diff --git a/swh/web/browse/snapshot_context.py b/swh/web/browse/snapshot_context.py index bcd6bcbd..e4d3c99a 100644 --- a/swh/web/browse/snapshot_context.py +++ b/swh/web/browse/snapshot_context.py @@ -1,1401 +1,1409 @@ # Copyright (C) 2018-2019 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 copy import copy from typing import Any, Dict, List, Optional, Union, Tuple from django.core.cache import cache from django.shortcuts import render from django.template.defaultfilters import filesizeformat from django.utils.html import escape import sentry_sdk from swh.model.identifiers import persistent_identifier, snapshot_identifier from swh.web.browse.utils import ( get_directory_entries, gen_directory_link, gen_revision_link, request_content, gen_content_link, prepare_content_for_display, content_display_max_size, format_log_entries, gen_revision_log_link, gen_release_link, get_readme_to_display, get_swh_persistent_ids, gen_snapshot_link, ) from swh.web.common import service, highlightjs from swh.web.common.exc import handle_view_exception, NotFoundExc from swh.web.common.origin_visits import get_origin_visit from swh.web.common.typing import ( OriginInfo, QueryParameters, SnapshotBranchInfo, SnapshotReleaseInfo, SnapshotContext, ) from swh.web.common.utils import ( reverse, gen_path_info, format_utc_iso_date, swh_object_icons, ) from swh.web.config import get_config _empty_snapshot_id = snapshot_identifier({"branches": {}}) def _get_branch(branches, branch_name, snapshot_id): """ Utility function to get a specific branch from a branches list. Its purpose is to get the default HEAD branch as some software origin (e.g those with svn type) does not have it. In that latter case, check if there is a master branch instead and returns it. """ filtered_branches = [b for b in branches if b["name"] == branch_name] if filtered_branches: return filtered_branches[0] elif branch_name == "HEAD": filtered_branches = [b for b in branches if b["name"].endswith("master")] if filtered_branches: return filtered_branches[0] elif branches: return branches[0] else: # case where a large branches list has been truncated snp = service.lookup_snapshot( snapshot_id, branches_from=branch_name, branches_count=1, target_types=["revision", "alias"], ) snp_branch, _ = process_snapshot_branches(snp) if snp_branch and snp_branch[0]["name"] == branch_name: branches.append(snp_branch[0]) return snp_branch[0] def _get_release(releases, release_name, snapshot_id): """ Utility function to get a specific release from a releases list. Returns None if the release can not be found in the list. """ filtered_releases = [r for r in releases if r["name"] == release_name] if filtered_releases: return filtered_releases[0] else: # case where a large branches list has been truncated for branch_name in (release_name, f"refs/tags/{release_name}"): snp = service.lookup_snapshot( snapshot_id, branches_from=branch_name, branches_count=1, target_types=["release"], ) _, snp_release = process_snapshot_branches(snp) if snp_release and snp_release[0]["name"] == release_name: releases.append(snp_release[0]) return snp_release[0] def _branch_not_found( branch_type, branch, snapshot_id, snapshot_sizes, origin_info, timestamp, visit_id ): """ Utility function to raise an exception when a specified branch/release can not be found. """ if branch_type == "branch": branch_type = "Branch" branch_type_plural = "branches" target_type = "revision" else: branch_type = "Release" branch_type_plural = "releases" target_type = "release" if snapshot_id and snapshot_sizes[target_type] == 0: msg = "Snapshot with id %s has an empty list" " of %s!" % ( snapshot_id, branch_type_plural, ) elif snapshot_id: msg = "%s %s for snapshot with id %s" " not found!" % ( branch_type, branch, snapshot_id, ) elif visit_id and snapshot_sizes[target_type] == 0: msg = ( "Origin with url %s" " for visit with id %s has an empty list" " of %s!" % (origin_info["url"], visit_id, branch_type_plural) ) elif visit_id: msg = ( "%s %s associated to visit with" " id %s for origin with url %s" " not found!" % (branch_type, branch, visit_id, origin_info["url"]) ) elif snapshot_sizes[target_type] == 0: msg = ( "Origin with url %s" " for visit with timestamp %s has an empty list" " of %s!" % (origin_info["url"], timestamp, branch_type_plural) ) else: msg = ( "%s %s associated to visit with" " timestamp %s for origin with " "url %s not found!" % (branch_type, branch, timestamp, origin_info["url"]) ) raise NotFoundExc(escape(msg)) def process_snapshot_branches( snapshot: Dict[str, Any] ) -> Tuple[List[SnapshotBranchInfo], List[SnapshotReleaseInfo]]: """ Process a dictionary describing snapshot branches: extract those targeting revisions and releases, put them in two different lists, then sort those lists in lexicographical order of the branches' names. Args: snapshot: A dict describing a snapshot as returned for instance by :func:`swh.web.common.service.lookup_snapshot` Returns: A tuple whose first member is the sorted list of branches targeting revisions and second member the sorted list of branches targeting releases """ snapshot_branches = snapshot["branches"] branches: Dict[str, SnapshotBranchInfo] = {} branch_aliases: Dict[str, str] = {} releases: Dict[str, SnapshotReleaseInfo] = {} revision_to_branch = defaultdict(set) revision_to_release = defaultdict(set) release_to_branch = defaultdict(set) for branch_name, target in snapshot_branches.items(): if not target: # FIXME: display branches with an unknown target anyway continue target_id = target["target"] target_type = target["target_type"] if target_type == "revision": branches[branch_name] = SnapshotBranchInfo( name=branch_name, revision=target_id, date=None, directory=None, message=None, url=None, ) revision_to_branch[target_id].add(branch_name) elif target_type == "release": release_to_branch[target_id].add(branch_name) elif target_type == "alias": branch_aliases[branch_name] = target_id # FIXME: handle pointers to other object types def _add_release_info(branch, release): releases[branch] = SnapshotReleaseInfo( name=release["name"], 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): branches[branch] = SnapshotBranchInfo( name=branch, revision=revision["id"], directory=revision["directory"], date=format_utc_iso_date(revision["date"]), message=revision["message"], url=None, ) releases_info = service.lookup_release_multiple(release_to_branch.keys()) for release in releases_info: 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 = service.