diff --git a/swh/web/browse/snapshot_context.py b/swh/web/browse/snapshot_context.py
index 56c8ddc7..33c5c02a 100644
--- a/swh/web/browse/snapshot_context.py
+++ b/swh/web/browse/snapshot_context.py
@@ -1,1379 +1,1429 @@
# Copyright (C) 2018-2022 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.http import HttpRequest, HttpResponse
-from django.shortcuts import render
+from django.shortcuts import redirect, 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 (
format_log_entries,
gen_release_link,
gen_revision_link,
gen_revision_log_link,
gen_revision_url,
gen_snapshot_link,
get_directory_entries,
get_readme_to_display,
)
from swh.web.config import get_config
from swh.web.utils import (
archive,
django_cache,
format_utc_iso_date,
gen_path_info,
reverse,
swh_object_icons,
)
from swh.web.utils.exc import BadInputExc, NotFoundExc, http_status_code_message
from swh.web.utils.identifiers import get_swhids_info
from swh.web.utils.origin_visits import get_origin_visit
from swh.web.utils.typing import (
DirectoryMetadata,
OriginInfo,
SnapshotBranchInfo,
SnapshotContext,
SnapshotReleaseInfo,
SWHObjectInfo,
)
_empty_snapshot_id = Snapshot(branches={}).id.hex()
def _get_branch(
branches: List[SnapshotBranchInfo], branch_name: str, snapshot_id: str
) -> Optional[SnapshotBranchInfo]:
"""
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]
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"],
+ target_types=["revision", "alias", "content", "directory"],
# 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]
return None
def _get_release(
releases: List[SnapshotReleaseInfo], release_name: Optional[str], snapshot_id: str
) -> Optional[SnapshotReleaseInfo]:
"""
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]
elif release_name:
# 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", "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]
return None
def _branch_not_found(
branch_type: str,
branch: str,
snapshot_id: str,
snapshot_sizes: Dict[str, int],
origin_info: Optional[OriginInfo],
timestamp: Optional[str],
visit_id: Optional[int],
) -> None:
"""
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"
+ target_type = "branch"
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 and origin_info:
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 and origin_info:
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 and origin_info and timestamp:
msg = (
"Origin with url %s"
" for visit with timestamp %s has an empty list"
" of %s!" % (origin_info["url"], timestamp, branch_type_plural)
)
elif origin_info and timestamp:
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.utils.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":
+ if target_type in ("content", "directory", "revision"):
branches[branch_name] = SnapshotBranchInfo(
name=branch_name,
alias=False,
- revision=target_id,
+ target_type=target_type,
+ target=target_id,
date=None,
directory=None,
message=None,
url=None,
)
- revision_to_branch[target_id].add(branch_name)
+ if target_type == "revision":
+ 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"],
+ target_type="revision",
+ target=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():
+ for branch_alias, _ 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)
+ elif target_type in ("content", "directory"):
+ branches[branch_name] = SnapshotBranchInfo(
+ name=branch_alias,
+ alias=True,
+ target_type=target_type,
+ target=target,
+ date=None,
+ directory=None,
+ message=None,
+ url=None,
+ )
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
@django_cache()
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
"""
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)
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.utils.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
+ visit_info["url"] = browse_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)
+ browse_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))
@django_cache()
def _get_snapshot_sizes(snapshot_id):
return archive.lookup_snapshot_sizes(snapshot_id)
snapshot_sizes = _get_snapshot_sizes(snapshot_id)
- is_empty = (snapshot_sizes["release"] + snapshot_sizes["revision"]) == 0
+ snapshot_total_size = sum(
+ v for k, v in snapshot_sizes.items() if k not in ("alias", "branch")
+ )
+
+ is_empty = snapshot_total_size == 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,
+ target_type="revision",
+ target=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:
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, 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"]
+ if branch["target_type"] == "revision":
+ revision_id = branch["target"]
+ root_directory = branch["directory"]
+ elif branch["target_type"] == "directory":
+ root_directory = branch["target"]
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"])
+ if head["target_type"] in ("content", "directory", "revision"):
branch_name = "HEAD"
- revision_id = head_rev["id"]
- root_directory = head_rev["directory"]
- else:
+ if head["target_type"] == "revision":
+ # HEAD alias targets a revision
+ head_rev = archive.lookup_revision(head["target"])
+ revision_id = head_rev["id"]
+ root_directory = head_rev["directory"]
+ elif head["target_type"] == "directory":
+ # HEAD alias targets a directory
+ root_directory = head["target"]
+ elif head["target_type"] == "release":
# 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 is None:
_branch_not_found(
"release",
str(release_name),
snapshot_id,
snapshot_sizes,
origin_info,
timestamp,
visit_id,
)
elif 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"]
if head_rel is not None:
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"]
+ revision_id = (
+ branch["target"] if branch["target_type"] == "revision" else None
+ )
+ if branch["target_type"] == "revision":
+ root_directory = branch["directory"]
+ elif branch["target_type"] == "directory":
+ root_directory = branch["target"]
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"]:
+ if b["name"] != b["target"]:
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,
+ browse_url=browse_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_info["id"], snapshot_context
)
return snapshot_context
def _build_breadcrumbs(
snapshot_context: SnapshotContext, path: Optional[str]
) -> List[Dict[str, 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: Optional[str], origin_url: Optional[str]) -> None:
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: HttpRequest,
snapshot_id: Optional[str] = None,
origin_url: Optional[str] = None,
timestamp: Optional[str] = None,
path: Optional[str] = None,
) -> HttpResponse:
"""
Django view implementation for browsing a directory in a snapshot context.
"""
_check_origin_url(snapshot_id, origin_url)
visit_id = int(request.GET.get("visit_id", 0))
snapshot_context = get_snapshot_context(
snapshot_id=snapshot_id,
origin_url=origin_url,
timestamp=timestamp,
visit_id=visit_id or None,
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"]
+
+ if root_directory is None and snapshot_context["branch"] is not None:
+ branch_info = [
+ branch
+ for branch in snapshot_context["branches"]
+ if branch["name"] == snapshot_context["branch"]
+ ]
+ # special case where the branch to browse targets a content instead of a directory
+ if branch_info and branch_info[0]["target_type"] == "content":
+ # redirect to browse content view
+ if "origin_url" not in snapshot_context["query_params"]:
+ snapshot_id = snapshot_context["snapshot_id"]
+ snapshot_context["query_params"]["snapshot"] = snapshot_id
+ return redirect(
+ reverse(
+ "browse-content",
+ url_args={"query_string": f"sha1_git:{branch_info[0]['target']}"},
+ query_params=snapshot_context["query_params"],
+ )
+ )
+
sha1_git = root_directory
error_info: Dict[str, Any] = {
"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: Dict[str, Any] = {
"directory_context": False,
"directory_swhid": None,
"revision_context": False,
"revision_swhid": None,
}
revision_found = False
if revision_id is not None:
try:
archive.lookup_revision(revision_id)
except NotFoundExc:
pass
else:
revision_found = True
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"],
)
PER_PAGE = 100
def browse_snapshot_log(
request: HttpRequest,
snapshot_id: Optional[str] = None,
origin_url: Optional[str] = None,
timestamp: Optional[str] = None,
) -> HttpResponse:
"""
Django view implementation for browsing a revision history in a
snapshot context.
"""
_check_origin_url(snapshot_id, origin_url)
visit_id = int(request.GET.get("visit_id", 0))
snapshot_context = get_snapshot_context(
snapshot_id=snapshot_id,
origin_url=origin_url,
timestamp=timestamp,
visit_id=visit_id or None,
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"]
if revision_id is None:
raise NotFoundExc("No revisions history found in the current snapshot context.")
