(cur) query = """SELECT * FROM (VALUES %s) AS t (%s) WHERE not exists (SELECT 1 FROM skipped_content s WHERE s.sha1 is not distinct from t.sha1::sha1 and s.sha1_git is not distinct from t.sha1_git::sha1 and s.sha256 is not distinct from t.sha256::bytea);""" % ( (", ".join("%s" for _ in contents)), ", ".join(self.content_hash_keys), ) cur.execute( query, [tuple(cont[key] for key in self.content_hash_keys) for cont in contents], ) yield from cur def snapshot_exists(self, snapshot_id, cur=None): """Check whether a snapshot with the given id exists""" cur = self._cursor(cur) cur.execute("""SELECT 1 FROM snapshot where id=%s""", (snapshot_id,)) return bool(cur.fetchone()) def snapshot_missing_from_list(self, snapshots, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT id FROM (VALUES %s) as t(id) WHERE NOT EXISTS ( SELECT 1 FROM snapshot d WHERE d.id = t.id ) """, ((id,) for id in snapshots), ) def snapshot_add(self, snapshot_id, cur=None): """Add a snapshot from the temporary table""" cur = self._cursor(cur) cur.execute("""SELECT swh_snapshot_add(%s)""", (snapshot_id,)) snapshot_count_cols = ["target_type", "count"] def snapshot_count_branches( self, snapshot_id, branch_name_exclude_prefix=None, cur=None, ): cur = self._cursor(cur) query = """\ SELECT %s FROM swh_snapshot_count_branches(%%s, %%s) """ % ", ".join( self.snapshot_count_cols ) cur.execute(query, (snapshot_id, branch_name_exclude_prefix)) yield from cur snapshot_get_cols = ["snapshot_id", "name", "target", "target_type"] def snapshot_get_by_id( self, snapshot_id, branches_from=b"", branches_count=None, target_types=None, branch_name_include_substring=None, branch_name_exclude_prefix=None, cur=None, ): cur = self._cursor(cur) query = """\ SELECT %s FROM swh_snapshot_get_by_id(%%s, %%s, %%s, %%s :: snapshot_target[], %%s, %%s) """ % ", ".join( self.snapshot_get_cols ) cur.execute( query, ( snapshot_id, branches_from, branches_count, target_types, branch_name_include_substring, branch_name_exclude_prefix, ), ) yield from cur def snapshot_get_random(self, cur=None): return self._get_random_row_from_table("snapshot", ["id"], "id", cur) content_find_cols = [ "sha1", "sha1_git", "sha256", "blake2s256", "length", "ctime", "status", ] def content_find( self, sha1: Optional[bytes] = None, sha1_git: Optional[bytes] = None, sha256: Optional[bytes] = None, blake2s256: Optional[bytes] = None, cur=None, ): """Find the content optionally on a combination of the following checksums sha1, sha1_git, sha256 or blake2s256. Args: sha1: sha1 content git_sha1: the sha1 computed `a la git` sha1 of the content sha256: sha256 content blake2s256: blake2s256 content Returns: The tuple (sha1, sha1_git, sha256, blake2s256) if found or None. """ cur = self._cursor(cur) checksum_dict = { "sha1": sha1, "sha1_git": sha1_git, "sha256": sha256, "blake2s256": blake2s256, } query_parts = [f"SELECT {','.join(self.content_find_cols)} FROM content WHERE "] query_params = [] where_parts = [] # Adds only those keys which have values exist for algorithm in checksum_dict: if checksum_dict[algorithm] is not None: where_parts.append(f"{algorithm} = %s") query_params.append(checksum_dict[algorithm]) query_parts.append(" AND ".join(where_parts)) query = "\n".join(query_parts) cur.execute(query, query_params) content = cur.fetchall() return content def content_get_random(self, cur=None): return self._get_random_row_from_table("content", ["sha1_git"], "sha1_git", cur) def directory_missing_from_list(self, directories, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT id FROM (VALUES %s) as t(id) WHERE NOT EXISTS ( SELECT 1 FROM directory d WHERE d.id = t.id ) """, ((id,) for id in directories), ) directory_ls_cols = [ "dir_id", "type", "target", "name", "perms", "status", "sha1", "sha1_git", "sha256", "length", ] def directory_walk_one(self, directory, cur=None): cur = self._cursor(cur) cols = ", ".join(self.directory_ls_cols) query = "SELECT %s FROM swh_directory_walk_one(%%s)" % cols cur.execute(query, (directory,)) yield from cur def directory_walk(self, directory, cur=None): cur = self._cursor(cur) cols = ", ".join(self.directory_ls_cols) query = "SELECT %s FROM swh_directory_walk(%%s)" % cols cur.execute(query, (directory,)) yield from cur def directory_entry_get_by_path(self, directory, paths, cur=None): """Retrieve a directory entry by path. """ cur = self._cursor(cur) cols = ", ".join(self.directory_ls_cols) query = "SELECT %s FROM swh_find_directory_entry_by_path(%%s, %%s)" % cols cur.execute(query, (directory, paths)) data = cur.fetchone() if set(data) == {None}: return None return data def directory_get_random(self, cur=None): return self._get_random_row_from_table("directory", ["id"], "id", cur) def revision_missing_from_list(self, revisions, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT id FROM (VALUES %s) as t(id) WHERE NOT EXISTS ( SELECT 1 FROM revision r WHERE r.id = t.id ) """, ((id,) for id in revisions), ) revision_add_cols = [ "id", "date", "date_offset", "date_neg_utc_offset", "committer_date", "committer_date_offset", "committer_date_neg_utc_offset", "type", "directory", "message", "author_fullname", "author_name", "author_email", "committer_fullname", "committer_name", "committer_email", "metadata", "synthetic", "extra_headers", ] revision_get_cols = revision_add_cols + ["parents"] def origin_visit_add(self, origin, ts, type, cur=None): """Add a new origin_visit for origin origin at timestamp ts. Args: origin: origin concerned by the visit ts: the date of the visit type: type of loader for the visit Returns: The new visit index step for that origin """ cur = self._cursor(cur) self._cursor(cur).execute( "SELECT swh_origin_visit_add(%s, %s, %s)", (origin, ts, type) ) return cur.fetchone()[0] origin_visit_status_cols = [ "origin", "visit", "date", "type", "status", "snapshot", "metadata", ] def origin_visit_status_add( self, visit_status: OriginVisitStatus, cur=None ) -> None: """Add new origin visit status """ assert self.origin_visit_status_cols[0] == "origin" assert self.origin_visit_status_cols[-1] == "metadata" cols = self.origin_visit_status_cols[1:-1] cur = self._cursor(cur) cur.execute( f"WITH origin_id as (select id from origin where url=%s) " f"INSERT INTO origin_visit_status " f"(origin, {', '.join(cols)}, metadata) " f"VALUES ((select id from origin_id), " f"{', '.join(['%s']*len(cols))}, %s) " f"ON CONFLICT (origin, visit, date) do nothing", [visit_status.origin] + [getattr(visit_status, key) for key in cols] + [jsonize(visit_status.metadata)], ) origin_visit_cols = ["origin", "visit", "date", "type"] def origin_visit_add_with_id(self, origin_visit: OriginVisit, cur=None) -> None: """Insert origin visit when id are already set """ ov = origin_visit assert ov.visit is not None cur = self._cursor(cur) query = """INSERT INTO origin_visit ({cols}) VALUES ((select id from origin where url=%s), {values}) ON CONFLICT (origin, visit) DO NOTHING""".format( cols=", ".join(self.origin_visit_cols), values=", ".join("%s" for col in self.origin_visit_cols[1:]), ) cur.execute(query, (ov.origin, ov.visit, ov.date, ov.type)) origin_visit_get_cols = [ "origin", "visit", "date", "type", "status", "metadata", "snapshot", ] origin_visit_select_cols = [ "o.url AS origin", "ov.visit", "ov.date", "ov.type AS type", "ovs.status", "ovs.metadata", "ovs.snapshot", ] origin_visit_status_select_cols = [ "o.url AS origin", "ovs.visit", "ovs.date", "ovs.type AS type", "ovs.status", "ovs.snapshot", "ovs.metadata", ] def _make_origin_visit_status( self, row: Optional[Tuple[Any]] ) -> Optional[Dict[str, Any]]: """Make an origin_visit_status dict out of a row """ if not row: return None return dict(zip(self.origin_visit_status_cols, row)) def origin_visit_status_get_latest( self, origin_url: str, visit: int, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, cur=None, ) -> Optional[Dict[str, Any]]: """Given an origin visit id, return its latest origin_visit_status """ cur = self._cursor(cur) query_parts = [ "SELECT %s" % ", ".join(self.origin_visit_status_select_cols), "FROM origin_visit_status ovs ", "INNER JOIN origin o ON o.id = ovs.origin", ] query_parts.append("WHERE o.url = %s") query_params: List[Any] = [origin_url] query_parts.append("AND ovs.visit = %s") query_params.append(visit) if require_snapshot: query_parts.append("AND ovs.snapshot is not null") if allowed_statuses: query_parts.append("AND ovs.status IN %s") query_params.append(tuple(allowed_statuses)) query_parts.append("ORDER BY ovs.date DESC LIMIT 1") query = "\n".join(query_parts) cur.execute(query, tuple(query_params)) row = cur.fetchone() return self._make_origin_visit_status(row) def origin_visit_status_get_range( self, origin: str, visit: int, date_from: Optional[datetime.datetime], order: ListOrder, limit: int, cur=None, ): """Retrieve visit_status rows for visit (origin, visit) in a paginated way. """ cur = self._cursor(cur) query_parts = [ f"SELECT {', '.join(self.origin_visit_status_select_cols)} " "FROM origin_visit_status ovs ", "INNER JOIN origin o ON o.id = ovs.origin ", ] query_parts.append("WHERE o.url = %s AND ovs.visit = %s ") query_params: List[Any] = [origin, visit] if date_from is not None: op_comparison = ">=" if order == ListOrder.ASC else "<=" query_parts.append(f"and ovs.date {op_comparison} %s ") query_params.append(date_from) if order == ListOrder.ASC: query_parts.append("ORDER BY ovs.date ASC ") elif order == ListOrder.DESC: query_parts.append("ORDER BY ovs.date DESC ") else: assert False query_parts.append("LIMIT %s") query_params.append(limit) query = "\n".join(query_parts) cur.execute(query, tuple(query_params)) yield from cur def origin_visit_get_range( self, origin: str, visit_from: int, order: ListOrder, limit: int, cur=None, ): cur = self._cursor(cur) origin_visit_cols = ["o.url as origin", "ov.visit", "ov.date", "ov.type"] query_parts = [ f"SELECT {', '.join(origin_visit_cols)} FROM origin_visit ov ", "INNER JOIN origin o ON o.id = ov.origin ", ] query_parts.append("WHERE o.url = %s") query_params: List[Any] = [origin] if visit_from > 0: op_comparison = ">" if order == ListOrder.ASC else "<" query_parts.append(f"and ov.visit {op_comparison} %s") query_params.append(visit_from) if order == ListOrder.ASC: query_parts.append("ORDER BY ov.visit ASC") elif order == ListOrder.DESC: query_parts.append("ORDER BY ov.visit DESC") query_parts.append("LIMIT %s") query_params.append(limit) query = "\n".join(query_parts) cur.execute(query, tuple(query_params)) yield from cur def origin_visit_get(self, origin_id, visit_id, cur=None): """Retrieve information on visit visit_id of origin origin_id. Args: origin_id: the origin concerned visit_id: The visit step for that origin Returns: The origin_visit information """ cur = self._cursor(cur) query = """\ SELECT %s FROM origin_visit ov INNER JOIN origin o ON o.id = ov.origin INNER JOIN origin_visit_status ovs USING (origin, visit) WHERE o.url = %%s AND ov.visit = %%s ORDER BY ovs.date DESC LIMIT 1 """ % ( ", ".join(self.origin_visit_select_cols) ) cur.execute(query, (origin_id, visit_id)) r = cur.fetchall() if not r: return None return r[0] def origin_visit_find_by_date(self, origin, visit_date, cur=None): cur = self._cursor(cur) cur.execute( "SELECT * FROM swh_visit_find_by_date(%s, %s)", (origin, visit_date) ) rows = cur.fetchall() if rows: visit = dict(zip(self.origin_visit_get_cols, rows[0])) visit["origin"] = origin return visit def origin_visit_exists(self, origin_id, visit_id, cur=None): """Check whether an origin visit with the given ids exists""" cur = self._cursor(cur) query = "SELECT 1 FROM origin_visit where origin = %s AND visit = %s" cur.execute(query, (origin_id, visit_id)) return bool(cur.fetchone()) def origin_visit_get_latest( self, origin_id: str, type: Optional[str], allowed_statuses: Optional[Iterable[str]], require_snapshot: bool, cur=None, ): """Retrieve the most recent origin_visit of the given origin, with optional filters. Args: origin_id: the origin concerned type: Optional visit type to filter on allowed_statuses: the visit statuses allowed for the returned visit require_snapshot (bool): If True, only a visit with a known snapshot will be returned. Returns: The origin_visit information, or None if no visit matches. """ cur = self._cursor(cur) query_parts = [ "SELECT %s" % ", ".join(self.origin_visit_select_cols), "FROM origin_visit ov ", "INNER JOIN origin o ON o.id = ov.origin", "INNER JOIN origin_visit_status ovs USING (origin, visit)", ] query_parts.append("WHERE o.url = %s") query_params: List[Any] = [origin_id] if type is not None: query_parts.append("AND ov.type = %s") query_params.append(type) if require_snapshot: query_parts.append("AND ovs.snapshot is not null") if allowed_statuses: query_parts.append("AND ovs.status IN %s") query_params.append(tuple(allowed_statuses)) query_parts.append( "ORDER BY ov.date DESC, ov.visit DESC, ovs.date DESC LIMIT 1" ) query = "\n".join(query_parts) cur.execute(query, tuple(query_params)) r = cur.fetchone() if not r: return None return r def origin_visit_get_random(self, type, cur=None): """Randomly select one origin visit that was full and in the last 3 months """ cur = self._cursor(cur) columns = ",".join(self.origin_visit_select_cols) query = f"""select {columns} from origin_visit ov inner join origin o on ov.origin=o.id inner join origin_visit_status ovs using (origin, visit) where ovs.status='full' and ov.type=%s and ov.date > now() - '3 months'::interval and random() < 0.1 limit 1 """ cur.execute(query, (type,)) return cur.fetchone() @staticmethod def mangle_query_key(key, main_table): if key == "id": return "t.id" if key == "parents": return """ ARRAY( SELECT rh.parent_id::bytea FROM revision_history rh WHERE rh.id = t.id ORDER BY rh.parent_rank )""" if "_" not in key: return "%s.