def _gen_diff_link(idx, diff_anchor, link_text):
if idx < _max_displayed_file_diffs:
return gen_link(diff_anchor, link_text)
else:
return link_text
def release_browse(request, sha1_git):
"""
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 = {}
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 = request.GET.get("visit_id")
if origin_url:
try:
snapshot_context = get_snapshot_context(
snapshot_id,
origin_url,
timestamp,
visit_id,
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"]
)
snapshot_id = snapshot_context.get("snapshot_id", None)
release_metadata = ReleaseMetadata(
object_type=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=RELEASE, object_id=sha1_git)]
vault_cooking = None
rev_directory = None
target_link = None
if release["target_type"] == REVISION:
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_id": rev_directory,
"revision_context": True,
"revision_id": release["target"],
}
swh_objects.append(
SWHObjectInfo(object_type=REVISION, object_id=release["target"])
)
swh_objects.append(
SWHObjectInfo(object_type=DIRECTORY, object_id=rev_directory)
)
except Exception as exc:
sentry_sdk.capture_exception(exc)
elif release["target_type"] == DIRECTORY:
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_id": release["target"],
"revision_context": False,
"revision_id": None,
}
swh_objects.append(
SWHObjectInfo(object_type=DIRECTORY, object_id=release["target"])
)
except Exception as exc:
sentry_sdk.capture_exception(exc)
elif release["target_type"] == CONTENT:
target_link = gen_content_link(
release["target"],
snapshot_context=snapshot_context,
link_text=None,
link_attrs=None,
)
swh_objects.append(
SWHObjectInfo(object_type=CONTENT, object_id=release["target"])
)
elif release["target_type"] == RELEASE:
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=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",
{
"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,
},
)
def revision_browse(request, sha1_git):
"""
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 = request.GET.get("visit_id")
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,
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 = {"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=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_id": None,
"revision_context": True,
"revision_id": sha1_git,
}
swh_objects = [SWHObjectInfo(object_type=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=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["length"] is not None:
f["length"] = filesizeformat(f["length"])
if f["name"].lower().startswith("readme"):
readmes[f["name"]] = f["checksums"]["sha1"]
readme_name, readme_url, readme_html = get_readme_to_display(readmes)
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_id"] = dir_id
swh_objects.append(
SWHObjectInfo(object_type=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=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",
{
"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"],
)
def release_browse(request, sha1_git):
"""
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)/`.
"""
try:
release = service.lookup_release(sha1_git)
snapshot_context = None
origin_info = None
snapshot_id = request.GET.get('snapshot_id', None)
origin_type = request.GET.get('origin_type', 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 = request.GET.get('visit_id', None)
if origin_url:
try:
snapshot_context = \
get_snapshot_context(snapshot_id, origin_type,
origin_url, timestamp,
visit_id)
except Exception:
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)))
raise NotFoundExc(error_message)
origin_info = snapshot_context['origin_info']
elif snapshot_id:
snapshot_context = get_snapshot_context(snapshot_id)
except Exception as exc:
return handle_view_exception(request, exc)
release_data = {}
author_name = 'None'
release_data['author'] = 'None'
if release['author']:
author_name = release['author']['name'] or \
release['author']['fullname']
release_data['author'] = \
gen_person_link(release['author']['id'], author_name,
snapshot_context)
release_data['date'] = format_utc_iso_date(release['date'])
release_data['id'] = sha1_git
release_data['name'] = release['name']
release_data['synthetic'] = release['synthetic']
release_data['target type'] = release['target_type']
if release['target_type'] == 'revision':
release_data['target'] = \
gen_revision_link(release['target'],
snapshot_context=snapshot_context)
elif release['target_type'] == 'content':
content_url = \
reverse('browse-content',
url_args={'query_string': 'sha1_git:' + release['target']})
release_data['target'] = gen_link(content_url, release['target'])
elif release['target_type'] == 'directory':
directory_url = \
reverse('browse-directory',
url_args={'sha1_git': release['target']})
release_data['target'] = gen_link(directory_url, release['target'])
elif release['target_type'] == 'release':
release_url = \
reverse('browse-release',
url_args={'sha1_git': release['target']})
release_data['target'] = gen_link(release_url, release['target'])
release_note_lines = []
if release['message']:
release_note_lines = release['message'].split('\n')
vault_cooking = None
query_params = {}
if snapshot_id:
query_params = {'snapshot_id': snapshot_id}
elif origin_info:
query_params = {'origin': origin_info['url']}
target_url = ''
if release['target_type'] == 'revision':
target_url = reverse('browse-revision',
url_args={'sha1_git': release['target']},
query_params=query_params)
try:
revision = service.lookup_revision(release['target'])
vault_cooking = {
'directory_context': True,
'directory_id': revision['directory'],
'revision_context': True,