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"] = str(per_page)
revs_ordering = request.GET.get("revs_ordering", "")
if revs_ordering:
query_params["revs_ordering"] = revs_ordering
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"] = str(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"] = str(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 and visit_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: HttpRequest,
snapshot_id: Optional[str] = None,
origin_url: Optional[str] = None,
timestamp: Optional[str] = None,
branch_name_include: Optional[str] = None,
) -> HttpResponse:
"""
Django view implementation for browsing a list of branches in a snapshot
context.
"""
_check_origin_url(snapshot_id, origin_url)
visit_id = int(request.GET.get("visit_id", 0))
snapshot_context = get_snapshot_context(
snapshot_id=snapshot_id,
origin_url=origin_url,
timestamp=timestamp,
visit_id=visit_id or None,
)
branches_bc_str = request.GET.get("branches_breadcrumbs", "")
branches_bc = branches_bc_str.split(",") if branches_bc_str 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: List[Dict[str, Any]] = []
if snapshot:
branches, _, _ = process_snapshot_branches(snapshot)
displayed_branches = [dict(branch) for branch in branches]
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"]},
+ url_args={"sha1_git": branch["target"]},
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: HttpRequest,
snapshot_id: Optional[str] = None,
origin_url: Optional[str] = None,
timestamp: Optional[str] = None,
release_name_include: Optional[str] = None,
):
"""
Django view implementation for browsing a list of releases in a snapshot
context.
"""
_check_origin_url(snapshot_id, origin_url)
visit_id = int(request.GET.get("visit_id", 0))
snapshot_context = get_snapshot_context(
snapshot_id=snapshot_id,
origin_url=origin_url,
timestamp=timestamp,
visit_id=visit_id or None,
)
rel_bc_str = request.GET.get("releases_breadcrumbs", "")
rel_bc = rel_bc_str.split(",") if rel_bc_str 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: List[Dict[str, Any]] = []
if snapshot:
_, releases, _ = process_snapshot_branches(snapshot)
displayed_releases = [dict(release) for release in releases]
for release in displayed_releases:
query_params_tgt = {"snapshot": snapshot_id, "release": release["name"]}
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/templates/includes/snapshot-context.html b/swh/web/browse/templates/includes/snapshot-context.html
index 0a33640d..6aa83df5 100644
--- a/swh/web/browse/templates/includes/snapshot-context.html
+++ b/swh/web/browse/templates/includes/snapshot-context.html
@@ -1,87 +1,87 @@
{% 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.visit_info %}
{% endif %}
-
+
Code
- {% if not snapshot_context.snapshot_sizes.revision %}
+ {% if not snapshot_context.snapshot_sizes.branch %}
Branches (0)
{% else %}
- Branches ({{ snapshot_context.snapshot_sizes.revision}})
+ Branches ({{ snapshot_context.snapshot_sizes.branch}})
{% endif %}
{% if not snapshot_context.snapshot_sizes.release %}
Releases (0)
{% else %}
Releases ({{ snapshot_context.snapshot_sizes.release }})
{% endif %}
{% if snapshot_context.origin_info %}
Visits
{% endif %}
diff --git a/swh/web/browse/templates/includes/top-navigation.html b/swh/web/browse/templates/includes/top-navigation.html
index b5637993..370cb176 100644
--- a/swh/web/browse/templates/includes/top-navigation.html
+++ b/swh/web/browse/templates/includes/top-navigation.html
@@ -1,157 +1,157 @@
{% comment %}
Copyright (C) 2017-2022 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 %}
{% if snapshot_context.branch %}
{% if snapshot_context.branch_alias %}
{% else %}
{% endif %}
Branch: {{ snapshot_context.branch }}
{% elif snapshot_context.release %}
{% if snapshot_context.release_alias %}
{% else %}
{% endif %}
Release: {{ snapshot_context.release }}
{% elif snapshot_context.revision_id %}
Revision: {{ snapshot_context.revision_id }}
{% endif %}
{% endif %}
{% endif %}
{% include "./breadcrumbs.html" %}
{% if top_right_link %}
{% if top_right_link.icon %}
{% endif %}
{{ top_right_link.text }}
{% endif %}
{% if available_languages %}
{% for lang in available_languages %}
{{ lang }}
{% endfor %}
{% endif %}
{% if show_actions %}
{% if "swh.web.vault" in SWH_DJANGO_APPS %}
{% if not snapshot_context or not snapshot_context.is_empty %}
{% include "includes/vault-create-tasks.html" %}
{% endif %}
{% endif %}
{% if "swh.web.save_code_now" in SWH_DJANGO_APPS %}
{% include "includes/take-new-snapshot.html" %}
{% endif %}
{% include "./show-metadata.html" %}
{% endif %}
{% include "./show-swhids.html" %}
diff --git a/swh/web/tests/browse/test_snapshot_context.py b/swh/web/tests/browse/test_snapshot_context.py
index 1a17bc4c..0ddaeae7 100644
--- a/swh/web/tests/browse/test_snapshot_context.py
+++ b/swh/web/tests/browse/test_snapshot_context.py
@@ -1,524 +1,526 @@
# 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.utils import format_utc_iso_date, reverse
from swh.web.utils.identifiers import gen_swhid
from swh.web.utils.origin_visits import get_origin_visit, get_origin_visits
from swh.web.utils.typing import (
SnapshotBranchInfo,
SnapshotContext,
SnapshotReleaseInfo,
)
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"],
+ target_type="revision",
+ target=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"]
+ revision_id = branch["target"]
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,
+ browse_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"]
+ revision_id = branch["target"]
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,
+ browse_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_id"] = branch["target"]
expected_branch["revision_info"] = _get_revision_info(
- archive_data, branch["revision"]
+ archive_data, branch["target"]
)
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
+ branch["target"], 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_log = archive_data.revision_log(branch["target"])
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"],
+ target_type="revision",
+ target=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
diff --git a/swh/web/tests/browse/views/test_content.py b/swh/web/tests/browse/views/test_content.py
index 2acd5cfa..65e05cbb 100644
--- a/swh/web/tests/browse/views/test_content.py
+++ b/swh/web/tests/browse/views/test_content.py
@@ -1,1115 +1,1115 @@
# Copyright (C) 2017-2022 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
import re
import pytest
from django.utils.html import escape
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 process_snapshot_branches
from swh.web.browse.utils import (
get_mimetype_and_encoding_for_content,
prepare_content_for_display,
re_encode_content,
)
from swh.web.tests.data import get_content
from swh.web.tests.django_asserts import assert_contains, assert_not_contains
from swh.web.tests.helpers import check_html_get_response, check_http_get_response
from swh.web.utils import (
format_utc_iso_date,
gen_path_info,
parse_iso8601_date_to_utc,
reverse,
)
from swh.web.utils.exc import NotFoundExc
from swh.web.utils.identifiers import gen_swhid
def test_content_view_text(client, archive_data, content_text):
sha1_git = content_text["sha1_git"]
