diff --git a/swh/web/browse/snapshot_context.py b/swh/web/browse/snapshot_context.py index 1baf66b7..56c8ddc7 100644 --- a/swh/web/browse/snapshot_context.py +++ b/swh/web/browse/snapshot_context.py @@ -1,1379 +1,1379 @@ # 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.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"], # 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" 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": branches[branch_name] = SnapshotBranchInfo( name=branch_name, alias=False, revision=target_id, date=None, directory=None, message=None, url=None, ) revision_to_branch[target_id].add(branch_name) elif target_type == "release": release_to_branch[target_id].add(branch_name) elif target_type == "alias": branch_aliases[branch_name] = target_id # FIXME: handle pointers to other object types def _add_release_info(branch, release, alias=False): releases[branch] = SnapshotReleaseInfo( name=release["name"], alias=alias, branch_name=branch, date=format_utc_iso_date(release["date"]), directory=None, id=release["id"], message=release["message"], target_type=release["target_type"], target=release["target"], url=None, ) def _add_branch_info(branch, revision, alias=False): branches[branch] = SnapshotBranchInfo( name=branch, alias=alias, revision=revision["id"], directory=revision["directory"], date=format_utc_iso_date(revision["date"]), message=revision["message"], url=None, ) releases_info = archive.lookup_release_multiple(release_to_branch.keys()) for release in releases_info: if release is None: continue branches_to_update = release_to_branch[release["id"]] for branch in branches_to_update: _add_release_info(branch, release) if release["target_type"] == "revision": revision_to_release[release["target"]].update(branches_to_update) revisions = archive.lookup_revision_multiple( set(revision_to_branch.keys()) | set(revision_to_release.keys()) ) for revision in revisions: if not revision: continue for branch in revision_to_branch[revision["id"]]: _add_branch_info(branch, revision) for release_id in revision_to_release[revision["id"]]: releases[release_id]["directory"] = revision["directory"] resolved_aliases = {} for branch_alias, branch_target in branch_aliases.items(): resolved_alias = archive.lookup_snapshot_alias(snapshot["id"], branch_alias) resolved_aliases[branch_alias] = resolved_alias if resolved_alias is None: continue target_type = resolved_alias["target_type"] target = resolved_alias["target"] if target_type == "revision": revision = archive.lookup_revision(target) _add_branch_info(branch_alias, revision, alias=True) elif target_type == "release": release = archive.lookup_release(target) _add_release_info(branch_alias, release, alias=True) if branch_alias in branches: branches[branch_alias]["name"] = branch_alias ret_branches = list(sorted(branches.values(), key=lambda b: b["name"])) ret_releases = list(sorted(releases.values(), key=lambda b: b["name"])) return ret_branches, ret_releases, resolved_aliases @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 branches_url = reverse("browse-origin-branches", query_params=query_params) releases_url = reverse("browse-origin-releases", query_params=query_params) else: assert snapshot_id is not None branches, releases, aliases = get_snapshot_content(snapshot_id) url_args = {"snapshot_id": snapshot_id} directory_url = reverse("browse-snapshot-directory", url_args=url_args) branches_url = reverse("browse-snapshot-branches", url_args=url_args) releases_url = reverse("browse-snapshot-releases", url_args=url_args) releases = list(reversed(releases)) @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 swh_snp_id = str( CoreSWHID(object_type=ObjectType.SNAPSHOT, object_id=hash_to_bytes(snapshot_id)) ) if visit_info: timestamp = format_utc_iso_date(visit_info["date"]) if origin_info: browse_view_name = f"browse-origin-{browse_context}" else: browse_view_name = f"browse-snapshot-{browse_context}" release_id = None root_directory = None snapshot_total_size = snapshot_sizes["release"] + snapshot_sizes["revision"] if path is not None: query_params["path"] = path if snapshot_total_size and revision_id is not None: # browse specific revision for a snapshot requested revision = archive.lookup_revision(revision_id) root_directory = revision["directory"] branches.append( SnapshotBranchInfo( name=revision_id, alias=False, revision=revision_id, directory=root_directory, date=revision["date"], message=revision["message"], url=None, ) ) query_params["revision"] = revision_id elif snapshot_total_size and release_name: # browse specific release for a snapshot requested release = _get_release(releases, release_name, snapshot_id) if release is None: _branch_not_found( "release", release_name, snapshot_id, snapshot_sizes, origin_info, timestamp, visit_id, ) else: 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"] elif head is not None: # otherwise, browse branch targeted by the HEAD alias if it exists if head["target_type"] == "revision": # HEAD alias targets a revision head_rev = archive.lookup_revision(head["target"]) branch_name = "HEAD" revision_id = head_rev["id"] root_directory = head_rev["directory"] else: # HEAD alias targets a release release_name = archive.lookup_release(head["target"])["name"] head_rel = _get_release(releases, release_name, snapshot_id) if head_rel 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"] elif releases: # fallback to browse last release otherwise release = releases[-1] if release["target_type"] == "revision": revision = archive.lookup_revision(release["target"]) root_directory = revision["directory"] revision_id = release["target"] elif release["target_type"] == "directory": root_directory = release["target"] release_id = release["id"] release_name = release["name"] for b in branches: branch_query_params = dict(query_params) branch_query_params.pop("release", None) if b["name"] != b["revision"]: branch_query_params.pop("revision", None) branch_query_params["branch"] = b["name"] b["url"] = reverse( browse_view_name, url_args=url_args, query_params=branch_query_params ) for r in releases: release_query_params = dict(query_params) release_query_params.pop("branch", None) release_query_params.pop("revision", None) release_query_params["release"] = r["name"] r["url"] = reverse( browse_view_name, url_args=url_args, query_params=release_query_params, ) revision_info = None if revision_id: try: revision_info = archive.lookup_revision(revision_id) except NotFoundExc: pass else: revision_info["date"] = format_utc_iso_date(revision_info["date"]) revision_info["committer_date"] = format_utc_iso_date( revision_info["committer_date"] ) if revision_info["message"]: message_lines = revision_info["message"].split("\n") revision_info["message_header"] = message_lines[0] else: revision_info["message_header"] = "" snapshot_context = SnapshotContext( directory_url=directory_url, branch=branch_name, branch_alias=branch_name in aliases, branches=branches, branches_url=branches_url, is_empty=is_empty, origin_info=origin_info, origin_visits_url=origin_visits_url, release=release_name, release_alias=release_name in aliases, release_id=release_id, query_params=query_params, releases=releases, releases_url=releases_url, revision_id=revision_id, revision_info=revision_info, root_directory=root_directory, snapshot_id=snapshot_id, snapshot_sizes=snapshot_sizes, snapshot_swhid=swh_snp_id, url_args=url_args, visit_info=visit_info, ) if revision_info: revision_info["revision_url"] = gen_revision_url( revision_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"] 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", + "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", + "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"]}, 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", + "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", + "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/templates/browse/branches.html b/swh/web/browse/templates/browse-branches.html similarity index 96% rename from swh/web/templates/browse/branches.html rename to swh/web/browse/templates/browse-branches.html index ab84814b..7a01cf35 100644 --- a/swh/web/templates/browse/branches.html +++ b/swh/web/browse/templates/browse-branches.html @@ -1,87 +1,87 @@ {% extends "./browse.html" %} {% comment %} -Copyright (C) 2017-2019 The Software Heritage developers +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 %} {% block swh-browse-content %}
- {% include "includes/branch-search.html" %} + {% include "./includes/branch-search.html" %}
{% if displayed_branches|length > 0 %} {% for branch in displayed_branches %} {% endfor %} {% else %} {% endif %}
Name Revision Message Date
{% if branch.alias %} {% else %} {% endif %} {{ branch.name }} {{ branch.revision|slice:":7" }} {{ branch.message }} {{ branch.date }}
{% if search_string %} No branch names containing {{search_string}} have been found! {% else %} The list of branches is empty! {% endif %}
