diff --git a/swh/web/browse/snapshot_context.py b/swh/web/browse/snapshot_context.py index b01bc948..9e4c9ef4 100644 --- a/swh/web/browse/snapshot_context.py +++ b/swh/web/browse/snapshot_context.py @@ -1,1491 +1,1527 @@ # Copyright (C) 2018-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU Affero General Public License version 3, or any later version # See top-level LICENSE file for more information # Utility module for browsing the archive in a snapshot context. from collections import defaultdict from typing import Any, Dict, List, Optional, Tuple from django.core.cache import cache from django.shortcuts import render from django.utils.html import escape from swh.model.hashutil import hash_to_bytes from swh.model.model import Snapshot from swh.model.swhids import CoreSWHID, ObjectType from swh.web.browse.utils import ( content_display_max_size, format_log_entries, gen_content_link, gen_release_link, gen_revision_link, gen_revision_log_link, gen_revision_url, gen_snapshot_link, get_directory_entries, get_readme_to_display, prepare_content_for_display, request_content, ) from swh.web.common import archive, highlightjs from swh.web.common.exc import BadInputExc, NotFoundExc, http_status_code_message from swh.web.common.identifiers import get_swhids_info from swh.web.common.origin_visits import get_origin_visit from swh.web.common.typing import ( ContentMetadata, DirectoryMetadata, OriginInfo, SnapshotBranchInfo, SnapshotContext, SnapshotReleaseInfo, SWHObjectInfo, ) from swh.web.common.utils import ( format_utc_iso_date, gen_path_info, reverse, swh_object_icons, ) from swh.web.config import get_config _empty_snapshot_id = Snapshot(branches={}).id.hex() def _get_branch(branches, branch_name, snapshot_id): """ - Utility function to get a specific branch from a branches list. - Its purpose is to get the default HEAD branch as some software origin - (e.g those with svn type) does not have it. In that latter case, check - if there is a master branch instead and returns it. + Utility function to get a specific branch from a snapshot. + Returns None if the branch cannot be found. """ filtered_branches = [b for b in branches if b["name"] == branch_name] if filtered_branches: return filtered_branches[0] - elif branch_name == "HEAD": - filtered_branches = [b for b in branches if b["name"].endswith("master")] - if filtered_branches: - return filtered_branches[0] - elif branches: - return branches[0] else: # case where a large branches list has been truncated snp = archive.lookup_snapshot( snapshot_id, branches_from=branch_name, branches_count=1, target_types=["revision", "alias"], # pull request branches must be browsable even if they are hidden # by default in branches list branch_name_exclude_prefix=None, ) snp_branch, _, _ = process_snapshot_branches(snp) if snp_branch and snp_branch[0]["name"] == branch_name: branches.append(snp_branch[0]) return snp_branch[0] def _get_release(releases, release_name, snapshot_id): """ - Utility function to get a specific release from a releases list. - Returns None if the release can not be found in the list. + Utility function to get a specific release from a snapshot. + Returns None if the release cannot be found. """ filtered_releases = [r for r in releases if r["name"] == release_name] if filtered_releases: return filtered_releases[0] else: # case where a large branches list has been truncated try: # git origins have specific branches for releases snp = archive.lookup_snapshot( snapshot_id, branches_from=f"refs/tags/{release_name}", branches_count=1, target_types=["release"], ) except NotFoundExc: snp = archive.lookup_snapshot( snapshot_id, branches_from=release_name, branches_count=1, - target_types=["release"], + target_types=["release", "alias"], ) _, snp_release, _ = process_snapshot_branches(snp) if snp_release and snp_release[0]["name"] == release_name: releases.append(snp_release[0]) return snp_release[0] def _branch_not_found( branch_type, branch, snapshot_id, snapshot_sizes, origin_info, timestamp, visit_id ): """ Utility function to raise an exception when a specified branch/release can not be found. """ if branch_type == "branch": branch_type = "Branch" branch_type_plural = "branches" target_type = "revision" else: branch_type = "Release" branch_type_plural = "releases" target_type = "release" if snapshot_id and snapshot_sizes[target_type] == 0: msg = "Snapshot with id %s has an empty list" " of %s!" % ( snapshot_id, branch_type_plural, ) elif snapshot_id: msg = "%s %s for snapshot with id %s" " not found!" % ( branch_type, branch, snapshot_id, ) elif visit_id and snapshot_sizes[target_type] == 0: msg = ( "Origin with url %s" " for visit with id %s has an empty list" " of %s!" % (origin_info["url"], visit_id, branch_type_plural) ) elif visit_id: msg = ( "%s %s associated to visit with" " id %s for origin with url %s" " not found!" % (branch_type, branch, visit_id, origin_info["url"]) ) elif snapshot_sizes[target_type] == 0: msg = ( "Origin with url %s" " for visit with timestamp %s has an empty list" " of %s!" % (origin_info["url"], timestamp, branch_type_plural) ) else: msg = ( "%s %s associated to visit with" " timestamp %s for origin with " "url %s not found!" % (branch_type, branch, timestamp, origin_info["url"]) ) raise NotFoundExc(escape(msg)) def process_snapshot_branches( snapshot: Dict[str, Any] ) -> Tuple[List[SnapshotBranchInfo], List[SnapshotReleaseInfo], Dict[str, Any]]: """ Process a dictionary describing snapshot branches: extract those targeting revisions and releases, put them in two different lists, then sort those lists in lexicographical order of the branches' names. Args: snapshot: A dict describing a snapshot as returned for instance by :func:`swh.web.common.archive.lookup_snapshot` Returns: A tuple whose first member is the sorted list of branches targeting revisions, second member the sorted list of branches targeting releases and third member a dict mapping resolved branch aliases to their real target. """ snapshot_branches = snapshot["branches"] branches: Dict[str, SnapshotBranchInfo] = {} branch_aliases: Dict[str, str] = {} releases: Dict[str, SnapshotReleaseInfo] = {} revision_to_branch = defaultdict(set) revision_to_release = defaultdict(set) release_to_branch = defaultdict(set) for branch_name, target in snapshot_branches.items(): if not target: # FIXME: display branches with an unknown target anyway continue target_id = target["target"] target_type = target["target_type"] if target_type == "revision": branches[branch_name] = SnapshotBranchInfo( name=branch_name, alias=False, revision=target_id, date=None, directory=None, message=None, url=None, ) revision_to_branch[target_id].add(branch_name) elif target_type == "release": release_to_branch[target_id].add(branch_name) elif target_type == "alias": branch_aliases[branch_name] = target_id # FIXME: handle pointers to other object types def _add_release_info(branch, release, alias=False): releases[branch] = SnapshotReleaseInfo( name=release["name"], alias=alias, branch_name=branch, date=format_utc_iso_date(release["date"]), directory=None, id=release["id"], message=release["message"], target_type=release["target_type"], target=release["target"], url=None, ) def _add_branch_info(branch, revision, alias=False): branches[branch] = SnapshotBranchInfo( name=branch, alias=alias, revision=revision["id"], directory=revision["directory"], date=format_utc_iso_date(revision["date"]), message=revision["message"], url=None, ) releases_info = archive.lookup_release_multiple(release_to_branch.keys()) for release in releases_info: if release is None: continue branches_to_update = release_to_branch[release["id"]] for branch in branches_to_update: _add_release_info(branch, release) if release["target_type"] == "revision": revision_to_release[release["target"]].update(branches_to_update) revisions = archive.lookup_revision_multiple( set(revision_to_branch.keys()) | set(revision_to_release.keys()) ) for revision in revisions: if not revision: continue for branch in revision_to_branch[revision["id"]]: _add_branch_info(branch, revision) for release_id in revision_to_release[revision["id"]]: releases[release_id]["directory"] = revision["directory"] resolved_aliases = {} for branch_alias, branch_target in branch_aliases.items(): resolved_alias = archive.lookup_snapshot_alias(snapshot["id"], branch_alias) resolved_aliases[branch_alias] = resolved_alias if resolved_alias is None: continue target_type = resolved_alias["target_type"] target = resolved_alias["target"] if target_type == "revision": revision = archive.lookup_revision(target) _add_branch_info(branch_alias, revision, alias=True) elif target_type == "release": release = archive.lookup_release(target) _add_release_info(branch_alias, release, alias=True) if branch_alias in branches: branches[branch_alias]["name"] = branch_alias ret_branches = list(sorted(branches.values(), key=lambda b: b["name"])) ret_releases = list(sorted(releases.values(), key=lambda b: b["name"])) return ret_branches, ret_releases, resolved_aliases def get_snapshot_content( snapshot_id: str, ) -> Tuple[List[SnapshotBranchInfo], List[SnapshotReleaseInfo], Dict[str, Any]]: """Returns the lists of branches and releases associated to a swh snapshot. That list is put in cache in order to speedup the navigation in the swh-web/browse ui. .. warning:: At most 1000 branches contained in the snapshot will be returned for performance reasons. Args: snapshot_id: hexadecimal representation of the snapshot identifier Returns: A tuple with three members. The first one is a list of dict describing the snapshot branches. The second one is a list of dict describing the snapshot releases. The third one is a dict mapping resolved branch aliases to their real target. Raises: NotFoundExc if the snapshot does not exist """ cache_entry_id = "swh_snapshot_%s" % snapshot_id cache_entry = cache.get(cache_entry_id) if cache_entry: return ( cache_entry["branches"], cache_entry["releases"], cache_entry.get("aliases", {}), ) branches: List[SnapshotBranchInfo] = [] releases: List[SnapshotReleaseInfo] = [] aliases: Dict[str, Any] = {} snapshot_content_max_size = get_config()["snapshot_content_max_size"] if snapshot_id: snapshot = archive.lookup_snapshot( snapshot_id, branches_count=snapshot_content_max_size ) branches, releases, aliases = process_snapshot_branches(snapshot) cache.set( cache_entry_id, {"branches": branches, "releases": releases, "aliases": aliases} ) return branches, releases, aliases def get_origin_visit_snapshot( origin_info: OriginInfo, visit_ts: Optional[str] = None, visit_id: Optional[int] = None, snapshot_id: Optional[str] = None, ) -> Tuple[List[SnapshotBranchInfo], List[SnapshotReleaseInfo], Dict[str, Any]]: """Returns the lists of branches and releases associated to an origin for a given visit. The visit is expressed by either: * a snapshot identifier * a timestamp, if no visit with that exact timestamp is found, the closest one from the provided timestamp will be used. If no visit parameter is provided, it returns the list of branches found for the latest visit. That list is put in cache in order to speedup the navigation in the swh-web/browse ui. .. warning:: At most 1000 branches contained in the snapshot will be returned for performance reasons. Args: origin_info: a dict filled with origin information visit_ts: an ISO 8601 datetime string to parse visit_id: visit id for disambiguation in case several visits have the same timestamp snapshot_id: if provided, visit associated to the snapshot will be processed Returns: A tuple with three members. The first one is a list of dict describing the origin branches for the given visit. The second one is a list of dict describing the origin releases for the given visit. The third one is a dict mapping resolved branch aliases to their real target. Raises: NotFoundExc if the origin or its visit are not found """ visit_info = get_origin_visit(origin_info, visit_ts, visit_id, snapshot_id) return get_snapshot_content(visit_info["snapshot"]) def get_snapshot_context( snapshot_id: Optional[str] = None, origin_url: Optional[str] = None, timestamp: Optional[str] = None, visit_id: Optional[int] = None, branch_name: Optional[str] = None, release_name: Optional[str] = None, revision_id: Optional[str] = None, path: Optional[str] = None, browse_context: str = "directory", ) -> SnapshotContext: """ Utility function to compute relevant information when navigating the archive in a snapshot context. The snapshot is either referenced by its id or it will be retrieved from an origin visit. Args: snapshot_id: hexadecimal representation of a snapshot identifier origin_url: an origin_url timestamp: a datetime string for retrieving the closest visit of the origin visit_id: optional visit id for disambiguation in case of several visits with the same timestamp branch_name: optional branch name set when browsing the snapshot in that scope (will default to "HEAD" if not provided) release_name: optional release name set when browsing the snapshot in that scope revision_id: optional revision identifier set when browsing the snapshot in that scope path: optional path of the object currently browsed in the snapshot browse_context: indicates which type of object is currently browsed Returns: A dict filled with snapshot context information. Raises: swh.web.common.exc.NotFoundExc: if no snapshot is found for the visit of an origin. """ assert origin_url is not None or snapshot_id is not None origin_info = None visit_info = None url_args = {} query_params: Dict[str, Any] = {} origin_visits_url = None if origin_url: if visit_id is not None: query_params["visit_id"] = visit_id elif snapshot_id is not None: query_params["snapshot"] = snapshot_id origin_info = archive.lookup_origin({"url": origin_url}) visit_info = get_origin_visit(origin_info, timestamp, visit_id, snapshot_id) formatted_date = format_utc_iso_date(visit_info["date"]) visit_info["formatted_date"] = formatted_date snapshot_id = visit_info["snapshot"] if not snapshot_id: raise NotFoundExc( "No snapshot associated to the visit of origin " "%s on %s" % (escape(origin_url), formatted_date) ) # provided timestamp is not necessarily equals to the one # of the retrieved visit, so get the exact one in order # to use it in the urls generated below if timestamp: timestamp = visit_info["date"] branches, releases, aliases = get_origin_visit_snapshot( origin_info, timestamp, visit_id, snapshot_id ) query_params["origin_url"] = origin_info["url"] origin_visits_url = reverse( "browse-origin-visits", query_params={"origin_url": origin_info["url"]} ) if timestamp is not None: query_params["timestamp"] = format_utc_iso_date( timestamp, "%Y-%m-%dT%H:%M:%SZ" ) visit_url = reverse("browse-origin-directory", query_params=query_params) visit_info["url"] = directory_url = visit_url branches_url = reverse("browse-origin-branches", query_params=query_params) releases_url = reverse("browse-origin-releases", query_params=query_params) else: assert snapshot_id is not None branches, releases, aliases = get_snapshot_content(snapshot_id) url_args = {"snapshot_id": snapshot_id} directory_url = reverse("browse-snapshot-directory", url_args=url_args) branches_url = reverse("browse-snapshot-branches", url_args=url_args) releases_url = reverse("browse-snapshot-releases", url_args=url_args) releases = list(reversed(releases)) snapshot_sizes_cache_id = f"swh_snapshot_{snapshot_id}_sizes" snapshot_sizes = cache.get(snapshot_sizes_cache_id) if snapshot_sizes is None: snapshot_sizes = archive.lookup_snapshot_sizes(snapshot_id) cache.set(snapshot_sizes_cache_id, snapshot_sizes) is_empty = (snapshot_sizes["release"] + snapshot_sizes["revision"]) == 0 swh_snp_id = str( CoreSWHID(object_type=ObjectType.SNAPSHOT, object_id=hash_to_bytes(snapshot_id)) ) if visit_info: timestamp = format_utc_iso_date(visit_info["date"]) if origin_info: browse_view_name = f"browse-origin-{browse_context}" else: browse_view_name = f"browse-snapshot-{browse_context}" release_id = None root_directory = None snapshot_total_size = snapshot_sizes["release"] + snapshot_sizes["revision"] if path is not None: query_params["path"] = path if snapshot_total_size and revision_id is not None: + # browse specific revision for a snapshot requested revision = archive.lookup_revision(revision_id) root_directory = revision["directory"] branches.append( SnapshotBranchInfo( name=revision_id, alias=False, revision=revision_id, directory=root_directory, date=revision["date"], message=revision["message"], url=None, ) ) query_params["revision"] = revision_id elif snapshot_total_size and release_name: + # browse specific release for a snapshot requested release = _get_release(releases, release_name, snapshot_id) if release is None: _branch_not_found( "release", release_name, snapshot_id, snapshot_sizes, origin_info, timestamp, visit_id, ) else: - root_directory = release["directory"] - revision_id = release["target"] + if release["target_type"] == "revision": + revision = archive.lookup_revision(release["target"]) + root_directory = revision["directory"] + revision_id = release["target"] + elif release["target_type"] == "directory": + root_directory = release["target"] release_id = release["id"] query_params["release"] = release_name elif snapshot_total_size: + head = aliases.get("HEAD") if branch_name: + # browse specific branch for a snapshot requested query_params["branch"] = branch_name - branch = _get_branch(branches, branch_name or "HEAD", snapshot_id) - if branch is None: - _branch_not_found( - "branch", - branch_name, - snapshot_id, - snapshot_sizes, - origin_info, - timestamp, - visit_id, - ) - else: + branch = _get_branch(branches, branch_name, snapshot_id) + if branch is None: + _branch_not_found( + "branch", + branch_name, + snapshot_id, + snapshot_sizes, + origin_info, + timestamp, + visit_id, + ) + else: + branch_name = branch["name"] + revision_id = branch["revision"] + root_directory = branch["directory"] + elif head is not None: + # otherwise, browse branch targeted by the HEAD alias if it exists + if head["target_type"] == "revision": + # HEAD alias targets a revision + head_rev = archive.lookup_revision(head["target"]) + branch_name = "HEAD" + revision_id = head_rev["id"] + root_directory = head_rev["directory"] + else: + # HEAD alias targets a release + release_name = archive.lookup_release(head["target"])["name"] + head_rel = _get_release(releases, release_name, snapshot_id) + if head_rel["target_type"] == "revision": + revision = archive.lookup_revision(head_rel["target"]) + root_directory = revision["directory"] + revision_id = head_rel["target"] + elif head_rel["target_type"] == "directory": + root_directory = head_rel["target"] + release_id = head_rel["id"] + elif branches: + # fallback to browse first branch otherwise + branch = branches[0] branch_name = branch["name"] revision_id = branch["revision"] root_directory = branch["directory"] + elif releases: + # fallback to browse last release otherwise + release = releases[-1] + if release["target_type"] == "revision": + revision = archive.lookup_revision(release["target"]) + root_directory = revision["directory"] + revision_id = release["target"] + elif release["target_type"] == "directory": + root_directory = release["target"] + release_id = release["id"] + release_name = release["name"] for b in branches: branch_query_params = dict(query_params) branch_query_params.pop("release", None) if b["name"] != b["revision"]: branch_query_params.pop("revision", None) branch_query_params["branch"] = b["name"] b["url"] = reverse( browse_view_name, url_args=url_args, query_params=branch_query_params ) for r in releases: release_query_params = dict(query_params) release_query_params.pop("branch", None) release_query_params.pop("revision", None) release_query_params["release"] = r["name"] r["url"] = reverse( browse_view_name, url_args=url_args, query_params=release_query_params, ) revision_info = None if revision_id: try: revision_info = archive.lookup_revision(revision_id) except NotFoundExc: pass else: revision_info["date"] = format_utc_iso_date(revision_info["date"]) revision_info["committer_date"] = format_utc_iso_date( revision_info["committer_date"] ) if revision_info["message"]: message_lines = revision_info["message"].split("\n") revision_info["message_header"] = message_lines[0] else: revision_info["message_header"] = "" snapshot_context = SnapshotContext( directory_url=directory_url, branch=branch_name, branch_alias=branch_name in aliases, branches=branches, branches_url=branches_url, is_empty=is_empty, origin_info=origin_info, origin_visits_url=origin_visits_url, release=release_name, release_alias=release_name in aliases, release_id=release_id, query_params=query_params, releases=releases, releases_url=releases_url, revision_id=revision_id, revision_info=revision_info, root_directory=root_directory, snapshot_id=snapshot_id, snapshot_sizes=snapshot_sizes, snapshot_swhid=swh_snp_id, url_args=url_args, visit_info=visit_info, ) if revision_info: revision_info["revision_url"] = gen_revision_url(revision_id, snapshot_context) return snapshot_context def _build_breadcrumbs(snapshot_context: SnapshotContext, path: str): origin_info = snapshot_context["origin_info"] url_args = snapshot_context["url_args"] query_params = dict(snapshot_context["query_params"]) root_directory = snapshot_context["root_directory"] path_info = gen_path_info(path) if origin_info: browse_view_name = "browse-origin-directory" else: browse_view_name = "browse-snapshot-directory" breadcrumbs = [] if root_directory: query_params.pop("path", None) breadcrumbs.append( { "name": root_directory[:7], "url": reverse( browse_view_name, url_args=url_args, query_params=query_params ), } ) for pi in path_info: query_params["path"] = pi["path"] breadcrumbs.append( { "name": pi["name"], "url": reverse( browse_view_name, url_args=url_args, query_params=query_params ), } ) return breadcrumbs def _check_origin_url(snapshot_id, origin_url): if snapshot_id is None and origin_url is None: raise BadInputExc("An origin URL must be provided as query parameter.") def browse_snapshot_directory( request, snapshot_id=None, origin_url=None, timestamp=None, path=None ): """ Django view implementation for browsing a directory in a snapshot context. """ _check_origin_url(snapshot_id, origin_url) snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=timestamp, visit_id=request.GET.get("visit_id"), path=path, browse_context="directory", branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), revision_id=request.GET.get("revision"), ) root_directory = snapshot_context["root_directory"] sha1_git = root_directory error_info = { "status_code": 200, "description": None, } if root_directory and path: try: dir_info = archive.lookup_directory_with_path(root_directory, path) sha1_git = dir_info["target"] except NotFoundExc as e: sha1_git = None error_info["status_code"] = 404 error_info["description"] = f"NotFoundExc: {str(e)}" dirs = [] files = [] if sha1_git: dirs, files = get_directory_entries(sha1_git) origin_info = snapshot_context["origin_info"] visit_info = snapshot_context["visit_info"] url_args = snapshot_context["url_args"] query_params = dict(snapshot_context["query_params"]) revision_id = snapshot_context["revision_id"] snapshot_id = snapshot_context["snapshot_id"] if origin_info: browse_view_name = "browse-origin-directory" else: browse_view_name = "browse-snapshot-directory" breadcrumbs = _build_breadcrumbs(snapshot_context, path) path = "" if path is None else (path + "/") for d in dirs: if d["type"] == "rev": d["url"] = reverse("browse-revision", url_args={"sha1_git": d["target"]}) else: query_params["path"] = path + d["name"] d["url"] = reverse( browse_view_name, url_args=url_args, query_params=query_params ) sum_file_sizes = 0 readmes = {} if origin_info: browse_view_name = "browse-origin-content" else: browse_view_name = "browse-snapshot-content" for f in files: query_params["path"] = path + f["name"] f["url"] = reverse( browse_view_name, url_args=url_args, query_params=query_params ) if f["length"] is not None: sum_file_sizes += f["length"] if f["name"].lower().startswith("readme"): readmes[f["name"]] = f["checksums"]["sha1"] readme_name, readme_url, readme_html = get_readme_to_display(readmes) if origin_info: browse_view_name = "browse-origin-log" else: browse_view_name = "browse-snapshot-log" history_url = None if snapshot_id != _empty_snapshot_id: query_params.pop("path", None) history_url = reverse( browse_view_name, url_args=url_args, query_params=query_params ) nb_files = None nb_dirs = None dir_path = None if root_directory: nb_files = len(files) nb_dirs = len(dirs) dir_path = "/" + path swh_objects = [] vault_cooking = {} revision_found = True if sha1_git is None and revision_id is not None: try: archive.lookup_revision(revision_id) except NotFoundExc: revision_found = False if sha1_git is not None: swh_objects.append( SWHObjectInfo(object_type=ObjectType.DIRECTORY, object_id=sha1_git) ) vault_cooking.update( {"directory_context": True, "directory_swhid": f"swh:1:dir:{sha1_git}",} ) if