url = reverse(
"browse-content",
url_args={"query_string": content_text["sha1"]},
query_params={"path": content_text["path"]},
)
url_raw = reverse(
"browse-content-raw", url_args={"query_string": content_text["sha1"]}
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-content.html"
)
content_display = _process_content_for_display(archive_data, content_text)
mimetype = content_display["mimetype"]
if mimetype.startswith("text/"):
assert_contains(resp, '' % content_display["language"])
assert_contains(resp, escape(content_display["content_data"]))
assert_contains(resp, url_raw)
swh_cnt_id = gen_swhid(ObjectType.CONTENT, sha1_git)
swh_cnt_id_url = reverse("browse-swhid", url_args={"swhid": swh_cnt_id})
assert_contains(resp, swh_cnt_id)
assert_contains(resp, swh_cnt_id_url)
assert_not_contains(resp, "swh-metadata-popover")
def test_content_view_no_highlight(
client, archive_data, content_application_no_highlight, content_text_no_highlight
):
for content_ in (content_application_no_highlight, content_text_no_highlight):
content = content_
sha1_git = content["sha1_git"]
url = reverse("browse-content", url_args={"query_string": content["sha1"]})
url_raw = reverse(
"browse-content-raw", url_args={"query_string": content["sha1"]}
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-content.html"
)
content_display = _process_content_for_display(archive_data, content)
if content["encoding"] != "binary":
assert_contains(resp, '')
assert_contains(resp, escape(content_display["content_data"]))
assert_contains(resp, url_raw)
swh_cnt_id = gen_swhid(ObjectType.CONTENT, sha1_git)
swh_cnt_id_url = reverse("browse-swhid", url_args={"swhid": swh_cnt_id})
assert_contains(resp, swh_cnt_id)
assert_contains(resp, swh_cnt_id_url)
def test_content_view_no_utf8_text(client, archive_data, content_text_non_utf8):
sha1_git = content_text_non_utf8["sha1_git"]
url = reverse(
"browse-content", url_args={"query_string": content_text_non_utf8["sha1"]}
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-content.html"
)
content_display = _process_content_for_display(archive_data, content_text_non_utf8)
swh_cnt_id = gen_swhid(ObjectType.CONTENT, sha1_git)
swh_cnt_id_url = reverse("browse-swhid", url_args={"swhid": swh_cnt_id})
assert_contains(resp, swh_cnt_id_url)
assert_contains(resp, escape(content_display["content_data"]))
def test_content_view_image(client, archive_data, content_image_type):
url = reverse(
"browse-content", url_args={"query_string": content_image_type["sha1"]}
)
url_raw = reverse(
"browse-content-raw", url_args={"query_string": content_image_type["sha1"]}
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-content.html"
)
content_display = _process_content_for_display(archive_data, content_image_type)
mimetype = content_display["mimetype"]
content_data = content_display["content_data"]
assert_contains(resp, ' ' % (mimetype, content_data))
assert_contains(resp, url_raw)
def test_content_view_image_no_rendering(
client, archive_data, content_unsupported_image_type_rendering
):
url = reverse(
"browse-content",
url_args={"query_string": content_unsupported_image_type_rendering["sha1"]},
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-content.html"
)
mimetype = content_unsupported_image_type_rendering["mimetype"]
encoding = content_unsupported_image_type_rendering["encoding"]
assert_contains(
resp,
(
f"Content with mime type {mimetype} and encoding {encoding} "
"cannot be displayed."
),
)
def test_content_view_text_with_path(client, archive_data, content_text):
path = content_text["path"]
url = reverse(
"browse-content",
url_args={"query_string": content_text["sha1"]},
query_params={"path": path},
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-content.html"
)
assert_contains(resp, '' % hljs_language)
assert_contains(resp, escape(content_display["content_data"]))
split_path = path.split("/")
root_dir_sha1 = split_path[0]
filename = split_path[-1]
path = path.replace(root_dir_sha1 + "/", "").replace(filename, "")
swhid_context = {
"anchor": gen_swhid(ObjectType.DIRECTORY, root_dir_sha1),
"path": f"/{path}{filename}",
}
swh_cnt_id = gen_swhid(
ObjectType.CONTENT, content_text["sha1_git"], metadata=swhid_context
)
swh_cnt_id_url = reverse("browse-swhid", url_args={"swhid": swh_cnt_id})
assert_contains(resp, swh_cnt_id)
assert_contains(resp, swh_cnt_id_url)
path_info = gen_path_info(path)
root_dir_url = reverse("browse-directory", url_args={"sha1_git": root_dir_sha1})
assert_contains(resp, '', count=len(path_info) + 1)
assert_contains(
resp, '' + root_dir_sha1[:7] + " "
)
for p in path_info:
dir_url = reverse(
"browse-directory",
url_args={"sha1_git": root_dir_sha1},
query_params={"path": p["path"]},
)
assert_contains(resp, '' + p["name"] + " ")
assert_contains(resp, " " + filename + " ")
url_raw = reverse(
"browse-content-raw",
url_args={"query_string": content_text["sha1"]},
query_params={"filename": filename},
)
assert_contains(resp, url_raw)
url = reverse(
"browse-content",
url_args={"query_string": content_text["sha1"]},
query_params={"path": filename},
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-content.html"
)
assert_not_contains(resp, '{branch_info['name']}")
cnt_swhid = gen_swhid(
ObjectType.CONTENT,
directory_file["checksums"]["sha1_git"],
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
- "anchor": gen_swhid(ObjectType.REVISION, branch_info["revision"]),
+ "anchor": gen_swhid(ObjectType.REVISION, branch_info["target"]),
"path": f"/{directory_file['name']}",
},
)
assert_contains(resp, cnt_swhid)
dir_swhid = gen_swhid(
ObjectType.DIRECTORY,
directory,
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
- "anchor": gen_swhid(ObjectType.REVISION, branch_info["revision"]),
+ "anchor": gen_swhid(ObjectType.REVISION, branch_info["target"]),
},
)
assert_contains(resp, dir_swhid)
rev_swhid = gen_swhid(
ObjectType.REVISION,
- branch_info["revision"],
+ branch_info["target"],
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
},
)
assert_contains(resp, rev_swhid)
snp_swhid = gen_swhid(
ObjectType.SNAPSHOT,
snapshot["id"],
metadata={
"origin": origin_url,
},
)
assert_contains(resp, snp_swhid)
def test_content_origin_snapshot_release_browse(
client, archive_data, origin_with_multiple_visits
):
origin_url = origin_with_multiple_visits["url"]
visits = archive_data.origin_visit_get(origin_url)
visit = random.choice(visits)
snapshot = archive_data.snapshot_get(visit["snapshot"])
snapshot_sizes = archive_data.snapshot_count_branches(visit["snapshot"])
branches, releases, _ = process_snapshot_branches(snapshot)
release_info = random.choice(releases)
directory_content = archive_data.directory_ls(release_info["directory"])
directory_file = random.choice(
[e for e in directory_content if e["type"] == "file"]
)
url = reverse(
"browse-content",
url_args={"query_string": directory_file["checksums"]["sha1"]},
query_params={
"origin_url": origin_url,
"snapshot": snapshot["id"],
"release": release_info["name"],
"path": directory_file["name"],
},
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-content.html"
)
_check_origin_snapshot_related_html(
resp, origin_with_multiple_visits, snapshot, snapshot_sizes, branches, releases
)
assert_contains(resp, directory_file["name"])