{% endblock %} {% block swh-browse-after-content %} {% if prev_branches_url or next_branches_url %} {% endif %} {% endblock %} diff --git a/swh/web/templates/browse/content.html b/swh/web/browse/templates/browse-content.html similarity index 66% rename from swh/web/templates/browse/content.html rename to swh/web/browse/templates/browse-content.html index 9e06fde7..d20b27cf 100644 --- a/swh/web/templates/browse/content.html +++ b/swh/web/browse/templates/browse-content.html @@ -1,13 +1,13 @@ {% extends "./browse.html" %} {% comment %} -Copyright (C) 2017-2018 The Software Heritage developers +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 %} {% block swh-browse-content %} -{% include "includes/top-navigation.html" %} -{% include "includes/content-display.html" %} +{% include "./includes/top-navigation.html" %} +{% include "./includes/content-display.html" %} {% endblock %} diff --git a/swh/web/templates/browse/directory.html b/swh/web/browse/templates/browse-directory.html similarity index 63% rename from swh/web/templates/browse/directory.html rename to swh/web/browse/templates/browse-directory.html index 7b35c76e..97818a4c 100644 --- a/swh/web/templates/browse/directory.html +++ b/swh/web/browse/templates/browse-directory.html @@ -1,17 +1,17 @@ {% extends "./browse.html" %} {% comment %} -Copyright (C) 2017-2018 The Software Heritage developers +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 %} {% block swh-browse-content %} -{% include "includes/top-navigation.html" %} -{% include "includes/directory-display.html" %} +{% include "./includes/top-navigation.html" %} +{% include "./includes/directory-display.html" %} {% endblock %} {% block swh-browse-after-content %} -{% include "includes/readme-display.html" %} +{% include "./includes/readme-display.html" %} {% endblock %} diff --git a/swh/web/templates/browse/help.html b/swh/web/browse/templates/browse-help.html similarity index 100% rename from swh/web/templates/browse/help.html rename to swh/web/browse/templates/browse-help.html diff --git a/swh/web/templates/browse/origin-visits.html b/swh/web/browse/templates/browse-origin-visits.html similarity index 100% rename from swh/web/templates/browse/origin-visits.html rename to swh/web/browse/templates/browse-origin-visits.html diff --git a/swh/web/templates/browse/release.html b/swh/web/browse/templates/browse-release.html similarity index 90% rename from swh/web/templates/browse/release.html rename to swh/web/browse/templates/browse-release.html index 8bdbd55c..652c7474 100644 --- a/swh/web/templates/browse/release.html +++ b/swh/web/browse/templates/browse-release.html @@ -1,30 +1,30 @@ {% extends "./browse.html" %} {% comment %} -Copyright (C) 2017-2019 The Software Heritage developers +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 %} {% block swh-browse-content %} -{% include "includes/top-navigation.html" %} +{% include "./includes/top-navigation.html" %}
Release {{ swh_object_metadata.name }} created by {{ swh_object_metadata.author_url }} on {{ swh_object_metadata.date }}
 
{{ release.note_header }}
{{ release.note_body }}
Target: {{ release.target_link}} {% if release.directory_link %}
Directory: {{ release.directory_link}} {% endif %}
{% endblock %} diff --git a/swh/web/templates/browse/releases.html b/swh/web/browse/templates/browse-releases.html similarity index 96% rename from swh/web/templates/browse/releases.html rename to swh/web/browse/templates/browse-releases.html index beb74425..8341336a 100644 --- a/swh/web/templates/browse/releases.html +++ b/swh/web/browse/templates/browse-releases.html @@ -1,85 +1,85 @@ {% extends "./browse.html" %} {% comment %} -Copyright (C) 2017-2021 The Software Heritage developers +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 %} {% block swh-browse-content %}
- {% include "includes/branch-search.html" %} + {% include "./includes/branch-search.html" %}
{% if displayed_releases|length > 0 %} {% for release in displayed_releases %} {% endfor %} {% else %} {% endif %}
Name Target Message Date
{% if release.alias %} {% else %} {% endif %} {{ release.name }} {{ release.target|slice:":7" }} {{ release.message }} {{ release.date }}
{% if search_string %} No release names containing {{search_string}} have been found! {% else %} The list of releases is empty! {% endif %}
{% endblock %} {% block swh-browse-after-content %} {% if prev_releases_url or next_releases_url %} {% endif %} {% endblock %} diff --git a/swh/web/templates/browse/revision-log.html b/swh/web/browse/templates/browse-revision-log.html similarity index 96% rename from swh/web/templates/browse/revision-log.html rename to swh/web/browse/templates/browse-revision-log.html index 8863d189..512a6c6a 100644 --- a/swh/web/templates/browse/revision-log.html +++ b/swh/web/browse/templates/browse-revision-log.html @@ -1,119 +1,119 @@ {% extends "./browse.html" %} {% comment %} -Copyright (C) 2017-2019 The Software Heritage developers +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 render_bundle from webpack_loader %} {% load swh_templatetags %} {% block header %} {{ block.super }} {% render_bundle 'revision' %} {% endblock %} {% block swh-browse-content %} {% if snapshot_context %} - {% include "includes/top-navigation.html" %} + {% include "./includes/top-navigation.html" %} {% endif %} {% if snapshot_context and snapshot_context.is_empty %} - {% include "includes/empty-snapshot.html" %} + {% include "./includes/empty-snapshot.html" %} {% else %}
sort by:
{% for rev in revision_log %} {% endfor %}
Revision Author Date Message Commit Date
{{ rev.id }} {{ rev.message }} {{ rev.commit_date }}
{% endif %} {% endblock %} {% block swh-browse-after-content %} {% if not snapshot_context or not snapshot_context.is_empty %} {% endif %} {% endblock %} diff --git a/swh/web/templates/browse/revision.html b/swh/web/browse/templates/browse-revision.html similarity index 94% rename from swh/web/templates/browse/revision.html rename to swh/web/browse/templates/browse-revision.html index 37c6fb10..34e80306 100644 --- a/swh/web/templates/browse/revision.html +++ b/swh/web/browse/templates/browse-revision.html @@ -1,111 +1,111 @@ {% extends "./browse.html" %} {% comment %} -Copyright (C) 2017-2019 The Software Heritage developers +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 static %} {% load swh_templatetags %} {% load render_bundle from webpack_loader %} {% block header %} {{ block.super }} {% render_bundle 'revision' %} {% endblock %} {% block swh-browse-content %}
Revision {{ swh_object_metadata.revision }} authored by {{ swh_object_metadata.author_url }} on {{ swh_object_metadata.date }}, committed by {{ swh_object_metadata.committer_url }} on {{ swh_object_metadata.committer_date }}
{{ message_header }}
{% if message_body %}
{{ message_body }}
{% endif %}
{{ parents|length }} parent{% if parents|length > 1 %}s{% endif %} {% for parent in parents %} {{ parent.id|slice:":7" }} {% if not forloop.last %} + {% endif %} {% endfor %}
- {% include "includes/top-navigation.html" %} + {% include "./includes/top-navigation.html" %} {% if content_size %} - {% include "includes/content-display.html" %} + {% include "./includes/content-display.html" %} {% else %} - {% include "includes/directory-display.html" %} + {% include "./includes/directory-display.html" %} {% endif %}
{% endblock %} {% block swh-browse-after-content %} -{% include "includes/readme-display.html" %} +{% include "./includes/readme-display.html" %} {% endblock %} diff --git a/swh/web/templates/browse/search.html b/swh/web/browse/templates/browse-search.html similarity index 90% rename from swh/web/templates/browse/search.html rename to swh/web/browse/templates/browse-search.html index d557d2b1..42ff8d58 100644 --- a/swh/web/templates/browse/search.html +++ b/swh/web/browse/templates/browse-search.html @@ -1,52 +1,52 @@ -{% extends "./layout.html" %} +{% extends "./browse.html" %} {% comment %} -Copyright (C) 2017-2020 The Software Heritage developers +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 static %} {% block navbar-content %}

Search archived software

{% endblock %} {% block browse-content %} -{% include "includes/origin-search-form.html" %} +{% include "./includes/origin-search-form.html" %}

Searching origins ...

{% endblock %} diff --git a/swh/web/templates/browse/vault-ui.html b/swh/web/browse/templates/browse-vault-ui.html similarity index 92% rename from swh/web/templates/browse/vault-ui.html rename to swh/web/browse/templates/browse-vault-ui.html index 2e82539a..2c729755 100644 --- a/swh/web/templates/browse/vault-ui.html +++ b/swh/web/browse/templates/browse-vault-ui.html @@ -1,51 +1,51 @@ -{% extends "./layout.html" %} +{% extends "./browse.html" %} {% comment %} -Copyright (C) 2017-2019 The Software Heritage developers +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 render_bundle from webpack_loader %} {% block navbar-content %}

Download archived software

{% endblock %} {% block browse-content %}

This interface enables you to track the status of the different Software Heritage Vault cooking tasks created while browsing the archive.

Once a cooking task is finished, a link will be made available in order to download the associated archive.

Origin Bundle type Object info Cooking status
-{% include "includes/vault-common.html" %} +{% include "./includes/vault-common.html" %} {% endblock %} diff --git a/swh/web/templates/browse/browse.html b/swh/web/browse/templates/browse.html similarity index 67% rename from swh/web/templates/browse/browse.html rename to swh/web/browse/templates/browse.html index 1e98e7cc..442aff74 100644 --- a/swh/web/templates/browse/browse.html +++ b/swh/web/browse/templates/browse.html @@ -1,38 +1,47 @@ -{% extends "./layout.html" %} +{% extends "layout.html" %} {% comment %} -Copyright (C) 2017-2020 The Software Heritage developers +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 %} +{% load render_bundle from webpack_loader %} {% block title %}{{ heading }} – Software Heritage archive{% endblock %} +{% block header %} +{% render_bundle 'browse' %} +{% render_bundle 'vault' %} +{% render_bundle 'save' %} +{% endblock %} + {% block navbar-content %}

Browse the archive

{% endblock %} +{% block content %} {% block browse-content %} {% block swh-browse-before-content %} {% if snapshot_context %} - {% include "includes/snapshot-context.html" %} + {% include "./includes/snapshot-context.html" %} {% endif %} {% endblock %} {% block swh-browse-content %}{% endblock %} {% block swh-browse-after-content %}{% endblock %} {% endblock %} +{% endblock %} diff --git a/swh/web/templates/includes/branch-search.html b/swh/web/browse/templates/includes/branch-search.html similarity index 100% rename from swh/web/templates/includes/branch-search.html rename to swh/web/browse/templates/includes/branch-search.html diff --git a/swh/web/templates/includes/breadcrumbs.html b/swh/web/browse/templates/includes/breadcrumbs.html similarity index 100% rename from swh/web/templates/includes/breadcrumbs.html rename to swh/web/browse/templates/includes/breadcrumbs.html diff --git a/swh/web/templates/includes/content-display.html b/swh/web/browse/templates/includes/content-display.html similarity index 97% rename from swh/web/templates/includes/content-display.html rename to swh/web/browse/templates/includes/content-display.html index aa1ec399..c612189b 100644 --- a/swh/web/templates/includes/content-display.html +++ b/swh/web/browse/templates/includes/content-display.html @@ -1,85 +1,85 @@ {% 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 %} -{% include "includes/revision-info.html" %} +{% include "./revision-info.html" %} {% if snapshot_context and snapshot_context.is_empty %} - {% include "includes/empty-snapshot.html" %} + {% include "./empty-snapshot.html" %} {% else %} {% if not iframe_mode %}