'revision_id': release['target']
}
except Exception:
pass
elif release['target_type'] == 'directory':
target_url = reverse('browse-directory',
url_args={'sha1_git': release['target']},
query_params=query_params)
try:
revision = service.lookup_directory(release['target'])
vault_cooking = {
'directory_context': True,
'directory_id': revision['directory'],
'revision_context': False,
'revision_id': None
}
except Exception:
pass
elif release['target_type'] == 'content':
target_url = reverse('browse-content',
url_args={'query_string': release['target']},
query_params=query_params)
elif release['target_type'] == 'release':
target_url = reverse('browse-release',
url_args={'sha1_git': release['target']},
query_params=query_params)
release['target_url'] = target_url
if snapshot_context:
release_data['snapshot id'] = snapshot_context['snapshot_id']
if origin_info:
release_url = reverse('browse-release',
url_args={'sha1_git': release['id']})
release_data['context-independent release'] = \
gen_link(release_url, link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
release_data['origin id'] = origin_info['id']
release_data['origin type'] = origin_info['type']
release_data['origin url'] = gen_link(origin_info['url'],
origin_info['url'])
browse_snapshot_link = \
gen_snapshot_link(snapshot_context['snapshot_id'],
link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
release_data['snapshot'] = browse_snapshot_link
swh_objects = [{'type': 'release',
'id': sha1_git}]
if snapshot_context:
snapshot_id = snapshot_context['snapshot_id']
if snapshot_id:
swh_objects.append({'type': 'snapshot',
'id': snapshot_id})
swh_ids = get_swh_persistent_ids(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',
{'heading': heading,
'swh_object_id': swh_ids[0]['swh_id'],
'swh_object_name': 'Release',
'swh_object_metadata': release_data,
'release': release,
'snapshot_context': snapshot_context,
'show_actions_menu': True,
'breadcrumbs': None,
'vault_cooking': vault_cooking,
'top_right_link': None,
'swh_ids': swh_ids})
def directory_browse(request, sha1_git, path=None):
"""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)/]`
""" # noqa
root_sha1_git = sha1_git
try:
if path:
dir_info = service.lookup_directory_with_path(sha1_git, path)
# some readme files can reference assets reachable from the
# browsed directory, handle that special case in order to
# correctly displayed them
if dir_info and dir_info['type'] == 'file':
file_raw_url = reverse(
'browse-content-raw',
url_args={'query_string': dir_info['checksums']['sha1']})
return redirect(file_raw_url)
sha1_git = dir_info['target']
dirs, files = get_directory_entries(sha1_git)
origin_type = request.GET.get('origin_type', None)
origin_url = request.GET.get('origin_url', None)
if not origin_url:
origin_url = request.GET.get('origin', None)
snapshot_context = None
if origin_url:
try:
snapshot_context = get_snapshot_context(
None, origin_type, origin_url)
except Exception:
raw_dir_url = reverse('browse-directory',
url_args={'sha1_git': sha1_git})
error_message = \
('The Software Heritage archive has a directory '
'with the hash you provided but the origin '
'mentioned in your request appears broken: %s. '
'Please check the URL and try again.\n\n'
'Nevertheless, you can still browse the directory '
'without origin information: %s'
% (gen_link(origin_url), gen_link(raw_dir_url)))
raise NotFoundExc(error_message)
if snapshot_context:
snapshot_context['visit_info'] = None
except Exception as exc:
return handle_view_exception(request, exc)
path_info = gen_path_info(path)
query_params = {'origin': origin_url}
breadcrumbs = []
breadcrumbs.append({
'name':
root_sha1_git[:7],
'url':
reverse('browse-directory',
url_args={'sha1_git': root_sha1_git},
query_params=query_params)
})
for pi in path_info:
breadcrumbs.append({
'name':
pi['name'],
'url':
reverse('browse-directory',
url_args={
'sha1_git': root_sha1_git,
'path': pi['path']
},
query_params=query_params)
})
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,
'path': path + d['name']
},
query_params=query_params)
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={
'path': root_sha1_git + '/' + path + f['name'],
'origin': origin_url
})
if f['length'] is not None:
sum_file_sizes += f['length']
f['length'] = filesizeformat(f['length'])
if f['name'].lower().startswith('readme'):
readmes[f['name']] = f['checksums']['sha1']
readme_name, readme_url, readme_html = get_readme_to_display(readmes)
sum_file_sizes = filesizeformat(sum_file_sizes)
dir_metadata = {
'id': sha1_git,
'number of regular files': len(files),
'number of subdirectories': len(dirs),
'sum of regular file sizes': sum_file_sizes
}
vault_cooking = {
'directory_context': True,
'directory_id': sha1_git,
'revision_context': False,
'revision_id': None
}
swh_ids = get_swh_persistent_ids([{'type': 'directory', 'id': sha1_git}])
heading = 'Directory - %s' % sha1_git
if breadcrumbs:
dir_path = '/'.join([bc['name'] for bc in breadcrumbs]) + '/'
heading += ' - %s' % dir_path
return render(
request, 'browse/directory.html', {
'heading': heading,
'swh_object_id': swh_ids[0]['swh_id'],
'swh_object_name': 'Directory',
'swh_object_metadata': dir_metadata,
'dirs': dirs,
'files': files,
'breadcrumbs': breadcrumbs,
'top_right_link': None,
'readme_name': readme_name,
'readme_url': readme_url,
'readme_html': readme_html,
'snapshot_context': snapshot_context,
'vault_cooking': vault_cooking,
'show_actions_menu': True,
'swh_ids': swh_ids
})
def browse_snapshot_log(request,
snapshot_id=None,
origin_type=None,
origin_url=None,
timestamp=None):
"""
Django view implementation for browsing a revision history in a
snapshot context.