revision_id is not None and revision_found: swh_objects.append( SWHObjectInfo(object_type=ObjectType.REVISION, object_id=revision_id) ) vault_cooking.update( {"revision_context": True, "revision_swhid": f"swh:1:rev:{revision_id}",} ) swh_objects.append( SWHObjectInfo(object_type=ObjectType.SNAPSHOT, object_id=snapshot_id) ) visit_date = None visit_type = None if visit_info: visit_date = format_utc_iso_date(visit_info["date"]) visit_type = visit_info["type"] release_id = snapshot_context["release_id"] if release_id: swh_objects.append( SWHObjectInfo(object_type=ObjectType.RELEASE, object_id=release_id) ) dir_metadata = DirectoryMetadata( object_type=ObjectType.DIRECTORY, object_id=sha1_git, directory=sha1_git, nb_files=nb_files, nb_dirs=nb_dirs, sum_file_sizes=sum_file_sizes, root_directory=root_directory, path=dir_path, revision=revision_id, revision_found=revision_found, release=release_id, snapshot=snapshot_id, origin_url=origin_url, visit_date=visit_date, visit_type=visit_type, ) swhids_info = get_swhids_info(swh_objects, snapshot_context, dir_metadata) dir_path = "/".join([bc["name"] for bc in breadcrumbs]) + "/" context_found = "snapshot: %s" % snapshot_context["snapshot_id"] if origin_info: context_found = "origin: %s" % origin_info["url"] heading = "Directory - %s - %s - %s" % ( dir_path, snapshot_context["branch"], context_found, ) top_right_link = None if not snapshot_context["is_empty"] and revision_found: top_right_link = { "url": history_url, "icon": swh_object_icons["revisions history"], "text": "History", } return render( request, "browse/directory.html", { "heading": heading, "swh_object_name": "Directory", "swh_object_metadata": dir_metadata, "dirs": dirs, "files": files, "breadcrumbs": breadcrumbs if root_directory else [], "top_right_link": top_right_link, "readme_name": readme_name, "readme_url": readme_url, "readme_html": readme_html, "snapshot_context": snapshot_context, "vault_cooking": vault_cooking, "show_actions": True, "swhids_info": swhids_info, "error_code": error_info["status_code"], "error_message": http_status_code_message.get(error_info["status_code"]), "error_description": error_info["description"], }, status=error_info["status_code"], ) def browse_snapshot_content( request, snapshot_id=None, origin_url=None, timestamp=None, path=None, selected_language=None, ): """ Django view implementation for browsing a content in a snapshot context. """ _check_origin_url(snapshot_id, origin_url) if path is None: raise BadInputExc("The path of a content must be given as query parameter.") snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=timestamp, visit_id=request.GET.get("visit_id"), path=path, browse_context="content", branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), revision_id=request.GET.get("revision"), ) root_directory = snapshot_context["root_directory"] sha1_git = None query_string = None content_data = {} directory_id = None split_path = path.split("/") filename = split_path[-1] filepath = path[: -len(filename)] error_info = { "status_code": 200, "description": None, } if root_directory: try: content_info = archive.lookup_directory_with_path(root_directory, path) sha1_git = content_info["target"] query_string = "sha1_git:" + sha1_git content_data = request_content(query_string) if filepath: dir_info = archive.lookup_directory_with_path(root_directory, filepath) directory_id = dir_info["target"] else: directory_id = root_directory except NotFoundExc as e: error_info["status_code"] = 404 error_info["description"] = f"NotFoundExc: {str(e)}" revision_id = snapshot_context["revision_id"] origin_info = snapshot_context["origin_info"] visit_info = snapshot_context["visit_info"] snapshot_id = snapshot_context["snapshot_id"] if content_data.get("raw_data") is not None: content_display_data = prepare_content_for_display( content_data["raw_data"], content_data["mimetype"], path ) content_data.update(content_display_data) # Override language with user-selected language if selected_language is not None: content_data["language"] = selected_language available_languages = None if content_data.get("mimetype") is not None and "text/" in content_data["mimetype"]: available_languages = highlightjs.get_supported_languages() breadcrumbs = _build_breadcrumbs(snapshot_context, filepath) breadcrumbs.append({"name": filename, "url": None}) browse_content_link = gen_content_link(sha1_git) content_raw_url = None if query_string: content_raw_url = reverse( "browse-content-raw", url_args={"query_string": query_string}, query_params={"filename": filename}, ) content_checksums = content_data.get("checksums", {}) sha1_git = content_checksums.get("sha1_git") swh_objects = [] if sha1_git is not None: swh_objects.append( SWHObjectInfo(object_type=ObjectType.CONTENT, object_id=sha1_git) ) if directory_id is not None: swh_objects.append( SWHObjectInfo(object_type=ObjectType.DIRECTORY, object_id=directory_id) ) if revision_id is not None: swh_objects.append( SWHObjectInfo(object_type=ObjectType.REVISION, object_id=revision_id) ) swh_objects.append( SWHObjectInfo(object_type=ObjectType.SNAPSHOT, object_id=snapshot_id) ) visit_date = None visit_type = None if visit_info: visit_date = format_utc_iso_date(visit_info["date"]) visit_type = visit_info["type"] release_id = snapshot_context["release_id"] if release_id: swh_objects.append( SWHObjectInfo(object_type=ObjectType.RELEASE, object_id=release_id) ) content_metadata = ContentMetadata( object_type=ObjectType.CONTENT, object_id=content_checksums.get("sha1_git"), sha1=content_checksums.get("sha1"), sha1_git=content_checksums.get("sha1_git"), sha256=content_checksums.get("sha256"), blake2s256=content_checksums.get("blake2s256"), content_url=browse_content_link, mimetype=content_data.get("mimetype"), encoding=content_data.get("encoding"), size=content_data.get("length", 0), language=content_data.get("language"), root_directory=root_directory, path=f"/{filepath}", filename=filename, directory=directory_id, revision=revision_id, release=release_id, snapshot=snapshot_id, origin_url=origin_url, visit_date=visit_date, visit_type=visit_type, ) swhids_info = get_swhids_info(swh_objects, snapshot_context, content_metadata) content_path = "/".join([bc["name"] for bc in breadcrumbs]) context_found = "snapshot: %s" % snapshot_context["snapshot_id"] if origin_info: context_found = "origin: %s" % origin_info["url"] heading = "Content - %s - %s - %s" % ( content_path, snapshot_context["branch"], context_found, ) top_right_link = None if not snapshot_context["is_empty"]: top_right_link = { "url": content_raw_url, "icon": swh_object_icons["content"], "text": "Raw File", } return render( request, "browse/content.html", { "heading": heading, "swh_object_name": "Content", "swh_object_metadata": content_metadata, "content": content_data.get("content_data"), "content_size": content_data.get("length"), "max_content_size": content_display_max_size, "filename": filename, "encoding": content_data.get("encoding"), "mimetype": content_data.get("mimetype"), "language": content_data.get("language"), "available_languages": available_languages, "breadcrumbs": breadcrumbs if root_directory else [], "top_right_link": top_right_link, "snapshot_context": snapshot_context, "vault_cooking": None, "show_actions": True, "swhids_info": swhids_info, "error_code": error_info["status_code"], "error_message": http_status_code_message.get(error_info["status_code"]), "error_description": error_info["description"], }, status=error_info["status_code"], ) PER_PAGE = 100 def browse_snapshot_log(request, snapshot_id=None, origin_url=None, timestamp=None): """ Django view implementation for browsing a revision history in a snapshot context. """ _check_origin_url(snapshot_id, origin_url) snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=timestamp, visit_id=request.GET.get("visit_id"), browse_context="log", branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), revision_id=request.GET.get("revision"), ) revision_id = snapshot_context["revision_id"] per_page = int(request.GET.get("per_page", PER_PAGE)) offset = int(request.GET.get("offset", 0)) revs_ordering = request.GET.get("revs_ordering", "committer_date") session_key = "rev_%s_log_ordering_%s" % (revision_id, revs_ordering) rev_log_session = request.session.get(session_key, None) rev_log = [] revs_walker_state = None if rev_log_session: rev_log = rev_log_session["rev_log"] revs_walker_state = rev_log_session["revs_walker_state"] if len(rev_log) < offset + per_page: revs_walker = archive.get_revisions_walker( revs_ordering, revision_id, max_revs=offset + per_page + 1, state=revs_walker_state, ) rev_log += [rev["id"] for rev in revs_walker] revs_walker_state = revs_walker.export_state() revs = rev_log[offset : offset + per_page] revision_log = archive.lookup_revision_multiple(revs) request.session[session_key] = { "rev_log": rev_log, "revs_walker_state": revs_walker_state, } origin_info = snapshot_context["origin_info"] visit_info = snapshot_context["visit_info"] url_args = snapshot_context["url_args"] query_params = snapshot_context["query_params"] snapshot_id = snapshot_context["snapshot_id"] query_params["per_page"] = per_page revs_ordering = request.GET.get("revs_ordering", "") query_params["revs_ordering"] = revs_ordering or None if origin_info: browse_view_name = "browse-origin-log" else: browse_view_name = "browse-snapshot-log" prev_log_url = None if len(rev_log) > offset + per_page: query_params["offset"] = offset + per_page prev_log_url = reverse( browse_view_name, url_args=url_args, query_params=query_params ) next_log_url = None if offset != 0: query_params["offset"] = offset - per_page next_log_url = reverse( browse_view_name, url_args=url_args, query_params=query_params ) revision_log_data = format_log_entries(revision_log, per_page, snapshot_context) browse_rev_link = gen_revision_link(revision_id) browse_log_link = gen_revision_log_link(revision_id) browse_snp_link = gen_snapshot_link(snapshot_id) revision_metadata = { "context-independent revision": browse_rev_link, "context-independent revision history": browse_log_link, "context-independent snapshot": browse_snp_link, "snapshot": snapshot_id, } if origin_info: revision_metadata["origin url"] = origin_info["url"] revision_metadata["origin visit date"] = format_utc_iso_date(visit_info["date"]) revision_metadata["origin visit type"] = visit_info["type"] swh_objects = [ SWHObjectInfo(object_type=ObjectType.REVISION, object_id=revision_id), SWHObjectInfo(object_type=ObjectType.SNAPSHOT, object_id=snapshot_id), ] release_id = snapshot_context["release_id"] if release_id: swh_objects.append( SWHObjectInfo(object_type=ObjectType.RELEASE, object_id=release_id) ) browse_rel_link = gen_release_link(release_id) revision_metadata["release"] = release_id revision_metadata["context-independent release"] = browse_rel_link swhids_info = get_swhids_info(swh_objects, snapshot_context) context_found = "snapshot: %s" % snapshot_context["snapshot_id"] if origin_info: context_found = "origin: %s" % origin_info["url"] heading = "Revision history - %s - %s" % (snapshot_context["branch"], context_found) return render( request, "browse/revision-log.html", { "heading": heading, "swh_object_name": "Revisions history", "swh_object_metadata": revision_metadata, "revision_log": revision_log_data, "revs_ordering": revs_ordering, "next_log_url": next_log_url, "prev_log_url": prev_log_url, "breadcrumbs": None, "top_right_link": None, "snapshot_context": snapshot_context, "vault_cooking": None, "show_actions": True, "swhids_info": swhids_info, }, ) def browse_snapshot_branches( request, snapshot_id=None, origin_url=None, timestamp=None, branch_name_include=None ): """ Django view implementation for browsing a list of branches in a snapshot context. """ _check_origin_url(snapshot_id, origin_url) snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=timestamp, visit_id=request.GET.get("visit_id"), ) branches_bc = request.GET.get("branches_breadcrumbs", "") branches_bc = branches_bc.split(",") if branches_bc else [] branches_from = branches_bc[-1] if branches_bc else "" origin_info = snapshot_context["origin_info"] url_args = snapshot_context["url_args"] query_params = snapshot_context["query_params"] if origin_info: browse_view_name = "browse-origin-directory" else: browse_view_name = "browse-snapshot-directory" snapshot = archive.lookup_snapshot( snapshot_context["snapshot_id"], branches_from, PER_PAGE + 1, target_types=["revision", "alias"], branch_name_include_substring=branch_name_include, ) displayed_branches = [] if snapshot: displayed_branches, _, _ = process_snapshot_branches(snapshot) for branch in displayed_branches: rev_query_params = {} if origin_info: rev_query_params["origin_url"] = origin_info["url"] revision_url = reverse( "browse-revision", url_args={"sha1_git": branch["revision"]}, query_params=query_params, ) query_params["branch"] = branch["name"] directory_url = reverse( browse_view_name, url_args=url_args, query_params=query_params ) del query_params["branch"] branch["revision_url"] = revision_url branch["directory_url"] = directory_url if origin_info: browse_view_name = "browse-origin-branches" else: browse_view_name = "browse-snapshot-branches" prev_branches_url = None next_branches_url = None if branches_bc: query_params_prev = dict(query_params) query_params_prev["branches_breadcrumbs"] = ",".join(branches_bc[:-1]) prev_branches_url = reverse( browse_view_name, url_args=url_args, query_params=query_params_prev ) elif branches_from: prev_branches_url = reverse( browse_view_name, url_args=url_args, query_params=query_params ) if snapshot and snapshot["next_branch"] is not None: query_params_next = dict(query_params) next_branch = displayed_branches[-1]["name"] del displayed_branches[-1] branches_bc.append(next_branch) query_params_next["branches_breadcrumbs"] = ",".join(branches_bc) next_branches_url = reverse( browse_view_name, url_args=url_args, query_params=query_params_next ) heading = "Branches - " if origin_info: heading += "origin: %s" % origin_info["url"] else: heading += "snapshot: %s" % snapshot_id return render( request, "browse/branches.html", { "heading": heading, "swh_object_name": "Branches", "swh_object_metadata": {}, "top_right_link": None, "displayed_branches": displayed_branches, "prev_branches_url": prev_branches_url, "next_branches_url": next_branches_url, "snapshot_context": snapshot_context, "search_string": branch_name_include or "", }, ) def browse_snapshot_releases( request, snapshot_id=None, origin_url=None, timestamp=None, release_name_include=None, ): """ Django view implementation for browsing a list of releases in a snapshot context. """ _check_origin_url(snapshot_id, origin_url) snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=timestamp, visit_id=request.GET.get("visit_id"), ) rel_bc = request.GET.get("releases_breadcrumbs", "") rel_bc = rel_bc.split(",") if rel_bc else [] rel_from = rel_bc[-1] if rel_bc else "" origin_info = snapshot_context["origin_info"] url_args = snapshot_context["url_args"] query_params = snapshot_context["query_params"] snapshot = archive.lookup_snapshot( snapshot_context["snapshot_id"], rel_from, PER_PAGE + 1, target_types=["release", "alias"], branch_name_include_substring=release_name_include, ) displayed_releases = [] if snapshot: _, displayed_releases, _ = process_snapshot_branches(snapshot) for release in displayed_releases: query_params_tgt = {"snapshot": snapshot_id} if origin_info: query_params_tgt["origin_url"] = origin_info["url"] release_url = reverse( "browse-release", url_args={"sha1_git": release["id"]}, query_params=query_params_tgt, ) target_url = "" tooltip = ( f"The release {release['name']} targets " f"{release['target_type']} {release['target']}" ) if release["target_type"] == "revision": target_url = reverse( "browse-revision", url_args={"sha1_git": release["target"]}, query_params=query_params_tgt, ) elif release["target_type"] == "directory": target_url = reverse( "browse-directory", url_args={"sha1_git": release["target"]}, query_params=query_params_tgt, ) elif release["target_type"] == "content": target_url = reverse( "browse-content", url_args={"query_string": release["target"]}, query_params=query_params_tgt, ) elif release["target_type"] == "release": target_url = reverse( "browse-release", url_args={"sha1_git": release["target"]}, query_params=query_params_tgt, ) tooltip = ( f"The release {release['name']} " f"is an alias for release {release['target']}" ) release["release_url"] = release_url release["target_url"] = target_url release["tooltip"] = tooltip if origin_info: browse_view_name = "browse-origin-releases" else: browse_view_name = "browse-snapshot-releases" prev_releases_url = None next_releases_url = None if rel_bc: query_params_prev = dict(query_params) query_params_prev["releases_breadcrumbs"] = ",".join(rel_bc[:-1]) prev_releases_url = reverse( browse_view_name, url_args=url_args, query_params=query_params_prev ) elif rel_from: prev_releases_url = reverse( browse_view_name, url_args=url_args, query_params=query_params ) if snapshot and snapshot["next_branch"] is not None: query_params_next = dict(query_params) next_rel = displayed_releases[-1]["branch_name"] del displayed_releases[-1] rel_bc.append(next_rel) query_params_next["releases_breadcrumbs"] = ",".join(rel_bc) next_releases_url = reverse( browse_view_name, url_args=url_args, query_params=query_params_next ) heading = "Releases - " if origin_info: heading += "origin: %s" % origin_info["url"] else: heading += "snapshot: %s" % snapshot_id return render( request, "browse/releases.html", { "heading": heading, "top_panel_visible": False, "top_panel_collapsible": False, "swh_object_name": "Releases", "swh_object_metadata": {}, "top_right_link": None, "displayed_releases": displayed_releases, "prev_releases_url": prev_releases_url, "next_releases_url": next_releases_url, "snapshot_context": snapshot_context, "vault_cooking": None, "show_actions": False, "search_string": release_name_include or "", }, ) diff --git a/swh/web/browse/views/directory.py b/swh/web/browse/views/directory.py index 22428510..7b1f8aae 100644 --- a/swh/web/browse/views/directory.py +++ b/swh/web/browse/views/directory.py @@ -1,281 +1,285 @@ # Copyright (C) 2017-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU Affero General Public License version 3, or any later version # See top-level LICENSE file for more information import os import sentry_sdk from django.http import HttpResponse from django.shortcuts import redirect, render from swh.model.swhids import ObjectType from swh.web.browse.browseurls import browse_route from swh.web.browse.snapshot_context import get_snapshot_context from swh.web.browse.utils import gen_link, get_directory_entries, get_readme_to_display from swh.web.common import archive from swh.web.common.exc import NotFoundExc, http_status_code_message from swh.web.common.identifiers import get_swhids_info from swh.web.common.typing import DirectoryMetadata, SWHObjectInfo from swh.web.common.utils import gen_path_info, reverse, swh_object_icons def _directory_browse(request, sha1_git, path=None): root_sha1_git = sha1_git error_info = {"status_code": 200, "description": None} if path: try: dir_info = archive.lookup_directory_with_path(sha1_git, path) sha1_git = dir_info["target"] except NotFoundExc as e: error_info["status_code"] = 404 error_info["description"] = f"NotFoundExc: {str(e)}" sha1_git = None dirs, files = [], [] if sha1_git is not None: dirs, files = get_directory_entries(sha1_git) origin_url = request.GET.get("origin_url") if not origin_url: origin_url = request.GET.get("origin") snapshot_id = request.GET.get("snapshot") snapshot_context = None if origin_url is not None or snapshot_id is not None: try: snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), revision_id=request.GET.get("revision"), path=path, ) except NotFoundExc as e: if str(e).startswith("Origin"): raw_dir_url = reverse( "browse-directory", url_args={"sha1_git": sha1_git} ) error_message = ( "The Software Heritage archive has a directory " "with the hash you provided but the origin " "mentioned in your request appears broken: %s. " "Please check the URL and try again.\n\n" "Nevertheless, you can still browse the directory " "without origin information: %s" % (gen_link(origin_url), gen_link(raw_dir_url)) ) raise NotFoundExc(error_message) else: raise e path_info = gen_path_info(path) query_params = snapshot_context["query_params"] if snapshot_context else {} breadcrumbs = [] breadcrumbs.append( { "name": root_sha1_git[:7], "url": reverse( "browse-directory", url_args={"sha1_git": root_sha1_git}, query_params={**query_params, "path": None}, ), } ) for pi in path_info: breadcrumbs.append( { "name": pi["name"], "url": reverse( "browse-directory", url_args={"sha1_git": root_sha1_git}, query_params={**query_params, "path": pi["path"],}, ), } ) path = "" if path is None else (path + "/") for d in dirs: if d["type"] == "rev": d["url"] = reverse( "browse-revision", url_args={"sha1_git": d["target"]}, query_params=query_params, ) else: d["url"] = reverse( "browse-directory", url_args={"sha1_git": root_sha1_git}, query_params={**query_params, "path": path + d["name"],}, ) sum_file_sizes = 0 readmes = {} for f in files: query_string = "sha1_git:" + f["target"] f["url"] = reverse( "browse-content", url_args={"query_string": query_string}, query_params={ **query_params, "path": root_sha1_git + "/" + path + f["name"], }, ) if f["length"] is not None: sum_file_sizes += f["length"] if f["name"].lower().startswith("readme"): readmes[f["name"]] = f["checksums"]["sha1"] readme_name, readme_url, readme_html = get_readme_to_display(readmes) dir_metadata = DirectoryMetadata( object_type=ObjectType.DIRECTORY, object_id=sha1_git, directory=root_sha1_git, nb_files=len(files), nb_dirs=len(dirs), sum_file_sizes=sum_file_sizes, root_directory=root_sha1_git, path=f"/{path}" if path else None, revision=None, revision_found=None, release=None, snapshot=None, ) vault_cooking = { "directory_context": True, "directory_swhid": f"swh:1:dir:{sha1_git}", "revision_context": False, "revision_swhid": None, } swh_objects = [SWHObjectInfo(object_type=ObjectType.DIRECTORY, object_id=sha1_git)] if snapshot_context: swh_objects.append( SWHObjectInfo( object_type=ObjectType.REVISION, object_id=snapshot_context["revision_id"], ) ) swh_objects.append( SWHObjectInfo( object_type=ObjectType.SNAPSHOT, object_id=snapshot_context["snapshot_id"], ) ) if snapshot_context["release_id"]: swh_objects.append( SWHObjectInfo( object_type=ObjectType.RELEASE, object_id=snapshot_context["release_id"], ) ) swhids_info = get_swhids_info(swh_objects, snapshot_context, dir_metadata) heading = "Directory - %s" % sha1_git if breadcrumbs: dir_path = "/".join([bc["name"] for bc in breadcrumbs]) + "/" heading += " - %s" % dir_path top_right_link = None - if snapshot_context is not None and not snapshot_context["is_empty"]: + if ( + snapshot_context is not None + and not snapshot_context["is_empty"] + and snapshot_context["revision_id"] is not None + ): history_url = reverse( "browse-revision-log", url_args={"sha1_git": snapshot_context["revision_id"]}, query_params=query_params, ) top_right_link = { "url": history_url, "icon": swh_object_icons["revisions history"], "text": "History", } return render( request, "browse/directory.html", { "heading": heading, "swh_object_id": swhids_info[0]["swhid"], "swh_object_name": "Directory", "swh_object_metadata": dir_metadata, "dirs": dirs, "files": files, "breadcrumbs": breadcrumbs, "top_right_link": top_right_link, "readme_name": readme_name, "readme_url": readme_url, "readme_html": readme_html, "snapshot_context": snapshot_context, "vault_cooking": vault_cooking, "show_actions": True, "swhids_info": swhids_info, "error_code": error_info["status_code"], "error_message": http_status_code_message.get(error_info["status_code"]), "error_description": error_info["description"], }, status=error_info["status_code"], ) @browse_route( r"directory/(?P[0-9a-f]+)/", view_name="browse-directory", checksum_args=["sha1_git"], ) def directory_browse(request, sha1_git): """Django view for browsing the content of a directory identified by its sha1_git value. The url that points to it is :http:get:`/browse/directory/(sha1_git)/` """ return _directory_browse(request, sha1_git, request.GET.get("path")) @browse_route( r"directory/(?P[0-9a-f]+)/(?P.+)/", view_name="browse-directory-legacy", checksum_args=["sha1_git"], ) def directory_browse_legacy(request, sha1_git, path): """Django view for browsing the content of a directory identified by its sha1_git value. The url that points to it is :http:get:`/browse/directory/(sha1_git)/(path)/` """ return _directory_browse(request, sha1_git, path) @browse_route( r"directory/resolve/content-path/(?P[0-9a-f]+)/", view_name="browse-directory-resolve-content-path", checksum_args=["sha1_git"], ) def _directory_resolve_content_path(request, sha1_git): """ Internal endpoint redirecting to data url for a specific file path relative to a root directory. """ try: path = os.path.normpath(request.GET.get("path")) if not path.startswith("../"): dir_info = archive.lookup_directory_with_path(sha1_git, path) if dir_info["type"] == "file": sha1 = dir_info["checksums"]["sha1"] data_url = reverse( "browse-content-raw", url_args={"query_string": sha1} ) return redirect(data_url) except Exception as exc: sentry_sdk.capture_exception(exc) return HttpResponse(status=404) diff --git a/swh/web/templates/includes/top-navigation.html b/swh/web/templates/includes/top-navigation.html index b9bd2241..0067abc5 100644 --- a/swh/web/templates/includes/top-navigation.html +++ b/swh/web/templates/includes/top-navigation.html @@ -1,153 +1,153 @@ {% comment %} Copyright (C) 2017-2020 The Software Heritage developers See the AUTHORS file at the top-level directory of this distribution License: GNU Affero General Public License version 3, or any later version See top-level LICENSE file for more information {% endcomment %} {% load swh_templatetags %}