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 in revision_to_release[revision["id"]]: releases[release]["directory"] = revision["directory"] for branch_alias, branch_target in branch_aliases.items(): if branch_target in branches: branches[branch_alias] = copy(branches[branch_target]) else: snp = service.lookup_snapshot( snapshot["id"], branches_from=branch_target, branches_count=1 ) if snp and branch_target in snp["branches"]: if snp["branches"][branch_target] is None: continue target_type = snp["branches"][branch_target]["target_type"] target = snp["branches"][branch_target]["target"] if target_type == "revision": branches[branch_alias] = snp["branches"][branch_target] revision = service.lookup_revision(target) _add_branch_info(branch_alias, revision) elif target_type == "release": release = service.lookup_release(target) _add_release_info(branch_alias, release) 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 def get_snapshot_content( snapshot_id: str, ) -> Tuple[List[SnapshotBranchInfo], List[SnapshotReleaseInfo]]: """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 two 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. Raises: NotFoundExc if the snapshot does not exist """ cache_entry_id = "swh_snapshot_%s" % snapshot_id cache_entry = cache.get(cache_entry_id) if cache_entry: return cache_entry["branches"], cache_entry["releases"] branches: List[SnapshotBranchInfo] = [] releases: List[SnapshotReleaseInfo] = [] snapshot_content_max_size = get_config()["snapshot_content_max_size"] if snapshot_id: snapshot = service.lookup_snapshot( snapshot_id, branches_count=snapshot_content_max_size ) branches, releases = process_snapshot_branches(snapshot) cache.set(cache_entry_id, {"branches": branches, "releases": releases,}) return branches, releases def get_origin_visit_snapshot( origin_info: OriginInfo, visit_ts: Optional[Union[int, str]] = None, visit_id: Optional[int] = None, snapshot_id: Optional[str] = None, ) -> Tuple[List[SnapshotBranchInfo], List[SnapshotReleaseInfo]]: """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 date string or Unix timestamp 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 two 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. Raises: NotFoundExc if the origin or its visit are not found """ visit_info = get_origin_visit(origin_info, visit_ts, visit_id, snapshot_id) return get_snapshot_content(visit_info["snapshot"]) def get_snapshot_context( snapshot_id: Optional[str] = None, origin_url: Optional[str] = None, timestamp: Optional[str] = None, visit_id: Optional[int] = None, branch_name: Optional[str] = None, release_name: Optional[str] = None, revision_id: Optional[str] = None, path: Optional[str] = None, browse_context: str = "directory", ) -> SnapshotContext: """ Utility function to compute relevant information when navigating the archive in a snapshot context. The snapshot is either referenced by its id or it will be retrieved from an origin visit. Args: snapshot_id: hexadecimal representation of a snapshot identifier origin_url: an origin_url timestamp: a datetime string for retrieving the closest visit of the origin visit_id: optional visit id for disambiguation in case of several visits with the same timestamp branch_name: optional branch name set when browsing the snapshot in that scope (will default to "HEAD" if not provided) release_name: optional release name set when browsing the snapshot in that scope revision_id: optional revision identifier set when browsing the snapshot in that scope path: optional path of the object currently browsed in the snapshot browse_context: indicates which type of object is currently browsed Returns: A dict filled with snapshot context information. Raises: swh.web.common.exc.NotFoundExc: if no snapshot is found for the visit of an origin. """ assert origin_url is not None or snapshot_id is not None origin_info = None visit_info = None url_args = {} query_params: QueryParameters = {} origin_visits_url = None if origin_url: origin_info = service.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 = get_origin_visit_snapshot( origin_info, timestamp, visit_id, snapshot_id ) url_args = {"origin_url": origin_info["url"]} query_params = {} if visit_id is not None: query_params["visit_id"] = visit_id origin_visits_url = reverse("browse-origin-visits", url_args=url_args) if timestamp: url_args["timestamp"] = format_utc_iso_date(timestamp, "%Y-%m-%dT%H:%M:%SZ") visit_url = reverse( "browse-origin-directory", url_args=url_args, query_params=query_params ) visit_info["url"] = visit_url branches_url = reverse( "browse-origin-branches", url_args=url_args, query_params=query_params ) releases_url = reverse( "browse-origin-releases", url_args=url_args, query_params=query_params ) else: assert snapshot_id is not None branches, releases = get_snapshot_content(snapshot_id) url_args = {"snapshot_id": snapshot_id} branches_url = reverse("browse-snapshot-branches", url_args=url_args) releases_url = reverse("browse-snapshot-releases", url_args=url_args) releases = list(reversed(releases)) snapshot_sizes = service.lookup_snapshot_sizes(snapshot_id) is_empty = sum(snapshot_sizes.values()) == 0 swh_snp_id = persistent_identifier("snapshot", 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 = sum(snapshot_sizes.values()) if snapshot_total_size and revision_id is not None: revision = service.lookup_revision(revision_id) root_directory = revision["directory"] branches.append( SnapshotBranchInfo( name=revision_id, revision=revision_id, directory=root_directory, date=revision["date"], message=revision["message"], url=None, ) ) branch_name = revision_id query_params["revision"] = revision_id elif snapshot_total_size and release_name: release = _get_release(releases, release_name, snapshot_id) try: root_directory = release["directory"] revision_id = release["target"] release_id = release["id"] query_params["release"] = release_name except Exception as exc: sentry_sdk.capture_exception(exc) _branch_not_found( "release", release_name, snapshot_id, snapshot_sizes, origin_info, timestamp, visit_id, ) elif snapshot_total_size: if branch_name: query_params["branch"] = branch_name branch = _get_branch(branches, branch_name or "HEAD", snapshot_id) try: branch_name = branch["name"] revision_id = branch["revision"] root_directory = branch["directory"] except Exception as exc: sentry_sdk.capture_exception(exc) _branch_not_found( "branch", branch_name, snapshot_id, snapshot_sizes, origin_info, timestamp, visit_id, ) for b in branches: branch_url_args = dict(url_args) branch_query_params = dict(query_params) branch_query_params.pop("release", None) if b["name"] != b["revision"]: branch_query_params.pop("revision", None) branch_query_params["branch"] = b["name"] if path: branch_url_args["path"] = path b["url"] = reverse( browse_view_name, url_args=branch_url_args, query_params=branch_query_params ) for r in releases: release_url_args = dict(url_args) release_query_params = dict(query_params) release_query_params.pop("branch", None) release_query_params.pop("revision", None) release_query_params["release"] = r["name"] if path: release_url_args["path"] = path r["url"] = reverse( browse_view_name, url_args=release_url_args, query_params=release_query_params, ) return