assert_contains(resp, f"Release: {release_info['name']} ")
cnt_swhid = gen_swhid(
ObjectType.CONTENT,
directory_file["checksums"]["sha1_git"],
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
"anchor": gen_swhid(ObjectType.RELEASE, release_info["id"]),
"path": f"/{directory_file['name']}",
},
)
assert_contains(resp, cnt_swhid)
dir_swhid = gen_swhid(
ObjectType.DIRECTORY,
release_info["directory"],
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
"anchor": gen_swhid(ObjectType.RELEASE, release_info["id"]),
},
)
assert_contains(resp, dir_swhid)
rev_swhid = gen_swhid(
ObjectType.REVISION,
release_info["target"],
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
},
)
assert_contains(resp, rev_swhid)
rel_swhid = gen_swhid(
ObjectType.RELEASE,
release_info["id"],
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
},
)
assert_contains(resp, rel_swhid)
snp_swhid = gen_swhid(
ObjectType.SNAPSHOT,
snapshot["id"],
metadata={
"origin": origin_url,
},
)
assert_contains(resp, snp_swhid)
def _check_origin_snapshot_related_html(
resp, origin, snapshot, snapshot_sizes, branches, releases
):
browse_origin_url = reverse(
"browse-origin", query_params={"origin_url": origin["url"]}
)
assert_contains(resp, f'href="{browse_origin_url}"')
origin_branches_url = reverse(
"browse-origin-branches",
query_params={"origin_url": origin["url"], "snapshot": snapshot["id"]},
)
assert_contains(resp, f'href="{escape(origin_branches_url)}"')
assert_contains(resp, f"Branches ({snapshot_sizes['revision']})")
origin_releases_url = reverse(
"browse-origin-releases",
query_params={"origin_url": origin["url"], "snapshot": snapshot["id"]},
)
assert_contains(resp, f'href="{escape(origin_releases_url)}"')
assert_contains(resp, f"Releases ({snapshot_sizes['release']})")
assert_contains(resp, '', count=len(branches))
assert_contains(resp, ' ', count=len(releases))
def _process_content_for_display(archive_data, content):
content_data = archive_data.content_get_data(content["sha1"])
mime_type, encoding = get_mimetype_and_encoding_for_content(content_data["data"])
mime_type, encoding, content_data = re_encode_content(
mime_type, encoding, content_data["data"]
)
content_display = prepare_content_for_display(
content_data, mime_type, content["path"]
)
assert type(content_display["content_data"]) == str
return content_display
def test_content_dispaly_empty_query_string_missing_path(client):
url = reverse(
"browse-content",
query_params={"origin_url": "http://example.com"},
)
resp = check_html_get_response(
client, url, status_code=400, template_used="error.html"
)
assert_contains(resp, "The path query parameter must be provided.", status_code=400)
def test_content_dispaly_empty_query_string_and_snapshot_origin(client):
url = reverse(
"browse-content",
query_params={"path": "test.txt"},
)
resp = check_html_get_response(
client,
url,
status_code=400,
)
assert_contains(
resp,
"The origin_url or snapshot query parameters must be provided.",
status_code=400,
)
def test_content_dispaly_empty_query_string_with_origin(
client, archive_data, origin_with_multiple_visits
):
origin_url = origin_with_multiple_visits["url"]
snapshot = archive_data.snapshot_get_latest(origin_url)
head_rev_id = archive_data.snapshot_get_head(snapshot)
head_rev = archive_data.revision_get(head_rev_id)
dir_content = archive_data.directory_ls(head_rev["directory"])
dir_files = [e for e in dir_content if e["type"] == "file"]
dir_file = random.choice(dir_files)
url = reverse(
"browse-content",
query_params={
"origin_url": origin_url,
"path": dir_file["name"],
},
)
resp = check_html_get_response(
client,
url,
status_code=302,
)
redict_url = reverse(
"browse-content",
url_args={"query_string": f"sha1_git:{dir_file['checksums']['sha1_git']}"},
query_params={
"origin_url": origin_url,
"path": dir_file["name"],
},
)
assert resp.url == redict_url
def test_content_dispaly_empty_query_string_with_snapshot(
client, archive_data, origin_with_multiple_visits
):
origin_url = origin_with_multiple_visits["url"]
snapshot = archive_data.snapshot_get_latest(origin_url)
head_rev_id = archive_data.snapshot_get_head(snapshot)
head_rev = archive_data.revision_get(head_rev_id)
dir_content = archive_data.directory_ls(head_rev["directory"])
dir_files = [e for e in dir_content if e["type"] == "file"]
dir_file = random.choice(dir_files)
url = reverse(
"browse-content",
query_params={
"snapshot": snapshot["id"],
"path": dir_file["name"],
},
)
resp = check_html_get_response(
client,
url,
status_code=302,
)
redict_url = reverse(
"browse-content",
url_args={"query_string": f"sha1_git:{dir_file['checksums']['sha1_git']}"},
query_params={
"snapshot": snapshot["id"],
"path": dir_file["name"],
},
)
assert resp.url == redict_url
def test_browse_origin_content_no_visit(client, mocker, origin):
mock_get_origin_visits = mocker.patch(
"swh.web.utils.origin_visits.get_origin_visits"
)
mock_get_origin_visits.return_value = []
mock_archive = mocker.patch("swh.web.utils.origin_visits.archive")
mock_archive.lookup_origin_visit_latest.return_value = None
url = reverse(
"browse-content",
query_params={"origin_url": origin["url"], "path": "foo"},
)
resp = check_html_get_response(
client, url, status_code=404, template_used="error.html"
)
assert_contains(resp, "No valid visit", status_code=404)
assert not mock_get_origin_visits.called
def test_browse_origin_content_unknown_visit(client, mocker, origin):
mock_get_origin_visits = mocker.patch(
"swh.web.utils.origin_visits.get_origin_visits"
)
mock_get_origin_visits.return_value = [{"visit": 1}]
url = reverse(
"browse-content",
query_params={"origin_url": origin["url"], "path": "foo", "visit_id": 2},
)
resp = check_html_get_response(
client, url, status_code=404, template_used="error.html"
)
assert re.search("Resource not found", resp.content.decode("utf-8"))
def test_browse_origin_content_not_found(client, origin):
url = reverse(
"browse-content",
query_params={"origin_url": origin["url"], "path": "/invalid/file/path"},
)
resp = check_html_get_response(
client, url, status_code=404, template_used="error.html"
)
assert re.search("Resource not found", resp.content.decode("utf-8"))
def test_browse_content_invalid_origin(client):
url = reverse(
"browse-content",
query_params={
"origin_url": "http://invalid-origin",
"path": "/invalid/file/path",
},
)
resp = check_html_get_response(
client, url, status_code=404, template_used="error.html"
)
assert re.search("Resource not found", resp.content.decode("utf-8"))
def test_origin_content_view(
client, archive_data, swh_scheduler, origin_with_multiple_visits
):
origin_visits = archive_data.origin_visit_get(origin_with_multiple_visits["url"])
def _get_archive_data(visit_idx):
snapshot = archive_data.snapshot_get(origin_visits[visit_idx]["snapshot"])
head_rev_id = archive_data.snapshot_get_head(snapshot)
head_rev = archive_data.revision_get(head_rev_id)
dir_content = archive_data.directory_ls(head_rev["directory"])
dir_files = [e for e in dir_content if e["type"] == "file"]
dir_file = random.choice(dir_files)
branches, releases, _ = process_snapshot_branches(snapshot)
return {
"branches": branches,
"releases": releases,
"root_dir_sha1": head_rev["directory"],
"content": get_content(dir_file["checksums"]["sha1"]),