{% if filename %}
{{ filename }}
{% endif %} {% endif %}
{% if content_size > max_content_size %} Content is too large to be displayed (size is greater than {{ max_content_size|filesizeformat }}). {% elif "inode/x-empty" == mimetype %} File is empty {% elif mimetype in browsers_supported_image_mimes and content %} {% elif "application/pdf" == mimetype %}
Page: /
{% elif filename and filename|default:""|slice:"-5:" == "ipynb" %}
{% elif "text/" in mimetype or "application/" in mimetype and encoding != "binary" %}
{{ content }}
{% elif content %} Content with mime type {{ mimetype }} and encoding {{ encoding }} cannot be displayed. {% else %} {% include "includes/http-error.html" %} {% endif %}
{% if not iframe_mode %}
{% endif %} {% if content %} {% endif %} {% endif %} diff --git a/swh/web/templates/includes/directory-display.html b/swh/web/browse/templates/includes/directory-display.html similarity index 93% rename from swh/web/templates/includes/directory-display.html rename to swh/web/browse/templates/includes/directory-display.html index 285adde1..dbc2a379 100644 --- a/swh/web/templates/includes/directory-display.html +++ b/swh/web/browse/templates/includes/directory-display.html @@ -1,69 +1,69 @@ {% comment %} -Copyright (C) 2017-2021 The Software Heritage developers +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 %} {% if not iframe_mode %} - {% include "includes/revision-info.html" %} + {% include "./revision-info.html" %} {% endif %} {% if snapshot_context and snapshot_context.is_empty %} - {% include "includes/empty-snapshot.html" %} + {% include "./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 }}
{{ f.name }} {{ f.perms }} {% if f.length is not None %}{{ f.length|filesizeformat }}{% endif %}
{% if not iframe_mode %}
{% endif %} {% 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 error_code != 200 %} {% include "includes/http-error.html" %} {% elif dirs|length == 0 and files|length == 0 %} Directory is empty {% endif %} diff --git a/swh/web/templates/includes/empty-snapshot.html b/swh/web/browse/templates/includes/empty-snapshot.html similarity index 100% rename from swh/web/templates/includes/empty-snapshot.html rename to swh/web/browse/templates/includes/empty-snapshot.html diff --git a/swh/web/templates/includes/origin-search-form.html b/swh/web/browse/templates/includes/origin-search-form.html similarity index 100% rename from swh/web/templates/includes/origin-search-form.html rename to swh/web/browse/templates/includes/origin-search-form.html diff --git a/swh/web/templates/includes/readme-display.html b/swh/web/browse/templates/includes/readme-display.html similarity index 100% rename from swh/web/templates/includes/readme-display.html rename to swh/web/browse/templates/includes/readme-display.html diff --git a/swh/web/templates/includes/revision-info.html b/swh/web/browse/templates/includes/revision-info.html similarity index 100% rename from swh/web/templates/includes/revision-info.html rename to swh/web/browse/templates/includes/revision-info.html diff --git a/swh/web/templates/includes/show-metadata.html b/swh/web/browse/templates/includes/show-metadata.html similarity index 100% rename from swh/web/templates/includes/show-metadata.html rename to swh/web/browse/templates/includes/show-metadata.html diff --git a/swh/web/templates/includes/show-swhids.html b/swh/web/browse/templates/includes/show-swhids.html similarity index 100% rename from swh/web/templates/includes/show-swhids.html rename to swh/web/browse/templates/includes/show-swhids.html diff --git a/swh/web/templates/includes/snapshot-context.html b/swh/web/browse/templates/includes/snapshot-context.html similarity index 100% rename from swh/web/templates/includes/snapshot-context.html rename to swh/web/browse/templates/includes/snapshot-context.html diff --git a/swh/web/templates/includes/take-new-snapshot.html b/swh/web/browse/templates/includes/take-new-snapshot.html similarity index 100% rename from swh/web/templates/includes/take-new-snapshot.html rename to swh/web/browse/templates/includes/take-new-snapshot.html diff --git a/swh/web/templates/includes/top-navigation.html b/swh/web/browse/templates/includes/top-navigation.html similarity index 95% rename from swh/web/templates/includes/top-navigation.html rename to swh/web/browse/templates/includes/top-navigation.html index 987216e9..b28fd4ee 100644 --- a/swh/web/templates/includes/top-navigation.html +++ b/swh/web/browse/templates/includes/top-navigation.html @@ -1,155 +1,155 @@ {% comment %} -Copyright (C) 2017-2020 The Software Heritage developers +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 %} {% endif %} {% endif %}
- {% include "includes/breadcrumbs.html" %} + {% include "./breadcrumbs.html" %}
{% if top_right_link %} {% if top_right_link.icon %} {% endif %} {{ top_right_link.text }} {% endif %} {% if available_languages %} {% endif %} {% if show_actions %} {% if not snapshot_context or not snapshot_context.is_empty %} - {% include "includes/vault-create-tasks.html" %} + {% include "./vault-create-tasks.html" %} {% endif %} {% if "swh.web.save_code_now" in SWH_DJANGO_APPS %} - {% include "includes/take-new-snapshot.html" %} + {% include "./take-new-snapshot.html" %} {% endif %} - {% include "includes/show-metadata.html" %} + {% include "./show-metadata.html" %} {% endif %}
-{% include "includes/show-swhids.html" %} +{% include "./show-swhids.html" %} diff --git a/swh/web/templates/includes/vault-common.html b/swh/web/browse/templates/includes/vault-common.html similarity index 100% rename from swh/web/templates/includes/vault-common.html rename to swh/web/browse/templates/includes/vault-common.html diff --git a/swh/web/templates/includes/vault-create-tasks.html b/swh/web/browse/templates/includes/vault-create-tasks.html similarity index 99% rename from swh/web/templates/includes/vault-create-tasks.html rename to swh/web/browse/templates/includes/vault-create-tasks.html index 811f5a3a..4ecaea6c 100644 --- a/swh/web/templates/includes/vault-create-tasks.html +++ b/swh/web/browse/templates/includes/vault-create-tasks.html @@ -1,177 +1,177 @@ {% 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 vault_cooking.directory_context or vault_cooking.revision_context %} {% if user.is_authenticated and user.is_staff or "swh.vault.git_bare.ui" in user.get_all_permissions %}
{% else %} {% endif %} - {% include "includes/vault-common.html" %} + {% include "./vault-common.html" %} {% endif %} diff --git a/swh/web/browse/urls.py b/swh/web/browse/urls.py index 2e9762f0..0bf42504 100644 --- a/swh/web/browse/urls.py +++ b/swh/web/browse/urls.py @@ -1,64 +1,64 @@ # 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 from django.http import HttpRequest, HttpResponse from django.shortcuts import redirect, render from django.urls import re_path as url from swh.web.browse.browseurls import BrowseUrls from swh.web.browse.identifiers import swhid_browse import swh.web.browse.views.content # noqa import swh.web.browse.views.directory # noqa import swh.web.browse.views.origin # noqa import swh.web.browse.views.release # noqa import swh.web.browse.views.revision # noqa import swh.web.browse.views.snapshot # noqa from swh.web.utils import origin_visit_types, reverse def _browse_help_view(request: HttpRequest) -> HttpResponse: return render( - request, "browse/help.html", {"heading": "How to browse the archive ?"} + request, "browse-help.html", {"heading": "How to browse the archive ?"} ) def _browse_search_view(request: HttpRequest) -> HttpResponse: return render( request, - "browse/search.html", + "browse-search.html", { "heading": "Search software origins to browse", "visit_types": origin_visit_types(), }, ) def _browse_vault_view(request: HttpRequest) -> HttpResponse: return render( request, - "browse/vault-ui.html", + "browse-vault-ui.html", {"heading": "Download archive content from the Vault"}, ) def _browse_origin_save_view(request: HttpRequest) -> HttpResponse: return redirect(reverse("origin-save")) urlpatterns = [ url(r"^browse/$", _browse_search_view), url(r"^browse/help/$", _browse_help_view, name="browse-help"), url(r"^browse/search/$", _browse_search_view, name="browse-search"), url(r"^browse/vault/$", _browse_vault_view, name="browse-vault"), # for backward compatibility url(r"^browse/origin/save/$", _browse_origin_save_view, name="browse-origin-save"), url( r"^browse/(?Pswh:[0-9]+:[a-z]+:[0-9a-f]+.*)/$", swhid_browse, name="browse-swhid-legacy", ), ] urlpatterns += BrowseUrls.get_url_patterns() diff --git a/swh/web/browse/views/content.py b/swh/web/browse/views/content.py index 9eb8dd70..e16af1d2 100644 --- a/swh/web/browse/views/content.py +++ b/swh/web/browse/views/content.py @@ -1,467 +1,467 @@ # 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 difflib from distutils.util import strtobool from typing import Any, Dict, Optional from django.http import HttpRequest, HttpResponse, JsonResponse from django.shortcuts import redirect, render from swh.model.hashutil import hash_to_hex from swh.model.swhids import ObjectType from swh.web.browse.browseurls import browse_route from swh.web.browse.snapshot_context import get_snapshot_context from swh.web.browse.utils import ( content_display_max_size, gen_link, prepare_content_for_display, request_content, ) from swh.web.utils import ( archive, gen_path_info, highlightjs, query, reverse, swh_object_icons, ) from swh.web.utils.exc import ( BadInputExc, NotFoundExc, http_status_code_message, sentry_capture_exception, ) from swh.web.utils.identifiers import get_swhids_info from swh.web.utils.typing import ContentMetadata, SWHObjectInfo @browse_route( r"content/(?P[0-9a-z_:]*[0-9a-f]+)/raw/", view_name="browse-content-raw", checksum_args=["query_string"], ) def content_raw(request: HttpRequest, query_string: str) -> HttpResponse: """Django view that produces a raw display of a content identified by its hash value. The url that points to it is :http:get:`/browse/content/[(algo_hash):](hash)/raw/` """ re_encode = bool(strtobool(request.GET.get("re_encode", "false"))) algo, checksum = query.parse_hash(query_string) checksum = hash_to_hex(checksum) content_data = request_content(query_string, max_size=None, re_encode=re_encode) filename = request.GET.get("filename", None) if not filename: filename = "%s_%s" % (algo, checksum) if ( content_data["mimetype"].startswith("text/") or content_data["mimetype"] == "inode/x-empty" ): response = HttpResponse(content_data["raw_data"], content_type="text/plain") response["Content-disposition"] = "filename=%s" % filename else: response = HttpResponse( content_data["raw_data"], content_type="application/octet-stream" ) response["Content-disposition"] = "attachment; filename=%s" % filename return response _auto_diff_size_limit = 20000 @browse_route( r"content/(?P.