SnapshotContext( branch=branch_name, branches=branches, branches_url=branches_url, is_empty=is_empty, origin_info=origin_info, origin_visits_url=origin_visits_url, release=release_name, release_id=release_id, query_params=query_params, releases=releases, releases_url=releases_url, revision_id=revision_id, 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, ) def _build_breadcrumbs(snapshot_context: SnapshotContext, path: str): origin_info = snapshot_context["origin_info"] url_args = snapshot_context["url_args"] query_params = 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: breadcrumbs.append( { "name": root_directory[:7], "url": reverse( browse_view_name, url_args=url_args, query_params=query_params ), } ) for pi in path_info: bc_url_args = dict(url_args) bc_url_args["path"] = pi["path"] breadcrumbs.append( { "name": pi["name"], "url": reverse( browse_view_name, url_args=bc_url_args, query_params=query_params ), } ) return breadcrumbs def browse_snapshot_directory( request, snapshot_id=None, origin_url=None, timestamp=None, path=None ): """ Django view implementation for browsing a directory in a snapshot context. """ try: snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=timestamp, visit_id=request.GET.get("visit_id"), path=path, browse_context="directory", branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), revision_id=request.GET.get("revision"), ) root_directory = snapshot_context["root_directory"] sha1_git = root_directory if root_directory and path: dir_info = service.lookup_directory_with_path(root_directory, path) sha1_git = dir_info["target"] dirs = [] files = [] if sha1_git: dirs, files = get_directory_entries(sha1_git) except Exception as exc: return handle_view_exception(request, exc) origin_info = snapshot_context["origin_info"] visit_info = snapshot_context["visit_info"] url_args = snapshot_context["url_args"] query_params = 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: bc_url_args = dict(url_args) bc_url_args["path"] = path + d["name"] d["url"] = reverse( browse_view_name, url_args=bc_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: bc_url_args = dict(url_args) bc_url_args["path"] = path + f["name"] f["url"] = reverse( browse_view_name, url_args=bc_url_args, query_params=query_params ) if f["length"] is not None: sum_file_sizes += f["length"] f["length"] = filesizeformat(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: 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) sum_file_sizes = filesizeformat(sum_file_sizes) dir_path = "/" + path browse_dir_link = gen_directory_link(sha1_git) browse_rev_link = gen_revision_link(revision_id) browse_snp_link = gen_snapshot_link(snapshot_id) + revision_found = True + if sha1_git is None: + try: + service.lookup_revision(revision_id) + except NotFoundExc: + revision_found = False + dir_metadata = { "directory": sha1_git, "context-independent directory": browse_dir_link, "number of regular files": nb_files, "number of subdirectories": nb_dirs, "sum of regular file sizes": sum_file_sizes, "path": dir_path, "revision": revision_id, + "revision_found": revision_found, "context-independent revision": browse_rev_link, "snapshot": snapshot_id, "context-independent snapshot": browse_snp_link, } if origin_info: dir_metadata["origin url"] = origin_info["url"] dir_metadata["origin visit date"] = format_utc_iso_date(visit_info["date"]) dir_metadata["origin visit type"] = visit_info["type"] vault_cooking = { "directory_context": True, "directory_id": sha1_git, "revision_context": True, "revision_id": revision_id, } swh_objects = [ {"type": "directory", "id": sha1_git}, {"type": "revision", "id": revision_id}, {"type": "snapshot", "id": snapshot_id}, ] release_id = snapshot_context["release_id"] if release_id: swh_objects.append({"type": "release", "id": release_id}) browse_rel_link = gen_release_link(release_id) dir_metadata["release"] = release_id dir_metadata["context-independent release"] = browse_rel_link swh_ids = get_swh_persistent_ids(swh_objects, snapshot_context) 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"]: 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_menu": True, "swh_ids": swh_ids, }, ) def browse_snapshot_content( request, snapshot_id=None, origin_url=None, timestamp=None, path=None, selected_language=None, ): """ Django view implementation for browsing a content in a snapshot context. """ try: snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=timestamp, visit_id=request.GET.get("visit_id"), path=path, browse_context="content", branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), revision_id=request.GET.get("revision"), ) root_directory = snapshot_context["root_directory"] sha1_git = None query_string = None content_data = None directory_id = None split_path = path.split("/") filename = split_path[-1] filepath = path[: -len(filename)] if root_directory: content_info = service.lookup_directory_with_path(root_directory, path) sha1_git = content_info["target"] query_string = "sha1_git:" + sha1_git content_data = request_content(query_string, raise_if_unavailable=False) if filepath: dir_info = service.lookup_directory_with_path(root_directory, filepath) directory_id = dir_info["target"] else: directory_id = root_directory except Exception as exc: return handle_view_exception(request, exc) revision_id = snapshot_context["revision_id"] origin_info = snapshot_context["origin_info"] visit_info = snapshot_context["visit_info"] snapshot_id = snapshot_context["snapshot_id"] content = None language = None mimetype = None if content_data and content_data["raw_data"] is not None: content_display_data = prepare_content_for_display( content_data["raw_data"], content_data["mimetype"], path ) content = content_display_data["content_data"] language = content_display_data["language"] mimetype = content_display_data["mimetype"] # Override language with user-selected language if selected_language is not None: language = selected_language available_languages = None if mimetype and "text/" in mimetype: available_languages = highlightjs.get_supported_languages() breadcrumbs = _build_breadcrumbs(snapshot_context, filepath) breadcrumbs.append({"name": filename, "url": None}) browse_content_link = gen_content_link(sha1_git) content_raw_url = None if query_string: content_raw_url = reverse( "browse-content-raw", url_args={"query_string": query_string}, query_params={"filename": filename}, ) browse_rev_link = gen_revision_link(revision_id) browse_dir_link = gen_directory_link(directory_id) content_metadata = { "context-independent content": browse_content_link, "path": None, "filename": None, "directory": directory_id, "context-independent directory": browse_dir_link, "revision": revision_id, "context-independent revision": browse_rev_link, "snapshot": snapshot_id, } cnt_sha1_git = None content_size = None error_code = 200 error_description = "" error_message = "" if content_data: for checksum in