"visit": origin_visits[visit_idx],
"snapshot_sizes": archive_data.snapshot_count_branches(snapshot["id"]),
}
tdata = _get_archive_data(-1)
_origin_content_view_test_helper(
client,
archive_data,
origin_with_multiple_visits,
origin_visits[-1],
tdata["snapshot_sizes"],
tdata["branches"],
tdata["releases"],
tdata["root_dir_sha1"],
tdata["content"],
)
_origin_content_view_test_helper(
client,
archive_data,
origin_with_multiple_visits,
origin_visits[-1],
tdata["snapshot_sizes"],
tdata["branches"],
tdata["releases"],
tdata["root_dir_sha1"],
tdata["content"],
timestamp=tdata["visit"]["date"],
)
_origin_content_view_test_helper(
client,
archive_data,
origin_with_multiple_visits,
origin_visits[-1],
tdata["snapshot_sizes"],
tdata["branches"],
tdata["releases"],
tdata["root_dir_sha1"],
tdata["content"],
snapshot_id=tdata["visit"]["snapshot"],
)
tdata = _get_archive_data(0)
_origin_content_view_test_helper(
client,
archive_data,
origin_with_multiple_visits,
origin_visits[0],
tdata["snapshot_sizes"],
tdata["branches"],
tdata["releases"],
tdata["root_dir_sha1"],
tdata["content"],
visit_id=tdata["visit"]["visit"],
)
_origin_content_view_test_helper(
client,
archive_data,
origin_with_multiple_visits,
origin_visits[0],
tdata["snapshot_sizes"],
tdata["branches"],
tdata["releases"],
tdata["root_dir_sha1"],
tdata["content"],
snapshot_id=tdata["visit"]["snapshot"],
)
def _origin_content_view_test_helper(
client,
archive_data,
origin_info,
origin_visit,
snapshot_sizes,
origin_branches,
origin_releases,
root_dir_sha1,
content,
visit_id=None,
timestamp=None,
snapshot_id=None,
):
content_path = "/".join(content["path"].split("/")[1:])
if not visit_id and not snapshot_id:
visit_id = origin_visit["visit"]
query_params = {"origin_url": origin_info["url"], "path": content_path}
if timestamp:
query_params["timestamp"] = timestamp
if visit_id:
query_params["visit_id"] = visit_id
elif snapshot_id:
query_params["snapshot"] = snapshot_id
url = reverse(
"browse-content",
url_args={"query_string": f"sha1_git:{content['sha1_git']}"},
query_params=query_params,
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-content.html"
)
assert type(content["data"]) == str
assert_contains(resp, '' % content["hljs_language"])
assert_contains(resp, escape(content["data"]))
split_path = content_path.split("/")
filename = split_path[-1]
path = content_path.replace(filename, "")[:-1]
path_info = gen_path_info(path)
del query_params["path"]
if timestamp:
query_params["timestamp"] = format_utc_iso_date(
parse_iso8601_date_to_utc(timestamp).isoformat(), "%Y-%m-%dT%H:%M:%SZ"
)
root_dir_url = reverse(
"browse-directory",
url_args={"sha1_git": root_dir_sha1},
query_params=query_params,
)
assert_contains(resp, '', count=len(path_info) + 1)
assert_contains(resp, '%s ' % (root_dir_url, root_dir_sha1[:7]))
for p in path_info:
query_params["path"] = p["path"]
dir_url = reverse("browse-origin-directory", query_params=query_params)
assert_contains(resp, '%s ' % (dir_url, p["name"]))
assert_contains(resp, " %s " % filename)
query_string = "sha1_git:" + content["sha1_git"]
url_raw = reverse(
"browse-content-raw",
url_args={"query_string": query_string},
query_params={"filename": filename},
)
assert_contains(resp, url_raw)
if "path" in query_params:
del query_params["path"]
origin_branches_url = reverse("browse-origin-branches", query_params=query_params)
assert_contains(resp, f'href="{escape(origin_branches_url)}"')
assert_contains(resp, f"Branches ({snapshot_sizes['revision']})")
origin_releases_url = reverse("browse-origin-releases", query_params=query_params)
assert_contains(resp, f'href="{escape(origin_releases_url)}">')
assert_contains(resp, f"Releases ({snapshot_sizes['release']})")
assert_contains(resp, '', count=len(origin_branches))
query_params["path"] = content_path
for branch in origin_branches:
root_dir_branch_url = reverse(
"browse-origin-content",
query_params={"branch": branch["name"], **query_params},
)
assert_contains(resp, '' % root_dir_branch_url)
assert_contains(resp, '', count=len(origin_releases))
query_params["branch"] = None
for release in origin_releases:
root_dir_release_url = reverse(
"browse-origin-content",
query_params={"release": release["name"], **query_params},
)
assert_contains(resp, '' % root_dir_release_url)
url = reverse(
"browse-content",
url_args={"query_string": query_string},
query_params=query_params,
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-content.html"
)
snapshot = archive_data.snapshot_get(origin_visit["snapshot"])
head_rev_id = archive_data.snapshot_get_head(snapshot)
swhid_context = {
"origin": origin_info["url"],
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
"anchor": gen_swhid(ObjectType.REVISION, head_rev_id),
"path": f"/{content_path}",
}
swh_cnt_id = gen_swhid(
ObjectType.CONTENT, content["sha1_git"], metadata=swhid_context
)
swh_cnt_id_url = reverse("browse-swhid", url_args={"swhid": swh_cnt_id})
assert_contains(resp, swh_cnt_id)
assert_contains(resp, swh_cnt_id_url)
assert_contains(resp, "swh-take-new-snapshot")
_check_origin_link(resp, origin_info["url"])
assert_not_contains(resp, "swh-metadata-popover")
def _check_origin_link(resp, origin_url):
browse_origin_url = reverse(
"browse-origin", query_params={"origin_url": origin_url}
)
assert_contains(resp, f'href="{browse_origin_url}"')
@pytest.mark.django_db
@pytest.mark.parametrize("staff_user_logged_in", [False, True])
def test_browse_content_snapshot_context_release_directory_target(
client, staff_user, archive_data, directory_with_files, staff_user_logged_in
):
if staff_user_logged_in:
client.force_login(staff_user)
release_name = "v1.0.0"
release = Release(
name=release_name.encode(),
message=f"release {release_name}".encode(),
target=hash_to_bytes(directory_with_files),
target_type=ModelObjectType.DIRECTORY,
synthetic=True,
)
archive_data.release_add([release])
snapshot = Snapshot(
branches={
release_name.encode(): SnapshotBranch(
target=release.id, target_type=TargetType.RELEASE
),
},
)
archive_data.snapshot_add([snapshot])
dir_content = archive_data.directory_ls(directory_with_files)
file_entry = random.choice(
[entry for entry in dir_content if entry["type"] == "file"]
)
sha1_git = file_entry["checksums"]["sha1_git"]
browse_url = reverse(
"browse-content",
url_args={"query_string": f"sha1_git:{sha1_git}"},
query_params={
"path": file_entry["name"],
"release": release_name,
"snapshot": snapshot.id.hex(),
},
)
check_html_get_response(
client, browse_url, status_code=200, template_used="browse-content.html"
)
diff --git a/swh/web/tests/browse/views/test_directory.py b/swh/web/tests/browse/views/test_directory.py
index a0f1272e..0e6a7439 100644
--- a/swh/web/tests/browse/views/test_directory.py
+++ b/swh/web/tests/browse/views/test_directory.py
@@ -1,565 +1,565 @@
# Copyright (C) 2017-2022 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 hypothesis import given
import pytest
from django.utils.html import escape
from swh.model.from_disk import DentryPerms
from swh.model.hashutil import hash_to_bytes, hash_to_hex
from swh.model.model import (
Origin,
OriginVisit,
OriginVisitStatus,
Release,
Revision,
RevisionType,
Snapshot,
SnapshotBranch,
TargetType,
TimestampWithTimezone,
)