"""
try:
snapshot_context = _process_snapshot_request(
request,
snapshot_id,
origin_type,
origin_url,
timestamp,
browse_context='log') # noqa
revision_id = snapshot_context['revision_id']
per_page = int(request.GET.get('per_page', PER_PAGE))
offset = int(request.GET.get('offset', 0))
revs_ordering = request.GET.get('revs_ordering', 'committer_date')
session_key = 'rev_%s_log_ordering_%s' % (revision_id, revs_ordering)
rev_log_session = request.session.get(session_key, None)
rev_log = []
revs_walker_state = None
if rev_log_session:
rev_log = rev_log_session['rev_log']
revs_walker_state = rev_log_session['revs_walker_state']
if len(rev_log) < offset + per_page:
revs_walker = \
service.get_revisions_walker(revs_ordering,
revision_id,
max_revs=offset+per_page+1,
state=revs_walker_state)
rev_log += list(revs_walker)
revs_walker_state = revs_walker.export_state()
revision_log = rev_log[offset:offset + per_page]
request.session[session_key] = {
'rev_log': rev_log,
'revs_walker_state': revs_walker_state
}
except Exception as exc:
return handle_view_exception(request, exc)
swh_type = snapshot_context['swh_type']
origin_info = snapshot_context['origin_info']
visit_info = snapshot_context['visit_info']
url_args = snapshot_context['url_args']
query_params = snapshot_context['query_params']
snapshot_id = snapshot_context['snapshot_id']
query_params['per_page'] = per_page
revs_ordering = request.GET.get('revs_ordering', '')
query_params['revs_ordering'] = revs_ordering
browse_view_name = 'browse-' + swh_type + '-log'
prev_log_url = None
if len(rev_log) > offset + per_page:
query_params['offset'] = offset + per_page
prev_log_url = reverse(browse_view_name,
url_args=url_args,
query_params=query_params)
next_log_url = None
if offset != 0:
query_params['offset'] = offset - per_page
next_log_url = reverse(browse_view_name,
url_args=url_args,
query_params=query_params)
revision_log_data = format_log_entries(revision_log, per_page,
snapshot_context)
browse_log_link = \
gen_revision_log_link(revision_id, link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
revision_metadata = {
'context-independent revision history': browse_log_link,
'snapshot id': snapshot_id
}
if origin_info:
revision_metadata['origin id'] = origin_info['id']
revision_metadata['origin type'] = origin_info['type']
revision_metadata['origin url'] = origin_info['url']
revision_metadata['origin visit date'] = format_utc_iso_date(
visit_info['date']) # noqa
revision_metadata['origin visit id'] = visit_info['visit']
browse_snapshot_url = reverse('browse-snapshot-log',
url_args={'snapshot_id': snapshot_id},
query_params=request.GET)
browse_snapshot_link = \
gen_link(browse_snapshot_url, link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
revision_metadata['snapshot context'] = browse_snapshot_link
swh_objects = [{
'type': 'revision',
'id': revision_id
}, {
'type': 'snapshot',
'id': snapshot_id
}]
release_id = snapshot_context['release_id']
if release_id:
swh_objects.append({'type': 'release', 'id': release_id})
swh_ids = get_swh_persistent_ids(swh_objects, snapshot_context)
context_found = 'snapshot: %s' % snapshot_context['snapshot_id']
if origin_info:
context_found = 'origin: %s' % origin_info['url']
heading = 'Revision history - %s - %s' %\
(snapshot_context['branch'], context_found)
return render(
request, 'browse/revision-log.html', {
'heading': heading,
'swh_object_name': 'Revisions history',
'swh_object_metadata': revision_metadata,
'revision_log': revision_log_data,
'revs_ordering': revs_ordering,
'next_log_url': next_log_url,
'prev_log_url': prev_log_url,
'breadcrumbs': None,
'top_right_link': None,
'snapshot_context': snapshot_context,
'vault_cooking': None,
'show_actions_menu': True,
'swh_ids': swh_ids
})
def test_gen_link():
assert (gen_link(
"https://www.softwareheritage.org/",
"swh") == '<a href="https://www.softwareheritage.org/">swh</a>')
def browse_snapshot_directory(request,
snapshot_id=None,
origin_type=None,
origin_url=None,
timestamp=None,
path=None):
"""
Django view implementation for browsing a directory in a snapshot context.