content_data["checksums"].keys(): content_metadata[checksum] = content_data["checksums"][checksum] content_metadata["mimetype"] = content_data["mimetype"] content_metadata["encoding"] = content_data["encoding"] content_metadata["size"] = filesizeformat(content_data["length"]) content_metadata["language"] = content_data["language"] content_metadata["licenses"] = content_data["licenses"] content_metadata["path"] = "/" + filepath content_metadata["filename"] = filename cnt_sha1_git = content_data["checksums"]["sha1_git"] content_size = content_data["length"] error_code = content_data["error_code"] error_message = content_data["error_message"] error_description = content_data["error_description"] if origin_info: content_metadata["origin url"] = origin_info["url"] content_metadata["origin visit date"] = format_utc_iso_date(visit_info["date"]) content_metadata["origin visit type"] = visit_info["type"] browse_snapshot_link = gen_snapshot_link(snapshot_id) content_metadata["context-independent snapshot"] = browse_snapshot_link swh_objects = [ {"type": "content", "id": cnt_sha1_git}, {"type": "directory", "id": directory_id}, {"type": "revision", "id": revision_id}, {"type": "snapshot", "id": snapshot_id}, ] release_id = snapshot_context["release_id"] if release_id: swh_objects.append({"type": "release", "id": release_id}) browse_rel_link = gen_release_link(release_id) content_metadata["release"] = release_id content_metadata["context-independent release"] = browse_rel_link swh_ids = get_swh_persistent_ids(swh_objects, snapshot_context) content_path = "/".join([bc["name"] for bc in breadcrumbs]) context_found = "snapshot: %s" % snapshot_context["snapshot_id"] if origin_info: context_found = "origin: %s" % origin_info["url"] heading = "Content - %s - %s - %s" % ( content_path, snapshot_context["branch"], context_found, ) top_right_link = None if not snapshot_context["is_empty"]: top_right_link = { "url": content_raw_url, "icon": swh_object_icons["content"], "text": "Raw File", } return render( request, "browse/content.html", { "heading": heading, "swh_object_name": "Content", "swh_object_metadata": content_metadata, "content": content, "content_size": content_size, "max_content_size": content_display_max_size, "mimetype": mimetype, "language": language, "available_languages": available_languages, "breadcrumbs": breadcrumbs if root_directory else [], "top_right_link": top_right_link, "snapshot_context": snapshot_context, "vault_cooking": None, "show_actions_menu": True, "swh_ids": swh_ids, "error_code": error_code, "error_message": error_message, "error_description": error_description, }, status=error_code, ) PER_PAGE = 100 def browse_snapshot_log(request, snapshot_id=None, origin_url=None, timestamp=None): """ Django view implementation for browsing a revision history in a snapshot context. """ try: snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=timestamp, visit_id=request.GET.get("visit_id"), browse_context="log", branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), revision_id=request.GET.get("revision"), ) revision_id = snapshot_context["revision_id"] per_page = int(request.GET.get("per_page", PER_PAGE)) offset = int(request.GET.get("offset", 0)) revs_ordering = request.GET.get("revs_ordering", "committer_date") session_key = "rev_%s_log_ordering_%s" % (revision_id, revs_ordering) rev_log_session = request.session.get(session_key, None) rev_log = [] revs_walker_state = None if rev_log_session: rev_log = rev_log_session["rev_log"] revs_walker_state = rev_log_session["revs_walker_state"] if len(rev_log) < offset + per_page: revs_walker = service.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 = service.lookup_revision_multiple(revs) request.session[session_key] = { "rev_log": rev_log, "revs_walker_state": revs_walker_state, } except Exception as exc: return handle_view_exception(request, exc) origin_info = snapshot_context["origin_info"] visit_info = snapshot_context["visit_info"] url_args = snapshot_context["url_args"] query_params = snapshot_context["query_params"] snapshot_id = snapshot_context["snapshot_id"] query_params["per_page"] = per_page revs_ordering = request.GET.get("revs_ordering", "") query_params["revs_ordering"] = revs_ordering if origin_info: browse_view_name = "browse-origin-log" else: browse_view_name = "browse-snapshot-log" prev_log_url = None if len(rev_log) > offset + per_page: query_params["offset"] = offset + per_page prev_log_url = reverse( browse_view_name, url_args=url_args, query_params=query_params ) next_log_url = None if offset != 0: query_params["offset"] = offset - per_page next_log_url = reverse( browse_view_name, url_args=url_args, query_params=query_params ) revision_log_data = format_log_entries(revision_log, per_page, snapshot_context) browse_rev_link = gen_revision_link(revision_id) browse_log_link = gen_revision_log_link(revision_id) browse_snp_link = gen_snapshot_link(snapshot_id) revision_metadata = { "context-independent revision": browse_rev_link, "context-independent revision history": browse_log_link, "context-independent snapshot": browse_snp_link, "snapshot": snapshot_id, } if origin_info: revision_metadata["origin url"] = origin_info["url"] revision_metadata["origin visit date"] = format_utc_iso_date(visit_info["date"]) revision_metadata["origin visit type"] = visit_info["type"] swh_objects = [ {"type": "revision", "id": revision_id}, {"type": "snapshot", "id": snapshot_id}, ] release_id = snapshot_context["release_id"] if release_id: swh_objects.append({"type": "release", "id": release_id}) browse_rel_link = gen_release_link(release_id) revision_metadata["release"] = release_id revision_metadata["context-independent release"] = browse_rel_link swh_ids = get_swh_persistent_ids(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_menu": True, "swh_ids": swh_ids, }, ) def browse_snapshot_branches( request, snapshot_id=None, origin_url=None, timestamp=None ): """ Django view implementation for browsing a list of branches in a snapshot context. """ try: snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=timestamp, visit_id=request.GET.get("visit_id"), ) branches_bc = request.GET.get("branches_breadcrumbs", "") branches_bc = branches_bc.split(",") if branches_bc else [] branches_from = branches_bc[-1] if branches_bc else "" origin_info = snapshot_context["origin_info"] url_args = snapshot_context["url_args"] query_params = snapshot_context["query_params"] if origin_info: browse_view_name = "browse-origin-directory" else: browse_view_name = "browse-snapshot-directory" snapshot = service.lookup_snapshot( snapshot_context["snapshot_id"], branches_from, PER_PAGE + 1, target_types=["revision", "alias"], ) displayed_branches, _ = process_snapshot_branches(snapshot) except Exception as exc: return handle_view_exception(request, exc) for branch in displayed_branches: if snapshot_id: revision_url = reverse( "browse-revision", url_args={"sha1_git": branch["revision"]}, query_params={"snapshot_id": snapshot_id}, ) else: revision_url = reverse( "browse-revision", url_args={"sha1_git": branch["revision"]}, query_params={"origin": origin_info["url"]}, ) 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["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, }, ) def browse_snapshot_releases( request, snapshot_id=None, origin_url=None, timestamp=None ): """ Django view implementation for browsing a list of releases in a snapshot context. """ try: snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=timestamp, visit_id=request.GET.get("visit_id"), ) rel_bc = request.GET.get("releases_breadcrumbs", "") rel_bc = rel_bc.split(",") if rel_bc else [] rel_from = rel_bc[-1] if rel_bc else "" origin_info = snapshot_context["origin_info"] url_args = snapshot_context["url_args"] query_params = snapshot_context["query_params"] snapshot = service.lookup_snapshot( snapshot_context["snapshot_id"], rel_from, PER_PAGE + 1, target_types=["release", "alias"], ) _, displayed_releases = process_snapshot_branches(snapshot) except Exception as exc: return handle_view_exception(request, exc) for release in displayed_releases: if snapshot_id: query_params_tgt = {"snapshot_id": snapshot_id} else: query_params_tgt = {"origin": origin_info["url"]} release_url = reverse( "browse-release", url_args={"sha1_git": release["id"]}, query_params=query_params_tgt, ) target_url = "" 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, ) release["release_url"] = release_url release["target_url"] = target_url 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["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_menu": False, }, ) diff --git a/swh/web/templates/includes/directory-display.html b/swh/web/templates/includes/directory-display.html index e0d2c9e7..b6cf8b1c 100644 --- a/swh/web/templates/includes/directory-display.html +++ b/swh/web/templates/includes/directory-display.html @@ -1,58 +1,62 @@ {% comment %} Copyright (C) 2017-2019 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 %} {% if snapshot_context and snapshot_context.is_empty %} {% include "includes/empty-snapshot.html" %} {% elif dirs|length > 0 or files|length > 0 %}
{% for d in dirs %} {% endfor %} {% for f in files %} {% endfor %}
File Mode Size
{{ d.name }} {{ d.perms }}
{{ f.name }} {{ f.perms }} {{ f.length }}
+{% elif "revision_found" in swh_object_metadata and swh_object_metadata.revision_found is False %} + Revision {{ swh_object_metadata.revision }} could not be found in the archive. +
+ Its associated directory can not be displayed. {% elif dirs|length == 0 and files|length == 0 %} Directory is empty {% endif %} diff --git a/swh/web/tests/browse/views/test_origin.py b/swh/web/tests/browse/views/test_origin.py index 2d04857e..f53114bc 100644 --- a/swh/web/tests/browse/views/test_origin.py +++ b/swh/web/tests/browse/views/test_origin.py @@ -1,1063 +1,1102 @@ # 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 +from datetime import datetime import random import re import string from django.utils.html import escape from hypothesis import given from swh.model.hashutil import hash_to_bytes -from swh.model.model import Snapshot +from swh.model.model import ( + Snapshot, + SnapshotBranch, + TargetType, +) from swh.web.browse.snapshot_context import process_snapshot_branches from swh.web.common.exc import NotFoundExc from swh.web.common.identifiers import get_swh_persistent_id from swh.web.common.utils import ( reverse, gen_path_info, format_utc_iso_date, parse_timestamp, ) from swh.web.config import get_config from swh.web.tests.data import get_content, random_sha1 from swh.web.tests.django_asserts import assert_contains, assert_template_used from swh.web.tests.strategies import ( origin, origin_with_multiple_visits, new_origin, new_snapshot, visit_dates, revisions, origin_with_releases, release as existing_release, + unknown_revision, ) @given(origin_with_multiple_visits()) def test_origin_visits_browse(client, archive_data, origin): url = reverse("browse-origin-visits", url_args={"origin_url": origin["url"]}) resp = client.get(url) assert resp.status_code == 200 assert_template_used(resp, "browse/origin-visits.html") url = reverse("browse-origin-visits", url_args={"origin_url": origin["url"]}) resp = client.get(url) assert resp.status_code == 200 assert_template_used(resp, "browse/origin-visits.html") visits = archive_data.origin_visit_get(origin["url"]) for v in visits: vdate = format_utc_iso_date(v["date"], "%Y-%m-%dT%H:%M:%SZ") browse_dir_url = reverse( "browse-origin-directory", url_args={"origin_url": origin["url"], "timestamp": vdate}, ) assert_contains(resp, browse_dir_url) @given(origin_with_multiple_visits()) def test_origin_content_view(client, archive_data, origin): origin_visits = archive_data.origin_visit_get(origin["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], } tdata = _get_archive_data(-1) _origin_content_view_test_helper( client, origin, origin_visits, tdata["branches"], tdata["releases"], tdata["root_dir_sha1"], tdata["content"], ) _origin_content_view_test_helper( client, origin, origin_visits, tdata["branches"], tdata["releases"], tdata["root_dir_sha1"], tdata["content"], timestamp=tdata["visit"]["date"], ) visit_unix_ts = parse_timestamp(tdata["visit"]["date"]).timestamp() visit_unix_ts = int(visit_unix_ts) _origin_content_view_test_helper( client, origin, origin_visits, tdata["branches"], tdata["releases"], tdata["root_dir_sha1"], tdata["content"], timestamp=visit_unix_ts, ) tdata = _get_archive_data(0) _origin_content_view_test_helper( client, origin, origin_visits, tdata["branches"], tdata["releases"], tdata["root_dir_sha1"], tdata["content"], visit_id=tdata["visit"]["visit"], ) @given(origin()) def test_origin_root_directory_view(client, archive_data, origin): origin_visits = archive_data.origin_visit_get(origin["url"]) visit = origin_visits[-1] snapshot = archive_data.snapshot_get(visit["snapshot"]) head_rev_id = archive_data.snapshot_get_head(snapshot) head_rev = archive_data.revision_get(head_rev_id) root_dir_sha1 = head_rev["directory"] dir_content = archive_data.directory_ls(root_dir_sha1) branches, releases = process_snapshot_branches(snapshot) visit_unix_ts = parse_timestamp(visit["date"]).timestamp() visit_unix_ts = int(visit_unix_ts) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, dir_content ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, dir_content, visit_id=visit["visit"], ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, dir_content, timestamp=visit_unix_ts, ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, dir_content, timestamp=visit["date"], ) origin = dict(origin) del origin["type"] _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, dir_content ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, dir_content, visit_id=visit["visit"], ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, dir_content, timestamp=visit_unix_ts, ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, dir_content, timestamp=visit["date"], ) @given(origin()) def test_origin_sub_directory_view(client, archive_data, origin): origin_visits = archive_data.origin_visit_get(origin["url"]) visit = origin_visits[-1] snapshot = archive_data.snapshot_get(visit["snapshot"]) head_rev_id = archive_data.snapshot_get_head(snapshot) head_rev = archive_data.revision_get(head_rev_id) root_dir_sha1 = head_rev["directory"] subdirs = [ e for e in archive_data.directory_ls(root_dir_sha1) if e["type"] == "dir" ] branches, releases = process_snapshot_branches(snapshot) visit_unix_ts = parse_timestamp(visit["date"]).timestamp() visit_unix_ts = int(visit_unix_ts) if len(subdirs) == 0: return subdir = random.choice(subdirs) subdir_content = archive_data.directory_ls(subdir["target"]) subdir_path = subdir["name"] _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, subdir_content, path=subdir_path, ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, subdir_content, path=subdir_path, visit_id=visit["visit"], ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, subdir_content, path=subdir_path, timestamp=visit_unix_ts, ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, subdir_content, path=subdir_path, timestamp=visit["date"], ) origin = dict(origin) del origin["type"] _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, subdir_content, path=subdir_path, ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, subdir_content, path=subdir_path, visit_id=visit["visit"], ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, subdir_content, path=subdir_path, timestamp=visit_unix_ts, ) _origin_directory_view_test_helper( client, origin, origin_visits, branches, releases, root_dir_sha1, subdir_content, path=subdir_path, timestamp=visit["date"], ) @given(origin()) def test_origin_branches(client, archive_data, origin): origin_visits = archive_data.origin_visit_get(origin["url"]) visit = origin_visits[-1] snapshot = archive_data.snapshot_get(visit["snapshot"]) snapshot_content = process_snapshot_branches(snapshot) _origin_branches_test_helper(client, origin, snapshot_content) origin = dict(origin) origin["type"] = None _origin_branches_test_helper(client, origin, snapshot_content) @given(origin()) def test_origin_releases(client, archive_data, origin): origin_visits = archive_data.origin_visit_get(origin["url"]) visit = origin_visits[-1] snapshot = archive_data.snapshot_get(visit["snapshot"]) snapshot_content = process_snapshot_branches(snapshot) _origin_releases_test_helper(client, origin, snapshot_content) origin = dict(origin) origin["type"] = None _origin_releases_test_helper(client, origin, snapshot_content) @given( new_origin(), new_snapshot(min_size=4, max_size=4), visit_dates(), revisions(min_size=3, max_size=3), ) def test_origin_snapshot_null_branch( client, archive_data, new_origin, new_snapshot, visit_dates, revisions ): snp_dict = new_snapshot.to_dict() new_origin = archive_data.origin_add([new_origin])[0] for i, branch in enumerate(snp_dict["branches"].keys()): if i == 0: snp_dict["branches"][branch] = None else: snp_dict["branches"][branch] = { "target_type": "revision", "target": hash_to_bytes(revisions[i - 1]), } archive_data.snapshot_add([Snapshot.from_dict(snp_dict)]) visit = archive_data.origin_visit_add(new_origin["url"], visit_dates[0], type="git") archive_data.origin_visit_update( new_origin["url"], visit.visit, status="partial", snapshot=snp_dict["id"] ) url = reverse("browse-origin-directory", url_args={"origin_url": new_origin["url"]}) rv = client.get(url) assert rv.status_code == 200 @given( new_origin(), new_snapshot(min_size=4, max_size=4), visit_dates(), revisions(min_size=4, max_size=4), ) def test_origin_snapshot_invalid_branch( client, archive_data, new_origin, new_snapshot, visit_dates, revisions ): snp_dict = new_snapshot.to_dict() new_origin = archive_data.origin_add([new_origin])[0] for i, branch in enumerate(snp_dict["branches"].keys()): snp_dict["branches"][branch] = { "target_type": "revision", "target": hash_to_bytes(revisions[i]), } archive_data.snapshot_add([Snapshot.from_dict(snp_dict)]) visit = archive_data.origin_visit_add(new_origin["url"], visit_dates[0], type="git") archive_data.origin_visit_update( new_origin["url"], visit.visit, status="full", snapshot=snp_dict["id"] ) url = reverse( "browse-origin-directory", url_args={"origin_url": new_origin["url"]}, query_params={"branch": "invalid_branch"}, ) rv = client.get(url) assert rv.status_code == 404 @given(new_origin()) def test_browse_visits_origin_not_found(client, new_origin): url = reverse("browse-origin-visits", url_args={"origin_url": new_origin.url}) resp = client.get(url) assert resp.status_code == 404 assert_template_used(resp, "error.html") assert_contains( resp, f"Origin with url {new_origin.url} not found", status_code=404 ) @given(origin()) def test_browse_origin_directory_no_visit(client, mocker, origin): mock_get_origin_visits = mocker.patch( "swh.web.common.origin_visits.get_origin_visits" ) mock_get_origin_visits.return_value = [] url = reverse("browse-origin-directory", url_args={"origin_url": origin["url"]}) resp = client.get(url) assert resp.status_code == 404 assert_template_used(resp, "error.html") assert_contains(resp, "No visit", status_code=404) assert mock_get_origin_visits.called @given(origin()) def test_browse_origin_directory_unknown_visit(client, mocker, origin): mock_get_origin_visits = mocker.patch( "swh.web.common.origin_visits.get_origin_visits" ) mock_get_origin_visits.return_value = [{"visit": 1}] url = reverse( "browse-origin-directory", url_args={"origin_url": origin["url"]}, query_params={"visit_id": 2}, ) resp = client.get(url) assert resp.status_code == 404 assert_template_used(resp, "error.html") assert re.search("Visit.*not found", resp.content.decode("utf-8")) assert mock_get_origin_visits.called @given(origin()) def test_browse_origin_directory_not_found(client, origin): url = reverse( "browse-origin-directory", url_args={"origin_url": origin["url"], "path": "/invalid/dir/path/"}, ) resp = client.get(url) assert resp.status_code == 404 assert_template_used(resp, "error.html") assert re.search("Directory.*not found", resp.content.decode("utf-8")) @given(origin()) def test_browse_origin_content_no_visit(client, mocker, origin): mock_get_origin_visits = mocker.patch( "swh.web.common.origin_visits.get_origin_visits" ) mock_get_origin_visits.return_value = [] url = reverse( "browse-origin-content", url_args={"origin_url": origin["url"], "path": "foo"} ) resp = client.get(url) assert resp.status_code == 404 assert_template_used(resp, "error.html") assert_contains(resp, "No visit", status_code=404) assert mock_get_origin_visits.called @given(origin()) def test_browse_origin_content_unknown_visit(client, mocker, origin): mock_get_origin_visits = mocker.patch( "swh.web.common.origin_visits.get_origin_visits" ) mock_get_origin_visits.return_value = [{"visit": 1}] url = reverse( "browse-origin-content", url_args={"origin_url": origin["url"], "path": "foo"}, query_params={"visit_id": 2}, ) resp = client.get(url) assert resp.status_code == 404 assert_template_used(resp, "error.html") assert re.search("Visit.