*)/diff/(?P.*)/", view_name="diff-contents", ) def _contents_diff( request: HttpRequest, from_query_string: str, to_query_string: str ) -> HttpResponse: """ Browse endpoint used to compute unified diffs between two contents. Diffs are generated only if the two contents are textual. By default, diffs whose size are greater than 20 kB will not be generated. To force the generation of large diffs, the 'force' boolean query parameter must be used. Args: request: input django http request from_query_string: a string of the form "[ALGO_HASH:]HASH" where optional ALGO_HASH can be either ``sha1``, ``sha1_git``, ``sha256``, or ``blake2s256`` (default to ``sha1``) and HASH the hexadecimal representation of the hash value identifying the first content to_query_string: same as above for identifying the second content Returns: A JSON object containing the unified diff. """ diff_data = {} content_from = None content_to = None content_from_size = 0 content_to_size = 0 content_from_lines = [] content_to_lines = [] force_str = request.GET.get("force", "false") path = request.GET.get("path", None) language = "plaintext" force = bool(strtobool(force_str)) if from_query_string == to_query_string: diff_str = "File renamed without changes" else: try: text_diff = True if from_query_string: content_from = request_content(from_query_string, max_size=None) content_from_display_data = prepare_content_for_display( content_from["raw_data"], content_from["mimetype"], path ) language = content_from_display_data["language"] content_from_size = content_from["length"] if not ( content_from["mimetype"].startswith("text/") or content_from["mimetype"] == "inode/x-empty" ): text_diff = False if text_diff and to_query_string: content_to = request_content(to_query_string, max_size=None) content_to_display_data = prepare_content_for_display( content_to["raw_data"], content_to["mimetype"], path ) language = content_to_display_data["language"] content_to_size = content_to["length"] if not ( content_to["mimetype"].startswith("text/") or content_to["mimetype"] == "inode/x-empty" ): text_diff = False diff_size = abs(content_to_size - content_from_size) if not text_diff: diff_str = "Diffs are not generated for non textual content" language = "plaintext" elif not force and diff_size > _auto_diff_size_limit: diff_str = "Large diffs are not automatically computed" language = "plaintext" else: if content_from: content_from_lines = ( content_from["raw_data"].decode("utf-8").splitlines(True) ) if content_from_lines and content_from_lines[-1][-1] != "\n": content_from_lines[-1] += "[swh-no-nl-marker]\n" if content_to: content_to_lines = ( content_to["raw_data"].decode("utf-8").splitlines(True) ) if content_to_lines and content_to_lines[-1][-1] != "\n": content_to_lines[-1] += "[swh-no-nl-marker]\n" diff_lines = difflib.unified_diff(content_from_lines, content_to_lines) diff_str = "".join(list(diff_lines)[2:]) except Exception as exc: sentry_capture_exception(exc) diff_str = str(exc) diff_data["diff_str"] = diff_str diff_data["language"] = language return JsonResponse(diff_data) def _get_content_from_request(request: HttpRequest) -> Dict[str, Any]: path = request.GET.get("path") if path is None: raise BadInputExc("The path query parameter must be provided.") snapshot = request.GET.get("snapshot") or request.GET.get("snapshot_id") origin_url = request.GET.get("origin_url") if snapshot is None and origin_url is None: raise BadInputExc( "The origin_url or snapshot query parameters must be provided." ) visit_id = int(request.GET.get("visit_id", 0)) snapshot_context = get_snapshot_context( snapshot_id=snapshot, origin_url=origin_url, path=path, timestamp=request.GET.get("timestamp"), visit_id=visit_id or None, branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), browse_context="content", ) root_directory = snapshot_context["root_directory"] assert root_directory is not None # to keep mypy happy return archive.lookup_directory_with_path(root_directory, path) @browse_route( r"content/(?P[0-9a-z_:]*[0-9a-f]+)/", r"content/", view_name="browse-content", checksum_args=["query_string"], ) def content_display( request: HttpRequest, query_string: Optional[str] = None ) -> HttpResponse: """Django view that produces an HTML display of a content identified by its hash value. The URLs that points to it are :http:get:`/browse/content/[(algo_hash):](hash)/` :http:get:`/browse/content/` """ if query_string is None: # this case happens when redirected from origin/content or snapshot/content content_data = _get_content_from_request(request) return redirect( reverse( "browse-content", url_args={"query_string": f"sha1_git:{content_data['target']}"}, query_params=request.GET, ), ) algo, checksum = query.parse_hash(query_string) checksum = hash_to_hex(checksum) origin_url = request.GET.get("origin_url") selected_language = request.GET.get("language") if not origin_url: origin_url = request.GET.get("origin") snapshot_id = request.GET.get("snapshot") or request.GET.get("snapshot_id") path = request.GET.get("path") content_data = {} error_info: Dict[str, Any] = {"status_code": 200, "description": None} try: content_data = request_content(query_string) except NotFoundExc as e: error_info["status_code"] = 404 error_info["description"] = f"NotFoundExc: {str(e)}" snapshot_context = None if origin_url is not None or snapshot_id is not None: try: visit_id = int(request.GET.get("visit_id", 0)) snapshot_context = get_snapshot_context( origin_url=origin_url, snapshot_id=snapshot_id, timestamp=request.GET.get("timestamp"), visit_id=visit_id or None, branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), revision_id=request.GET.get("revision"), path=path, browse_context="content", ) except NotFoundExc as e: if str(e).startswith("Origin") and origin_url is not None: raw_cnt_url = reverse( "browse-content", url_args={"query_string": query_string} ) error_message = ( "The Software Heritage archive has a content " "with the hash you provided but the origin " "mentioned in your request appears broken: %s. " "Please check the URL and try again.\n\n" "Nevertheless, you can still browse the content " "without origin information: %s" % (gen_link(origin_url), gen_link(raw_cnt_url)) ) raise NotFoundExc(error_message) else: raise e content = None language = None mimetype = None if content_data.get("raw_data") is not None: content_display_data = prepare_content_for_display( content_data["raw_data"], content_data["mimetype"], path ) content = 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() filename = None path_info = None directory_id = None root_dir = None if snapshot_context: root_dir = snapshot_context.get("root_directory") query_params = snapshot_context["query_params"] if snapshot_context else {} breadcrumbs = [] if path: split_path = path.split("/") root_dir = root_dir or split_path[0] filename = split_path[-1] if root_dir != path: path = path.replace(root_dir + "/", "") path = path[: -len(filename)] path_info = gen_path_info(path) query_params.pop("path", None) dir_url = reverse( "browse-directory", url_args={"sha1_git": root_dir}, query_params=query_params, ) breadcrumbs.append({"name": root_dir[:7], "url": dir_url}) for pi in path_info: query_params["path"] = pi["path"] dir_url = reverse( "browse-directory", url_args={"sha1_git": root_dir}, query_params=query_params, ) breadcrumbs.append({"name": pi["name"], "url": dir_url}) breadcrumbs.append({"name": filename, "url": ""}) if path and root_dir is not None and root_dir != path: dir_info = archive.lookup_directory_with_path(root_dir, path) directory_id = dir_info["target"] elif root_dir != path: directory_id = root_dir else: root_dir = None query_params = {"filename": filename} content_checksums = content_data.get("checksums", {}) content_url = reverse( "browse-content", url_args={"query_string": query_string}, ) content_raw_url = reverse( "browse-content-raw", url_args={"query_string": query_string}, query_params=query_params, ) content_metadata = ContentMetadata( object_type=ObjectType.CONTENT, object_id=content_checksums.get("sha1_git"), sha1=content_checksums.get("sha1"), sha1_git=content_checksums.get("sha1_git"), sha256=content_checksums.get("sha256"), blake2s256=content_checksums.get("blake2s256"), content_url=content_url, mimetype=content_data.get("mimetype", ""), encoding=content_data.get("encoding", ""), size=content_data.get("length", 0), language=content_data.get("language", ""), root_directory=root_dir, path=f"/{path}" if path else None, filename=filename or "", directory=directory_id, revision=None, release=None, snapshot=None, origin_url=origin_url, ) swh_objects = [] if content_checksums: swh_objects.append( SWHObjectInfo( object_type=ObjectType.CONTENT, object_id=content_checksums.get("sha1_git"), ) ) if directory_id: swh_objects.append( SWHObjectInfo(object_type=ObjectType.DIRECTORY, object_id=directory_id) ) if snapshot_context: if snapshot_context["revision_id"]: swh_objects.append( SWHObjectInfo( object_type=ObjectType.REVISION, object_id=snapshot_context["revision_id"], ) ) swh_objects.append( SWHObjectInfo( object_type=ObjectType.SNAPSHOT, object_id=snapshot_context["snapshot_id"], ) ) if snapshot_context["release_id"]: swh_objects.append( SWHObjectInfo( object_type=ObjectType.RELEASE, object_id=snapshot_context["release_id"], ) ) swhids_info = get_swhids_info( swh_objects, snapshot_context, extra_context=content_metadata, ) heading = "Content - %s" % content_checksums.get("sha1_git") if breadcrumbs: content_path = "/".join(bc["name"] for bc in breadcrumbs) heading += " - %s" % content_path return render( request, - "browse/content.html", + "browse-content.html", { "heading": heading, "swh_object_id": swhids_info[0]["swhid"] if swhids_info else "", "swh_object_name": "Content", "swh_object_metadata": content_metadata, "content": content, "content_size": content_data.get("length"), "max_content_size": content_display_max_size, "filename": filename, "encoding": content_data.get("encoding"), "mimetype": mimetype, "language": language, "available_languages": available_languages, "breadcrumbs": breadcrumbs, "top_right_link": { "url": content_raw_url, "icon": swh_object_icons["content"], "text": "Raw File", }, "snapshot_context": snapshot_context, "vault_cooking": None, "show_actions": True, "swhids_info": swhids_info, "error_code": error_info["status_code"], "error_message": http_status_code_message.get(error_info["status_code"]), "error_description": error_info["description"], }, status=error_info["status_code"], ) diff --git a/swh/web/browse/views/directory.py b/swh/web/browse/views/directory.py index 7cc22729..199a5e81 100644 --- a/swh/web/browse/views/directory.py +++ b/swh/web/browse/views/directory.py @@ -1,303 +1,303 @@ # 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 os from typing import Any, Dict, Optional from django.http import HttpRequest, HttpResponse from django.shortcuts import redirect, render from swh.model.swhids import ObjectType from swh.web.browse.browseurls import browse_route from swh.web.browse.snapshot_context import get_snapshot_context from swh.web.browse.utils import gen_link, get_directory_entries, get_readme_to_display from swh.web.utils import archive, gen_path_info, reverse, swh_object_icons from swh.web.utils.exc import ( NotFoundExc, http_status_code_message, sentry_capture_exception, ) from swh.web.utils.identifiers import get_swhids_info from swh.web.utils.typing import DirectoryMetadata, SWHObjectInfo def _directory_browse( request: HttpRequest, sha1_git: str, path: Optional[str] = None ) -> HttpResponse: root_sha1_git = sha1_git dir_sha1_git: Optional[str] = sha1_git error_info: Dict[str, Any] = {"status_code": 200, "description": None} if path: try: dir_info = archive.lookup_directory_with_path(sha1_git, path) dir_sha1_git = dir_info["target"] except NotFoundExc as e: error_info["status_code"] = 404 error_info["description"] = f"NotFoundExc: {str(e)}" dir_sha1_git = None dirs, files = [], [] if dir_sha1_git is not None: dirs, files = get_directory_entries(dir_sha1_git) origin_url = request.GET.get("origin_url") if not origin_url: origin_url = request.GET.get("origin") snapshot_id = request.GET.get("snapshot") snapshot_context = None if origin_url is not None or snapshot_id is not None: try: snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), revision_id=request.GET.get("revision"), path=path, ) except NotFoundExc as e: if str(e).startswith("Origin") and origin_url is not None: raw_dir_url = reverse( "browse-directory", url_args={"sha1_git": dir_sha1_git} ) error_message = ( "The Software Heritage archive has a directory " "with the hash you provided but the origin " "mentioned in your request appears broken: %s. " "Please check the URL and try again.\n\n" "Nevertheless, you can still browse the directory " "without origin information: %s" % (gen_link(origin_url), gen_link(raw_dir_url)) ) raise NotFoundExc(error_message) else: raise e path_info = gen_path_info(path) query_params = snapshot_context["query_params"] if snapshot_context else {} breadcrumbs = [] breadcrumbs.append( { "name": root_sha1_git[:7], "url": reverse( "browse-directory", url_args={"sha1_git": root_sha1_git}, query_params={**query_params, "path": None}, ), } ) for pi in path_info: breadcrumbs.append( { "name": pi["name"], "url": reverse( "browse-directory", url_args={"sha1_git": root_sha1_git}, query_params={ **query_params, "path": pi["path"], }, ), } ) path = "" if path is None else (path + "/") for d in dirs: if d["type"] == "rev": d["url"] = reverse( "browse-revision", url_args={"sha1_git": d["target"]}, query_params=query_params, ) else: d["url"] = reverse( "browse-directory", url_args={"sha1_git": root_sha1_git}, query_params={ **query_params, "path": path + d["name"], }, ) sum_file_sizes = 0 readmes = {} for f in files: query_string = "sha1_git:" + f["target"] f["url"] = reverse( "browse-content", url_args={"query_string": query_string}, query_params={ **query_params, "path": root_sha1_git + "/" + path + f["name"], }, ) if f["length"] is not None: sum_file_sizes += f["length"] if f["name"].lower().startswith("readme"): readmes[f["name"]] = f["checksums"]["sha1"] readme_name, readme_url, readme_html = get_readme_to_display(readmes) dir_metadata = DirectoryMetadata( object_type=ObjectType.DIRECTORY, object_id=dir_sha1_git, directory=root_sha1_git, nb_files=len(files), nb_dirs=len(dirs), sum_file_sizes=sum_file_sizes, root_directory=root_sha1_git, path=f"/{path}" if path else None, revision=None, revision_found=None, release=None, snapshot=None, ) vault_cooking = { "directory_context": True, "directory_swhid": f"swh:1:dir:{dir_sha1_git}", "revision_context": False, "revision_swhid": None, } swh_objects = [ SWHObjectInfo(object_type=ObjectType.DIRECTORY, object_id=dir_sha1_git) ] if snapshot_context: if snapshot_context["revision_id"]: swh_objects.append( SWHObjectInfo( object_type=ObjectType.REVISION, object_id=snapshot_context["revision_id"], ) ) swh_objects.append( SWHObjectInfo( object_type=ObjectType.SNAPSHOT, object_id=snapshot_context["snapshot_id"], ) ) if snapshot_context["release_id"]: swh_objects.append( SWHObjectInfo( object_type=ObjectType.RELEASE, object_id=snapshot_context["release_id"], ) ) swhids_info = get_swhids_info(swh_objects, snapshot_context, dir_metadata) heading = "Directory - %s" % dir_sha1_git if breadcrumbs: dir_path = "/".join([bc["name"] for bc in breadcrumbs]) + "/" heading += " - %s" % dir_path top_right_link = None if ( snapshot_context is not None and not snapshot_context["is_empty"] and snapshot_context["revision_id"] is not None ): history_url = reverse( "browse-revision-log", url_args={"sha1_git": snapshot_context["revision_id"]}, query_params=query_params, ) top_right_link = { "url": history_url, "icon": swh_object_icons["revisions history"], "text": "History", } return render( request, - "browse/directory.html", + "browse-directory.html", { "heading": heading, "swh_object_id": swhids_info[0]["swhid"], "swh_object_name": "Directory", "swh_object_metadata": dir_metadata, "dirs": dirs, "files": files, "breadcrumbs": breadcrumbs, "top_right_link": top_right_link, "readme_name": readme_name, "readme_url": readme_url, "readme_html": readme_html, "snapshot_context": snapshot_context, "vault_cooking": vault_cooking, "show_actions": True, "swhids_info": swhids_info, "error_code": error_info["status_code"], "error_message": http_status_code_message.get(error_info["status_code"]), "error_description": error_info["description"], }, status=error_info["status_code"], ) @browse_route( r"directory/(?P[0-9a-f]+)/", view_name="browse-directory", checksum_args=["sha1_git"], ) def directory_browse(request: HttpRequest, sha1_git: str) -> HttpResponse: """Django view for browsing the content of a directory identified by its sha1_git value. The url that points to it is :http:get:`/browse/directory/(sha1_git)/` """ return _directory_browse(request, sha1_git, request.GET.get("path")) @browse_route( r"directory/(?P[0-9a-f]+)/(?P.+)/", view_name="browse-directory-legacy", checksum_args=["sha1_git"], ) def directory_browse_legacy( request: HttpRequest, sha1_git: str, path: str ) -> HttpResponse: """Django view for browsing the content of a directory identified by its sha1_git value. The url that points to it is :http:get:`/browse/directory/(sha1_git)/(path)/` """ return _directory_browse(request, sha1_git, path) @browse_route( r"directory/resolve/content-path/(?P[0-9a-f]+)/", view_name="browse-directory-resolve-content-path", checksum_args=["sha1_git"], ) def _directory_resolve_content_path( request: HttpRequest, sha1_git: str ) -> HttpResponse: """ Internal endpoint redirecting to data url for a specific file path relative to a root directory. """ try: path = os.path.normpath(request.GET.get("path", "")) if not path.startswith("../"): dir_info = archive.lookup_directory_with_path(sha1_git, path) if dir_info["type"] == "file": sha1 = dir_info["checksums"]["sha1"] data_url = reverse( "browse-content-raw", url_args={"query_string": sha1} ) return redirect(data_url) except Exception as exc: sentry_capture_exception(exc) return HttpResponse(status=404) diff --git a/swh/web/browse/views/origin.py b/swh/web/browse/views/origin.py index 63c4817b..3c7ef0c8 100644 --- a/swh/web/browse/views/origin.py +++ b/swh/web/browse/views/origin.py @@ -1,332 +1,332 @@ # Copyright (C) 2021-2022 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 from typing import Any, Dict, List, Optional, cast from django.http import HttpRequest, HttpResponse from django.shortcuts import redirect, render from swh.web.browse.browseurls import browse_route from swh.web.browse.snapshot_context import ( browse_snapshot_directory, get_snapshot_context, ) from swh.web.utils import ( archive, format_utc_iso_date, parse_iso8601_date_to_utc, redirect_to_new_route, reverse, ) from swh.web.utils.exc import BadInputExc from swh.web.utils.origin_visits import get_origin_visits @browse_route( r"origin/directory/", view_name="browse-origin-directory", ) def origin_directory_browse(request: HttpRequest) -> HttpResponse: """Django view for browsing the content of a directory associated to an origin for a given visit. The URL that points to it is :http:get:`/browse/origin/directory/` """ return browse_snapshot_directory( request, origin_url=request.GET.get("origin_url"), snapshot_id=request.GET.get("snapshot"), timestamp=request.GET.get("timestamp"), path=request.GET.get("path"), ) @browse_route( r"origin/(?P.+)/visit/(?P.+)/directory/", r"origin/(?P.+)/visit/(?P.+)/directory/(?P.+)/", r"origin/(?P.+)/directory/(?P.+)/", r"origin/(?P.