%s" % (main_table, key) head, tail = key.split("_", 1) if head in ("author", "committer") and tail in ( "name", "email", "id", "fullname", ): return "%s.%s" % (head, tail) return "%s.%s" % (main_table, key) def revision_get_from_list(self, revisions, cur=None): cur = self._cursor(cur) query_keys = ", ".join( self.mangle_query_key(k, "revision") for k in self.revision_get_cols ) yield from execute_values_generator( cur, """ SELECT %s FROM (VALUES %%s) as t(sortkey, id) LEFT JOIN revision ON t.id = revision.id LEFT JOIN person author ON revision.author = author.id LEFT JOIN person committer ON revision.committer = committer.id ORDER BY sortkey """ % query_keys, ((sortkey, id) for sortkey, id in enumerate(revisions)), ) extid_cols = ["extid", "extid_type", "target", "target_type"] def extid_get_from_extid_list(self, extid_type, ids, cur=None): cur = self._cursor(cur) query_keys = ", ".join( self.mangle_query_key(k, "extid") for k in self.extid_cols ) sql = """ SELECT %s FROM (VALUES %%s) as t(sortkey, extid, extid_type) LEFT JOIN extid USING (extid, extid_type) ORDER BY sortkey """ % ( query_keys, ) yield from execute_values_generator( cur, sql, (((sortkey, extid, extid_type) for sortkey, extid in enumerate(ids))), ) def extid_get_from_swhid_list(self, target_type, ids, cur=None): cur = self._cursor(cur) target_type = ObjectType( target_type ).name.lower() # aka "rev" -> "revision", ... query_keys = ", ".join( self.mangle_query_key(k, "extid") for k in self.extid_cols ) sql = """ SELECT %s FROM (VALUES %%s) as t(sortkey, target, target_type) LEFT JOIN extid USING (target, target_type) ORDER BY sortkey """ % ( query_keys, ) yield from execute_values_generator( cur, sql, (((sortkey, target, target_type) for sortkey, target in enumerate(ids))), template=b"(%s,%s,%s::object_type)", ) def revision_log(self, root_revisions, limit=None, cur=None): cur = self._cursor(cur) query = """SELECT %s FROM swh_revision_log(%%s, %%s) """ % ", ".join( self.revision_get_cols ) cur.execute(query, (root_revisions, limit)) yield from cur revision_shortlog_cols = ["id", "parents"] def revision_shortlog(self, root_revisions, limit=None, cur=None): cur = self._cursor(cur) query = """SELECT %s FROM swh_revision_list(%%s, %%s) """ % ", ".join( self.revision_shortlog_cols ) cur.execute(query, (root_revisions, limit)) yield from cur def revision_get_random(self, cur=None): return self._get_random_row_from_table("revision", ["id"], "id", cur) def release_missing_from_list(self, releases, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ SELECT id FROM (VALUES %s) as t(id) WHERE NOT EXISTS ( SELECT 1 FROM release r WHERE r.id = t.id ) """, ((id,) for id in releases), ) object_find_by_sha1_git_cols = ["sha1_git", "type"] def object_find_by_sha1_git(self, ids, cur=None): cur = self._cursor(cur) yield from execute_values_generator( cur, """ WITH t (sha1_git) AS (VALUES %s), known_objects as (( select id as sha1_git, 'release'::object_type as type, object_id from release r where exists (select 1 from t where t.sha1_git = r.id) ) union all ( select id as sha1_git, 'revision'::object_type as type, object_id from revision r where exists (select 1 from t where t.sha1_git = r.id) ) union all ( select id as sha1_git, 'directory'::object_type as type, object_id from directory d where exists (select 1 from t where t.sha1_git = d.id) ) union all ( select sha1_git as sha1_git, 'content'::object_type as type, object_id from content c where exists (select 1 from t where t.sha1_git = c.sha1_git) )) select t.sha1_git as sha1_git, k.type from t left join known_objects k on t.sha1_git = k.sha1_git """, ((id,) for id in ids), ) def stat_counters(self, cur=None): cur = self._cursor(cur) cur.execute("SELECT * FROM swh_stat_counters()") yield from cur def origin_add(self, url, cur=None): """Insert a new origin and return the new identifier.""" insert = """INSERT INTO origin (url) values (%s) ON CONFLICT DO NOTHING """ cur.execute(insert, (url,)) return cur.rowcount origin_cols = ["url"] def origin_get_by_url(self, origins, cur=None): """Retrieve origin `(type, url)` from urls if found.""" cur = self._cursor(cur) query = """SELECT %s FROM (VALUES %%s) as t(url) LEFT JOIN origin ON t.url = origin.url """ % ",".join( "origin." + col for col in self.origin_cols ) yield from execute_values_generator(cur, query, ((url,) for url in origins)) def origin_get_by_sha1(self, sha1s, cur=None): """Retrieve origin urls from sha1s if found.""" cur = self._cursor(cur) query = """SELECT %s FROM (VALUES %%s) as t(sha1) LEFT JOIN origin ON t.sha1 = digest(origin.url, 'sha1') """ % ",".join( "origin." + col for col in self.origin_cols ) yield from execute_values_generator(cur, query, ((sha1,) for sha1 in sha1s)) def origin_id_get_by_url(self, origins, cur=None): """Retrieve origin `(type, url)` from urls if found.""" cur = self._cursor(cur) query = """SELECT id FROM (VALUES %s) as t(url) LEFT JOIN origin ON t.url = origin.url """ for row in execute_values_generator(cur, query, ((url,) for url in origins)): yield row[0] origin_get_range_cols = ["id", "url"] def origin_get_range(self, origin_from: int = 1, origin_count: int = 100, cur=None): """Retrieve ``origin_count`` origins whose ids are greater or equal than ``origin_from``. Origins are sorted by id before retrieving them. Args: origin_from: the minimum id of origins to retrieve origin_count: the maximum number of origins to retrieve """ cur = self._cursor(cur) query = """SELECT %s FROM origin WHERE id >= %%s ORDER BY id LIMIT %%s """ % ",".join( self.origin_get_range_cols ) cur.execute(query, (origin_from, origin_count)) yield from cur def _origin_query( self, url_pattern, count=False, offset=0, limit=50, regexp=False, with_visit=False, visit_types=None, cur=None, ): """ Method factorizing query creation for searching and counting origins. """ cur = self._cursor(cur) if count: origin_cols = "COUNT(*)" order_clause = "" else: origin_cols = ",".join(self.origin_cols) order_clause = "ORDER BY id" if not regexp: operator = "ILIKE" query_params = [f"%{url_pattern}%"] else: operator = "~*" query_params = [url_pattern] query = f""" WITH filtered_origins AS ( SELECT * FROM origin WHERE url {operator} %s {order_clause} ) SELECT {origin_cols} FROM filtered_origins AS o """ if with_visit or visit_types: visit_predicat = ( """ INNER JOIN origin_visit_status ovs USING (origin, visit) INNER JOIN snapshot ON ovs.snapshot=snapshot.id """ if with_visit else "" ) type_predicat = ( f"AND ov.type=any(ARRAY{visit_types})" if visit_types else "" ) query += f""" WHERE EXISTS ( SELECT 1 FROM origin_visit ov {visit_predicat} WHERE ov.origin=o.id {type_predicat} ) """ if not count: query += "OFFSET %s LIMIT %s" query_params.extend([offset, limit]) cur.execute(query, query_params) def origin_search( self, url_pattern: str, offset: int = 0, limit: int = 50, regexp: bool = False, with_visit: bool = False, visit_types: Optional[List[str]] = None, cur=None, ): """Search for origins whose urls contain a provided string pattern or match a provided regular expression. The search is performed in a case insensitive way. Args: url_pattern: the string pattern to search for in origin urls offset: number of found origins to skip before returning results limit: the maximum number of found origins to return regexp: if True, consider the provided pattern as a regular expression and returns origins whose urls match it with_visit: if True, filter out origins with no visit """ self._origin_query( url_pattern, offset=offset, limit=limit, regexp=regexp, with_visit=with_visit, visit_types=visit_types, cur=cur, ) yield from cur def origin_count(self, url_pattern, regexp=False, with_visit=False, cur=None): """Count origins whose urls contain a provided string pattern or match a provided regular expression. The pattern search in origin urls is performed in a case insensitive way. Args: url_pattern (str): the string pattern to search for in origin urls regexp (bool): if True, consider the provided pattern as a regular expression and returns origins whose urls match it with_visit (bool): if True, filter out origins with no visit """ self._origin_query( url_pattern, count=True, regexp=regexp, with_visit=with_visit, cur=cur ) return cur.fetchone()[0] release_add_cols = [ "id", "target", "target_type", "date", "date_offset", "date_neg_utc_offset", "name", "comment", "synthetic", "author_fullname", "author_name", "author_email", ] release_get_cols = release_add_cols def release_get_from_list(self, releases, cur=None): cur = self._cursor(cur) query_keys = ", ".join( self.mangle_query_key(k, "release") for k in self.release_get_cols ) yield from execute_values_generator( cur, """ SELECT %s FROM (VALUES %%s) as t(sortkey, id) LEFT JOIN release ON t.id = release.id LEFT JOIN person author ON release.author = author.id ORDER BY sortkey """ % query_keys, ((sortkey, id) for sortkey, id in enumerate(releases)), ) def release_get_random(self, cur=None): return self._get_random_row_from_table("release", ["id"], "id", cur) _raw_extrinsic_metadata_context_cols = [ "origin", "visit", "snapshot", "release", "revision", "path", "directory", ] """The list of context columns for all artifact types.""" _raw_extrinsic_metadata_insert_cols = [ "id", "type", "target", "authority_id", "fetcher_id", "discovery_date", "format", "metadata", *_raw_extrinsic_metadata_context_cols, ] """List of columns of the raw_extrinsic_metadata table, used when writing metadata.""" _raw_extrinsic_metadata_insert_query = f""" INSERT INTO raw_extrinsic_metadata ({', '.join(_raw_extrinsic_metadata_insert_cols)}) VALUES ({', '.join('%s' for _ in _raw_extrinsic_metadata_insert_cols)}) ON CONFLICT (id) DO NOTHING """ raw_extrinsic_metadata_get_cols = [ "raw_extrinsic_metadata.target", "raw_extrinsic_metadata.type", "discovery_date", "metadata_authority.type", "metadata_authority.url", "metadata_fetcher.id", "metadata_fetcher.name", "metadata_fetcher.version", *_raw_extrinsic_metadata_context_cols, "format", "raw_extrinsic_metadata.metadata", ] """List of columns of the raw_extrinsic_metadata, metadata_authority, and metadata_fetcher tables, used when reading object metadata.""" _raw_extrinsic_metadata_select_query = f""" SELECT {', '.join(raw_extrinsic_metadata_get_cols)} FROM raw_extrinsic_metadata INNER JOIN metadata_authority ON (metadata_authority.id=authority_id) INNER JOIN metadata_fetcher ON (metadata_fetcher.id=fetcher_id) WHERE raw_extrinsic_metadata.target=%s AND authority_id=%s """ def raw_extrinsic_metadata_add( self, id: bytes, type: str, target: str, discovery_date: datetime.datetime, authority_id: int, fetcher_id: int, format: str, metadata: bytes, origin: Optional[str], visit: Optional[int], snapshot: Optional[str], release: Optional[str], revision: Optional[str], path: Optional[bytes], directory: Optional[str], cur, ): query = self._raw_extrinsic_metadata_insert_query args: Dict[str, Any] = dict( id=id, type=type, target=target, authority_id=authority_id, fetcher_id=fetcher_id, discovery_date=discovery_date, format=format, metadata=metadata, origin=origin, visit=visit, snapshot=snapshot, release=release, revision=revision, path=path, directory=directory, ) params = [args[col] for col in self._raw_extrinsic_metadata_insert_cols] cur.execute(query, params) def raw_extrinsic_metadata_get( self, type: str, target: str, authority_id: int, after_time: Optional[datetime.datetime], after_fetcher: Optional[int], limit: int, cur, ): query_parts = [self._raw_extrinsic_metadata_select_query] args = [target, authority_id] if after_fetcher is not None: assert after_time query_parts.append("AND (discovery_date, fetcher_id) > (%s, %s)") args.extend([after_time, after_fetcher]) elif after_time is not None: query_parts.append("AND discovery_date > %s") args.append(after_time) query_parts.append("ORDER BY discovery_date, fetcher_id") if limit: query_parts.append("LIMIT %s") args.append(limit) cur.execute(" ".join(query_parts), args) yield from cur - metadata_fetcher_cols = ["name", "version", "metadata"] + metadata_fetcher_cols = ["name", "version"] - def metadata_fetcher_add( - self, name: str, version: str, metadata: bytes, cur=None - ) -> None: + def metadata_fetcher_add(self, name: str, version: str, cur=None) -> None: cur = self._cursor(cur) cur.execute( - "INSERT INTO metadata_fetcher (name, version, metadata) " - "VALUES (%s, %s, %s) ON CONFLICT DO NOTHING", - (name, version, jsonize(metadata)), + "INSERT INTO metadata_fetcher (name, version) " + "VALUES (%s, %s) ON CONFLICT DO NOTHING", + (name, version), ) def metadata_fetcher_get(self, name: str, version: str, cur=None): cur = self._cursor(cur) cur.execute( f"SELECT {', '.join(self.metadata_fetcher_cols)} " f"FROM metadata_fetcher " f"WHERE name=%s AND version=%s", (name, version), ) return cur.fetchone() def metadata_fetcher_get_id( self, name: str, version: str, cur=None ) -> Optional[int]: cur = self._cursor(cur) cur.execute( "SELECT id FROM metadata_fetcher WHERE name=%s AND version=%s", (name, version), ) row = cur.fetchone() if row: return row[0] else: return None - metadata_authority_cols = ["type", "url", "metadata"] + metadata_authority_cols = ["type", "url"] - def metadata_authority_add( - self, type: str, url: str, metadata: bytes, cur=None - ) -> None: + def metadata_authority_add(self, type: str, url: str, cur=None) -> None: cur = self._cursor(cur) cur.execute( - "INSERT INTO metadata_authority (type, url, metadata) " - "VALUES (%s, %s, %s) ON CONFLICT DO NOTHING", - (type, url, jsonize(metadata)), + "INSERT INTO metadata_authority (type, url) " + "VALUES (%s, %s) ON CONFLICT DO NOTHING", + (type, url), ) def metadata_authority_get(self, type: str, url: str, cur=None): cur = self._cursor(cur) cur.execute( f"SELECT {', '.join(self.metadata_authority_cols)} " f"FROM metadata_authority " f"WHERE type=%s AND url=%s", (type, url), ) return cur.fetchone() def metadata_authority_get_id(self, type: str, url: str, cur=None) -> Optional[int]: cur = self._cursor(cur) cur.execute( "SELECT id FROM metadata_authority WHERE type=%s AND url=%s", (type, url) ) row = cur.fetchone() if row: return row[0] else: return None def _get_random_row_from_table(self, table_name, cols, id_col, cur=None): random_sha1 = bytes(random.randint(0, 255) for _ in range(SHA1_SIZE)) cur = self._cursor(cur) query = """ (SELECT {cols} FROM {table} WHERE {id_col} >= %s ORDER BY {id_col} LIMIT 1) UNION (SELECT {cols} FROM {table} WHERE {id_col} < %s ORDER BY {id_col} DESC LIMIT 1) LIMIT 1 """.format( cols=", ".join(cols), table=table_name, id_col=id_col ) cur.execute(query, (random_sha1, random_sha1)) row = cur.fetchone() if row: return row[0] dbversion_cols = ["version", "release", "description"] def dbversion(self): with self.transaction() as cur: cur.execute( f""" SELECT {', '.join(self.dbversion_cols)} FROM dbversion ORDER BY version DESC LIMIT 1 """ ) return dict(zip(self.dbversion_cols, cur.fetchone())) def check_dbversion(self): dbversion = self.dbversion()["version"] if dbversion != self.current_version: logger.warning( "database dbversion (%s) != %s current_version (%s)", dbversion, __name__, self.current_version, ) return dbversion == self.current_version diff --git a/swh/storage/postgresql/storage.py b/swh/storage/postgresql/storage.py index b25ba78b..b8aafc68 100644 --- a/swh/storage/postgresql/storage.py +++ b/swh/storage/postgresql/storage.py @@ -1,1480 +1,1467 @@ # Copyright (C) 2015-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import base64 from collections import defaultdict import contextlib from contextlib import contextmanager import datetime import itertools from typing import Any, Counter, Dict, Iterable, List, Optional, Sequence, Tuple import attr import psycopg2 import psycopg2.errors import psycopg2.pool from swh.core.api.serializers import msgpack_dumps, msgpack_loads from swh.core.db.common import db_transaction, db_transaction_generator from swh.model.hashutil import DEFAULT_ALGORITHMS, hash_to_bytes, hash_to_hex from swh.model.identifiers import ExtendedObjectType, ExtendedSWHID, ObjectType from swh.model.model import ( SHA1_SIZE, Content, Directory, ExtID, MetadataAuthority, MetadataAuthorityType, MetadataFetcher, Origin, OriginVisit, OriginVisitStatus, RawExtrinsicMetadata, Release, Revision, Sha1, Sha1Git, SkippedContent, Snapshot, SnapshotBranch, TargetType, ) from swh.storage.exc import HashCollision, StorageArgumentException, StorageDBError from swh.storage.interface import ( VISIT_STATUSES, ListOrder, PagedResult, PartialBranches, ) from swh.storage.metrics import process_metrics, send_metric, timed from swh.storage.objstorage import ObjStorage from swh.storage.utils import ( extract_collision_hash, get_partition_bounds_bytes, map_optional, now, ) from swh.storage.writer import JournalWriter from . import converters from .db import Db # Max block size of contents to return BULK_BLOCK_CONTENT_LEN_MAX = 10000 EMPTY_SNAPSHOT_ID = hash_to_bytes("1a8893e6a86f444e8be8e7bda6cb34fb1735a00e") """Identifier for the empty snapshot""" VALIDATION_EXCEPTIONS = ( KeyError, TypeError, ValueError, psycopg2.errors.CheckViolation, psycopg2.errors.IntegrityError, psycopg2.errors.InvalidTextRepresentation, psycopg2.errors.NotNullViolation, psycopg2.errors.NumericValueOutOfRange, psycopg2.errors.UndefinedFunction, # (raised on wrong argument typs) ) """Exceptions raised by postgresql when validation of the arguments failed.""" @contextlib.contextmanager def convert_validation_exceptions(): """Catches postgresql errors related to invalid arguments, and re-raises a StorageArgumentException.""" try: yield except tuple(VALIDATION_EXCEPTIONS) as e: raise StorageArgumentException(str(e)) class Storage: """SWH storage proxy, encompassing DB and object storage """ def __init__( self, db, objstorage, min_pool_conns=1, max_pool_conns=10, journal_writer=None ): """ Args: db_conn: either a libpq connection string, or a psycopg2 connection obj_root: path to the root of the object storage """ try: if isinstance(db, psycopg2.extensions.connection): self._pool = None self._db = Db(db) else: self._pool = psycopg2.pool.ThreadedConnectionPool( min_pool_conns, max_pool_conns, db ) self._db = None except psycopg2.OperationalError as e: raise StorageDBError(e) self.journal_writer = JournalWriter(journal_writer) self.objstorage = ObjStorage(objstorage) def get_db(self): if self._db: return self._db else: return Db.from_pool(self._pool) def put_db(self, db): if db is not self._db: db.put_conn() @contextmanager def db(self): db = None try: db = self.get_db() yield db finally: if db: self.put_db(db) @timed @db_transaction() def check_config(self, *, check_write: bool, db=None, cur=None) -> bool: if not self.objstorage.check_config(check_write=check_write): return False if not db.check_dbversion(): return False # Check permissions on one of the tables if check_write: check = "INSERT" else: check = "SELECT" cur.execute("select has_table_privilege(current_user, 'content', %s)", (check,)) return cur.fetchone()[0] def _content_unique_key(self, hash, db): """Given a hash (tuple or dict), return a unique key from the aggregation of keys. """ keys = db.content_hash_keys if isinstance(hash, tuple): return hash return tuple([hash[k] for k in keys]) def _content_add_metadata(self, db, cur, content): """Add content to the postgresql database but not the object storage. """ # create temporary table for metadata injection db.mktemp("content", cur) db.copy_to( (c.to_dict() for c in content), "tmp_content", db.content_add_keys, cur ) # move metadata in place try: db.content_add_from_temp(cur) except psycopg2.IntegrityError as e: if e.diag.sqlstate == "23505" and e.diag.table_name == "content": message_detail = e.diag.message_detail if message_detail: hash_name, hash_id = extract_collision_hash(message_detail) collision_contents_hashes = [ c.hashes() for c in content if c.get_hash(hash_name) == hash_id ] else: constraint_to_hash_name = { "content_pkey": "sha1", "content_sha1_git_idx": "sha1_git", "content_sha256_idx": "sha256", } hash_name = constraint_to_hash_name.get(e.diag.constraint_name) hash_id = None collision_contents_hashes = None raise HashCollision( hash_name, hash_id, collision_contents_hashes ) from None else: raise @timed @process_metrics def content_add(self, content: List[Content]) -> Dict[str, int]: ctime = now() contents = [attr.evolve(c, ctime=ctime) for c in content] # Must add to the objstorage before the DB and journal. Otherwise: # 1. in case of a crash the DB may "believe" we have the content, but # we didn't have time to write to the objstorage before the crash # 2. the objstorage mirroring, which reads from the journal, may attempt to # read from the objstorage before we finished writing it objstorage_summary = self.objstorage.content_add(contents) with self.db() as db: with db.transaction() as cur: missing = list( self.content_missing( map(Content.to_dict, contents), key_hash="sha1_git", db=db, cur=cur, ) ) contents = [c for c in contents if c.sha1_git in missing] self.journal_writer.content_add(contents) self._content_add_metadata(db, cur, contents) return { "content:add": len(contents), "content:add:bytes": objstorage_summary["content:add:bytes"], } @timed @db_transaction() def content_update( self, contents: List[Dict[str, Any]], keys: List[str] = [], db=None, cur=None ) -> None: # TODO: Add a check on input keys. How to properly implement # this? We don't know yet the new columns. self.journal_writer.content_update(contents) db.mktemp("content", cur) select_keys = list(set(db.content_get_metadata_keys).union(set(keys))) with convert_validation_exceptions(): db.copy_to(contents, "tmp_content", select_keys, cur) db.content_update_from_temp(keys_to_update=keys, cur=cur) @timed @process_metrics @db_transaction() def content_add_metadata( self, content: List[Content], db=None, cur=None ) -> Dict[str, int]: missing = self.content_missing( (c.to_dict() for c in content), key_hash="sha1_git", db=db, cur=cur, ) contents = [c for c in content if c.sha1_git in missing] self.journal_writer.content_add_metadata(contents) self._content_add_metadata(db, cur, contents) return { "content:add": len(contents), } @timed def content_get_data(self, content: Sha1) -> Optional[bytes]: # FIXME: Make this method support slicing the `data` return self.objstorage.content_get(content) @timed @db_transaction() def content_get_partition( self, partition_id: int, nb_partitions: int, page_token: Optional[str] = None, limit: int = 1000, db=None, cur=None, ) -> PagedResult[Content]: if limit is None: raise StorageArgumentException("limit should not be None") (start, end) = get_partition_bounds_bytes( partition_id, nb_partitions, SHA1_SIZE ) if page_token: start = hash_to_bytes(page_token) if end is None: end = b"\xff" * SHA1_SIZE next_page_token: Optional[str] = None contents = [] for counter, row in enumerate(db.content_get_range(start, end, limit + 1, cur)): row_d = dict(zip(db.content_get_metadata_keys, row)) content = Content(**row_d) if counter >= limit: # take the last content for the next page starting from this next_page_token = hash_to_hex(content.sha1) break contents.append(content) assert len(contents) <= limit return PagedResult(results=contents, next_page_token=next_page_token) @timed @db_transaction(statement_timeout=500) def content_get( self, contents: List[Sha1], db=None, cur=None ) -> List[Optional[Content]]: contents_by_sha1: Dict[Sha1, Optional[Content]] = {} for row in db.content_get_metadata_from_sha1s(contents, cur): row_d = dict(zip(db.content_get_metadata_keys, row)) content = Content(**row_d) contents_by_sha1[content.sha1] = content return [contents_by_sha1.get(sha1) for sha1 in contents] @timed @db_transaction_generator() def content_missing( self, contents: List[Dict[str, Any]], key_hash: str = "sha1", db=None, cur=None ) -> Iterable[bytes]: if key_hash not in DEFAULT_ALGORITHMS: raise StorageArgumentException( "key_hash should be one of {','.join(DEFAULT_ALGORITHMS)}" ) keys = db.content_hash_keys key_hash_idx = keys.index(key_hash) for obj in db.content_missing_from_list(contents, cur): yield obj[key_hash_idx] @timed @db_transaction_generator() def content_missing_per_sha1( self, contents: List[bytes], db=None, cur=None ) -> Iterable[bytes]: for obj in db.content_missing_per_sha1(contents, cur): yield obj[0] @timed @db_transaction_generator() def content_missing_per_sha1_git( self, contents: List[bytes], db=None, cur=None ) -> Iterable[Sha1Git]: for obj in db.content_missing_per_sha1_git(contents, cur): yield obj[0] @timed @db_transaction() def content_find(self, content: Dict[str, Any], db=None, cur=None) -> List[Content]: if not set(content).intersection(DEFAULT_ALGORITHMS): raise StorageArgumentException( "content keys must contain at least one " f"of: {', '.join(sorted(DEFAULT_ALGORITHMS))}" ) rows = db.content_find( sha1=content.get("sha1"), sha1_git=content.get("sha1_git"), sha256=content.get("sha256"), blake2s256=content.get("blake2s256"), cur=cur, ) contents = [] for row in rows: row_d = dict(zip(db.content_find_cols, row)) contents.append(Content(**row_d)) return contents @timed @db_transaction() def content_get_random(self, db=None, cur=None) -> Sha1Git: return db.content_get_random(cur) @staticmethod def _skipped_content_normalize(d): d = d.copy() if d.get("status") is None: d["status"] = "absent" if d.get("length") is None: d["length"] = -1 return d def _skipped_content_add_metadata(self, db, cur, content: List[SkippedContent]): origin_ids = db.origin_id_get_by_url([cont.origin for cont in content], cur=cur) content = [ attr.evolve(c, origin=origin_id) for (c, origin_id) in zip(content, origin_ids) ] db.mktemp("skipped_content", cur) db.copy_to( [c.to_dict() for c in content], "tmp_skipped_content", db.skipped_content_keys, cur, ) # move metadata in place db.skipped_content_add_from_temp(cur) @timed @process_metrics @db_transaction() def skipped_content_add( self, content: List[SkippedContent], db=None, cur=None ) -> Dict[str, int]: ctime = now() content = [attr.evolve(c, ctime=ctime) for c in content] missing_contents = self.skipped_content_missing( (c.to_dict() for c in content), db=db, cur=cur, ) content = [ c for c in content if any( all( c.get_hash(algo) == missing_content.get(algo) for algo in DEFAULT_ALGORITHMS ) for missing_content in missing_contents ) ] self.journal_writer.skipped_content_add(content) self._skipped_content_add_metadata(db, cur, content) return { "skipped_content:add": len(content), } @timed @db_transaction_generator() def skipped_content_missing( self, contents: List[Dict[str, Any]], db=None, cur=None ) -> Iterable[Dict[str, Any]]: contents = list(contents) for content in db.skipped_content_missing(contents, cur): yield dict(zip(db.content_hash_keys, content)) @timed @process_metrics @db_transaction() def directory_add( self, directories: List[Directory], db=None, cur=None ) -> Dict[str, int]: summary = {"directory:add": 0} dirs = set() dir_entries: Dict[str, defaultdict] = { "file": defaultdict(list), "dir": defaultdict(list), "rev": defaultdict(list), } for cur_dir in directories: dir_id = cur_dir.id dirs.add(dir_id) for src_entry in cur_dir.entries: entry = src_entry.to_dict() entry["dir_id"] = dir_id dir_entries[entry["type"]][dir_id].append(entry) dirs_missing = set(self.directory_missing(dirs, db=db, cur=cur)) if not dirs_missing: return summary self.journal_writer.directory_add( dir_ for dir_ in directories if dir_.id in dirs_missing ) # Copy directory ids dirs_missing_dict = ({"id": dir} for dir in dirs_missing) db.mktemp("directory", cur) db.copy_to(dirs_missing_dict, "tmp_directory", ["id"], cur) # Copy entries for entry_type, entry_list in dir_entries.items(): entries = itertools.chain.from_iterable( entries_for_dir for dir_id, entries_for_dir in entry_list.items() if dir_id in dirs_missing ) db.mktemp_dir_entry(entry_type) db.copy_to( entries, "tmp_directory_entry_%s" % entry_type, ["target", "name", "perms", "dir_id"], cur, ) # Do the final copy db.directory_add_from_temp(cur) summary["directory:add"] = len(dirs_missing) return summary @timed @db_transaction_generator() def directory_missing( self, directories: List[Sha1Git], db=None, cur=None ) -> Iterable[Sha1Git]: for obj in db.directory_missing_from_list(directories, cur): yield obj[0] @timed @db_transaction_generator(statement_timeout=20000) def directory_ls( self, directory: Sha1Git, recursive: bool = False, db=None, cur=None ) -> Iterable[Dict[str, Any]]: if recursive: res_gen = db.directory_walk(directory, cur=cur) else: res_gen = db.directory_walk_one(directory, cur=cur) for line in res_gen: yield dict(zip(db.directory_ls_cols, line)) @timed @db_transaction(statement_timeout=2000) def directory_entry_get_by_path( self, directory: Sha1Git, paths: List[bytes], db=None, cur=None ) -> Optional[Dict[str, Any]]: res = db.directory_entry_get_by_path(directory, paths, cur) return dict(zip(db.directory_ls_cols, res)) if res else None @timed @db_transaction() def directory_get_random(self, db=None, cur=None) -> Sha1Git: return db.directory_get_random(cur) @timed @process_metrics @db_transaction() def revision_add( self, revisions: List[Revision], db=None, cur=None ) -> Dict[str, int]: summary = {"revision:add": 0} revisions_missing = set( self.revision_missing( set(revision.id for revision in revisions), db=db, cur=cur ) ) if not revisions_missing: return summary db.mktemp_revision(cur) revisions_filtered = [ revision for revision in revisions if revision.id in revisions_missing ] self.journal_writer.revision_add(revisions_filtered) db_revisions_filtered = list(map(converters.revision_to_db, revisions_filtered)) parents_filtered: List[bytes] = [] with convert_validation_exceptions(): db.copy_to( db_revisions_filtered, "tmp_revision", db.revision_add_cols, cur, lambda rev: parents_filtered.extend(rev["parents"]), ) db.revision_add_from_temp(cur) db.copy_to( parents_filtered, "revision_history", ["id", "parent_id", "parent_rank"], cur, ) return {"revision:add": len(revisions_missing)} @timed @db_transaction_generator() def revision_missing( self, revisions: List[Sha1Git], db=None, cur=None ) -> Iterable[Sha1Git]: if not revisions: return None for obj in db.revision_missing_from_list(revisions, cur): yield obj[0] @timed @db_transaction(statement_timeout=1000) def revision_get( self, revision_ids: List[Sha1Git], db=None, cur=None ) -> List[Optional[Revision]]: revisions = [] for line in db.revision_get_from_list(revision_ids, cur): revision = converters.db_to_revision(dict(zip(db.revision_get_cols, line))) revisions.append(revision) return revisions @timed @db_transaction_generator(statement_timeout=2000) def revision_log( self, revisions: List[Sha1Git], limit: Optional[int] = None, db=None, cur=None ) -> Iterable[Optional[Dict[str, Any]]]: for line in db.revision_log(revisions, limit, cur): data = converters.db_to_revision(dict(zip(db.revision_get_cols, line))) if not data: yield None continue yield data.to_dict() @timed @db_transaction_generator(statement_timeout=2000) def revision_shortlog( self, revisions: List[Sha1Git], limit: Optional[int] = None, db=None, cur=None ) -> Iterable[Optional[Tuple[Sha1Git, Tuple[Sha1Git, ...]]]]: yield from db.revision_shortlog(revisions, limit, cur) @timed @db_transaction() def revision_get_random(self, db=None, cur=None) -> Sha1Git: return db.revision_get_random(cur) @timed @db_transaction() def extid_get_from_extid( self, id_type: str, ids: List[bytes], db=None, cur=None ) -> List[ExtID]: extids = [] for row in db.extid_get_from_extid_list(id_type, ids, cur): if row[0] is not None: extids.append(converters.db_to_extid(dict(zip(db.extid_cols, row)))) return extids @timed @db_transaction() def extid_get_from_target( self, target_type: ObjectType, ids: List[Sha1Git], db=None, cur=None ) -> List[ExtID]: extids = [] for row in db.extid_get_from_swhid_list(target_type.value, ids, cur): if row[0] is not None: extids.append(converters.db_to_extid(dict(zip(db.extid_cols, row)))) return extids @timed @db_transaction() def extid_add(self, ids: List[ExtID], db=None, cur=None) -> Dict[str, int]: extid = [ { "extid": extid.extid, "extid_type": extid.extid_type, "target": extid.target.object_id, "target_type": extid.target.object_type.name.lower(), # arghh } for extid in ids ] db.mktemp("extid", cur) self.journal_writer.extid_add(ids) db.copy_to(extid, "tmp_extid", db.extid_cols, cur) # move metadata in place db.extid_add_from_temp(cur) return {"extid:add": len(extid)} @timed @process_metrics @db_transaction() def release_add(self, releases: List[Release], db=None, cur=None) -> Dict[str, int]: summary = {"release:add": 0} release_ids = set(release.id for release in releases) releases_missing = set(self.release_missing(release_ids, db=db, cur=cur)) if not releases_missing: return summary db.mktemp_release(cur) releases_filtered = [ release for release in releases if release.id in releases_missing ] self.journal_writer.release_add(releases_filtered) db_releases_filtered = list(map(converters.release_to_db, releases_filtered)) with convert_validation_exceptions(): db.copy_to(db_releases_filtered, "tmp_release", db.release_add_cols, cur) db.release_add_from_temp(cur) return {"release:add": len(releases_missing)} @timed @db_transaction_generator() def release_missing( self, releases: List[Sha1Git], db=None, cur=None ) -> Iterable[Sha1Git]: if not releases: return for obj in db.release_missing_from_list(releases, cur): yield obj[0] @timed @db_transaction(statement_timeout=500) def release_get( self, releases: List[Sha1Git], db=None, cur=None ) -> List[Optional[Release]]: rels = [] for release in db.release_get_from_list(releases, cur): data = converters.db_to_release(dict(zip(db.release_get_cols, release))) rels.append(data if data else None) return rels @timed @db_transaction() def release_get_random(self, db=None, cur=None) -> Sha1Git: return db.release_get_random(cur) @timed @process_metrics @db_transaction() def snapshot_add( self, snapshots: List[Snapshot], db=None, cur=None ) -> Dict[str, int]: created_temp_table = False count = 0 for snapshot in snapshots: if not db.snapshot_exists(snapshot.id, cur): if not created_temp_table: db.mktemp_snapshot_branch(cur) created_temp_table = True with convert_validation_exceptions(): db.copy_to( ( { "name": name, "target": info.target if info else None, "target_type": ( info.target_type.value if info else None ), } for name, info in snapshot.branches.items() ), "tmp_snapshot_branch", ["name", "target", "target_type"], cur, ) self.journal_writer.snapshot_add([snapshot]) db.snapshot_add(snapshot.id, cur) count += 1 return {"snapshot:add": count} @timed @db_transaction_generator() def snapshot_missing( self, snapshots: List[Sha1Git], db=None, cur=None ) -> Iterable[Sha1Git]: for obj in db.snapshot_missing_from_list(snapshots, cur): yield obj[0] @timed @db_transaction(statement_timeout=2000) def snapshot_get( self, snapshot_id: Sha1Git, db=None, cur=None ) -> Optional[Dict[str, Any]]: d = self.snapshot_get_branches(snapshot_id) if d is None: return d return { "id": d["id"], "branches": { name: branch.to_dict() if branch else None for (name, branch) in d["branches"].items() }, "next_branch": d["next_branch"], } @timed @db_transaction(statement_timeout=2000) def snapshot_count_branches( self, snapshot_id: Sha1Git, branch_name_exclude_prefix: Optional[bytes] = None, db=None, cur=None, ) -> Optional[Dict[Optional[str], int]]: return dict( [ bc for bc in db.snapshot_count_branches( snapshot_id, branch_name_exclude_prefix, cur, ) ] ) @timed @db_transaction(statement_timeout=2000) def snapshot_get_branches( self, snapshot_id: Sha1Git, branches_from: bytes = b"", branches_count: int = 1000, target_types: Optional[List[str]] = None, branch_name_include_substring: Optional[bytes] = None, branch_name_exclude_prefix: Optional[bytes] = None, db=None, cur=None, ) -> Optional[PartialBranches]: if snapshot_id == EMPTY_SNAPSHOT_ID: return PartialBranches(id=snapshot_id, branches={}, next_branch=None,) branches = {} next_branch = None fetched_branches = list( db.snapshot_get_by_id( snapshot_id, branches_from=branches_from, # the underlying SQL query can be quite expensive to execute for small # branches_count value, so we ensure a minimum branches limit of 10 for # optimal performances branches_count=max(branches_count + 1, 10), target_types=target_types, branch_name_include_substring=branch_name_include_substring, branch_name_exclude_prefix=branch_name_exclude_prefix, cur=cur, ) ) for row in fetched_branches[:branches_count]: branch_d = dict(zip(db.snapshot_get_cols, row)) del branch_d["snapshot_id"] name = branch_d.pop("name") if branch_d["target"] is None and branch_d["target_type"] is None: branch = None else: assert branch_d["target_type"] is not None branch = SnapshotBranch( target=branch_d["target"], target_type=TargetType(branch_d["target_type"]), ) branches[name] = branch if len(fetched_branches) > branches_count: next_branch = dict( zip(db.snapshot_get_cols, fetched_branches[branches_count]) )["name"] if branches: return PartialBranches( id=snapshot_id, branches=branches, next_branch=next_branch, ) return None @timed @db_transaction() def snapshot_get_random(self, db=None, cur=None) -> Sha1Git: return db.snapshot_get_random(cur) @timed @db_transaction() def origin_visit_add( self, visits: List[OriginVisit], db=None, cur=None ) -> Iterable[OriginVisit]: for visit in visits: origin = self.origin_get([visit.origin], db=db, cur=cur)[0] if not origin: # Cannot add a visit without an origin raise StorageArgumentException("Unknown origin %s", visit.origin) all_visits = [] nb_visits = 0 for visit in visits: nb_visits += 1 if not visit.visit: with convert_validation_exceptions(): visit_id = db.origin_visit_add( visit.origin, visit.date, visit.type, cur=cur ) visit = attr.evolve(visit, visit=visit_id) else: db.origin_visit_add_with_id(visit, cur=cur) assert visit.visit is not None all_visits.append(visit) # Forced to write after for the case when the visit has no id self.journal_writer.origin_visit_add([visit]) visit_status = OriginVisitStatus( origin=visit.origin, visit=visit.visit, date=visit.date, type=visit.type, status="created", snapshot=None, ) self._origin_visit_status_add(visit_status, db=db, cur=cur) send_metric("origin_visit:add", count=nb_visits, method_name="origin_visit") return all_visits def _origin_visit_status_add( self, visit_status: OriginVisitStatus, db, cur ) -> None: """Add an origin visit status""" self.journal_writer.origin_visit_status_add([visit_status]) db.origin_visit_status_add(visit_status, cur=cur) @timed @process_metrics @db_transaction() def origin_visit_status_add( self, visit_statuses: List[OriginVisitStatus], db=None, cur=None, ) -> Dict[str, int]: visit_statuses_ = [] # First round to check existence (fail early if any is ko) for visit_status in visit_statuses: origin_url = self.origin_get([visit_status.origin], db=db, cur=cur)[0] if not origin_url: raise StorageArgumentException(f"Unknown origin {visit_status.origin}") if visit_status.type is None: origin_visit = self.origin_visit_get_by( visit_status.origin, visit_status.visit, db=db, cur=cur ) if origin_visit is None: raise StorageArgumentException( f"Unknown origin visit {visit_status.visit} " f"of origin {visit_status.origin}" ) origin_visit_status = attr.evolve(visit_status, type=origin_visit.type) else: origin_visit_status = visit_status visit_statuses_.append(origin_visit_status) for visit_status in visit_statuses_: self._origin_visit_status_add(visit_status, db, cur) return {"origin_visit_status:add": len(visit_statuses_)} @timed @db_transaction() def origin_visit_status_get_latest( self, origin_url: str, visit: int, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, db=None, cur=None, ) -> Optional[OriginVisitStatus]: if allowed_statuses and not set(allowed_statuses).intersection(VISIT_STATUSES): raise StorageArgumentException( f"Unknown allowed statuses {','.join(allowed_statuses)}, only " f"{','.join(VISIT_STATUSES)} authorized" ) row_d = db.origin_visit_status_get_latest( origin_url, visit, allowed_statuses, require_snapshot, cur=cur ) if not row_d: return None return OriginVisitStatus(**row_d) @timed @db_transaction(statement_timeout=500) def origin_visit_get( self, origin: str, page_token: Optional[str] = None, order: ListOrder = ListOrder.ASC, limit: int = 10, db=None, cur=None, ) -> PagedResult[OriginVisit]: page_token = page_token or "0" if not isinstance(order, ListOrder): raise StorageArgumentException("order must be a ListOrder value") if not isinstance(page_token, str): raise StorageArgumentException("page_token must be a string.") next_page_token = None visit_from = int(page_token) visits: List[OriginVisit] = [] extra_limit = limit + 1 for row in db.origin_visit_get_range( origin, visit_from=visit_from, order=order, limit=extra_limit, cur=cur ): row_d = dict(zip(db.origin_visit_cols, row)) visits.append( OriginVisit( origin=row_d["origin"], visit=row_d["visit"], date=row_d["date"], type=row_d["type"], ) ) assert len(visits) <= extra_limit if len(visits) == extra_limit: visits = visits[:limit] next_page_token = str(visits[-1].visit) return PagedResult(results=visits, next_page_token=next_page_token) @timed @db_transaction(statement_timeout=500) def origin_visit_find_by_date( self, origin: str, visit_date: datetime.datetime, db=None, cur=None ) -> Optional[OriginVisit]: row_d = db.origin_visit_find_by_date(origin, visit_date, cur=cur) if not row_d: return None return OriginVisit( origin=row_d["origin"], visit=row_d["visit"], date=row_d["date"], type=row_d["type"], ) @timed @db_transaction(statement_timeout=500) def origin_visit_get_by( self, origin: str, visit: int, db=None, cur=None ) -> Optional[OriginVisit]: row = db.origin_visit_get(origin, visit, cur) if row: row_d = dict(zip(db.origin_visit_get_cols, row)) return OriginVisit( origin=row_d["origin"], visit=row_d["visit"], date=row_d["date"], type=row_d["type"], ) return None @timed @db_transaction(statement_timeout=4000) def origin_visit_get_latest( self, origin: str, type: Optional[str] = None, allowed_statuses: Optional[List[str]] = None, require_snapshot: bool = False, db=None, cur=None, ) -> Optional[OriginVisit]: if allowed_statuses and not set(allowed_statuses).intersection(VISIT_STATUSES): raise StorageArgumentException( f"Unknown allowed statuses {','.join(allowed_statuses)}, only " f"{','.join(VISIT_STATUSES)} authorized" ) row = db.origin_visit_get_latest( origin, type=type, allowed_statuses=allowed_statuses, require_snapshot=require_snapshot, cur=cur, ) if row: row_d = dict(zip(db.origin_visit_get_cols, row)) visit = OriginVisit( origin=row_d["origin"], visit=row_d["visit"], date=row_d["date"], type=row_d["type"], ) return visit return None @timed @db_transaction(statement_timeout=500) def origin_visit_status_get( self, origin: str, visit: int, page_token: Optional[str] = None, order: ListOrder = ListOrder.ASC, limit: int = 10, db=None, cur=None, ) -> PagedResult[OriginVisitStatus]: next_page_token = None date_from = None if page_token is not None: date_from = datetime.datetime.fromisoformat(page_token) visit_statuses: List[OriginVisitStatus] = [] # Take one more visit status so we can reuse it as the next page token if any for row in db.origin_visit_status_get_range( origin, visit, date_from=date_from, order=order, limit=limit + 1, cur=cur, ): row_d = dict(zip(db.origin_visit_status_cols, row)) visit_statuses.append(OriginVisitStatus(**row_d)) if len(visit_statuses) > limit: # last visit status date is the next page token next_page_token = str(visit_statuses[-1].date) # excluding that visit status from the result to respect the limit size visit_statuses = visit_statuses[:limit] return PagedResult(results=visit_statuses, next_page_token=next_page_token) @timed @db_transaction() def origin_visit_status_get_random( self, type: str, db=None, cur=None ) -> Optional[OriginVisitStatus]: row = db.origin_visit_get_random(type, cur) if row is not None: row_d = dict(zip(db.origin_visit_status_cols, row)) return OriginVisitStatus(**row_d) return None @timed @db_transaction(statement_timeout=2000) def object_find_by_sha1_git( self, ids: List[Sha1Git], db=None, cur=None ) -> Dict[Sha1Git, List[Dict]]: ret: Dict[Sha1Git, List[Dict]] = {id: [] for id in ids} for retval in db.object_find_by_sha1_git(ids, cur=cur): if retval[1]: ret[retval[0]].append( dict(zip(db.object_find_by_sha1_git_cols, retval)) ) return ret @timed @db_transaction(statement_timeout=500) def origin_get( self, origins: List[str], db=None, cur=None ) -> Iterable[Optional[Origin]]: rows = db.origin_get_by_url(origins, cur) result: List[Optional[Origin]] = [] for row in rows: origin_d = dict(zip(db.origin_cols, row)) url = origin_d["url"] result.append(None if url is None else Origin(url=url)) return result @timed @db_transaction(statement_timeout=500) def origin_get_by_sha1( self, sha1s: List[bytes], db=None, cur=None ) -> List[Optional[Dict[str, Any]]]: return [ dict(zip(db.origin_cols, row)) if row[0] else None for row in db.origin_get_by_sha1(sha1s, cur) ] @timed @db_transaction_generator() def origin_get_range(self, origin_from=1, origin_count=100, db=None, cur=None): for origin in db.origin_get_range(origin_from, origin_count, cur): yield dict(zip(db.origin_get_range_cols, origin)) @timed @db_transaction() def origin_list( self, page_token: Optional[str] = None, limit: int = 100, *, db=None, cur=None ) -> PagedResult[Origin]: page_token = page_token or "0" if not isinstance(page_token, str): raise StorageArgumentException("page_token must be a string.") origin_from = int(page_token) next_page_token = None origins: List[Origin] = [] # Take one more origin so we can reuse it as the next page token if any for row_d in self.origin_get_range(origin_from, limit + 1, db=db, cur=cur): origins.append(Origin(url=row_d["url"])) # keep the last_id for the pagination if needed last_id = row_d["id"] if len(origins) > limit: # data left for subsequent call # last origin id is the next page token next_page_token = str(last_id) # excluding that origin from the result to respect the limit size origins = origins[:limit] assert len(origins) <= limit return PagedResult(results=origins, next_page_token=next_page_token) @timed @db_transaction() def origin_search( self, url_pattern: str, page_token: Optional[str] = None, limit: int = 50, regexp: bool = False, with_visit: bool = False, visit_types: Optional[List[str]] = None, db=None, cur=None, ) -> PagedResult[Origin]: next_page_token = None offset = int(page_token) if page_token else 0 origins = [] # Take one more origin so we can reuse it as the next page token if any for origin in db.origin_search( url_pattern, offset, limit + 1, regexp, with_visit, visit_types, cur ): row_d = dict(zip(db.origin_cols, origin)) origins.append(Origin(url=row_d["url"])) if len(origins) > limit: # next offset next_page_token = str(offset + limit) # excluding that origin from the result to respect the limit size origins = origins[:limit] assert len(origins) <= limit return PagedResult(results=origins, next_page_token=next_page_token) @timed @db_transaction() def origin_count( self, url_pattern: str, regexp: bool = False, with_visit: bool = False, db=None, cur=None, ) -> int: return db.origin_count(url_pattern, regexp, with_visit, cur) @timed @process_metrics @db_transaction() def origin_add(self, origins: List[Origin], db=None, cur=None) -> Dict[str, int]: urls = [o.url for o in origins] known_origins = set(url for (url,) in db.origin_get_by_url(urls, cur)) # keep only one occurrence of each given origin while keeping the list # sorted as originally given to_add = sorted(set(urls) - known_origins, key=urls.index) self.journal_writer.origin_add([Origin(url=url) for url in to_add]) added = 0 for url in to_add: if db.origin_add(url, cur): added += 1 return {"origin:add": added} @db_transaction(statement_timeout=500) def stat_counters(self, db=None, cur=None): return {k: v for (k, v) in db.stat_counters()} @db_transaction() def refresh_stat_counters(self, db=None, cur=None): keys = [ "content", "directory", "directory_entry_dir", "directory_entry_file", "directory_entry_rev", "origin", "origin_visit", "person", "release", "revision", "revision_history", "skipped_content", "snapshot", ] for key in keys: cur.execute("select * from swh_update_counter(%s)", (key,)) @timed @process_metrics @db_transaction() def raw_extrinsic_metadata_add( self, metadata: List[RawExtrinsicMetadata], db, cur, ) -> Dict[str, int]: metadata = list(metadata) self.journal_writer.raw_extrinsic_metadata_add(metadata) counter = Counter[ExtendedObjectType]() for metadata_entry in metadata: authority_id = self._get_authority_id(metadata_entry.authority, db, cur) fetcher_id = self._get_fetcher_id(metadata_entry.fetcher, db, cur) db.raw_extrinsic_metadata_add( id=metadata_entry.id, type=metadata_entry.target.object_type.name.lower(), target=str(metadata_entry.target), discovery_date=metadata_entry.discovery_date, authority_id=authority_id, fetcher_id=fetcher_id, format=metadata_entry.format, metadata=metadata_entry.metadata, origin=metadata_entry.origin, visit=metadata_entry.visit, snapshot=map_optional(str, metadata_entry.snapshot), release=map_optional(str, metadata_entry.release), revision=map_optional(str, metadata_entry.revision), path=metadata_entry.path, directory=map_optional(str, metadata_entry.directory), cur=cur, ) counter[metadata_entry.target.object_type] += 1 return { f"{type.value}_metadata:add": count for (type, count) in counter.items() } @db_transaction() def raw_extrinsic_metadata_get( self, target: ExtendedSWHID, authority: MetadataAuthority, after: Optional[datetime.datetime] = None, page_token: Optional[bytes] = None, limit: int = 1000, db=None, cur=None, ) -> PagedResult[RawExtrinsicMetadata]: if page_token: (after_time, after_fetcher) = msgpack_loads(base64.b64decode(page_token)) if after and after_time < after: raise StorageArgumentException( "page_token is inconsistent with the value of 'after'." ) else: after_time = after after_fetcher = None authority_id = self._get_authority_id(authority, db, cur) if not authority_id: return PagedResult(next_page_token=None, results=[],) rows = db.raw_extrinsic_metadata_get( type, str(target), authority_id, after_time, after_fetcher, limit + 1, cur, ) rows = [dict(zip(db.raw_extrinsic_metadata_get_cols, row)) for row in rows] results = [] for row in rows: assert str(target) == row["raw_extrinsic_metadata.target"] results.append(converters.db_to_raw_extrinsic_metadata(row)) if len(results) > limit: results.pop() assert len(results) == limit last_returned_row = rows[-2] # rows[-1] corresponds to the popped result next_page_token: Optional[str] = base64.b64encode( msgpack_dumps( ( last_returned_row["discovery_date"], last_returned_row["metadata_fetcher.id"], ) ) ).decode() else: next_page_token = None return PagedResult(next_page_token=next_page_token, results=results,) @timed @process_metrics @db_transaction() def metadata_fetcher_add( self, fetchers: List[MetadataFetcher], db=None, cur=None ) -> Dict[str, int]: fetchers = list(fetchers) self.journal_writer.metadata_fetcher_add(fetchers) count = 0 for fetcher in fetchers: - if fetcher.metadata is None: - raise StorageArgumentException( - "MetadataFetcher.metadata may not be None in metadata_fetcher_add." - ) - db.metadata_fetcher_add( - fetcher.name, fetcher.version, dict(fetcher.metadata), cur=cur - ) + db.metadata_fetcher_add(fetcher.name, fetcher.version, cur=cur) count += 1 return {"metadata_fetcher:add": count} @timed @db_transaction(statement_timeout=500) def metadata_fetcher_get( self, name: str, version: str, db=None, cur=None ) -> Optional[MetadataFetcher]: row = db.metadata_fetcher_get(name, version, cur=cur) if not row: return None return MetadataFetcher.from_dict(dict(zip(db.metadata_fetcher_cols, row))) @timed @process_metrics @db_transaction() def metadata_authority_add( self, authorities: List[MetadataAuthority], db=None, cur=None ) -> Dict[str, int]: authorities = list(authorities) self.journal_writer.metadata_authority_add(authorities) count = 0 for authority in authorities: - if authority.metadata is None: - raise StorageArgumentException( - "MetadataAuthority.metadata may not be None in " - "metadata_authority_add." - ) - db.metadata_authority_add( - authority.type.value, authority.url, dict(authority.metadata), cur=cur - ) + db.metadata_authority_add(authority.type.value, authority.url, cur=cur) count += 1 return {"metadata_authority:add": count} @timed @db_transaction() def metadata_authority_get( self, type: MetadataAuthorityType, url: str, db=None, cur=None ) -> Optional[MetadataAuthority]: row = db.metadata_authority_get(type.value, url, cur=cur) if not row: return None return MetadataAuthority.from_dict(dict(zip(db.metadata_authority_cols, row))) def clear_buffers(self, object_types: Sequence[str] = ()) -> None: """Do nothing """ return None def flush(self, object_types: Sequence[str] = ()) -> Dict[str, int]: return {} def _get_authority_id(self, authority: MetadataAuthority, db, cur): authority_id = db.metadata_authority_get_id( authority.type.value, authority.url, cur ) if not authority_id: raise StorageArgumentException(f"Unknown authority {authority}") return authority_id def _get_fetcher_id(self, fetcher: MetadataFetcher, db, cur): fetcher_id = db.metadata_fetcher_get_id(fetcher.name, fetcher.version, cur) if not fetcher_id: raise StorageArgumentException(f"Unknown fetcher {fetcher}") return fetcher_id diff --git a/swh/storage/sql/30-schema.sql b/swh/storage/sql/30-schema.sql index a10c9a71..25d2487a 100644 --- a/swh/storage/sql/30-schema.sql +++ b/swh/storage/sql/30-schema.sql @@ -1,519 +1,515 @@ --- --- SQL implementation of the Software Heritage data model --- -- schema versions create table dbversion ( version int primary key, release timestamptz, description text ); comment on table dbversion is 'Details of current db version'; comment on column dbversion.version is 'SQL schema version'; comment on column dbversion.release is 'Version deployment timestamp'; comment on column dbversion.description is 'Release description'; -- latest schema version insert into dbversion(version, release, description) - values(173, now(), 'Work In Progress'); + values(174, now(), 'Work In Progress'); -- a SHA1 checksum create domain sha1 as bytea check (length(value) = 20); -- a Git object ID, i.e., a Git-style salted SHA1 checksum create domain sha1_git as bytea check (length(value) = 20); -- a SHA256 checksum create domain sha256 as bytea check (length(value) = 32); -- a blake2 checksum create domain blake2s256 as bytea check (length(value) = 32); -- UNIX path (absolute, relative, individual path component, etc.) create domain unix_path as bytea; -- a set of UNIX-like access permissions, as manipulated by, e.g., chmod create domain file_perms as int; -- an SWHID create domain swhid as text check (value ~ '^swh:[0-9]+:.