from swh.model.model import Directory, DirectoryEntry
from swh.model.model import ObjectType as ModelObjectType
from swh.model.swhids import ObjectType
from swh.storage.utils import now
from swh.web.browse.snapshot_context import process_snapshot_branches
from swh.web.tests.django_asserts import assert_contains, assert_not_contains
from swh.web.tests.helpers import check_html_get_response
from swh.web.tests.strategies import new_person, new_swh_date
from swh.web.utils import gen_path_info, reverse
from swh.web.utils.identifiers import gen_swhid
def test_root_directory_view(client, archive_data, directory):
_directory_view_checks(client, directory, archive_data.directory_ls(directory))
def test_sub_directory_view(client, archive_data, directory_with_subdirs):
dir_content = archive_data.directory_ls(directory_with_subdirs)
subdir = random.choice([e for e in dir_content if e["type"] == "dir"])
subdir_content = archive_data.directory_ls(subdir["target"])
_directory_view_checks(
client, directory_with_subdirs, subdir_content, subdir["name"]
)
@given(new_person(), new_swh_date())
def test_sub_directory_view_origin_context(
client, archive_data, empty_directory, person, date
):
origin_url = "test_sub_directory_view_origin_context"
subdir = Directory(
entries=(
DirectoryEntry(
name=b"foo",
type="dir",
target=hash_to_bytes(empty_directory),
perms=DentryPerms.directory,
),
DirectoryEntry(
name=b"bar",
type="dir",
target=hash_to_bytes(empty_directory),
perms=DentryPerms.directory,
),
)
)
parentdir = Directory(
entries=(
DirectoryEntry(
name=b"baz",
type="dir",
target=subdir.id,
perms=DentryPerms.directory,
),
)
)
archive_data.directory_add([subdir, parentdir])
revision = Revision(
directory=parentdir.id,
author=person,
committer=person,
message=b"commit message",
date=TimestampWithTimezone.from_datetime(date),
committer_date=TimestampWithTimezone.from_datetime(date),
synthetic=False,
type=RevisionType.GIT,
)
archive_data.revision_add([revision])
snapshot = Snapshot(
branches={
b"HEAD": SnapshotBranch(
target="refs/head/master".encode(),
target_type=TargetType.ALIAS,
),
b"refs/head/master": SnapshotBranch(
target=revision.id,
target_type=TargetType.REVISION,
),
}
)
archive_data.snapshot_add([snapshot])
archive_data.origin_add([Origin(url=origin_url)])
date = now()
visit = OriginVisit(origin=origin_url, date=date, type="git")
visit = archive_data.origin_visit_add([visit])[0]
visit_status = OriginVisitStatus(
origin=origin_url,
visit=visit.visit,
date=date,
status="full",
snapshot=snapshot.id,
)
archive_data.origin_visit_status_add([visit_status])
dir_content = archive_data.directory_ls(hash_to_hex(parentdir.id))
subdir = dir_content[0]
subdir_content = archive_data.directory_ls(subdir["target"])
_directory_view_checks(
client,
hash_to_hex(parentdir.id),
subdir_content,
subdir["name"],
origin_url,
hash_to_hex(snapshot.id),
hash_to_hex(revision.id),
)
def test_directory_request_errors(client, invalid_sha1, unknown_directory):
dir_url = reverse("browse-directory", url_args={"sha1_git": invalid_sha1})
check_html_get_response(
client, dir_url, status_code=400, template_used="error.html"
)
dir_url = reverse("browse-directory", url_args={"sha1_git": unknown_directory})
check_html_get_response(
client, dir_url, status_code=404, template_used="error.html"
)
def test_directory_with_invalid_path(client, directory):
path = "foo/bar"
dir_url = reverse(
"browse-directory",
url_args={"sha1_git": directory},
query_params={"path": path},
)
resp = check_html_get_response(
client, dir_url, status_code=404, template_used="browse-directory.html"
)
error_message = (
f"Directory entry with path {path} from root directory {directory} not found"
)
assert_contains(resp, error_message, status_code=404)
def test_directory_uppercase(client, directory):
url = reverse(
"browse-directory-uppercase-checksum", url_args={"sha1_git": directory.upper()}
)
resp = check_html_get_response(client, url, status_code=302)
redirect_url = reverse("browse-directory", url_args={"sha1_git": directory})
assert resp["location"] == redirect_url
def test_permalink_box_context(client, tests_data, directory):
origin_url = random.choice(tests_data["origins"])["url"]
url = reverse(
"browse-directory",
url_args={"sha1_git": directory},
query_params={"origin_url": origin_url},
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-directory.html"
)
assert_contains(resp, 'id="swhid-context-option-directory"')
def test_directory_origin_snapshot_branch_browse(
client, archive_data, origin_with_multiple_visits
):
origin_url = origin_with_multiple_visits["url"]
visits = archive_data.origin_visit_get(origin_url)
visit = random.choice(visits)
snapshot = archive_data.snapshot_get(visit["snapshot"])
snapshot_sizes = archive_data.snapshot_count_branches(visit["snapshot"])
branches, releases, _ = process_snapshot_branches(snapshot)
branch_info = next(
branch for branch in branches if branch["name"] == "refs/heads/master"
)
- directory = archive_data.revision_get(branch_info["revision"])["directory"]
+ directory = archive_data.revision_get(branch_info["target"])["directory"]
directory_content = archive_data.directory_ls(directory)
directory_subdir = random.choice(
[e for e in directory_content if e["type"] == "dir"]
)
url = reverse(
"browse-directory",
url_args={"sha1_git": directory},
query_params={
"origin_url": origin_url,
"snapshot": snapshot["id"],
"branch": branch_info["name"],
"path": directory_subdir["name"],
},
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-directory.html"
)
_check_origin_snapshot_related_html(
resp, origin_with_multiple_visits, snapshot, snapshot_sizes, branches, releases
)
assert_contains(resp, directory_subdir["name"])
assert_contains(resp, f"Branch: {branch_info['name']} ")
dir_swhid = gen_swhid(
ObjectType.DIRECTORY,
directory_subdir["target"],
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
- "anchor": gen_swhid(ObjectType.REVISION, branch_info["revision"]),
+ "anchor": gen_swhid(ObjectType.REVISION, branch_info["target"]),
"path": "/",
},
)
assert_contains(resp, dir_swhid)
rev_swhid = gen_swhid(
ObjectType.REVISION,
- branch_info["revision"],
+ branch_info["target"],
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
},
)
assert_contains(resp, rev_swhid)
snp_swhid = gen_swhid(
ObjectType.SNAPSHOT,
snapshot["id"],
metadata={
"origin": origin_url,
},
)
assert_contains(resp, snp_swhid)
def test_drectory_origin_snapshot_release_browse(
client, archive_data, origin_with_multiple_visits
):
origin_url = origin_with_multiple_visits["url"]
visits = archive_data.origin_visit_get(origin_url)
visit = random.choice(visits)
snapshot = archive_data.snapshot_get(visit["snapshot"])
snapshot_sizes = archive_data.snapshot_count_branches(visit["snapshot"])
branches, releases, _ = process_snapshot_branches(snapshot)
release_info = random.choice(releases)
directory = release_info["directory"]
directory_content = archive_data.directory_ls(directory)
directory_subdir = random.choice(
[e for e in directory_content if e["type"] == "dir"]
)
url = reverse(
"browse-directory",
url_args={"sha1_git": directory},
query_params={
"origin_url": origin_url,
"snapshot": snapshot["id"],
"release": release_info["name"],
"path": directory_subdir["name"],
},