*not found", resp.content.decode("utf-8")) assert mock_get_origin_visits.called @given(origin()) def test_browse_origin_content_empty_snapshot(client, mocker, origin): mock_snapshot_service = mocker.patch("swh.web.browse.snapshot_context.service") mock_get_origin_visit_snapshot = mocker.patch( "swh.web.browse.snapshot_context.get_origin_visit_snapshot" ) mock_get_origin_visit_snapshot.return_value = ([], []) mock_snapshot_service.lookup_origin.return_value = origin mock_snapshot_service.lookup_snapshot_sizes.return_value = { "revision": 0, "release": 0, } url = reverse( "browse-origin-content", url_args={"origin_url": origin["url"], "path": "baz"} ) resp = client.get(url) assert resp.status_code == 200 assert_template_used(resp, "browse/content.html") assert re.search("snapshot.*is empty", resp.content.decode("utf-8")) assert mock_get_origin_visit_snapshot.called assert mock_snapshot_service.lookup_origin.called assert mock_snapshot_service.lookup_snapshot_sizes.called @given(origin()) def test_browse_origin_content_not_found(client, origin): url = reverse( "browse-origin-content", url_args={"origin_url": origin["url"], "path": "/invalid/file/path"}, ) resp = client.get(url) assert resp.status_code == 404 assert_template_used(resp, "error.html") assert re.search("Directory entry.*not found", resp.content.decode("utf-8")) @given(origin()) def test_browse_directory_snapshot_not_found(client, mocker, origin): mock_get_snapshot_context = mocker.patch( "swh.web.browse.snapshot_context.get_snapshot_context" ) mock_get_snapshot_context.side_effect = NotFoundExc("Snapshot not found") url = reverse("browse-origin-directory", url_args={"origin_url": origin["url"]}) resp = client.get(url) assert resp.status_code == 404 assert_template_used(resp, "error.html") assert_contains(resp, "Snapshot not found", status_code=404) assert mock_get_snapshot_context.called @given(origin()) def test_origin_empty_snapshot(client, mocker, origin): mock_service = mocker.patch("swh.web.browse.snapshot_context.service") mock_get_origin_visit_snapshot = mocker.patch( "swh.web.browse.snapshot_context.get_origin_visit_snapshot" ) mock_get_origin_visit_snapshot.return_value = ([], []) mock_service.lookup_snapshot_sizes.return_value = { "revision": 0, "release": 0, } mock_service.lookup_origin.return_value = origin url = reverse("browse-origin-directory", url_args={"origin_url": origin["url"]}) resp = client.get(url) assert resp.status_code == 200 assert_template_used(resp, "browse/directory.html") resp_content = resp.content.decode("utf-8") assert re.search("snapshot.*is empty", resp_content) assert not re.search("swh-tr-link", resp_content) assert mock_get_origin_visit_snapshot.called assert mock_service.lookup_snapshot_sizes.called @given(origin_with_releases()) def test_origin_release_browse(client, archive_data, origin): # for swh.web.browse.snapshot_context.get_snapshot_content to only return one branch config = get_config() snapshot_max_size = int(config["snapshot_content_max_size"]) config["snapshot_content_max_size"] = 1 try: snapshot = archive_data.snapshot_get_latest(origin["url"]) release = [ b for b in snapshot["branches"].values() if b["target_type"] == "release" ][-1] release_data = archive_data.release_get(release["target"]) url = reverse( "browse-origin-directory", url_args={"origin_url": origin["url"]}, query_params={"release": release_data["name"]}, ) resp = client.get(url) assert resp.status_code == 200 assert_contains(resp, release_data["name"]) assert_contains(resp, release["target"]) finally: config["snapshot_content_max_size"] = snapshot_max_size @given(origin_with_releases()) def test_origin_release_browse_not_found(client, origin): invalid_release_name = "swh-foo-bar" url = reverse( "browse-origin-directory", url_args={"origin_url": origin["url"]}, query_params={"release": invalid_release_name}, ) resp = client.get(url) assert resp.status_code == 404 assert re.search( f"Release {invalid_release_name}.*not found", resp.content.decode("utf-8") ) +@given(new_origin(), unknown_revision()) +def test_origin_browse_directory_branch_with_non_resolvable_revision( + client, archive_data, new_origin, unknown_revision +): + branch_name = "master" + snapshot = Snapshot( + branches={ + branch_name.encode(): SnapshotBranch( + target=hash_to_bytes(unknown_revision), target_type=TargetType.REVISION, + ) + } + ) + new_origin = archive_data.origin_add([new_origin])[0] + archive_data.snapshot_add([snapshot]) + visit = archive_data.origin_visit_add(new_origin["url"], datetime.now(), type="git") + archive_data.origin_visit_update( + new_origin["url"], visit.visit, status="full", snapshot=snapshot.id + ) + + url = reverse( + "browse-origin-directory", + url_args={"origin_url": new_origin["url"]}, + query_params={"branch": branch_name}, + ) + + resp = client.get(url) + + assert resp.status_code == 200 + assert_contains( + resp, f"Revision {unknown_revision } could not be found in the archive." + ) + + def _origin_content_view_test_helper( client, origin_info, origin_visits, origin_branches, origin_releases, root_dir_sha1, content, visit_id=None, timestamp=None, ): content_path = "/".join(content["path"].split("/")[1:]) url_args = {"origin_url": origin_info["url"], "path": content_path} if not visit_id: visit_id = origin_visits[-1]["visit"] query_params = {} if timestamp: url_args["timestamp"] = timestamp if visit_id: query_params["visit_id"] = visit_id url = reverse("browse-origin-content", url_args=url_args, query_params=query_params) resp = client.get(url) assert resp.status_code == 200 assert_template_used(resp, "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 url_args["path"] if timestamp: url_args["timestamp"] = format_utc_iso_date( parse_timestamp(timestamp).isoformat(), "%Y-%m-%dT%H:%M:%SZ" ) root_dir_url = reverse( "browse-origin-directory", url_args=url_args, 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: url_args["path"] = p["path"] dir_url = reverse( "browse-origin-directory", url_args=url_args, 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 "args" in url_args: del url_args["path"] origin_branches_url = reverse( "browse-origin-branches", url_args=url_args, query_params=query_params ) assert_contains( resp, 'Branches (%s)' % (origin_branches_url, len(origin_branches)), ) origin_releases_url = reverse( "browse-origin-releases", url_args=url_args, query_params=query_params ) assert_contains( resp, 'Releases (%s)' % (origin_releases_url, len(origin_releases)), ) assert_contains(resp, '