{% if snapshot_context %} {% if snapshot_context.branch or snapshot_context.release or snapshot_context.revision_id %} {% endif %} {% endif %}
{% include "includes/breadcrumbs.html" %}
{% if top_right_link %} {% if top_right_link.icon %} {% endif %} {{ top_right_link.text }} {% endif %} {% if available_languages %} {% endif %} {% if show_actions %} {% if not snapshot_context or not snapshot_context.is_empty %} {% include "includes/vault-create-tasks.html" %} {% endif %} {% include "includes/take-new-snapshot.html" %} {% include "includes/show-metadata.html" %} {% endif %}
{% include "includes/show-swhids.html" %} diff --git a/swh/web/tests/browse/test_snapshot_context.py b/swh/web/tests/browse/test_snapshot_context.py index a84bcbce..56c14a0d 100644 --- a/swh/web/tests/browse/test_snapshot_context.py +++ b/swh/web/tests/browse/test_snapshot_context.py @@ -1,406 +1,507 @@ # Copyright (C) 2020-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU Affero General Public License version 3, or any later version # See top-level LICENSE file for more information import random +from swh.model.hashutil import hash_to_bytes +from swh.model.model import ObjectType as ModelObjectType +from swh.model.model import Release, Snapshot, SnapshotBranch, TargetType from swh.model.swhids import ObjectType from swh.web.browse.snapshot_context import ( _get_release, get_origin_visit_snapshot, get_snapshot_content, get_snapshot_context, ) from swh.web.browse.utils import gen_revision_url from swh.web.common.identifiers import gen_swhid from swh.web.common.origin_visits import get_origin_visit, get_origin_visits from swh.web.common.typing import ( SnapshotBranchInfo, SnapshotContext, SnapshotReleaseInfo, ) from swh.web.common.utils import format_utc_iso_date, reverse def test_get_origin_visit_snapshot_simple(archive_data, origin_with_multiple_visits): visits = archive_data.origin_visit_get(origin_with_multiple_visits["url"]) for visit in visits: snapshot = archive_data.snapshot_get(visit["snapshot"]) branches = [] releases = [] def _process_branch_data(branch, branch_data, alias=False): if branch_data["target_type"] == "revision": rev_data = archive_data.revision_get(branch_data["target"]) branches.append( SnapshotBranchInfo( name=branch, alias=alias, revision=branch_data["target"], directory=rev_data["directory"], date=format_utc_iso_date(rev_data["date"]), message=rev_data["message"], url=None, ) ) elif branch_data["target_type"] == "release": rel_data = archive_data.release_get(branch_data["target"]) rev_data = archive_data.revision_get(rel_data["target"]) releases.append( SnapshotReleaseInfo( name=rel_data["name"], alias=alias, branch_name=branch, date=format_utc_iso_date(rel_data["date"]), id=rel_data["id"], message=rel_data["message"], target_type=rel_data["target_type"], target=rel_data["target"], directory=rev_data["directory"], url=None, ) ) aliases = {} for branch in sorted(snapshot["branches"].keys()): branch_data = snapshot["branches"][branch] if branch_data["target_type"] == "alias": target_data = snapshot["branches"][branch_data["target"]] aliases[branch] = target_data _process_branch_data(branch, target_data, alias=True) else: _process_branch_data(branch, branch_data) assert branches and releases, "Incomplete test data." origin_visit_branches = get_origin_visit_snapshot( origin_with_multiple_visits, visit_id=visit["visit"] ) assert origin_visit_branches == (branches, releases, aliases) def test_get_snapshot_context_no_origin(archive_data, snapshot): for browse_context, kwargs in ( ("content", {"snapshot_id": snapshot, "path": "/some/path"}), ("directory", {"snapshot_id": snapshot}), ("log", {"snapshot_id": snapshot}), ): url_args = {"snapshot_id": snapshot} query_params = dict(kwargs) query_params.pop("snapshot_id") snapshot_context = get_snapshot_context(**kwargs, browse_context=browse_context) branches, releases, _ = get_snapshot_content(snapshot) releases = list(reversed(releases)) revision_id = None root_directory = None for branch in branches: if branch["name"] == "HEAD": revision_id = branch["revision"] root_directory = branch["directory"] branch["url"] = reverse( f"browse-snapshot-{browse_context}", url_args=url_args, query_params={"branch": branch["name"], **query_params}, ) for release in releases: release["url"] = reverse( f"browse-snapshot-{browse_context}", url_args=url_args, query_params={"release": release["name"], **query_params}, ) branches_url = reverse("browse-snapshot-branches", url_args=url_args) releases_url = reverse("browse-snapshot-releases", url_args=url_args) directory_url = reverse("browse-snapshot-directory", url_args=url_args) is_empty = not branches and not releases snapshot_swhid = gen_swhid(ObjectType.SNAPSHOT, snapshot) snapshot_sizes = archive_data.snapshot_count_branches(snapshot) expected = SnapshotContext( branch="HEAD", branch_alias=True, branches=branches, branches_url=branches_url, is_empty=is_empty, origin_info=None, origin_visits_url=None, release=None, release_alias=False, release_id=None, query_params=query_params, releases=releases, releases_url=releases_url, revision_id=revision_id, revision_info=_get_revision_info(archive_data, revision_id), root_directory=root_directory, snapshot_id=snapshot, snapshot_sizes=snapshot_sizes, snapshot_swhid=snapshot_swhid, url_args=url_args, visit_info=None, directory_url=directory_url, ) if revision_id: expected["revision_info"]["revision_url"] = gen_revision_url( revision_id, snapshot_context ) assert snapshot_context == expected _check_branch_release_revision_parameters( archive_data, expected, browse_context, kwargs, branches, releases ) def test_get_snapshot_context_with_origin(archive_data, origin_with_multiple_visits): origin_visits = get_origin_visits(origin_with_multiple_visits) timestamp = format_utc_iso_date(origin_visits[0]["date"], "%Y-%m-%dT%H:%M:%SZ") visit_id = origin_visits[1]["visit"] origin_url = origin_with_multiple_visits["url"] for browse_context, kwargs in ( ("content", {"origin_url": origin_url, "path": "/some/path"}), ("directory", {"origin_url": origin_url}), ("log", {"origin_url": origin_url}), ("directory", {"origin_url": origin_url, "timestamp": timestamp,},), ("directory", {"origin_url": origin_url, "visit_id": visit_id,},), ): visit_id = kwargs["visit_id"] if "visit_id" in kwargs else None visit_ts = kwargs["timestamp"] if "timestamp" in kwargs else None visit_info = get_origin_visit( {"url": kwargs["origin_url"]}, visit_ts=visit_ts, visit_id=visit_id ) snapshot = visit_info["snapshot"] snapshot_context = get_snapshot_context(**kwargs, browse_context=browse_context) query_params = dict(kwargs) branches, releases, _ = get_snapshot_content(snapshot) releases = list(reversed(releases)) revision_id = None root_directory = None for branch in branches: if branch["name"] == "HEAD": revision_id = branch["revision"] root_directory = branch["directory"] branch["url"] = reverse( f"browse-origin-{browse_context}", query_params={"branch": branch["name"], **query_params}, ) for release in releases: release["url"] = reverse( f"browse-origin-{browse_context}", query_params={"release": release["name"], **query_params}, ) query_params.pop("path", None) branches_url = reverse("browse-origin-branches", query_params=query_params) releases_url = reverse("browse-origin-releases", query_params=query_params) origin_visits_url = reverse( "browse-origin-visits", query_params={"origin_url": kwargs["origin_url"]} ) is_empty = not branches and not releases snapshot_swhid = gen_swhid(ObjectType.SNAPSHOT, snapshot) snapshot_sizes = archive_data.snapshot_count_branches(snapshot) visit_info["url"] = directory_url = reverse( "browse-origin-directory", query_params=query_params ) visit_info["formatted_date"] = format_utc_iso_date(visit_info["date"]) if "path" in kwargs: query_params["path"] = kwargs["path"] expected = SnapshotContext( branch="HEAD", branch_alias=True, branches=branches, branches_url=branches_url, is_empty=is_empty, origin_info={"url": origin_url}, origin_visits_url=origin_visits_url, release=None, release_alias=False, release_id=None, query_params=query_params, releases=releases, releases_url=releases_url, revision_id=revision_id, revision_info=_get_revision_info(archive_data, revision_id), root_directory=root_directory, snapshot_id=snapshot, snapshot_sizes=snapshot_sizes, snapshot_swhid=snapshot_swhid, url_args={}, visit_info=visit_info, directory_url=directory_url, ) if revision_id: expected["revision_info"]["revision_url"] = gen_revision_url( revision_id, snapshot_context ) assert snapshot_context == expected _check_branch_release_revision_parameters( archive_data, expected, browse_context, kwargs, branches, releases ) def _check_branch_release_revision_parameters( archive_data, base_expected_context, browse_context, kwargs, branches, releases, ): branch = random.choice(branches) snapshot_context = get_snapshot_context( **kwargs, browse_context=browse_context, branch_name=branch["name"] ) url_args = dict(kwargs) url_args.pop("path", None) url_args.pop("timestamp", None) url_args.pop("visit_id", None) url_args.pop("origin_url", None) query_params = dict(kwargs) query_params.pop("snapshot_id", None) expected_branch = dict(base_expected_context) expected_branch["branch"] = branch["name"] expected_branch["branch_alias"] = branch["alias"] expected_branch["revision_id"] = branch["revision"] expected_branch["revision_info"] = _get_revision_info( archive_data, branch["revision"] ) expected_branch["root_directory"] = branch["directory"] expected_branch["query_params"] = {"branch": branch["name"], **query_params} expected_branch["revision_info"]["revision_url"] = gen_revision_url( branch["revision"], expected_branch ) assert snapshot_context == expected_branch if releases: release = random.choice(releases) snapshot_context = get_snapshot_context( **kwargs, browse_context=browse_context, release_name=release["name"] ) expected_release = dict(base_expected_context) expected_release["branch"] = None expected_release["branch_alias"] = False expected_release["release"] = release["name"] expected_release["release_id"] = release["id"] if release["target_type"] == "revision": expected_release["revision_id"] = release["target"] expected_release["revision_info"] = _get_revision_info( archive_data, release["target"] ) expected_release["root_directory"] = release["directory"] expected_release["query_params"] = {"release": release["name"], **query_params} expected_release["revision_info"]["revision_url"] = gen_revision_url( release["target"], expected_release ) assert snapshot_context == expected_release revision_log = archive_data.revision_log(branch["revision"]) revision = revision_log[-1] snapshot_context = get_snapshot_context( **kwargs, browse_context=browse_context, revision_id=revision["id"] ) if "origin_url" in kwargs: view_name = f"browse-origin-{browse_context}" else: view_name = f"browse-snapshot-{browse_context}" kwargs.pop("visit_id", None) revision_browse_url = reverse( view_name, url_args=url_args, query_params={"revision": revision["id"], **query_params}, ) branches.append( SnapshotBranchInfo( name=revision["id"], alias=False, revision=revision["id"], directory=revision["directory"], date=revision["date"], message=revision["message"], url=revision_browse_url, ) ) expected_revision = dict(base_expected_context) expected_revision["branch"] = None expected_revision["branch_alias"] = False expected_revision["branches"] = branches expected_revision["revision_id"] = revision["id"] expected_revision["revision_info"] = _get_revision_info( archive_data, revision["id"] ) expected_revision["root_directory"] = revision["directory"] expected_revision["query_params"] = {"revision": revision["id"], **query_params} expected_revision["revision_info"]["revision_url"] = gen_revision_url( revision["id"], expected_revision ) assert snapshot_context == expected_revision def test_get_release_large_snapshot(archive_data, origin_with_releases): snapshot = archive_data.snapshot_get_latest(origin_with_releases["url"]) release_id = random.choice( [ v["target"] for v in snapshot["branches"].values() if v["target_type"] == "release" ] ) release_data = archive_data.release_get(release_id) # simulate large snapshot processing by providing releases parameter # as an empty list release = _get_release( releases=[], release_name=release_data["name"], snapshot_id=snapshot["id"] ) assert release_data["name"] == release["name"] assert release_data["id"] == release["id"] def _get_revision_info(archive_data, revision_id): revision_info = None if revision_id: revision_info = archive_data.revision_get(revision_id) revision_info["message_header"] = revision_info["message"].split("\n")[0] revision_info["date"] = format_utc_iso_date(revision_info["date"]) revision_info["committer_date"] = format_utc_iso_date( revision_info["committer_date"] ) return revision_info + + +def test_get_snapshot_context_revision_release(archive_data, revision): + release_name = "v1.0.0" + release = Release( + name=release_name.encode(), + message=f"release {release_name}".encode(), + target=hash_to_bytes(revision), + target_type=ModelObjectType.REVISION, + synthetic=True, + ) + archive_data.release_add([release]) + + snapshot = Snapshot( + branches={ + b"HEAD": SnapshotBranch( + target=release_name.encode(), target_type=TargetType.ALIAS + ), + release_name.encode(): SnapshotBranch( + target=release.id, target_type=TargetType.RELEASE + ), + }, + ) + archive_data.snapshot_add([snapshot]) + + snapshot_no_head = Snapshot( + branches={ + release_name.encode(): SnapshotBranch( + target=release.id, target_type=TargetType.RELEASE + ), + }, + ) + archive_data.snapshot_add([snapshot_no_head]) + + revision_data = archive_data.revision_get(revision) + + for params in ( + {"snapshot_id": snapshot.id.hex()}, + {"snapshot_id": snapshot.id.hex(), "release_name": release_name}, + {"snapshot_id": snapshot_no_head.id.hex()}, + ): + + snapshot_context = get_snapshot_context(**params) + + assert snapshot_context["branches"] == [] + assert snapshot_context["releases"] != [] + assert snapshot_context["release"] == release_name + assert snapshot_context["release_id"] == release.id.hex() + assert snapshot_context["revision_id"] == revision + assert snapshot_context["root_directory"] == revision_data["directory"] + + +def test_get_snapshot_context_directory_release(archive_data, directory): + release_name = "v1.0.0" + release = Release( + name=release_name.encode(), + message=f"release {release_name}".encode(), + target=hash_to_bytes(directory), + target_type=ModelObjectType.DIRECTORY, + synthetic=True, + ) + archive_data.release_add([release]) + + snapshot = Snapshot( + branches={ + b"HEAD": SnapshotBranch( + target=release_name.encode(), target_type=TargetType.ALIAS + ), + release_name.encode(): SnapshotBranch( + target=release.id, target_type=TargetType.RELEASE + ), + }, + ) + archive_data.snapshot_add([snapshot]) + + snapshot_no_head = Snapshot( + branches={ + release_name.encode(): SnapshotBranch( + target=release.id, target_type=TargetType.RELEASE + ), + }, + ) + archive_data.snapshot_add([snapshot_no_head]) + + for params in ( + {"snapshot_id": snapshot.id.hex()}, + {"snapshot_id": snapshot.id.hex(), "release_name": release_name}, + {"snapshot_id": snapshot_no_head.id.hex()}, + ): + + snapshot_context = get_snapshot_context(**params) + + assert snapshot_context["branches"] == [] + assert snapshot_context["releases"] != [] + assert snapshot_context["release"] == release_name + assert snapshot_context["release_id"] == release.id.hex() + assert snapshot_context["revision_id"] is None + assert snapshot_context["root_directory"] == directory