"""
try:
snapshot_context = _process_snapshot_request(
request,
snapshot_id,
origin_type,
origin_url,
timestamp,
path,
browse_context='directory') # noqa
root_sha1_git = snapshot_context['root_sha1_git']
sha1_git = root_sha1_git
if root_sha1_git and path:
dir_info = service.lookup_directory_with_path(root_sha1_git, path)
# some readme files can reference assets reachable from the
# browsed directory, handle that special case in order to
# correctly displayed them
if dir_info and dir_info['type'] == 'file':
file_raw_url = reverse(
'browse-content-raw',
url_args={'query_string': dir_info['checksums']['sha1']})
return redirect(file_raw_url)
sha1_git = dir_info['target']
dirs = []
files = []
if sha1_git:
dirs, files = get_directory_entries(sha1_git)
except Exception as exc:
return handle_view_exception(request, exc)
swh_type = snapshot_context['swh_type']
origin_info = snapshot_context['origin_info']
visit_info = snapshot_context['visit_info']
url_args = snapshot_context['url_args']
query_params = snapshot_context['query_params']
revision_id = snapshot_context['revision_id']
snapshot_id = snapshot_context['snapshot_id']
path_info = gen_path_info(path)
browse_view_name = 'browse-' + swh_type + '-directory'
breadcrumbs = []
if root_sha1_git:
breadcrumbs.append({
'name':
root_sha1_git[:7],
'url':
reverse(browse_view_name,
url_args=url_args,
query_params=query_params)
})
for pi in path_info:
bc_url_args = dict(url_args)
bc_url_args['path'] = pi['path']
breadcrumbs.append({
'name':
pi['name'],
'url':
reverse(browse_view_name,
url_args=bc_url_args,
query_params=query_params)
})
path = '' if path is None else (path + '/')
for d in dirs:
if d['type'] == 'rev':
d['url'] = reverse('browse-revision',
url_args={'sha1_git': d['target']})
else:
bc_url_args = dict(url_args)
bc_url_args['path'] = path + d['name']
d['url'] = reverse(browse_view_name,
url_args=bc_url_args,
query_params=query_params)
sum_file_sizes = 0
readmes = {}
browse_view_name = 'browse-' + swh_type + '-content'
for f in files:
bc_url_args = dict(url_args)
bc_url_args['path'] = path + f['name']
f['url'] = reverse(browse_view_name,
url_args=bc_url_args,
query_params=query_params)
if f['length'] is not None:
sum_file_sizes += f['length']
f['length'] = filesizeformat(f['length'])
if f['name'].lower().startswith('readme'):
readmes[f['name']] = f['checksums']['sha1']
readme_name, readme_url, readme_html = get_readme_to_display(readmes)
browse_view_name = 'browse-' + swh_type + '-log'
history_url = None
if snapshot_id != _empty_snapshot_id:
history_url = reverse(browse_view_name,
url_args=url_args,
query_params=query_params)
nb_files = None
nb_dirs = None
sum_file_sizes = None
dir_path = None
if root_sha1_git:
nb_files = len(files)
nb_dirs = len(dirs)
sum_file_sizes = filesizeformat(sum_file_sizes)
dir_path = '/' + path
browse_dir_link = \
gen_directory_link(sha1_git, link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
browse_rev_link = \
gen_revision_link(revision_id,
snapshot_context=snapshot_context,
link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
dir_metadata = {
'id': sha1_git,
'context-independent directory': browse_dir_link,
'number of regular files': nb_files,
'number of subdirectories': nb_dirs,
'sum of regular file sizes': sum_file_sizes,
'path': dir_path,
'revision id': revision_id,
'revision': browse_rev_link,
'snapshot id': snapshot_id
}
if origin_info:
dir_metadata['origin id'] = origin_info['id']
dir_metadata['origin type'] = origin_info['type']
dir_metadata['origin url'] = origin_info['url']
dir_metadata['origin visit date'] = format_utc_iso_date(
visit_info['date']) # noqa
dir_metadata['origin visit id'] = visit_info['visit']
snapshot_context_url = reverse('browse-snapshot-directory',
url_args={'snapshot_id': snapshot_id},
query_params=request.GET)
browse_snapshot_link = \
gen_link(snapshot_context_url, link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
dir_metadata['snapshot context'] = browse_snapshot_link
vault_cooking = {
'directory_context': True,
'directory_id': sha1_git,
'revision_context': True,
'revision_id': revision_id
}
swh_objects = [{
'type': 'directory',
'id': sha1_git
}, {
'type': 'revision',
'id': revision_id
}, {
'type': 'snapshot',
'id': snapshot_id
}]
release_id = snapshot_context['release_id']
if release_id:
swh_objects.append({'type': 'release', 'id': release_id})
swh_ids = get_swh_persistent_ids(swh_objects, snapshot_context)
dir_path = '/'.join([bc['name'] for bc in breadcrumbs]) + '/'
context_found = 'snapshot: %s' % snapshot_context['snapshot_id']
if origin_info:
context_found = 'origin: %s' % origin_info['url']
heading = 'Directory - %s - %s - %s' %\
(dir_path, snapshot_context['branch'], context_found)
return render(
request, 'browse/directory.html', {
'heading': heading,
'swh_object_name': 'Directory',
'swh_object_metadata': dir_metadata,
'dirs': dirs,
'files': files,
'breadcrumbs': breadcrumbs if root_sha1_git else [],
'top_right_link': {
'url': history_url,
'icon': swh_object_icons['revisions history'],
'text': 'History'
},
'readme_name': readme_name,
'readme_url': readme_url,
'readme_html': readme_html,
'snapshot_context': snapshot_context,
'vault_cooking': vault_cooking,
'show_actions_menu': True,
'swh_ids': swh_ids
})
def browse_snapshot_content(request,
snapshot_id=None,
origin_type=None,
origin_url=None,
timestamp=None,
path=None):
"""
Django view implementation for browsing a content in a snapshot context.