  • ', count=len(origin_branches)) url_args["path"] = content_path for branch in origin_branches: query_params["branch"] = branch["name"] root_dir_branch_url = reverse( "browse-origin-content", url_args=url_args, query_params=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: query_params["release"] = release["name"] root_dir_release_url = reverse( "browse-origin-content", url_args=url_args, query_params=query_params ) assert_contains(resp, '' % root_dir_release_url) url = reverse("browse-origin-content", url_args=url_args, query_params=query_params) resp = client.get(url) assert resp.status_code == 200 assert_template_used(resp, "browse/content.html") swh_cnt_id = get_swh_persistent_id("content", content["sha1_git"]) swh_cnt_id_url = reverse("browse-swh-id", url_args={"swh_id": swh_cnt_id}) assert_contains(resp, swh_cnt_id) assert_contains(resp, swh_cnt_id_url) assert_contains(resp, "swh-take-new-snapshot") def _origin_directory_view_test_helper( client, origin_info, origin_visits, origin_branches, origin_releases, root_directory_sha1, directory_entries, visit_id=None, timestamp=None, path=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"] if not visit_id: visit_id = origin_visits[-1]["visit"] url_args = {"origin_url": origin_info["url"]} query_params = {} if timestamp: url_args["timestamp"] = timestamp else: query_params["visit_id"] = visit_id if path: url_args["path"] = path url = reverse( "browse-origin-directory", url_args=url_args, query_params=query_params ) resp = client.get(url) assert resp.status_code == 200 assert_template_used(resp, "browse/directory.html") assert resp.status_code == 200 assert_template_used(resp, "browse/directory.html") assert_contains(resp, '', count=len(dirs)) assert_contains(resp, '', count=len(files)) if timestamp: url_args["timestamp"] = format_utc_iso_date( parse_timestamp(timestamp).isoformat(), "%Y-%m-%dT%H:%M:%SZ" ) 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_args = dict(url_args) dir_url_args["path"] = dir_path dir_url = reverse( "browse-origin-directory", url_args=dir_url_args, query_params=query_params, ) assert_contains(resp, dir_url) for f in files: file_path = f["name"] if path: file_path = "%s/%s" % (path, f["name"]) file_url_args = dict(url_args) file_url_args["path"] = file_path file_url = reverse( "browse-origin-content", url_args=file_url_args, query_params=query_params ) assert_contains(resp, file_url) if "path" in url_args: del url_args["path"] root_dir_branch_url = reverse( "browse-origin-directory", url_args=url_args, query_params=query_params ) nb_bc_paths = 1 if path: nb_bc_paths = len(path.split("/")) + 1 assert_contains(resp, '
  • ', count=nb_bc_paths) assert_contains( resp, '%s' % (root_dir_branch_url, root_directory_sha1[:7]) ) origin_branches_url = reverse( "browse-origin-branches", url_args=url_args, query_params=query_params ) assert_contains( resp, 'Branches (%s)' % (origin_branches_url, len(origin_branches)), ) origin_releases_url = reverse( "browse-origin-releases", url_args=url_args, query_params=query_params ) nb_releases = len(origin_releases) if nb_releases > 0: assert_contains( resp, 'Releases (%s)' % (origin_releases_url, nb_releases) ) if path: url_args["path"] = path assert_contains(resp, '
  • ', count=len(origin_branches)) for branch in origin_branches: query_params["branch"] = branch["name"] root_dir_branch_url = reverse( "browse-origin-directory", url_args=url_args, query_params=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: query_params["release"] = release["name"] root_dir_release_url = reverse( "browse-origin-directory", url_args=url_args, query_params=query_params ) assert_contains(resp, '' % root_dir_release_url) assert_contains(resp, "vault-cook-directory") assert_contains(resp, "vault-cook-revision") swh_dir_id = get_swh_persistent_id("directory", directory_entries[0]["dir_id"]) swh_dir_id_url = reverse("browse-swh-id", url_args={"swh_id": swh_dir_id}) assert_contains(resp, swh_dir_id) assert_contains(resp, swh_dir_id_url) assert_contains(resp, "swh-take-new-snapshot") def _origin_branches_test_helper(client, origin_info, origin_snapshot): url_args = {"origin_url": origin_info["url"]} url = reverse("browse-origin-branches", url_args=url_args) resp = client.get(url) assert resp.status_code == 200 assert_template_used(resp, "browse/branches.html") origin_branches = origin_snapshot[0] origin_releases = origin_snapshot[1] origin_branches_url = reverse("browse-origin-branches", url_args=url_args) assert_contains( resp, 'Branches (%s)' % (origin_branches_url, len(origin_branches)), ) origin_releases_url = reverse("browse-origin-releases", url_args=url_args) nb_releases = len(origin_releases) if nb_releases > 0: assert_contains( resp, 'Releases (%s)' % (origin_releases_url, nb_releases) ) assert_contains(resp, '' % escape(browse_branch_url)) browse_revision_url = reverse( "browse-revision", url_args={"sha1_git": branch["revision"]}, query_params={"origin": origin_info["url"]}, ) assert_contains(resp, '' % escape(browse_revision_url)) def _origin_releases_test_helper(client, origin_info, origin_snapshot): url_args = {"origin_url": origin_info["url"]} url = reverse("browse-origin-releases", url_args=url_args) resp = client.get(url) assert resp.status_code == 200 assert_template_used(resp, "browse/releases.html") origin_branches = origin_snapshot[0] origin_releases = origin_snapshot[1] origin_branches_url = reverse("browse-origin-branches", url_args=url_args) assert_contains( resp, 'Branches (%s)' % (origin_branches_url, len(origin_branches)), ) origin_releases_url = reverse("browse-origin-releases", url_args=url_args) nb_releases = len(origin_releases) if nb_releases > 0: assert_contains( resp, 'Releases (%s)' % (origin_releases_url, nb_releases) ) assert_contains(resp, '' % escape(browse_release_url)) assert_contains(resp, '' % escape(browse_revision_url)) @given( new_origin(), visit_dates(), revisions(min_size=10, max_size=10), existing_release() ) def test_origin_branches_pagination_with_alias( client, archive_data, mocker, new_origin, visit_dates, revisions, existing_release ): """ When a snapshot contains a branch or a release alias, pagination links in the branches / releases view should be displayed. """ 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 = "".join(random.choices(string.ascii_lowercase, k=8)) snp_dict["branches"][b"RELEASE_ALIAS"] = { "target_type": "alias", "target": release.encode(), } snp_dict["branches"][release.encode()] = { "target_type": "release", "target": hash_to_bytes(existing_release), } new_origin = archive_data.origin_add([new_origin])[0] archive_data.snapshot_add([Snapshot.from_dict(snp_dict)]) visit = archive_data.origin_visit_add(new_origin["url"], visit_dates[0], type="git") archive_data.origin_visit_update( new_origin["url"], visit.visit, status="full", snapshot=snp_dict["id"] ) url = reverse("browse-origin-branches", url_args={"origin_url": new_origin["url"]}) resp = client.get(url) assert resp.status_code == 200 assert_template_used(resp, "browse/branches.html") assert_contains(resp, '