*'); -- Checksums about actual file content. Note that the content itself is not -- stored in the DB, but on external (key-value) storage. A single checksum is -- used as key there, but the other can be used to verify that we do not inject -- content collisions not knowingly. create table content ( sha1 sha1 not null, sha1_git sha1_git not null, sha256 sha256 not null, blake2s256 blake2s256 not null, length bigint not null, ctime timestamptz not null default now(), -- creation time, i.e. time of (first) injection into the storage status content_status not null default 'visible', object_id bigserial ); comment on table content is 'Checksums of file content which is actually stored externally'; comment on column content.sha1 is 'Content sha1 hash'; comment on column content.sha1_git is 'Git object sha1 hash'; comment on column content.sha256 is 'Content Sha256 hash'; comment on column content.blake2s256 is 'Content blake2s hash'; comment on column content.length is 'Content length'; comment on column content.ctime is 'First seen time'; comment on column content.status is 'Content status (absent, visible, hidden)'; comment on column content.object_id is 'Content identifier'; -- An origin is a place, identified by an URL, where software source code -- artifacts can be found. We support different kinds of origins, e.g., git and -- other VCS repositories, web pages that list tarballs URLs (e.g., -- http://www.kernel.org), indirect tarball URLs (e.g., -- http://www.example.org/latest.tar.gz), etc. The key feature of an origin is -- that it can be *fetched* from (wget, git clone, svn checkout, etc.) to -- retrieve all the contained software. create table origin ( id bigserial not null, url text not null ); comment on column origin.id is 'Artifact origin id'; comment on column origin.url is 'URL of origin'; -- Content blobs observed somewhere, but not ingested into the archive for -- whatever reason. This table is separate from the content table as we might -- not have the sha1 checksum of skipped contents (for instance when we inject -- git repositories, objects that are too big will be skipped here, and we will -- only know their sha1_git). 'reason' contains the reason the content was -- skipped. origin is a nullable column allowing to find out which origin -- contains that skipped content. create table skipped_content ( sha1 sha1, sha1_git sha1_git, sha256 sha256, blake2s256 blake2s256, length bigint not null, ctime timestamptz not null default now(), status content_status not null default 'absent', reason text not null, origin bigint, object_id bigserial ); comment on table skipped_content is 'Content blobs observed, but not ingested in the archive'; comment on column skipped_content.sha1 is 'Skipped content sha1 hash'; comment on column skipped_content.sha1_git is 'Git object sha1 hash'; comment on column skipped_content.sha256 is 'Skipped content sha256 hash'; comment on column skipped_content.blake2s256 is 'Skipped content blake2s hash'; comment on column skipped_content.length is 'Skipped content length'; comment on column skipped_content.ctime is 'First seen time'; comment on column skipped_content.status is 'Skipped content status (absent, visible, hidden)'; comment on column skipped_content.reason is 'Reason for skipping'; comment on column skipped_content.origin is 'Origin table identifier'; comment on column skipped_content.object_id is 'Skipped content identifier'; -- A file-system directory. A directory is a list of directory entries (see -- tables: directory_entry_{dir,file}). -- -- To list the contents of a directory: -- 1. list the contained directory_entry_dir using array dir_entries -- 2. list the contained directory_entry_file using array file_entries -- 3. list the contained directory_entry_rev using array rev_entries -- 4. UNION -- -- Synonyms/mappings: -- * git: tree create table directory ( id sha1_git not null, dir_entries bigint[], -- sub-directories, reference directory_entry_dir file_entries bigint[], -- contained files, reference directory_entry_file rev_entries bigint[], -- mounted revisions, reference directory_entry_rev object_id bigserial -- short object identifier ); comment on table directory is 'Contents of a directory, synonymous to tree (git)'; comment on column directory.id is 'Git object sha1 hash'; comment on column directory.dir_entries is 'Sub-directories, reference directory_entry_dir'; comment on column directory.file_entries is 'Contained files, reference directory_entry_file'; comment on column directory.rev_entries is 'Mounted revisions, reference directory_entry_rev'; comment on column directory.object_id is 'Short object identifier'; -- A directory entry pointing to a (sub-)directory. create table directory_entry_dir ( id bigserial, target sha1_git not null, -- id of target directory name unix_path not null, -- path name, relative to containing dir perms file_perms not null -- unix-like permissions ); comment on table directory_entry_dir is 'Directory entry for directory'; comment on column directory_entry_dir.id is 'Directory identifier'; comment on column directory_entry_dir.target is 'Target directory identifier'; comment on column directory_entry_dir.name is 'Path name, relative to containing directory'; comment on column directory_entry_dir.perms is 'Unix-like permissions'; -- A directory entry pointing to a file content. create table directory_entry_file ( id bigserial, target sha1_git not null, -- id of target file name unix_path not null, -- path name, relative to containing dir perms file_perms not null -- unix-like permissions ); comment on table directory_entry_file is 'Directory entry for file'; comment on column directory_entry_file.id is 'File identifier'; comment on column directory_entry_file.target is 'Target file identifier'; comment on column directory_entry_file.name is 'Path name, relative to containing directory'; comment on column directory_entry_file.perms is 'Unix-like permissions'; -- A directory entry pointing to a revision. create table directory_entry_rev ( id bigserial, target sha1_git not null, -- id of target revision name unix_path not null, -- path name, relative to containing dir perms file_perms not null -- unix-like permissions ); comment on table directory_entry_rev is 'Directory entry for revision'; comment on column directory_entry_dir.id is 'Revision identifier'; comment on column directory_entry_dir.target is 'Target revision in identifier'; comment on column directory_entry_dir.name is 'Path name, relative to containing directory'; comment on column directory_entry_dir.perms is 'Unix-like permissions'; -- A person referenced by some source code artifacts, e.g., a VCS revision or -- release metadata. create table person ( id bigserial, name bytea, -- advisory: not null if we managed to parse a name email bytea, -- advisory: not null if we managed to parse an email fullname bytea not null -- freeform specification; what is actually used in the checksums -- will usually be of the form 'name ' ); comment on table person is 'Person referenced in code artifact release metadata'; comment on column person.id is 'Person identifier'; comment on column person.name is 'Name'; comment on column person.email is 'Email'; comment on column person.fullname is 'Full name (raw name)'; -- The state of a source code tree at a specific point in time. -- -- Synonyms/mappings: -- * git / subversion / etc: commit -- * tarball: a specific tarball -- -- Revisions are organized as DAGs. Each revision points to 0, 1, or more (in -- case of merges) parent revisions. Each revision points to a directory, i.e., -- a file-system tree containing files and directories. create table revision ( id sha1_git not null, date timestamptz, date_offset smallint, committer_date timestamptz, committer_date_offset smallint, type revision_type not null, directory sha1_git, -- source code 'root' directory message bytea, author bigint, committer bigint, synthetic boolean not null default false, -- true iff revision has been created by Software Heritage metadata jsonb, -- extra metadata (tarball checksums, extra commit information, etc...) object_id bigserial, date_neg_utc_offset boolean, committer_date_neg_utc_offset boolean, extra_headers bytea[][] not null -- extra headers (used in hash computation) ); comment on table revision is 'A revision represents the state of a source code tree at a specific point in time'; comment on column revision.id is 'Git-style SHA1 commit identifier'; comment on column revision.date is 'Author timestamp as UNIX epoch'; comment on column revision.date_offset is 'Author timestamp timezone, as minute offsets from UTC'; comment on column revision.date_neg_utc_offset is 'True indicates a -0 UTC offset on author timestamp'; comment on column revision.committer_date is 'Committer timestamp as UNIX epoch'; comment on column revision.committer_date_offset is 'Committer timestamp timezone, as minute offsets from UTC'; comment on column revision.committer_date_neg_utc_offset is 'True indicates a -0 UTC offset on committer timestamp'; comment on column revision.type is 'Type of revision'; comment on column revision.directory is 'Directory identifier'; comment on column revision.message is 'Commit message'; comment on column revision.author is 'Author identity'; comment on column revision.committer is 'Committer identity'; comment on column revision.synthetic is 'True iff revision has been synthesized by Software Heritage'; comment on column revision.metadata is 'Extra revision metadata'; comment on column revision.object_id is 'Non-intrinsic, sequential object identifier'; comment on column revision.extra_headers is 'Extra revision headers; used in revision hash computation'; -- either this table or the sha1_git[] column on the revision table create table revision_history ( id sha1_git not null, parent_id sha1_git not null, parent_rank int not null default 0 -- parent position in merge commits, 0-based ); comment on table revision_history is 'Sequence of revision history with parent and position in history'; comment on column revision_history.id is 'Revision history git object sha1 checksum'; comment on column revision_history.parent_id is 'Parent revision git object identifier'; comment on column revision_history.parent_rank is 'Parent position in merge commits, 0-based'; -- Crawling history of software origins visited by Software Heritage. Each -- visit is a 3-way mapping between a software origin, a timestamp, and a -- snapshot object capturing the full-state of the origin at visit time. create table origin_visit ( origin bigint not null, visit bigint not null, date timestamptz not null, type text not null ); comment on column origin_visit.origin is 'Visited origin'; comment on column origin_visit.visit is 'Sequential visit number for the origin'; comment on column origin_visit.date is 'Visit timestamp'; comment on column origin_visit.type is 'Type of loader that did the visit (hg, git, ...)'; -- Crawling history of software origin visits by Software Heritage. Each -- visit see its history change through new origin visit status updates create table origin_visit_status ( origin bigint not null, visit bigint not null, date timestamptz not null, type text not null, status origin_visit_state not null, metadata jsonb, snapshot sha1_git ); comment on column origin_visit_status.origin is 'Origin concerned by the visit update'; comment on column origin_visit_status.visit is 'Visit concerned by the visit update'; comment on column origin_visit_status.date is 'Visit update timestamp'; comment on column origin_visit_status.type is 'Type of loader that did the visit (hg, git, ...)'; comment on column origin_visit_status.status is 'Visit status (ongoing, failed, full)'; comment on column origin_visit_status.metadata is 'Optional origin visit metadata'; comment on column origin_visit_status.snapshot is 'Optional, possibly partial, snapshot of the origin visit. It can be partial.'; -- A snapshot represents the entire state of a software origin as crawled