"""
try:
snapshot_context = _process_snapshot_request(request,
snapshot_id,
origin_type,
origin_url,
timestamp,
path,
browse_context='content')
root_sha1_git = snapshot_context['root_sha1_git']
sha1_git = None
query_string = None
content_data = None
if root_sha1_git:
content_info = service.lookup_directory_with_path(
root_sha1_git, path)
sha1_git = content_info['target']
query_string = 'sha1_git:' + sha1_git
content_data = request_content(query_string,
raise_if_unavailable=False)
except Exception as exc:
return handle_view_exception(request, exc)
swh_type = snapshot_context['swh_type']
url_args = snapshot_context['url_args']
query_params = snapshot_context['query_params']
revision_id = snapshot_context['revision_id']
origin_info = snapshot_context['origin_info']
visit_info = snapshot_context['visit_info']
snapshot_id = snapshot_context['snapshot_id']
content = None
language = None
mimetype = None
if content_data and content_data['raw_data'] is not None:
content_display_data = prepare_content_for_display(
content_data['raw_data'], content_data['mimetype'], path)
content = content_display_data['content_data']
language = content_display_data['language']
mimetype = content_display_data['mimetype']
filename = None
path_info = None
browse_view_name = 'browse-' + swh_type + '-directory'
breadcrumbs = []
split_path = path.split('/')
filename = split_path[-1]
path_info = gen_path_info(path[:-len(filename)])
if root_sha1_git:
breadcrumbs.append({
'name':
root_sha1_git[:7],
'url':
reverse(browse_view_name,
url_args=url_args,
query_params=query_params)
})
for pi in path_info:
bc_url_args = dict(url_args)
bc_url_args['path'] = pi['path']
breadcrumbs.append({
'name':
pi['name'],
'url':
reverse(browse_view_name,
url_args=bc_url_args,
query_params=query_params)
})
breadcrumbs.append({'name': filename, 'url': None})
browse_content_link = \
gen_content_link(sha1_git, link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
content_raw_url = None
if query_string:
content_raw_url = reverse('browse-content-raw',
url_args={'query_string': query_string},
query_params={'filename': filename})
browse_rev_link = \
gen_revision_link(revision_id,
snapshot_context=snapshot_context,
link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
content_metadata = {
'context-independent content': browse_content_link,
'path': None,
'filename': None,
'revision id': revision_id,
'revision': browse_rev_link,
'snapshot id': snapshot_id
}
cnt_sha1_git = None
content_size = None
error_code = 200
error_description = ''
error_message = ''
if content_data:
content_metadata['sha1 checksum'] = \
content_data['checksums']['sha1']
content_metadata['sha1_git checksum'] = \
content_data['checksums']['sha1_git']
content_metadata['sha256 checksum'] = \
content_data['checksums']['sha256']
content_metadata['blake2s256 checksum'] = \
content_data['checksums']['blake2s256']
content_metadata['mime type'] = content_data['mimetype']
content_metadata['encoding'] = content_data['encoding']
content_metadata['size'] = filesizeformat(content_data['length'])
content_metadata['language'] = content_data['language']
content_metadata['licenses'] = content_data['licenses']
content_metadata['path'] = '/' + path[:-len(filename)]
content_metadata['filename'] = filename
cnt_sha1_git = content_data['checksums']['sha1_git']
content_size = content_data['length']
error_code = content_data['error_code']
error_message = content_data['error_message']
error_description = content_data['error_description']
if origin_info:
content_metadata['origin id'] = origin_info['id']
content_metadata['origin type'] = origin_info['type']
content_metadata['origin url'] = origin_info['url']
content_metadata['origin visit date'] = format_utc_iso_date(
visit_info['date']) # noqa
content_metadata['origin visit id'] = visit_info['visit']
browse_snapshot_url = reverse('browse-snapshot-content',
url_args={
'snapshot_id': snapshot_id,
'path': path
},
query_params=request.GET)
browse_snapshot_link = \
gen_link(browse_snapshot_url, link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
content_metadata['snapshot context'] = browse_snapshot_link
swh_objects = [{
'type': 'content',
'id': cnt_sha1_git
}, {
'type': 'revision',
'id': revision_id
}, {
'type': 'snapshot',
'id': snapshot_id
}]
release_id = snapshot_context['release_id']
if release_id:
swh_objects.append({'type': 'release', 'id': release_id})
swh_ids = get_swh_persistent_ids(swh_objects, snapshot_context)
content_path = '/'.join([bc['name'] for bc in breadcrumbs])
context_found = 'snapshot: %s' % snapshot_context['snapshot_id']
if origin_info:
context_found = 'origin: %s' % origin_info['url']
heading = 'Content - %s - %s - %s' %\
(content_path, snapshot_context['branch'], context_found)
return render(request,
'browse/content.html', {
'heading': heading,
'swh_object_name': 'Content',
'swh_object_metadata': content_metadata,
'content': content,
'content_size': content_size,
'max_content_size': content_display_max_size,
'mimetype': mimetype,
'language': language,
'breadcrumbs': breadcrumbs if root_sha1_git else [],
'top_right_link': {
'url': content_raw_url,
'icon': swh_object_icons['content'],
'text': 'Raw File'
},
'snapshot_context': snapshot_context,
'vault_cooking': None,
'show_actions_menu': True,
'swh_ids': swh_ids,
'error_code': error_code,
'error_message': error_message,
'error_description': error_description
},
status=error_code)
def content_display(request, query_string):
"""Django view that produces an HTML display of a content identified
by its hash value.