+)/directory/", view_name="browse-origin-directory-legacy", ) def origin_directory_browse_legacy( request: HttpRequest, origin_url: str, timestamp: Optional[str] = None, path: Optional[str] = None, ) -> HttpResponse: """Django view for browsing the content of a directory associated to an origin for a given visit. The URLs that point to it are :http:get:`/browse/origin/(origin_url)/directory/[(path)/]` and :http:get:`/browse/origin/(origin_url)/visit/(timestamp)/directory/[(path)/]` """ return browse_snapshot_directory( request, origin_url=origin_url, snapshot_id=request.GET.get("snapshot"), timestamp=timestamp, path=path, ) @browse_route( r"origin/content/", view_name="browse-origin-content", ) def origin_content_browse(request: HttpRequest) -> HttpResponse: """ This route is deprecated; use http:get:`/browse/content` instead Django view that produces an HTML display of a content associated to an origin for a given visit. The URL that points to it is :http:get:`/browse/origin/content/` """ return redirect_to_new_route(request, "browse-content") @browse_route( r"origin/(?P.+)/visit/(?P.+)/content/(?P.+)/", r"origin/(?P.+)/content/(?P.+)/", r"origin/(?P.+)/content/", view_name="browse-origin-content-legacy", ) def origin_content_browse_legacy( request: HttpRequest, origin_url: str, path: Optional[str] = None, timestamp: Optional[str] = None, ) -> HttpResponse: """ This route is deprecated; use http:get:`/browse/content` instead Django view that produces an HTML display of a content associated to an origin for a given visit. The URLs that point to it are :http:get:`/browse/origin/(origin_url)/content/(path)/` and :http:get:`/browse/origin/(origin_url)/visit/(timestamp)/content/(path)/` """ return redirect_to_new_route(request, "browse-content") @browse_route( r"origin/log/", view_name="browse-origin-log", ) def origin_log_browse(request: HttpRequest) -> HttpResponse: """ This route is deprecated; use http:get:`/browse/snapshot/log` instead Django view that produces an HTML display of revisions history (aka the commit log) associated to a software origin. The URL that points to it is :http:get:`/browse/origin/log/` """ return redirect_to_new_route(request, "browse-snapshot-log") @browse_route( r"origin/(?P.+)/visit/(?P.+)/log/", r"origin/(?P.+)/log/", view_name="browse-origin-log-legacy", ) def origin_log_browse_legacy( request: HttpRequest, origin_url: str, timestamp: Optional[str] = None ) -> HttpResponse: """ This route is deprecated; use http:get:`/browse/snapshot/log` instead Django view that produces an HTML display of revisions history (aka the commit log) associated to a software origin. The URLs that point to it are :http:get:`/browse/origin/(origin_url)/log/` and :http:get:`/browse/origin/(origin_url)/visit/(timestamp)/log/` """ return redirect_to_new_route( request, "browse-snapshot-log", ) @browse_route( r"origin/branches/", view_name="browse-origin-branches", ) def origin_branches_browse(request: HttpRequest) -> HttpResponse: """ This route is deprecated; use http:get:`/browse/snapshot/branches` instead Django view that produces an HTML display of the list of branches associated to an origin for a given visit. The URL that points to it is :http:get:`/browse/origin/branches/` """ return redirect_to_new_route(request, "browse-snapshot-branches") @browse_route( r"origin/(?P.+)/visit/(?P.+)/branches/", r"origin/(?P.+)/branches/", view_name="browse-origin-branches-legacy", ) def origin_branches_browse_legacy( request: HttpRequest, origin_url: str, timestamp: Optional[str] = None ) -> HttpResponse: """ This route is deprecated; use http:get:`/browse/snapshot/branches` instead Django view that produces an HTML display of the list of branches associated to an origin for a given visit. The URLs that point to it are :http:get:`/browse/origin/(origin_url)/branches/` and :http:get:`/browse/origin/(origin_url)/visit/(timestamp)/branches/` """ return redirect_to_new_route(request, "browse-snapshot-branches") @browse_route( r"origin/releases/", view_name="browse-origin-releases", ) def origin_releases_browse(request: HttpRequest) -> HttpResponse: """ This route is deprecated; use http:get:`/browse/snapshot/releases` instead Django view that produces an HTML display of the list of releases associated to an origin for a given visit. The URL that points to it is :http:get:`/browse/origin/releases/` """ return redirect_to_new_route(request, "browse-snapshot-releases") @browse_route( r"origin/(?P.+)/visit/(?P.+)/releases/", r"origin/(?P.+)/releases/", view_name="browse-origin-releases-legacy", ) def origin_releases_browse_legacy( request: HttpRequest, origin_url: str, timestamp: Optional[str] = None ) -> HttpResponse: """ This route is deprecated; use http:get:`/browse/snapshot/releases` instead Django view that produces an HTML display of the list of releases associated to an origin for a given visit. The URLs that point to it are :http:get:`/browse/origin/(origin_url)/releases/` and :http:get:`/browse/origin/(origin_url)/visit/(timestamp)/releases/` """ return redirect_to_new_route(request, "browse-snapshot-releases") def _origin_visits_browse( request: HttpRequest, origin_url: Optional[str] ) -> HttpResponse: if origin_url is None: raise BadInputExc("An origin URL must be provided as query parameter.") origin_info = archive.lookup_origin({"url": origin_url}) origin_visits = cast(List[Dict[str, Any]], get_origin_visits(origin_info)) snapshot_context = get_snapshot_context(origin_url=origin_url) for i, visit in enumerate(origin_visits): url_date = format_utc_iso_date(visit["date"], "%Y-%m-%dT%H:%M:%SZ") visit["formatted_date"] = format_utc_iso_date(visit["date"]) query_params = {"origin_url": origin_url, "timestamp": url_date} if i < len(origin_visits) - 1: if visit["date"] == origin_visits[i + 1]["date"]: query_params = {"visit_id": visit["visit"]} if i > 0: if visit["date"] == origin_visits[i - 1]["date"]: query_params = {"visit_id": visit["visit"]} snapshot = visit["snapshot"] if visit["snapshot"] else "" visit["url"] = reverse( "browse-origin-directory", query_params=query_params, ) if not snapshot: visit["snapshot"] = "" visit["date"] = parse_iso8601_date_to_utc(visit["date"]).timestamp() heading = "Origin visits - %s" % origin_url return render( request, - "browse/origin-visits.html", + "browse-origin-visits.html", { "heading": heading, "swh_object_name": "Visits", "swh_object_metadata": origin_info, "origin_visits": origin_visits, "origin_info": origin_info, "snapshot_context": snapshot_context, "vault_cooking": None, "show_actions": False, }, ) @browse_route(r"origin/visits/", view_name="browse-origin-visits") def origin_visits_browse(request: HttpRequest) -> HttpResponse: """Django view that produces an HTML display of visits reporting for a given origin. The URL that points to it is :http:get:`/browse/origin/visits/`. """ return _origin_visits_browse(request, request.GET.get("origin_url")) @browse_route( r"origin/(?P.+)/visits/", view_name="browse-origin-visits-legacy" ) def origin_visits_browse_legacy(request: HttpRequest, origin_url: str) -> HttpResponse: """Django view that produces an HTML display of visits reporting for a given origin. The URL that points to it is :http:get:`/browse/origin/(origin_url)/visits/`. """ return _origin_visits_browse(request, origin_url) @browse_route(r"origin/", view_name="browse-origin") def origin_browse(request: HttpRequest) -> HttpResponse: """Django view that redirects to the display of the latest archived snapshot for a given software origin. """ last_snapshot_url = reverse( "browse-origin-directory", query_params=request.GET, ) return redirect(last_snapshot_url) @browse_route(r"origin/(?P.+)/", view_name="browse-origin-legacy") def origin_browse_legacy(request: HttpRequest, origin_url: str) -> HttpResponse: """Django view that redirects to the display of the latest archived snapshot for a given software origin. """ last_snapshot_url = reverse( "browse-origin-directory", query_params={"origin_url": origin_url, **request.GET}, ) return redirect(last_snapshot_url) diff --git a/swh/web/browse/views/release.py b/swh/web/browse/views/release.py index d41d6a91..1a811a9c 100644 --- a/swh/web/browse/views/release.py +++ b/swh/web/browse/views/release.py @@ -1,246 +1,246 @@ # 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 from typing import Optional from django.http import HttpRequest, HttpResponse from django.shortcuts import render from swh.model.swhids import ObjectType from swh.web.browse.browseurls import browse_route from swh.web.browse.snapshot_context import get_snapshot_context from swh.web.browse.utils import ( gen_content_link, gen_directory_link, gen_link, gen_person_mail_link, gen_release_link, gen_revision_link, ) from swh.web.utils import archive, format_utc_iso_date, reverse from swh.web.utils.exc import NotFoundExc, sentry_capture_exception from swh.web.utils.identifiers import get_swhids_info from swh.web.utils.typing import ReleaseMetadata, SnapshotContext, SWHObjectInfo @browse_route( r"release/(?P[0-9a-f]+)/", view_name="browse-release", checksum_args=["sha1_git"], ) def release_browse(request: HttpRequest, sha1_git: str) -> HttpResponse: """ Django view that produces an HTML display of a release identified by its id. The url that points to it is :http:get:`/browse/release/(sha1_git)/`. """ release = archive.lookup_release(sha1_git) snapshot_context: Optional[SnapshotContext] = None origin_info = None snapshot_id = request.GET.get("snapshot_id") if not snapshot_id: snapshot_id = request.GET.get("snapshot") origin_url = request.GET.get("origin_url") if not origin_url: origin_url = request.GET.get("origin") timestamp = request.GET.get("timestamp") visit_id = int(request.GET.get("visit_id", 0)) if origin_url: try: snapshot_context = get_snapshot_context( snapshot_id, origin_url, timestamp, visit_id or None, release_name=release["name"], ) except NotFoundExc as e: raw_rel_url = reverse("browse-release", url_args={"sha1_git": sha1_git}) error_message = ( "The Software Heritage archive has a release " "with the hash you provided but the origin " "mentioned in your request appears broken: %s. " "Please check the URL and try again.