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-directory.html"
)
_check_origin_snapshot_related_html(
resp, origin_with_multiple_visits, snapshot, snapshot_sizes, branches, releases
)
assert_contains(resp, directory_subdir["name"])
assert_contains(resp, f"Release: {release_info['name']} ")
dir_swhid = gen_swhid(
ObjectType.DIRECTORY,
directory_subdir["target"],
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
"anchor": gen_swhid(ObjectType.RELEASE, release_info["id"]),
"path": "/",
},
)
assert_contains(resp, dir_swhid)
rev_swhid = gen_swhid(
ObjectType.REVISION,
release_info["target"],
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
},
)
assert_contains(resp, rev_swhid)
rel_swhid = gen_swhid(
ObjectType.RELEASE,
release_info["id"],
metadata={
"origin": origin_url,
"visit": gen_swhid(ObjectType.SNAPSHOT, snapshot["id"]),
},
)
assert_contains(resp, rel_swhid)
snp_swhid = gen_swhid(
ObjectType.SNAPSHOT,
snapshot["id"],
metadata={
"origin": origin_url,
},
)
assert_contains(resp, snp_swhid)
def test_directory_origin_snapshot_revision_browse(
client, archive_data, origin_with_multiple_visits
):
origin_url = origin_with_multiple_visits["url"]
visits = archive_data.origin_visit_get(origin_url)
visit = random.choice(visits)
snapshot = archive_data.snapshot_get(visit["snapshot"])
branches, releases, _ = process_snapshot_branches(snapshot)
branch_info = next(
branch for branch in branches if branch["name"] == "refs/heads/master"
)
- directory = archive_data.revision_get(branch_info["revision"])["directory"]
+ directory = archive_data.revision_get(branch_info["target"])["directory"]
directory_content = archive_data.directory_ls(directory)
directory_subdir = random.choice(
[e for e in directory_content if e["type"] == "dir"]
)
url = reverse(
"browse-directory",
url_args={"sha1_git": directory},
query_params={
"origin_url": origin_url,
"snapshot": snapshot["id"],
- "revision": branch_info["revision"],
+ "revision": branch_info["target"],
"path": directory_subdir["name"],
},
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-directory.html"
)
- assert_contains(resp, f"Revision: {branch_info['revision']} ")
+ assert_contains(resp, f"Revision: {branch_info['target']} ")
def _check_origin_snapshot_related_html(
resp, origin, snapshot, snapshot_sizes, branches, releases
):
browse_origin_url = reverse(
"browse-origin", query_params={"origin_url": origin["url"]}
)
assert_contains(resp, f'href="{browse_origin_url}"')
origin_branches_url = reverse(
"browse-origin-branches",
query_params={"origin_url": origin["url"], "snapshot": snapshot["id"]},
)
assert_contains(resp, f'href="{escape(origin_branches_url)}"')
assert_contains(resp, f"Branches ({snapshot_sizes['revision']})")
origin_releases_url = reverse(
"browse-origin-releases",
query_params={"origin_url": origin["url"], "snapshot": snapshot["id"]},
)
assert_contains(resp, f'href="{escape(origin_releases_url)}"')
assert_contains(resp, f"Releases ({snapshot_sizes['release']})")
assert_contains(resp, '', count=len(branches))
assert_contains(resp, ' ', count=len(releases))
def _directory_view_checks(
client,
root_directory_sha1,
directory_entries,
path=None,
origin_url=None,
snapshot_id=None,
revision_id=None,
):
dirs = [e for e in directory_entries if e["type"] in ("dir", "rev")]
files = [e for e in directory_entries if e["type"] == "file"]
url_args = {"sha1_git": root_directory_sha1}
query_params = {"origin_url": origin_url, "snapshot": snapshot_id}
url = reverse(
"browse-directory",
url_args=url_args,
query_params={**query_params, "path": path},
)
root_dir_url = reverse(
"browse-directory",
url_args=url_args,
query_params=query_params,
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-directory.html"
)
assert_contains(
resp,
'' + root_directory_sha1[:7] + " ",
)
assert_contains(resp, '', count=len(dirs))
assert_contains(resp, ' ', count=len(files))
for d in dirs:
if d["type"] == "rev":
dir_url = reverse("browse-revision", url_args={"sha1_git": d["target"]})
else:
dir_path = d["name"]
if path:
dir_path = "%s/%s" % (path, d["name"])
dir_url = reverse(
"browse-directory",
url_args={"sha1_git": root_directory_sha1},
query_params={**query_params, "path": dir_path},
)
assert_contains(resp, dir_url)
for f in files:
file_path = "%s/%s" % (root_directory_sha1, f["name"])
if path:
file_path = "%s/%s/%s" % (root_directory_sha1, path, f["name"])
query_string = "sha1_git:" + f["target"]
file_url = reverse(
"browse-content",
url_args={"query_string": query_string},
query_params={**query_params, "path": file_path},
)
assert_contains(resp, file_url)
path_info = gen_path_info(path)
assert_contains(resp, '', count=len(path_info) + 1)
assert_contains(
resp, '%s ' % (root_dir_url, root_directory_sha1[:7])
)
for p in path_info:
dir_url = reverse(
"browse-directory",
url_args={"sha1_git": root_directory_sha1},
query_params={**query_params, "path": p["path"]},
)
assert_contains(resp, '%s ' % (dir_url, p["name"]))
assert_contains(resp, "vault-cook-directory")
swh_dir_id = gen_swhid(ObjectType.DIRECTORY, directory_entries[0]["dir_id"])
swh_dir_id_url = reverse("browse-swhid", url_args={"swhid": swh_dir_id})
swhid_context = {}
if origin_url:
swhid_context["origin"] = origin_url
if snapshot_id:
swhid_context["visit"] = gen_swhid(ObjectType.SNAPSHOT, snapshot_id)
if root_directory_sha1 != directory_entries[0]["dir_id"]:
swhid_context["anchor"] = gen_swhid(ObjectType.DIRECTORY, root_directory_sha1)
if root_directory_sha1 != directory_entries[0]["dir_id"]:
swhid_context["anchor"] = gen_swhid(ObjectType.DIRECTORY, root_directory_sha1)
if revision_id:
swhid_context["anchor"] = gen_swhid(ObjectType.REVISION, revision_id)
swhid_context["path"] = f"/{path}/" if path else None
swh_dir_id = gen_swhid(
ObjectType.DIRECTORY, directory_entries[0]["dir_id"], metadata=swhid_context
)
swh_dir_id_url = reverse("browse-swhid", url_args={"swhid": swh_dir_id})
assert_contains(resp, swh_dir_id)
assert_contains(resp, swh_dir_id_url)
assert_not_contains(resp, "swh-metadata-popover")
@pytest.mark.django_db
@pytest.mark.parametrize("staff_user_logged_in", [False, True])
def test_browse_directory_snapshot_context_release_directory_target(
client, staff_user, archive_data, directory_with_subdirs, staff_user_logged_in
):
if staff_user_logged_in:
client.force_login(staff_user)
release_name = "v1.0.0"
release = Release(
name=release_name.encode(),
message=f"release {release_name}".encode(),
target=hash_to_bytes(directory_with_subdirs),
target_type=ModelObjectType.DIRECTORY,
synthetic=True,
)
archive_data.release_add([release])
snapshot = Snapshot(
branches={
release_name.encode(): SnapshotBranch(
target=release.id, target_type=TargetType.RELEASE
),
},
)
archive_data.snapshot_add([snapshot])
dir_content = archive_data.directory_ls(directory_with_subdirs)
dir_entry = random.choice(
[entry for entry in dir_content if entry["type"] == "dir"]
)
browse_url = reverse(
"browse-directory",
url_args={"sha1_git": directory_with_subdirs},
query_params={
"path": dir_entry["name"],
"release": release_name,
"snapshot": snapshot.id.hex(),
},
)
check_html_get_response(
client, browse_url, status_code=200, template_used="browse-directory.html"
)
diff --git a/swh/web/tests/browse/views/test_snapshot.py b/swh/web/tests/browse/views/test_snapshot.py