The url that points to it is
:http:get:`/browse/content/[(algo_hash):](hash)/`
"""
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")
path = request.GET.get("path")
content_data = {}
error_info = {"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:
snapshot_context = get_snapshot_context(
origin_url=origin_url,
snapshot_id=snapshot_id,
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"):
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": None})
if path 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=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=filesizeformat(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 = [
SWHObjectInfo(object_type=CONTENT,
object_id=content_checksums.get("sha1_git"))
]
if directory_id:
swh_objects.append(
SWHObjectInfo(object_type=DIRECTORY, object_id=directory_id))
if snapshot_context:
swh_objects.append(
SWHObjectInfo(object_type=REVISION,
object_id=snapshot_context["revision_id"]))
swh_objects.append(
SWHObjectInfo(object_type=SNAPSHOT,
object_id=snapshot_context["snapshot_id"]))
if snapshot_context["release_id"]:
swh_objects.append(
SWHObjectInfo(object_type=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",
{
"heading":
heading,
"swh_object_id":
swhids_info[0]["swhid"],
"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"],
)
def content_display(request, query_string):
"""Django view that produces an HTML display of a content identified
by its hash value.
The url that points to it is :http:get:`/browse/content/[(algo_hash):](hash)/`
""" # noqa
try:
algo, checksum = query.parse_hash(query_string)
checksum = hash_to_hex(checksum)
content_data = request_content(query_string,
raise_if_unavailable=False)
origin_type = request.GET.get('origin_type', None)
origin_url = request.GET.get('origin_url', None)
if not origin_url:
origin_url = request.GET.get('origin', None)
snapshot_context = None
if origin_url:
try:
snapshot_context = get_snapshot_context(
None, origin_type, origin_url)
except Exception:
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)
if snapshot_context:
snapshot_context['visit_info'] = None
except Exception as exc:
return handle_view_exception(request, exc)
path = request.GET.get('path', None)
content = None
language = None
mimetype = None
if content_data['raw_data'] is not None:
content_display_data = prepare_content_for_display(
content_data['raw_data'], content_data['mimetype'], path)
content = content_display_data['content_data']
language = content_display_data['language']
mimetype = content_display_data['mimetype']
root_dir = None
filename = None
path_info = None
query_params = {'origin': origin_url}
breadcrumbs = []
if path:
split_path = path.split('/')
root_dir = 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)
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:
dir_url = reverse('browse-directory',
url_args={
'sha1_git': root_dir,
'path': pi['path']
},
query_params=query_params)
breadcrumbs.append({'name': pi['name'], 'url': dir_url})
breadcrumbs.append({'name': filename, 'url': None})
query_params = {'filename': filename}
content_raw_url = reverse('browse-content-raw',
url_args={'query_string': query_string},
query_params=query_params)
content_metadata = {
'sha1 checksum': content_data['checksums']['sha1'],
'sha1_git checksum': content_data['checksums']['sha1_git'],
'sha256 checksum': content_data['checksums']['sha256'],
'blake2s256 checksum': content_data['checksums']['blake2s256'],
'mime type': content_data['mimetype'],
'encoding': content_data['encoding'],
'size': filesizeformat(content_data['length']),
'language': content_data['language'],
'licenses': content_data['licenses'],
'filename': filename
}
if filename:
content_metadata['filename'] = filename
sha1_git = content_data['checksums']['sha1_git']
swh_ids = get_swh_persistent_ids([{'type': 'content', 'id': sha1_git}])
heading = 'Content - %s' % sha1_git
if breadcrumbs:
content_path = '/'.join([bc['name'] for bc in breadcrumbs])
heading += ' - %s' % content_path
return render(request,
'browse/content.html', {
'heading': heading,
'swh_object_id': swh_ids[0]['swh_id'],
'swh_object_name': 'Content',
'swh_object_metadata': content_metadata,
'content': content,
'content_size': content_data['length'],
'max_content_size': content_display_max_size,
'mimetype': mimetype,
'language': language,
'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_menu': True,
'swh_ids': swh_ids,
'error_code': content_data['error_code'],
'error_message': content_data['error_message'],
'error_description': content_data['error_description']
},
status=content_data['error_code'])
def test_gen_link(self):
self.assertEqual(
utils.gen_link('https://www.softwareheritage.org/', 'swh'),
'<a href="https://www.softwareheritage.org/">swh</a>')
def _directory_browse(request, sha1_git, path=None):
root_sha1_git = sha1_git
error_info = {"status_code": 200, "description": None}