by -- Software Heritage. This table is a simple mapping between (public) intrinsic -- snapshot identifiers and (private) numeric sequential identifiers. create table snapshot ( object_id bigserial not null, -- PK internal object identifier id sha1_git not null -- snapshot intrinsic identifier ); comment on table snapshot is 'State of a software origin as crawled by Software Heritage'; comment on column snapshot.object_id is 'Internal object identifier'; comment on column snapshot.id is 'Intrinsic snapshot identifier'; -- Each snapshot associate "branch" names to other objects in the Software -- Heritage Merkle DAG. This table describes branches as mappings between names -- and target typed objects. create table snapshot_branch ( object_id bigserial not null, -- PK internal object identifier name bytea not null, -- branch name, e.g., "master" or "feature/drag-n-drop" target bytea, -- target object identifier, e.g., a revision identifier target_type snapshot_target -- target object type, e.g., "revision" ); comment on table snapshot_branch is 'Associates branches with objects in Heritage Merkle DAG'; comment on column snapshot_branch.object_id is 'Internal object identifier'; comment on column snapshot_branch.name is 'Branch name'; comment on column snapshot_branch.target is 'Target object identifier'; comment on column snapshot_branch.target_type is 'Target object type'; -- Mapping between snapshots and their branches. create table snapshot_branches ( snapshot_id bigint not null, -- snapshot identifier, ref. snapshot.object_id branch_id bigint not null -- branch identifier, ref. snapshot_branch.object_id ); comment on table snapshot_branches is 'Mapping between snapshot and their branches'; comment on column snapshot_branches.snapshot_id is 'Snapshot identifier'; comment on column snapshot_branches.branch_id is 'Branch identifier'; -- A "memorable" point in time in the development history of a software -- project. -- -- Synonyms/mappings: -- * git: tag (of the annotated kind, otherwise they are just references) -- * tarball: the release version number create table release ( id sha1_git not null, target sha1_git, date timestamptz, date_offset smallint, name bytea, comment bytea, author bigint, synthetic boolean not null default false, -- true iff release has been created by Software Heritage object_id bigserial, target_type object_type not null, date_neg_utc_offset boolean ); comment on table release is 'Details of a software release, synonymous with a tag (git) or version number (tarball)'; comment on column release.id is 'Release git identifier'; comment on column release.target is 'Target git identifier'; comment on column release.date is 'Release timestamp'; comment on column release.date_offset is 'Timestamp offset from UTC'; comment on column release.name is 'Name'; comment on column release.comment is 'Comment'; comment on column release.author is 'Author'; comment on column release.synthetic is 'Indicates if created by Software Heritage'; comment on column release.object_id is 'Object identifier'; comment on column release.target_type is 'Object type (''content'', ''directory'', ''revision'', ''release'', ''snapshot'')'; comment on column release.date_neg_utc_offset is 'True indicates -0 UTC offset for release timestamp'; -- Tools create table metadata_fetcher ( id serial not null, name text not null, - version text not null, - metadata jsonb not null + version text not null ); comment on table metadata_fetcher is 'Tools used to retrieve metadata'; comment on column metadata_fetcher.id is 'Internal identifier of the fetcher'; comment on column metadata_fetcher.name is 'Fetcher name'; comment on column metadata_fetcher.version is 'Fetcher version'; -comment on column metadata_fetcher.metadata is 'Extra information about the fetcher'; create table metadata_authority ( id serial not null, type text not null, - url text not null, - metadata jsonb not null + url text not null ); comment on table metadata_authority is 'Metadata authority information'; comment on column metadata_authority.id is 'Internal identifier of the authority'; comment on column metadata_authority.type is 'Type of authority (deposit_client/forge/registry)'; comment on column metadata_authority.url is 'Authority''s uri'; -comment on column metadata_authority.metadata is 'Other metadata about authority'; -- Extrinsic metadata on a DAG objects and origins. create table raw_extrinsic_metadata ( id sha1_git not null, type text not null, target text not null, -- metadata source authority_id bigint not null, fetcher_id bigint not null, discovery_date timestamptz not null, -- metadata itself format text not null, metadata bytea not null, -- context origin text, visit bigint, snapshot swhid, release swhid, revision swhid, path bytea, directory swhid ); comment on table raw_extrinsic_metadata is 'keeps all metadata found concerning an object'; comment on column raw_extrinsic_metadata.type is 'the type of object (content/directory/revision/release/snapshot/origin) the metadata is on'; comment on column raw_extrinsic_metadata.target is 'the SWHID or origin URL for which the metadata was found'; comment on column raw_extrinsic_metadata.discovery_date is 'the date of retrieval'; comment on column raw_extrinsic_metadata.authority_id is 'the metadata provider: github, openhub, deposit, etc.'; comment on column raw_extrinsic_metadata.fetcher_id is 'the tool used for extracting metadata: loaders, crawlers, etc.'; comment on column raw_extrinsic_metadata.format is 'name of the format of metadata, used by readers to interpret it.'; comment on column raw_extrinsic_metadata.metadata is 'original metadata in opaque format'; -- Keep a cache of object counts create table object_counts ( object_type text, -- table for which we're counting objects (PK) value bigint, -- count of objects in the table last_update timestamptz, -- last update for the object count in this table single_update boolean -- whether we update this table standalone (true) or through bucketed counts (false) ); comment on table object_counts is 'Cache of object counts'; comment on column object_counts.object_type is 'Object type (''content'', ''directory'', ''revision'', ''release'', ''snapshot'')'; comment on column object_counts.value is 'Count of objects in the table'; comment on column object_counts.last_update is 'Last update for object count'; comment on column object_counts.single_update is 'standalone (true) or bucketed counts (false)'; create table object_counts_bucketed ( line serial not null, -- PK object_type text not null, -- table for which we're counting objects identifier text not null, -- identifier across which we're bucketing objects bucket_start bytea, -- lower bound (inclusive) for the bucket bucket_end bytea, -- upper bound (exclusive) for the bucket value bigint, -- count of objects in the bucket last_update timestamptz -- last update for the object count in this bucket ); comment on table object_counts_bucketed is 'Bucketed count for objects ordered by type'; comment on column object_counts_bucketed.line is 'Auto incremented idenitfier value'; comment on column object_counts_bucketed.object_type is 'Object type (''content'', ''directory'', ''revision'', ''release'', ''snapshot'')'; comment on column object_counts_bucketed.identifier is 'Common identifier for bucketed objects'; comment on column object_counts_bucketed.bucket_start is 'Lower bound (inclusive) for the bucket'; comment on column object_counts_bucketed.bucket_end is 'Upper bound (exclusive) for the bucket'; comment on column object_counts_bucketed.value is 'Count of objects in the bucket'; comment on column object_counts_bucketed.last_update is 'Last update for the object count in this bucket'; -- The ExtID (typ. original VCS) <-> swhid relation table create table extid ( extid_type text not null, extid bytea not null, target_type object_type not null, target sha1_git not null ); comment on table extid is 'Correspondance SWH object (SWHID) <-> original revision id (vcs id)'; comment on column extid.extid_type is 'ExtID type'; comment on column extid.extid is 'Intrinsic identifier of the object (e.g. hg revision)'; comment on column extid.target_type is 'Type of SWHID of the referenced SWH object'; comment on column extid.target is 'Value (hash) of SWHID of the refenced SWH object'; diff --git a/swh/storage/tests/storage_data.py b/swh/storage/tests/storage_data.py index 12ad3a38..5ed077b4 100644 --- a/swh/storage/tests/storage_data.py +++ b/swh/storage/tests/storage_data.py @@ -1,713 +1,705 @@ # Copyright (C) 2015-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime from typing import Tuple import attr from swh.model import from_disk from swh.model.hashutil import hash_to_bytes from swh.model.identifiers import CoreSWHID, ExtendedObjectType, ExtendedSWHID from swh.model.identifiers import ObjectType as SwhidObjectType from swh.model.model import ( Content, Directory, DirectoryEntry, ExtID, MetadataAuthority, MetadataAuthorityType, MetadataFetcher, ObjectType, Origin, OriginVisit, Person, RawExtrinsicMetadata, Release, Revision, RevisionType, SkippedContent, Snapshot, SnapshotBranch, TargetType, Timestamp, TimestampWithTimezone, ) class StorageData: """Data model objects to use within tests. """ content = Content( data=b"42\n", length=3, sha1=hash_to_bytes("34973274ccef6ab4dfaaf86599792fa9c3fe4689"), sha1_git=hash_to_bytes("d81cc0710eb6cf9efd5b920a8453e1e07157b6cd"), sha256=hash_to_bytes( "084c799cd551dd1d8d5c5f9a5d593b2e931f5e36122ee5c793c1d08a19839cc0" ), blake2s256=hash_to_bytes( "d5fe1939576527e42cfd76a9455a2432fe7f56669564577dd93c4280e76d661d" ), status="visible", ) content2 = Content( data=b"4242\n", length=5, sha1=hash_to_bytes("61c2b3a30496d329e21af70dd2d7e097046d07b7"), sha1_git=hash_to_bytes("36fade77193cb6d2bd826161a0979d64c28ab4fa"), sha256=hash_to_bytes( "859f0b154fdb2d630f45e1ecae4a862915435e663248bb8461d914696fc047cd" ), blake2s256=hash_to_bytes( "849c20fad132b7c2d62c15de310adfe87be94a379941bed295e8141c6219810d" ), status="visible", ) content3 = Content( data=b"424242\n", length=7, sha1=hash_to_bytes("3e21cc4942a4234c9e5edd8a9cacd1670fe59f13"), sha1_git=hash_to_bytes("c932c7649c6dfa4b82327d121215116909eb3bea"), sha256=hash_to_bytes( "92fb72daf8c6818288a35137b72155f507e5de8d892712ab96277aaed8cf8a36" ), blake2s256=hash_to_bytes( "76d0346f44e5a27f6bafdd9c2befd304aff83780f93121d801ab6a1d4769db11" ), status="visible", ctime=datetime.datetime(2019, 12, 1, tzinfo=datetime.timezone.utc), ) contents: Tuple[Content, ...] = (content, content2, content3) skipped_content = SkippedContent( length=1024 * 1024 * 200, sha1_git=hash_to_bytes("33e45d56f88993aae6a0198013efa80716fd8920"), sha1=hash_to_bytes("43e45d56f88993aae6a0198013efa80716fd8920"), sha256=hash_to_bytes( "7bbd052ab054ef222c1c87be60cd191addedd24cc882d1f5f7f7be61dc61bb3a" ), blake2s256=hash_to_bytes( "ade18b1adecb33f891ca36664da676e12c772cc193778aac9a137b8dc5834b9b" ), reason="Content too long", status="absent", origin="file:///dev/zero", ) skipped_content2 = SkippedContent( length=1024 * 1024 * 300, sha1_git=hash_to_bytes("44e45d56f88993aae6a0198013efa80716fd8921"), sha1=hash_to_bytes("54e45d56f88993aae6a0198013efa80716fd8920"), sha256=hash_to_bytes( "8cbd052ab054ef222c1c87be60cd191addedd24cc882d1f5f7f7be61dc61bb3a" ), blake2s256=hash_to_bytes( "9ce18b1adecb33f891ca36664da676e12c772cc193778aac9a137b8dc5834b9b" ), reason="Content too long", status="absent", ) skipped_contents: Tuple[SkippedContent, ...] = (skipped_content, skipped_content2) directory5 = Directory( id=hash_to_bytes("4b825dc642cb6eb9a060e54bf8d69288fbee4904"), entries=(), ) directory = Directory( id=hash_to_bytes("5256e856a0a0898966d6ba14feb4388b8b82d302"), entries=tuple( [ DirectoryEntry( name=b"foo", type="file", target=content.sha1_git, perms=from_disk.DentryPerms.content, ), DirectoryEntry( name=b"bar\xc3", type="dir", target=directory5.id, perms=from_disk.DentryPerms.directory, ), ], ), ) directory2 = Directory( id=hash_to_bytes("8505808532953da7d2581741f01b29c04b1cb9ab"), entries=tuple( [ DirectoryEntry( name=b"oof", type="file", target=content2.sha1_git, perms=from_disk.DentryPerms.content, ) ], ), ) directory3 = Directory( id=hash_to_bytes("13089e6e544f78df7c9a40a3059050d10dee686a"), entries=tuple( [ DirectoryEntry( name=b"foo", type="file", target=content.sha1_git, perms=from_disk.DentryPerms.content, ), DirectoryEntry( name=b"subdir", type="dir", target=directory.id, perms=from_disk.DentryPerms.directory, ), DirectoryEntry( name=b"hello", type="file", target=content2.sha1_git, perms=from_disk.DentryPerms.content, ), ], ), ) directory4 = Directory( id=hash_to_bytes("cd5dfd9c09d9e99ed123bc7937a0d5fddc3cd531"), entries=tuple( [ DirectoryEntry( name=b"subdir1", type="dir", target=directory3.id, perms=from_disk.DentryPerms.directory, ) ], ), ) directories: Tuple[Directory, ...] = ( directory2, directory, directory3, directory4, directory5, ) revision = Revision( id=hash_to_bytes("01a7114f36fddd5ef2511b2cadda237a68adbb12"), message=b"hello", author=Person( name=b"Nicolas Dandrimont", email=b"nicolas@example.com", fullname=b"Nicolas Dandrimont ", ), date=TimestampWithTimezone( timestamp=Timestamp(seconds=1234567890, microseconds=0), offset=120, negative_utc=False, ), committer=Person( name=b"St\xc3fano Zacchiroli", email=b"stefano@example.com", fullname=b"St\xc3fano Zacchiroli ", ), committer_date=TimestampWithTimezone( timestamp=Timestamp(seconds=1123456789, microseconds=0), offset=120, negative_utc=False, ), parents=(), type=RevisionType.GIT, directory=directory.id, metadata={ "checksums": {"sha1": "tarball-sha1", "sha256": "tarball-sha256",}, "signed-off-by": "some-dude", }, extra_headers=( (b"gpgsig", b"test123"), (b"mergetag", b"foo\\bar"), (b"mergetag", b"\x22\xaf\x89\x80\x01\x00"), ), synthetic=True, ) revision2 = Revision( id=hash_to_bytes("a646dd94c912829659b22a1e7e143d2fa5ebde1b"), message=b"hello again", author=Person( name=b"Roberto Dicosmo", email=b"roberto@example.com", fullname=b"Roberto Dicosmo ", ), date=TimestampWithTimezone( timestamp=Timestamp(seconds=1234567843, microseconds=220000,), offset=-720, negative_utc=False, ), committer=Person( name=b"tony", email=b"ar@dumont.fr", fullname=b"tony ", ), committer_date=TimestampWithTimezone( timestamp=Timestamp(seconds=1123456789, microseconds=220000,), offset=0, negative_utc=False, ), parents=tuple([revision.id]), type=RevisionType.GIT, directory=directory2.id, metadata=None, extra_headers=(), synthetic=False, ) revision3 = Revision( id=hash_to_bytes("beb2844dff30658e27573cb46eb55980e974b391"), message=b"a simple revision with no parents this time", author=Person( name=b"Roberto Dicosmo", email=b"roberto@example.com", fullname=b"Roberto Dicosmo ", ), date=TimestampWithTimezone( timestamp=Timestamp(seconds=1234567843, microseconds=220000,), offset=-720, negative_utc=False, ), committer=Person( name=b"tony", email=b"ar@dumont.fr", fullname=b"tony ", ), committer_date=TimestampWithTimezone( timestamp=Timestamp(seconds=1127351742, microseconds=220000,), offset=0, negative_utc=False, ), parents=tuple([revision.id, revision2.id]), type=RevisionType.GIT, directory=directory2.id, metadata=None, extra_headers=(), synthetic=True, ) revision4 = Revision( id=hash_to_bytes("ae860aec43700c7f5a295e2ef47e2ae41b535dfe"), message=b"parent of self.revision2", author=Person( name=b"me", email=b"me@soft.heri", fullname=b"me ", ), date=TimestampWithTimezone( timestamp=Timestamp(seconds=1234567843, microseconds=220000,), offset=-720, negative_utc=False, ), committer=Person( name=b"committer-dude", email=b"committer@dude.com", fullname=b"committer-dude ", ), committer_date=TimestampWithTimezone( timestamp=Timestamp(seconds=1244567843, microseconds=220000,), offset=-720, negative_utc=False, ), parents=tuple([revision3.id]), type=RevisionType.GIT, directory=directory.id, metadata=None, extra_headers=(), synthetic=False, ) git_revisions: Tuple[Revision, ...] = (revision, revision2, revision3, revision4) hg_revision = Revision( id=hash_to_bytes("951c9503541e7beaf002d7aebf2abd1629084c68"), message=b"hello", author=Person( name=b"Nicolas Dandrimont", email=b"nicolas@example.com", fullname=b"Nicolas Dandrimont ", ), date=TimestampWithTimezone( timestamp=Timestamp(seconds=1234567890, microseconds=0), offset=120, negative_utc=False, ), committer=Person( name=b"St\xc3fano Zacchiroli", email=b"stefano@example.com", fullname=b"St\xc3fano Zacchiroli ", ), committer_date=TimestampWithTimezone( timestamp=Timestamp(seconds=1123456789, microseconds=0), offset=120, negative_utc=False, ), parents=(), type=RevisionType.MERCURIAL, directory=directory.id, metadata={ "checksums": {"sha1": "tarball-sha1", "sha256": "tarball-sha256",}, "signed-off-by": "some-dude", "node": "a316dfb434af2b451c1f393496b7eaeda343f543", }, extra_headers=(), synthetic=True, ) hg_revision2 = Revision( id=hash_to_bytes("df4afb063236300eb13b96a0d7fff03f7b7cbbaf"), message=b"hello again", author=Person( name=b"Roberto Dicosmo", email=b"roberto@example.com", fullname=b"Roberto Dicosmo ", ), date=TimestampWithTimezone( timestamp=Timestamp(seconds=1234567843, microseconds=220000,), offset=-720, negative_utc=False, ), committer=Person( name=b"tony", email=b"ar@dumont.fr", fullname=b"tony ", ), committer_date=TimestampWithTimezone( timestamp=Timestamp(seconds=1123456789, microseconds=220000,), offset=0, negative_utc=False, ), parents=tuple([hg_revision.id]), type=RevisionType.MERCURIAL, directory=directory2.id, metadata=None, extra_headers=( (b"node", hash_to_bytes("fa1b7c84a9b40605b67653700f268349a6d6aca1")), ), synthetic=False, ) hg_revision3 = Revision( id=hash_to_bytes("84d8e7081b47ebb88cad9fa1f25de5f330872a37"), message=b"a simple revision with no parents this time", author=Person( name=b"Roberto Dicosmo", email=b"roberto@example.com", fullname=b"Roberto Dicosmo ", ), date=TimestampWithTimezone( timestamp=Timestamp(seconds=1234567843, microseconds=220000,), offset=-720, negative_utc=False, ), committer=Person( name=b"tony", email=b"ar@dumont.fr", fullname=b"tony ", ), committer_date=TimestampWithTimezone( timestamp=Timestamp(seconds=1127351742, microseconds=220000,), offset=0, negative_utc=False, ), parents=tuple([hg_revision.id, hg_revision2.id]), type=RevisionType.MERCURIAL, directory=directory2.id, metadata=None, extra_headers=( (b"node", hash_to_bytes("7f294a01c49065a90b3fe8b4ad49f08ce9656ef6")), ), synthetic=True, ) hg_revision4 = Revision( id=hash_to_bytes("4683324ba26dfe941a72cc7552e86eaaf7c27fe3"), message=b"parent of self.revision2", author=Person( name=b"me", email=b"me@soft.heri", fullname=b"me ", ), date=TimestampWithTimezone( timestamp=Timestamp(seconds=1234567843, microseconds=220000,), offset=-720, negative_utc=False, ), committer=Person( name=b"committer-dude", email=b"committer@dude.com", fullname=b"committer-dude ", ), committer_date=TimestampWithTimezone( timestamp=Timestamp(seconds=1244567843, microseconds=220000,), offset=-720, negative_utc=False, ), parents=tuple([hg_revision3.id]), type=RevisionType.MERCURIAL, directory=directory.id, metadata=None, extra_headers=( (b"node", hash_to_bytes("f4160af0485c85823d9e829bae2c00b00a2e6297")), ), synthetic=False, ) hg_revisions: Tuple[Revision, ...] = ( hg_revision, hg_revision2, hg_revision3, hg_revision4, ) revisions: Tuple[Revision, ...] = git_revisions + hg_revisions origins: Tuple[Origin, ...] = ( Origin(url="https://github.com/user1/repo1"), Origin(url="https://github.com/user2/repo1"), Origin(url="https://github.com/user3/repo1"), Origin(url="https://gitlab.com/user1/repo1"), Origin(url="https://gitlab.com/user2/repo1"), Origin(url="https://forge.softwareheritage.org/source/repo1"), ) origin, origin2 = origins[:2] metadata_authority = MetadataAuthority( - type=MetadataAuthorityType.DEPOSIT_CLIENT, - url="http://hal.inria.example.com/", - metadata={"location": "France"}, + type=MetadataAuthorityType.DEPOSIT_CLIENT, url="http://hal.inria.example.com/", ) metadata_authority2 = MetadataAuthority( - type=MetadataAuthorityType.REGISTRY, - url="http://wikidata.example.com/", - metadata={}, + type=MetadataAuthorityType.REGISTRY, url="http://wikidata.example.com/", ) authorities: Tuple[MetadataAuthority, ...] = ( metadata_authority, metadata_authority2, ) - metadata_fetcher = MetadataFetcher( - name="swh-deposit", version="0.0.1", metadata={"sword_version": "2"}, - ) - metadata_fetcher2 = MetadataFetcher( - name="swh-example", version="0.0.1", metadata={}, - ) + metadata_fetcher = MetadataFetcher(name="swh-deposit", version="0.0.1",) + metadata_fetcher2 = MetadataFetcher(name="swh-example", version="0.0.1",) fetchers: Tuple[MetadataFetcher, ...] = (metadata_fetcher, metadata_fetcher2) date_visit1 = datetime.datetime(2015, 1, 1, 23, 0, 0, tzinfo=datetime.timezone.utc) date_visit2 = datetime.datetime(2017, 1, 1, 23, 0, 0, tzinfo=datetime.timezone.utc) date_visit3 = datetime.datetime(2018, 1, 1, 23, 0, 0, tzinfo=datetime.timezone.utc) type_visit1 = "git" type_visit2 = "hg" type_visit3 = "deb" origin_visit = OriginVisit( origin=origin.url, visit=1, date=date_visit1, type=type_visit1, ) origin_visit2 = OriginVisit( origin=origin.url, visit=2, date=date_visit2, type=type_visit1, ) origin_visit3 = OriginVisit( origin=origin2.url, visit=1, date=date_visit1, type=type_visit2, ) origin_visits: Tuple[OriginVisit, ...] = ( origin_visit, origin_visit2, origin_visit3, ) release = Release( id=hash_to_bytes("f7f222093a18ec60d781070abec4a630c850b837"), name=b"v0.0.1", author=Person( name=b"olasd", email=b"nic@olasd.fr", fullname=b"olasd ", ), date=TimestampWithTimezone( timestamp=Timestamp(seconds=1234567890, microseconds=0), offset=42, negative_utc=False, ), target=revision.id, target_type=ObjectType.REVISION, message=b"synthetic release", synthetic=True, ) release2 = Release( id=hash_to_bytes("db81a26783a3f4a9db07b4759ffc37621f159bb2"), name=b"v0.0.2", author=Person( name=b"tony", email=b"ar@dumont.fr", fullname=b"tony ", ), date=TimestampWithTimezone( timestamp=Timestamp(seconds=1634366813, microseconds=0), offset=-120, negative_utc=False, ), target=revision2.id, target_type=ObjectType.REVISION, message=b"v0.0.2\nMisc performance improvements + bug fixes", synthetic=False, ) release3 = Release( id=hash_to_bytes("1c5d42e603ce2eea44917fadca76c78bad76aeb9"), name=b"v0.0.2", author=Person( name=b"tony", email=b"tony@ardumont.fr", fullname=b"tony ", ), date=TimestampWithTimezone( timestamp=Timestamp(seconds=1634366813, microseconds=0), offset=-120, negative_utc=False, ), target=revision3.id, target_type=ObjectType.REVISION, message=b"yet another synthetic release", synthetic=True, ) releases: Tuple[Release, ...] = (release, release2, release3) snapshot = Snapshot( id=hash_to_bytes("9b922e6d8d5b803c1582aabe5525b7b91150788e"), branches={ b"master": SnapshotBranch( target=revision.id, target_type=TargetType.REVISION, ), }, ) empty_snapshot = Snapshot( id=hash_to_bytes("1a8893e6a86f444e8be8e7bda6cb34fb1735a00e"), branches={}, ) complete_snapshot = Snapshot( id=hash_to_bytes("db99fda25b43dc5cd90625ee4b0744751799c917"), branches={ b"directory": SnapshotBranch( target=directory.id, target_type=TargetType.DIRECTORY, ), b"directory2": SnapshotBranch( target=directory2.id, target_type=TargetType.DIRECTORY, ), b"content": SnapshotBranch( target=content.sha1_git, target_type=TargetType.CONTENT, ), b"alias": SnapshotBranch(target=b"revision", target_type=TargetType.ALIAS,), b"revision": SnapshotBranch( target=revision.id, target_type=TargetType.REVISION, ), b"release": SnapshotBranch( target=release.id, target_type=TargetType.RELEASE, ), b"snapshot": SnapshotBranch( target=empty_snapshot.id, target_type=TargetType.SNAPSHOT, ), b"dangling": None, }, ) snapshots: Tuple[Snapshot, ...] = (snapshot, empty_snapshot, complete_snapshot) content_metadata1 = RawExtrinsicMetadata( target=ExtendedSWHID( object_type=ExtendedObjectType.CONTENT, object_id=content.sha1_git ), origin=origin.url, discovery_date=datetime.datetime( 2015, 1, 1, 21, 0, 0, tzinfo=datetime.timezone.utc ), - authority=attr.evolve(metadata_authority, metadata=None), - fetcher=attr.evolve(metadata_fetcher, metadata=None), + authority=metadata_authority, + fetcher=metadata_fetcher, format="json", metadata=b'{"foo": "bar"}', ) content_metadata2 = RawExtrinsicMetadata( target=ExtendedSWHID( object_type=ExtendedObjectType.CONTENT, object_id=content.sha1_git ), origin=origin2.url, discovery_date=datetime.datetime( 2017, 1, 1, 22, 0, 0, tzinfo=datetime.timezone.utc ), - authority=attr.evolve(metadata_authority, metadata=None), - fetcher=attr.evolve(metadata_fetcher, metadata=None), + authority=metadata_authority, + fetcher=metadata_fetcher, format="yaml", metadata=b"foo: bar", ) content_metadata3 = RawExtrinsicMetadata( target=ExtendedSWHID( object_type=ExtendedObjectType.CONTENT, object_id=content.sha1_git ), discovery_date=datetime.datetime( 2017, 1, 1, 22, 0, 0, tzinfo=datetime.timezone.utc ), authority=attr.evolve(metadata_authority2, metadata=None), fetcher=attr.evolve(metadata_fetcher2, metadata=None), format="yaml", metadata=b"foo: bar", origin=origin.url, visit=42, snapshot=snapshot.swhid(), release=release.swhid(), revision=revision.swhid(), directory=directory.swhid(), path=b"/foo/bar", ) content_metadata: Tuple[RawExtrinsicMetadata, ...] = ( content_metadata1, content_metadata2, content_metadata3, ) origin_metadata1 = RawExtrinsicMetadata( target=Origin(origin.url).swhid(), discovery_date=datetime.datetime( 2015, 1, 1, 21, 0, 0, tzinfo=datetime.timezone.utc ), authority=attr.evolve(metadata_authority, metadata=None), fetcher=attr.evolve(metadata_fetcher, metadata=None), format="json", metadata=b'{"foo": "bar"}', ) origin_metadata2 = RawExtrinsicMetadata( target=Origin(origin.url).swhid(), discovery_date=datetime.datetime( 2017, 1, 1, 22, 0, 0, tzinfo=datetime.timezone.utc ), authority=attr.evolve(metadata_authority, metadata=None), fetcher=attr.evolve(metadata_fetcher, metadata=None), format="yaml", metadata=b"foo: bar", ) origin_metadata3 = RawExtrinsicMetadata( target=Origin(origin.url).swhid(), discovery_date=datetime.datetime( 2017, 1, 1, 22, 0, 0, tzinfo=datetime.timezone.utc ), authority=attr.evolve(metadata_authority2, metadata=None), fetcher=attr.evolve(metadata_fetcher2, metadata=None), format="yaml", metadata=b"foo: bar", ) origin_metadata: Tuple[RawExtrinsicMetadata, ...] = ( origin_metadata1, origin_metadata2, origin_metadata3, ) extid1 = ExtID( target=CoreSWHID(object_type=SwhidObjectType.REVISION, object_id=revision.id), extid_type="git", extid=revision.id, ) extid2 = ExtID( target=CoreSWHID( object_type=SwhidObjectType.REVISION, object_id=hg_revision.id ), extid_type="mercurial", extid=hash_to_bytes("a316dfb434af2b451c1f393496b7eaeda343f543"), ) extid3 = ExtID( target=CoreSWHID(object_type=SwhidObjectType.DIRECTORY, object_id=directory.id), extid_type="directory", extid=b"something", ) extids: Tuple[ExtID, ...] = ( extid1, extid2, extid3, )