\n\n" "Nevertheless, you can still browse the release " "without origin information: %s" % (gen_link(origin_url), gen_link(raw_rel_url)) ) if str(e).startswith("Origin"): raise NotFoundExc(error_message) else: raise e origin_info = snapshot_context["origin_info"] elif snapshot_id: snapshot_context = get_snapshot_context( snapshot_id, release_name=release["name"] ) if snapshot_context is not None: snapshot_id = snapshot_context.get("snapshot_id", None) release_metadata = ReleaseMetadata( object_type=ObjectType.RELEASE, object_id=sha1_git, release=sha1_git, author=release["author"]["fullname"] if release["author"] else "None", author_url=gen_person_mail_link(release["author"]) if release["author"] else "None", date=format_utc_iso_date(release["date"]), name=release["name"], synthetic=release["synthetic"], target=release["target"], target_type=release["target_type"], snapshot=snapshot_id, origin_url=origin_url, ) release_note_lines = [] if release["message"]: release_note_lines = release["message"].split("\n") swh_objects = [SWHObjectInfo(object_type=ObjectType.RELEASE, object_id=sha1_git)] vault_cooking = None rev_directory = None target_link = None if release["target_type"] == ObjectType.REVISION.name.lower(): target_link = gen_revision_link( release["target"], snapshot_context=snapshot_context, link_text=None, link_attrs=None, ) try: revision = archive.lookup_revision(release["target"]) rev_directory = revision["directory"] vault_cooking = { "directory_context": True, "directory_swhid": f"swh:1:dir:{rev_directory}", "revision_context": True, "revision_swhid": f"swh:1:rev:{release['target']}", } swh_objects.append( SWHObjectInfo( object_type=ObjectType.REVISION, object_id=release["target"] ) ) swh_objects.append( SWHObjectInfo(object_type=ObjectType.DIRECTORY, object_id=rev_directory) ) except Exception as exc: sentry_capture_exception(exc) elif release["target_type"] == ObjectType.DIRECTORY.name.lower(): target_link = gen_directory_link( release["target"], snapshot_context=snapshot_context, link_text=None, link_attrs=None, ) try: # check directory exists archive.lookup_directory(release["target"]) vault_cooking = { "directory_context": True, "directory_swhid": f"swh:1:dir:{release['target']}", "revision_context": False, "revision_swhid": None, } swh_objects.append( SWHObjectInfo( object_type=ObjectType.DIRECTORY, object_id=release["target"] ) ) except Exception as exc: sentry_capture_exception(exc) elif release["target_type"] == ObjectType.CONTENT.name.lower(): target_link = gen_content_link( release["target"], snapshot_context=snapshot_context, link_text=None, link_attrs=None, ) swh_objects.append( SWHObjectInfo(object_type=ObjectType.CONTENT, object_id=release["target"]) ) elif release["target_type"] == ObjectType.RELEASE.name.lower(): target_link = gen_release_link( release["target"], snapshot_context=snapshot_context, link_text=None, link_attrs=None, ) rev_directory_url = None if rev_directory is not None: if origin_info: rev_directory_url = reverse( "browse-origin-directory", query_params={ "origin_url": origin_info["url"], "release": release["name"], "snapshot": snapshot_id, }, ) elif snapshot_id: rev_directory_url = reverse( "browse-snapshot-directory", url_args={"snapshot_id": snapshot_id}, query_params={"release": release["name"]}, ) else: rev_directory_url = reverse( "browse-directory", url_args={"sha1_git": rev_directory} ) directory_link = None if rev_directory_url is not None: directory_link = gen_link(rev_directory_url, rev_directory) release["directory_link"] = directory_link release["target_link"] = target_link if snapshot_context: snapshot_id = snapshot_context["snapshot_id"] if snapshot_id: swh_objects.append( SWHObjectInfo(object_type=ObjectType.SNAPSHOT, object_id=snapshot_id) ) swhids_info = get_swhids_info(swh_objects, snapshot_context) note_header = "None" if len(release_note_lines) > 0: note_header = release_note_lines[0] release["note_header"] = note_header release["note_body"] = "\n".join(release_note_lines[1:]) heading = "Release - %s" % release["name"] if snapshot_context: context_found = "snapshot: %s" % snapshot_context["snapshot_id"] if origin_info: context_found = "origin: %s" % origin_info["url"] heading += " - %s" % context_found return render( request, - "browse/release.html", + "browse-release.html", { "heading": heading, "swh_object_id": swhids_info[0]["swhid"], "swh_object_name": "Release", "swh_object_metadata": release_metadata, "release": release, "snapshot_context": snapshot_context, "show_actions": True, "breadcrumbs": None, "vault_cooking": vault_cooking, "top_right_link": None, "swhids_info": swhids_info, }, ) diff --git a/swh/web/browse/views/revision.py b/swh/web/browse/views/revision.py index 14a5820b..a78f721e 100644 --- a/swh/web/browse/views/revision.py +++ b/swh/web/browse/views/revision.py @@ -1,600 +1,600 @@ # 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 hashlib import json import textwrap from typing import Any, Dict, List, Optional from django.http import HttpRequest, HttpResponse, JsonResponse from django.shortcuts import render from django.utils.safestring import mark_safe from swh.model.hashutil import hash_to_bytes from swh.model.swhids import CoreSWHID, ObjectType from swh.web.browse.browseurls import browse_route from swh.web.browse.snapshot_context import get_snapshot_context from swh.web.browse.utils import ( content_display_max_size, format_log_entries, gen_link, gen_person_mail_link, gen_revision_url, get_directory_entries, get_readme_to_display, get_revision_log_url, prepare_content_for_display, request_content, ) from swh.web.utils import ( archive, format_utc_iso_date, gen_path_info, reverse, swh_object_icons, ) from swh.web.utils.exc import NotFoundExc, http_status_code_message from swh.web.utils.identifiers import get_swhids_info from swh.web.utils.typing import RevisionMetadata, SnapshotContext, SWHObjectInfo def _gen_content_url( revision: Dict[str, Any], query_string: str, path: str, snapshot_context: Optional[SnapshotContext], ) -> str: if snapshot_context: query_params = snapshot_context["query_params"] query_params["path"] = path query_params["revision"] = revision["id"] content_url = reverse("browse-origin-content", query_params=query_params) else: content_path = "%s/%s" % (revision["directory"], path) content_url = reverse( "browse-content", url_args={"query_string": query_string}, query_params={"path": content_path}, ) return content_url def _gen_diff_link(idx: int, diff_anchor: str, link_text: str) -> str: if idx < _max_displayed_file_diffs: return gen_link(diff_anchor, link_text) else: return link_text # TODO: put in conf _max_displayed_file_diffs = 1000 def _gen_revision_changes_list( revision: Dict[str, Any], changes: List[Dict[str, Any]], snapshot_context: Optional[SnapshotContext], ) -> str: """ Returns a HTML string describing the file changes introduced in a revision. As this string will be displayed in the browse revision view, links to adequate file diffs are also generated. Args: revision (str): hexadecimal representation of a revision identifier changes (list): list of file changes in the revision snapshot_context (dict): optional origin context used to reverse the content urls Returns: A string to insert in a revision HTML view. """ changes_msg = [] for i, change in enumerate(changes): hasher = hashlib.sha1() from_query_string = "" to_query_string = "" diff_id = "diff-" if change["from"]: from_query_string = "sha1_git:" + change["from"]["target"] diff_id += change["from"]["target"] + "-" + change["from_path"] diff_id += "-" if change["to"]: to_query_string = "sha1_git:" + change["to"]["target"] diff_id += change["to"]["target"] + change["to_path"] change["path"] = change["to_path"] or change["from_path"] url_args = { "from_query_string": from_query_string, "to_query_string": to_query_string, } query_params = {"path": change["path"]} change["diff_url"] = reverse( "diff-contents", url_args=url_args, query_params=query_params ) hasher.update(diff_id.encode("utf-8")) diff_id = hasher.hexdigest() change["id"] = diff_id diff_link = "#diff_" + diff_id if change["type"] == "modify": change["content_url"] = _gen_content_url( revision, to_query_string, change["to_path"], snapshot_context ) changes_msg.append( "modified: %s" % _gen_diff_link(i, diff_link, change["to_path"]) ) elif change["type"] == "insert": change["content_url"] = _gen_content_url( revision, to_query_string, change["to_path"], snapshot_context ) changes_msg.append( "new file: %s" % _gen_diff_link(i, diff_link, change["to_path"]) ) elif change["type"] == "delete": parent = archive.lookup_revision(revision["parents"][0]) change["content_url"] = _gen_content_url( parent, from_query_string, change["from_path"], snapshot_context ) changes_msg.append( "deleted: %s" % _gen_diff_link(i, diff_link, change["from_path"]) ) elif change["type"] == "rename": change["content_url"] = _gen_content_url( revision, to_query_string, change["to_path"], snapshot_context ) link_text = change["from_path"] + " → " + change["to_path"] changes_msg.append( "renamed: %s" % _gen_diff_link(i, diff_link, link_text) ) if not changes: changes_msg.append("No changes") return mark_safe("\n".join(changes_msg)) @browse_route( r"revision/(?P[0-9a-f]+)/diff/", view_name="diff-revision", checksum_args=["sha1_git"], ) def _revision_diff(request: HttpRequest, sha1_git: str) -> HttpResponse: """ Browse internal endpoint to compute revision diff """ revision = archive.lookup_revision(sha1_git) snapshot_context = None origin_url = request.GET.get("origin_url", None) if not origin_url: origin_url = request.GET.get("origin", None) timestamp = request.GET.get("timestamp", None) visit_id_str = request.GET.get("visit_id", None) visit_id = int(visit_id_str) if visit_id_str is not None else None if origin_url: snapshot_context = get_snapshot_context( origin_url=origin_url, timestamp=timestamp, visit_id=visit_id ) changes = archive.diff_revision(sha1_git) changes_msg = _gen_revision_changes_list(revision, changes, snapshot_context) diff_data = { "total_nb_changes": len(changes), "changes": changes[:_max_displayed_file_diffs], "changes_msg": changes_msg, } return JsonResponse(diff_data) NB_LOG_ENTRIES = 100 @browse_route( r"revision/(?P[0-9a-f]+)/log/", view_name="browse-revision-log", checksum_args=["sha1_git"], ) def revision_log_browse(request: HttpRequest, sha1_git: str) -> HttpResponse: """ Django view that produces an HTML display of the history log for a revision identified by its id. The url that points to it is :http:get:`/browse/revision/(sha1_git)/log/` """ origin_url = request.GET.get("origin_url") snapshot_id = request.GET.get("snapshot") snapshot_context = None if origin_url or snapshot_id: visit_id = int(request.GET.get("visit_id", 0)) snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=request.GET.get("timestamp"), visit_id=visit_id or None, branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), revision_id=sha1_git, ) per_page = int(request.GET.get("per_page", NB_LOG_ENTRIES)) offset = int(request.GET.get("offset", 0)) revs_ordering = request.GET.get("revs_ordering", "committer_date") session_key = "rev_%s_log_ordering_%s" % (sha1_git, 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, sha1_git, 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, } revs_ordering = request.GET.get("revs_ordering", "") prev_log_url = None if len(rev_log) > offset + per_page: prev_log_url = reverse( "browse-revision-log", url_args={"sha1_git": sha1_git}, query_params={ "per_page": str(per_page), "offset": str(offset + per_page), "revs_ordering": revs_ordering or None, }, ) next_log_url = None if offset != 0: next_log_url = reverse( "browse-revision-log", url_args={"sha1_git": sha1_git}, query_params={ "per_page": str(per_page), "offset": str(offset - per_page), "revs_ordering": revs_ordering or None, }, ) revision_log_data = format_log_entries(revision_log, per_page) swh_rev_id = str( CoreSWHID(object_type=ObjectType.REVISION, object_id=hash_to_bytes(sha1_git)) ) return render( request, - "browse/revision-log.html", + "browse-revision-log.html", { "heading": "Revision history", "swh_object_id": swh_rev_id, "swh_object_name": "Revisions history", "swh_object_metadata": None, "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": None, }, ) @browse_route( r"revision/(?P[0-9a-f]+)/", view_name="browse-revision", checksum_args=["sha1_git"], ) def revision_browse(request: HttpRequest, sha1_git: str) -> HttpResponse: """ Django view that produces an HTML display of a revision identified by its id. The url that points to it is :http:get:`/browse/revision/(sha1_git)/`. """ revision = archive.lookup_revision(sha1_git) origin_info = None snapshot_context = None origin_url = request.GET.get("origin_url") if not origin_url: origin_url = request.GET.get("origin") timestamp = request.GET.get("timestamp") visit_id = int(request.GET.get("visit_id", 0)) snapshot_id = request.GET.get("snapshot_id") if not snapshot_id: snapshot_id = request.GET.get("snapshot") path = request.GET.get("path") dir_id = None dirs, files = [], [] content_data = {} if origin_url: try: snapshot_context = get_snapshot_context( snapshot_id=snapshot_id, origin_url=origin_url, timestamp=timestamp, visit_id=visit_id or None, branch_name=request.GET.get("branch"), release_name=request.GET.get("release"), revision_id=sha1_git, path=path, ) except NotFoundExc as e: raw_rev_url = reverse("browse-revision", url_args={"sha1_git": sha1_git}) error_message = ( "The Software Heritage archive has a revision " "with the hash you provided but the origin " "mentioned in your request appears broken: %s. " "Please check the URL and try again.\n\n" "Nevertheless, you can still browse the revision " "without origin information: %s" % (gen_link(origin_url), gen_link(raw_rev_url)) ) if str(e).startswith("Origin"): raise NotFoundExc(error_message) else: raise e origin_info = snapshot_context["origin_info"] snapshot_id = snapshot_context["snapshot_id"] elif snapshot_id: snapshot_context = get_snapshot_context(snapshot_id) error_info: Dict[str, Any] = {"status_code": 200, "description": None} if path: try: file_info = archive.lookup_directory_with_path(revision["directory"], path) if file_info["type"] == "dir": dir_id = file_info["target"] else: query_string = "sha1_git:" + file_info["target"] content_data = request_content(query_string) except NotFoundExc as e: error_info["status_code"] = 404 error_info["description"] = f"NotFoundExc: {str(e)}" else: dir_id = revision["directory"] if dir_id: path = "" if path is None else (path + "/") dirs, files = get_directory_entries(dir_id) revision_metadata = RevisionMetadata( object_type=ObjectType.REVISION, object_id=sha1_git, revision=sha1_git, author=revision["author"]["fullname"] if revision["author"] else "None", author_url=gen_person_mail_link(revision["author"]) if revision["author"] else "None", committer=revision["committer"]["fullname"] if revision["committer"] else "None", committer_url=gen_person_mail_link(revision["committer"]) if revision["committer"] else "None", committer_date=format_utc_iso_date(revision["committer_date"]), date=format_utc_iso_date(revision["date"]), directory=revision["directory"], merge=revision["merge"], metadata=json.dumps( revision["metadata"], sort_keys=True, indent=4, separators=(",", ": ") ), parents=revision["parents"], synthetic=revision["synthetic"], type=revision["type"], snapshot=snapshot_id, origin_url=origin_url, ) message_lines = ["None"] if revision["message"]: message_lines = revision["message"].split("\n") parents = [] for p in revision["parents"]: parent_url = gen_revision_url(p, snapshot_context) parents.append({"id": p, "url": parent_url}) path_info = gen_path_info(path) query_params = snapshot_context["query_params"] if snapshot_context else {} breadcrumbs = [] breadcrumbs.append( { "name": revision["directory"][:7], "url": reverse( "browse-revision", url_args={"sha1_git": sha1_git}, query_params=query_params, ), } ) for pi in path_info: query_params["path"] = pi["path"] breadcrumbs.append( { "name": pi["name"], "url": reverse( "browse-revision", url_args={"sha1_git": sha1_git}, query_params=query_params, ), } ) vault_cooking = { "directory_context": False, "directory_swhid": None, "revision_context": True, "revision_swhid": f"swh:1:rev:{sha1_git}", } swh_objects = [SWHObjectInfo(object_type=ObjectType.REVISION, object_id=sha1_git)] content = None content_size = None filename = None mimetype = None language = None readme_name = None readme_url = None readme_html = None readmes = {} extra_context = dict(revision_metadata) extra_context["path"] = f"/{path}" if path else None if content_data: breadcrumbs[-1]["url"] = None content_size = content_data["length"] mimetype = content_data["mimetype"] if content_data["raw_data"]: 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"] if path: filename = path_info[-1]["name"] query_params["filename"] = filename filepath = "/".join(pi["name"] for pi in path_info[:-1]) extra_context["path"] = f"/{filepath}/" if filepath else "/" extra_context["filename"] = filename top_right_link = { "url": reverse( "browse-content-raw", url_args={"query_string": query_string}, query_params={"filename": filename}, ), "icon": swh_object_icons["content"], "text": "Raw File", } swh_objects.append( SWHObjectInfo(object_type=ObjectType.CONTENT, object_id=file_info["target"]) ) else: 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-revision", url_args={"sha1_git": sha1_git}, query_params=query_params, ) for f in files: query_params["path"] = path + f["name"] f["url"] = reverse( "browse-revision", url_args={"sha1_git": sha1_git}, query_params=query_params, ) if f["name"].lower().startswith("readme"): readmes[f["name"]] = f["checksums"]["sha1"] readme_name, readme_url, readme_html = get_readme_to_display(readmes) top_right_link = { "url": get_revision_log_url(sha1_git, snapshot_context), "icon": swh_object_icons["revisions history"], "text": "History", } vault_cooking["directory_context"] = True vault_cooking["directory_swhid"] = f"swh:1:dir:{dir_id}" swh_objects.append( SWHObjectInfo(object_type=ObjectType.DIRECTORY, object_id=dir_id) ) query_params.pop("path", None) diff_revision_url = reverse( "diff-revision", url_args={"sha1_git": sha1_git}, query_params=query_params, ) if snapshot_id: swh_objects.append( SWHObjectInfo(object_type=ObjectType.SNAPSHOT, object_id=snapshot_id) ) swhids_info = get_swhids_info(swh_objects, snapshot_context, extra_context) heading = "Revision - %s - %s" % ( sha1_git[:7], textwrap.shorten(message_lines[0], width=70), ) if snapshot_context: context_found = "snapshot: %s" % snapshot_context["snapshot_id"] if origin_info: context_found = "origin: %s" % origin_info["url"] heading += " - %s" % context_found return render( request, - "browse/revision.html", + "browse-revision.html", { "heading": heading, "swh_object_id": swhids_info[0]["swhid"], "swh_object_name": "Revision", "swh_object_metadata": revision_metadata, "message_header": message_lines[0], "message_body": "\n".join(message_lines[1:]), "parents": parents, "snapshot_context": snapshot_context, "dirs": dirs, "files": files, "content": content, "content_size": content_size, "max_content_size": content_display_max_size, "filename": filename, "encoding": content_data.get("encoding"), "mimetype": mimetype, "language": language, "readme_name": readme_name, "readme_url": readme_url, "readme_html": readme_html, "breadcrumbs": breadcrumbs, "top_right_link": top_right_link, "vault_cooking": vault_cooking, "diff_revision_url": diff_revision_url, "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"], ) diff --git a/swh/web/templates/browse/layout.html b/swh/web/templates/browse/layout.html deleted file mode 100644 index ac1ade77..00000000 --- a/swh/web/templates/browse/layout.html +++ /dev/null @@ -1,23 +0,0 @@ -{% extends "layout.html" %} - -{% 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 %} - -{% load swh_templatetags %} -{% load render_bundle from webpack_loader %} - -{% block title %}{{ heading }} – Software Heritage archive {% endblock %} - -{% block header %} -{% render_bundle 'browse' %} -{% render_bundle 'vault' %} -{% render_bundle 'save' %} -{% endblock %} - -{% block content %} -{% block browse-content %}{% endblock %} -{% endblock %} diff --git a/swh/web/tests/browse/views/test_content.py b/swh/web/tests/browse/views/test_content.py index ef83cb18..2acd5cfa 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) 2017-2021 The Software Heritage developers +# 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" + 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" + 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" + 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" + 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" + 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" + client, url, status_code=200, template_used="browse-content.html" ) assert_contains(resp, '