if path:
try:
dir_info = archive.lookup_directory_with_path(sha1_git, path)
sha1_git = dir_info["target"]
except NotFoundExc as e:
error_info["status_code"] = 404
error_info["description"] = f"NotFoundExc: {str(e)}"
sha1_git = None
dirs, files = [], []
if sha1_git is not None:
dirs, files = get_directory_entries(sha1_git)
origin_url = request.GET.get("origin_url")
if not origin_url:
origin_url = request.GET.get("origin")
snapshot_id = request.GET.get("snapshot")
snapshot_context = None
if origin_url is not None or snapshot_id is not None:
try:
snapshot_context = get_snapshot_context(
snapshot_id=snapshot_id,
origin_url=origin_url,
branch_name=request.GET.get("branch"),
release_name=request.GET.get("release"),
revision_id=request.GET.get("revision"),
path=path,
)
except NotFoundExc as e:
if str(e).startswith("Origin"):
raw_dir_url = reverse(
"browse-directory", url_args={"sha1_git": sha1_git}
)
error_message = (
"The Software Heritage archive has a directory "
"with the hash you provided but the origin "
"mentioned in your request appears broken: %s. "
"Please check the URL and try again.\n\n"
"Nevertheless, you can still browse the directory "
"without origin information: %s"
% (gen_link(origin_url), gen_link(raw_dir_url))
)
raise NotFoundExc(error_message)
else:
raise e
path_info = gen_path_info(path)
query_params = snapshot_context["query_params"] if snapshot_context else {}
breadcrumbs = []
breadcrumbs.append(
{
"name": root_sha1_git[:7],
"url": reverse(
"browse-directory",
url_args={"sha1_git": root_sha1_git},
query_params={**query_params, "path": None},
),
}
)
for pi in path_info:
breadcrumbs.append(
{
"name": pi["name"],
"url": reverse(
"browse-directory",
url_args={"sha1_git": root_sha1_git},
query_params={**query_params, "path": pi["path"],},
),
}
)
path = "" if path is None else (path + "/")
for d in dirs:
if d["type"] == "rev":
d["url"] = reverse(
"browse-revision",
url_args={"sha1_git": d["target"]},
query_params=query_params,
)
else:
d["url"] = reverse(
"browse-directory",
url_args={"sha1_git": root_sha1_git},
query_params={**query_params, "path": path + d["name"],},
)
sum_file_sizes = 0
readmes = {}
for f in files:
query_string = "sha1_git:" + f["target"]
f["url"] = reverse(
"browse-content",
url_args={"query_string": query_string},
query_params={
**query_params,
"path": root_sha1_git + "/" + path + f["name"],
},
)
if f["length"] is not None:
sum_file_sizes += f["length"]
f["length"] = filesizeformat(f["length"])
if f["name"].lower().startswith("readme"):
readmes[f["name"]] = f["checksums"]["sha1"]
readme_name, readme_url, readme_html = get_readme_to_display(readmes)
sum_file_sizes = filesizeformat(sum_file_sizes)
dir_metadata = DirectoryMetadata(
object_type=DIRECTORY,
object_id=sha1_git,
directory=root_sha1_git,
nb_files=len(files),
nb_dirs=len(dirs),
sum_file_sizes=sum_file_sizes,
root_directory=root_sha1_git,
path=f"/{path}" if path else None,
revision=None,
revision_found=None,
release=None,
snapshot=None,
)
vault_cooking = {
"directory_context": True,
"directory_id": sha1_git,
"revision_context": False,
"revision_id": None,
}
swh_objects = [SWHObjectInfo(object_type=DIRECTORY, object_id=sha1_git)]
if snapshot_context:
swh_objects.append(
SWHObjectInfo(
object_type=REVISION, object_id=snapshot_context["revision_id"]
)
)
swh_objects.append(
SWHObjectInfo(
object_type=SNAPSHOT, object_id=snapshot_context["snapshot_id"]
)
)
if snapshot_context["release_id"]:
swh_objects.append(
SWHObjectInfo(
object_type=RELEASE, object_id=snapshot_context["release_id"]
)
)
swhids_info = get_swhids_info(swh_objects, snapshot_context, dir_metadata)
heading = "Directory - %s" % sha1_git
if breadcrumbs:
dir_path = "/".join([bc["name"] for bc in breadcrumbs]) + "/"
heading += " - %s" % dir_path
top_right_link = None
if snapshot_context is not None and not snapshot_context["is_empty"]:
history_url = reverse(
"browse-revision-log",
url_args={"sha1_git": snapshot_context["revision_id"]},
query_params=query_params,
)
top_right_link = {
"url": history_url,
"icon": swh_object_icons["revisions history"],
"text": "History",
}
return render(
request,
"browse/directory.html",
{
"heading": heading,
"swh_object_id": swhids_info[0]["swhid"],
"swh_object_name": "Directory",
"swh_object_metadata": dir_metadata,
"dirs": dirs,
"files": files,
"breadcrumbs": breadcrumbs,
"top_right_link": top_right_link,
"readme_name": readme_name,
"readme_url": readme_url,
"readme_html": readme_html,
"snapshot_context": snapshot_context,
"vault_cooking": vault_cooking,
"show_actions": True,
"swhids_info": swhids_info,
"error_code": error_info["status_code"],
"error_message": http_status_code_message.get(error_info["status_code"]),
"error_description": error_info["description"],
},
status=error_info["status_code"],
)
def revision_browse(request, sha1_git, extra_path=None):
"""
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)/`.