index a5595375..f1689756 100644
--- a/swh/web/tests/browse/views/test_snapshot.py
+++ b/swh/web/tests/browse/views/test_snapshot.py
@@ -1,448 +1,651 @@
# Copyright (C) 2022 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
import re
import string
from dateutil import parser
from hypothesis import given
import pytest
from django.utils.html import escape
from swh.model.hashutil import hash_to_bytes
from swh.model.model import (
ObjectType,
+ Origin,
OriginVisit,
OriginVisitStatus,
Release,
Revision,
RevisionType,
Snapshot,
SnapshotBranch,
TargetType,
TimestampWithTimezone,
)
from swh.storage.utils import now
from swh.web.browse.snapshot_context import process_snapshot_branches
from swh.web.tests.data import random_sha1
from swh.web.tests.django_asserts import assert_contains, assert_not_contains
from swh.web.tests.helpers import check_html_get_response
from swh.web.tests.strategies import new_origin, new_person, new_swh_date, visit_dates
from swh.web.utils import reverse
@pytest.mark.parametrize(
"browse_context,template_used",
[
("log", "revision-log.html"),
("branches", "branches.html"),
("releases", "releases.html"),
],
)
def test_snapshot_browse_with_id(client, browse_context, template_used, snapshot):
url = reverse(
f"browse-snapshot-{browse_context}", url_args={"snapshot_id": snapshot}
)
resp = check_html_get_response(
client, url, status_code=200, template_used=f"browse-{template_used}"
)
assert_contains(resp, f"swh:1:snp:{snapshot}")
@pytest.mark.parametrize("browse_context", ["log", "branches", "releases"])
def test_snapshot_browse_with_id_and_origin(
client, browse_context, archive_data, origin
):
snapshot = archive_data.snapshot_get_latest(origin["url"])
url = reverse(
f"browse-snapshot-{browse_context}",
url_args={"snapshot_id": snapshot["id"]},
query_params={"origin_url": origin["url"]},
)
resp = check_html_get_response(
client, url, status_code=200, template_used="includes/snapshot-context.html"
)
assert_contains(resp, origin["url"])
@pytest.mark.parametrize("browse_context", ["log", "branches", "releases"])
def test_snapshot_browse_with_id_origin_and_timestamp(
client, browse_context, archive_data, origin_with_multiple_visits
):
visit = archive_data.origin_visit_get(origin_with_multiple_visits["url"])[0]
url = reverse(
f"browse-snapshot-{browse_context}",
url_args={"snapshot_id": visit["snapshot"]},
query_params={"origin_url": visit["origin"], "timestamp": visit["date"]},
)
resp = check_html_get_response(
client, url, status_code=200, template_used="includes/snapshot-context.html"
)
requested_time = parser.parse(visit["date"]).strftime("%d %B %Y, %H:%M")
assert_contains(resp, requested_time)
assert_contains(resp, visit["origin"])
@pytest.mark.parametrize("browse_context", ["log", "branches", "releases"])
def test_snapshot_browse_without_id(client, browse_context, archive_data, origin):
url = reverse(
f"browse-snapshot-{browse_context}", query_params={"origin_url": origin["url"]}
)
# This will be redirected to /snapshot//log
resp = check_html_get_response(
client,
url,
status_code=302,
)
snapshot = archive_data.snapshot_get_latest(origin["url"])
assert resp.url == reverse(
f"browse-snapshot-{browse_context}",
url_args={"snapshot_id": snapshot["id"]},
query_params={"origin_url": origin["url"]},
)
@pytest.mark.parametrize("browse_context", ["log", "branches", "releases"])
def test_snapshot_browse_without_id_and_origin(client, browse_context):
url = reverse(f"browse-snapshot-{browse_context}")
resp = check_html_get_response(
client,
url,
status_code=400,
)
# assert_contains works only with a success response, using re.search instead
assert re.search(
"An origin URL must be provided as a query parameter",
resp.content.decode("utf-8"),
)
def test_snapshot_browse_branches(client, archive_data, origin):
snapshot = archive_data.snapshot_get_latest(origin["url"])
snapshot_sizes = archive_data.snapshot_count_branches(snapshot["id"])
snapshot_content = process_snapshot_branches(snapshot)
_origin_branches_test_helper(
client, origin, snapshot_content, snapshot_sizes, snapshot_id=snapshot["id"]
)
def _origin_branches_test_helper(
client, origin_info, origin_snapshot, snapshot_sizes, snapshot_id
):
query_params = {"origin_url": origin_info["url"], "snapshot": snapshot_id}
url = reverse(
"browse-snapshot-branches",
url_args={"snapshot_id": snapshot_id},
query_params=query_params,
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-branches.html"
)
origin_branches = origin_snapshot[0]
origin_releases = origin_snapshot[1]
origin_branches_url = reverse("browse-origin-branches", query_params=query_params)
assert_contains(resp, f'href="{escape(origin_branches_url)}"')
assert_contains(resp, f"Branches ({snapshot_sizes['revision']})")
origin_releases_url = reverse("browse-origin-releases", query_params=query_params)
nb_releases = len(origin_releases)
if nb_releases > 0:
assert_contains(resp, f'href="{escape(origin_releases_url)}">')
assert_contains(resp, f"Releases ({snapshot_sizes['release']})")
assert_contains(resp, '' % escape(browse_branch_url))
browse_revision_url = reverse(
"browse-revision",
- url_args={"sha1_git": branch["revision"]},
+ url_args={"sha1_git": branch["target"]},
query_params=query_params,
)
assert_contains(resp, '' % escape(browse_revision_url))
_check_origin_link(resp, origin_info["url"])
def _check_origin_link(resp, origin_url):
browse_origin_url = reverse(
"browse-origin", query_params={"origin_url": origin_url}
)
assert_contains(resp, f'href="{browse_origin_url}"')
@given(
new_origin(),
visit_dates(),
)
def test_snapshot_branches_pagination_with_alias(
client,
archive_data,
mocker,
release,
revisions_list,
new_origin,
visit_dates,
):
"""
When a snapshot contains a branch or a release alias, pagination links
in the branches / releases view should be displayed.
"""
revisions = revisions_list(size=10)
mocker.patch("swh.web.browse.snapshot_context.PER_PAGE", len(revisions) / 2)
snp_dict = {"branches": {}, "id": hash_to_bytes(random_sha1())}
for i in range(len(revisions)):
branch = "".join(random.choices(string.ascii_lowercase, k=8))
snp_dict["branches"][branch.encode()] = {
"target_type": "revision",
"target": hash_to_bytes(revisions[i]),
}
release_name = "".join(random.choices(string.ascii_lowercase, k=8))
snp_dict["branches"][b"RELEASE_ALIAS"] = {
"target_type": "alias",
"target": release_name.encode(),
}
snp_dict["branches"][release_name.encode()] = {
"target_type": "release",
"target": hash_to_bytes(release),
}
archive_data.origin_add([new_origin])
archive_data.snapshot_add([Snapshot.from_dict(snp_dict)])
visit = archive_data.origin_visit_add(
[
OriginVisit(
origin=new_origin.url,
date=visit_dates[0],
type="git",
)
]
)[0]
visit_status = OriginVisitStatus(
origin=new_origin.url,
visit=visit.visit,
date=now(),
status="full",
snapshot=snp_dict["id"],
)
archive_data.origin_visit_status_add([visit_status])
snapshot = archive_data.snapshot_get_latest(new_origin.url)
url = reverse(
"browse-snapshot-branches",
url_args={"snapshot_id": snapshot["id"]},
query_params={"origin_url": new_origin.url},
)
resp = check_html_get_response(
client, url, status_code=200, template_used="browse-branches.html"
)
assert_contains(resp, '