"""
try:
revision = service.lookup_revision(sha1_git)
# some readme files can reference assets reachable from the
# browsed directory, handle that special case in order to
# correctly displayed them
if extra_path:
dir_info = \
service.lookup_directory_with_path(revision['directory'],
extra_path)
if dir_info and dir_info['type'] == 'file':
file_raw_url = reverse(
'browse-content-raw',
url_args={'query_string': dir_info['checksums']['sha1']})
return redirect(file_raw_url)
origin_info = None
snapshot_context = None
origin_type = request.GET.get('origin_type', 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 = request.GET.get('visit_id', None)
snapshot_id = request.GET.get('snapshot_id', None)
path = request.GET.get('path', None)
dir_id = None
dirs, files = None, None
content_data = None
if origin_url:
try:
snapshot_context = get_snapshot_context(
None, origin_type, origin_url, timestamp, visit_id)
except Exception:
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)))
raise NotFoundExc(error_message)
origin_info = snapshot_context['origin_info']
snapshot_id = snapshot_context['snapshot_id']
elif snapshot_id:
snapshot_context = get_snapshot_context(snapshot_id)
if path:
file_info = \
service.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,
raise_if_unavailable=False)
else:
dir_id = revision['directory']
if dir_id:
path = '' if path is None else (path + '/')
dirs, files = get_directory_entries(dir_id)
except Exception as exc:
return handle_view_exception(request, exc)
revision_data = {}
author_name = 'None'
revision_data['author'] = 'None'
if revision['author']:
author_name = revision['author']['name'] or \
revision['author']['fullname']
revision_data['author'] = \
gen_person_link(revision['author']['id'], author_name,
snapshot_context)
revision_data['committer'] = 'None'
if revision['committer']:
revision_data['committer'] = \
gen_person_link(revision['committer']['id'],
revision['committer']['name'], snapshot_context)
revision_data['committer date'] = \
format_utc_iso_date(revision['committer_date'])
revision_data['date'] = format_utc_iso_date(revision['date'])
if snapshot_context:
revision_data['snapshot id'] = snapshot_id
revision_data['directory'] = \
gen_snapshot_directory_link(snapshot_context, sha1_git,
link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm', # noqa
'role': 'button'})
else:
revision_data['directory'] = \
gen_directory_link(revision['directory'], link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
revision_data['id'] = sha1_git
revision_data['merge'] = revision['merge']
revision_data['metadata'] = escape(
json.dumps(revision['metadata'],
sort_keys=True,
indent=4,
separators=(',', ': ')))
if origin_info:
revision_data['context-independent revision'] = \
gen_revision_link(sha1_git, link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
revision_data['origin id'] = origin_info['id']
revision_data['origin type'] = origin_info['type']
revision_data['origin url'] = gen_link(origin_info['url'],
origin_info['url'])
browse_snapshot_link = \
gen_snapshot_link(snapshot_id, link_text='Browse',
link_attrs={'class': 'btn btn-default btn-sm',
'role': 'button'})
revision_data['snapshot'] = browse_snapshot_link
parents = ''
for p in revision['parents']:
parent_link = gen_revision_link(p, snapshot_context=snapshot_context)
parents += parent_link + '<br/>'
revision_data['parents'] = mark_safe(parents)
revision_data['synthetic'] = revision['synthetic']
revision_data['type'] = revision['type']
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_id': snapshot_id,
'origin_type': origin_type,
'origin': origin_url,
'timestamp': timestamp,
'visit_id': visit_id
}
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_id': None,
'revision_context': True,
'revision_id': sha1_git
}
swh_objects = [{'type': 'revision', 'id': sha1_git}]
content = None
content_size = None
mimetype = None
language = None
readme_name = None
readme_url = None
readme_html = None
readmes = {}
error_code = 200
error_message = ''
error_description = ''
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']
query_params = {}
if path:
filename = path_info[-1]['name']
query_params['filename'] = path_info[-1]['name']
revision_data['filename'] = filename
top_right_link = {
'url':
reverse('browse-content-raw',
url_args={'query_string': query_string},
query_params=query_params),
'icon':
swh_object_icons['content'],
'text':
'Raw File'
}
swh_objects.append({'type': 'content', 'id': file_info['target']})
error_code = content_data['error_code']
error_message = content_data['error_message']
error_description = content_data['error_description']
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['length'] is not None:
f['length'] = filesizeformat(f['length'])
if f['name'].lower().startswith('readme'):
readmes[f['name']] = f['checksums']['sha1']
readme_name, readme_url, readme_html = get_readme_to_display(readmes)
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_id'] = dir_id
swh_objects.append({'type': 'directory', 'id': dir_id})
diff_revision_url = reverse('diff-revision',
url_args={'sha1_git': sha1_git},
query_params={
'origin_type': origin_type,
'origin': origin_url,
'timestamp': timestamp,
'visit_id': visit_id
})
if snapshot_id:
swh_objects.append({'type': 'snapshot', 'id': snapshot_id})
swh_ids = get_swh_persistent_ids(swh_objects, snapshot_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', {
'heading': heading,
'swh_object_id': swh_ids[0]['swh_id'],
'swh_object_name': 'Revision',
'swh_object_metadata': revision_data,
'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,
'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_menu': True,
'swh_ids': swh_ids,
'error_code': error_code,
'error_message': error_message,
'error_description': error_description
},
status=error_code)
