diff --git a/swh/web/browse/urls.py b/swh/web/browse/urls.py index 57eb3e3d..31c8587d 100644 --- a/swh/web/browse/urls.py +++ b/swh/web/browse/urls.py @@ -1,45 +1,45 @@ # Copyright (C) 2017-2018 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.conf.urls import url from django.shortcuts import render import swh.web.browse.views.directory # noqa import swh.web.browse.views.content # noqa import swh.web.browse.views.origin # noqa import swh.web.browse.views.person # 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.browse.browseurls import BrowseUrls from swh.web.browse.identifiers import swh_id_browse def _browse_help_view(request): return render(request, 'browse-help.html', - {'heading': 'Help'}) + {'heading': 'How to browse the archive ?'}) def _browse_search_view(request): return render(request, 'browse-search.html', - {'heading': 'Search'}) + {'heading': 'Search software origins to browse'}) def _browse_vault_view(request): return render(request, 'browse-vault-ui.html', - {'heading': 'Vault'}) + {'heading': 'Download archive content from the Vault'}) urlpatterns = [ url(r'^$', _browse_search_view), url(r'^help/$', _browse_help_view, name='browse-help'), url(r'^search/$', _browse_search_view, name='browse-search'), url(r'^vault/$', _browse_vault_view, name='browse-vault'), # for backward compatibility url(r'^(?Pswh:[0-9]+:[a-z]+:[0-9a-f]+.*)/$', swh_id_browse) ] urlpatterns += BrowseUrls.get_url_patterns() diff --git a/swh/web/browse/views/content.py b/swh/web/browse/views/content.py index aef2ec70..3d4aaa94 100644 --- a/swh/web/browse/views/content.py +++ b/swh/web/browse/views/content.py @@ -1,257 +1,262 @@ # Copyright (C) 2017-2018 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 import json from distutils.util import strtobool from django.http import HttpResponse from django.utils.safestring import mark_safe from django.shortcuts import render from django.template.defaultfilters import filesizeformat from swh.model.hashutil import hash_to_hex from swh.web.common import query from swh.web.common.utils import ( reverse, gen_path_info ) from swh.web.common.exc import handle_view_exception from swh.web.browse.utils import ( request_content, prepare_content_for_display, content_display_max_size, get_snapshot_context, get_swh_persistent_ids ) from swh.web.browse.browseurls import browse_route @browse_route(r'content/(?P.+)/raw/', view_name='browse-content-raw') def content_raw(request, query_string): """Django view that produces a raw display of a SWH content identified by its hash value. The url that points to it is :http:get:`/browse/content/[(algo_hash):](hash)/raw/` """ # noqa try: algo, checksum = query.parse_hash(query_string) checksum = hash_to_hex(checksum) content_data = request_content(query_string, max_size=None) except Exception as exc: return handle_view_exception(request, exc) 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.*)', # noqa view_name='diff-contents') def _contents_diff(request, from_query_string, to_query_string): """ 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 = request.GET.get('force', 'false') path = request.GET.get('path', None) language = 'nohighlight-swh' force = bool(strtobool(force)) if from_query_string == to_query_string: diff_str = 'File renamed without changes' else: 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 = 'nohighlight-swh' elif not force and diff_size > _auto_diff_size_limit: diff_str = 'Large diffs are not automatically computed' language = 'nohighlight-swh' 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:]) diff_data['diff_str'] = diff_str diff_data['language'] = language diff_data_json = json.dumps(diff_data, separators=(',', ': ')) return HttpResponse(diff_data_json, content_type='application/json') @browse_route(r'content/(?P.+)/', view_name='browse-content') def content_display(request, query_string): """Django view that produces an HTML display of a SWH 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) 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: snapshot_context = get_snapshot_context(None, origin_type, origin_url) 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 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'] root_dir = None filename = None path_info = None breadcrumbs = [] if path: split_path = path.split('/') root_dir = split_path[0] filename = split_path[-1] path = path.replace(root_dir + '/', '') path = path[:-len(filename)] path_info = gen_path_info(path) breadcrumbs.append({'name': root_dir[:7], 'url': reverse('browse-directory', kwargs={'sha1_git': root_dir})}) for pi in path_info: breadcrumbs.append({'name': pi['name'], 'url': reverse('browse-directory', kwargs={'sha1_git': root_dir, 'path': pi['path']})}) breadcrumbs.append({'name': filename, 'url': None}) query_params = None if filename: query_params = {'filename': filename} content_raw_url = reverse('browse-content-raw', kwargs={'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'] } 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, 'content.html', - {'heading': 'Content', + {'heading': heading, 'swh_object_name': 'Content', 'swh_object_metadata': content_metadata, 'content': content, 'content_size': content_data['length'], 'max_content_size': content_display_max_size, 'mimetype': content_data['mimetype'], 'language': language, 'breadcrumbs': breadcrumbs, 'top_right_link': content_raw_url, 'top_right_link_text': mark_safe( 'Raw File'), 'snapshot_context': snapshot_context, 'vault_cooking': None, 'show_actions_menu': True, 'swh_ids': swh_ids}) diff --git a/swh/web/browse/views/directory.py b/swh/web/browse/views/directory.py index fc7defe1..c0abced7 100644 --- a/swh/web/browse/views/directory.py +++ b/swh/web/browse/views/directory.py @@ -1,128 +1,133 @@ # Copyright (C) 2017-2018 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.shortcuts import render, redirect from django.template.defaultfilters import filesizeformat from swh.web.common import service from swh.web.common.utils import ( reverse, gen_path_info ) from swh.web.common.exc import handle_view_exception from swh.web.browse.utils import ( get_directory_entries, get_snapshot_context, get_readme_to_display, get_swh_persistent_ids ) from swh.web.browse.browseurls import browse_route @browse_route(r'directory/(?P[0-9a-f]+)/', r'directory/(?P[0-9a-f]+)/(?P.+)/', view_name='browse-directory') def directory_browse(request, sha1_git, path=None): """Django view for browsing the content of a SWH 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', kwargs={'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: snapshot_context = get_snapshot_context(None, origin_type, origin_url) if snapshot_context: snapshot_context['visit_info'] = None except Exception as exc: return handle_view_exception(request, exc) path_info = gen_path_info(path) breadcrumbs = [] breadcrumbs.append({'name': root_sha1_git[:7], 'url': reverse('browse-directory', kwargs={'sha1_git': root_sha1_git})}) for pi in path_info: breadcrumbs.append({'name': pi['name'], 'url': reverse('browse-directory', kwargs={'sha1_git': root_sha1_git, 'path': pi['path']})}) path = '' if path is None else (path + '/') for d in dirs: d['url'] = reverse('browse-directory', kwargs={'sha1_git': root_sha1_git, 'path': path + d['name']}) sum_file_sizes = 0 readmes = {} for f in files: query_string = 'sha1_git:' + f['target'] f['url'] = reverse('browse-content', kwargs={'query_string': query_string}, query_params={'path': root_sha1_git + '/' + path + f['name']}) 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, 'directory.html', - {'heading': 'Directory', + {'heading': heading, 'swh_object_name': 'Directory', 'swh_object_metadata': dir_metadata, 'dirs': dirs, 'files': files, 'breadcrumbs': breadcrumbs, 'top_right_link': None, 'top_right_link_text': 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}) diff --git a/swh/web/browse/views/origin.py b/swh/web/browse/views/origin.py index 2fa7533f..f4a6c468 100644 --- a/swh/web/browse/views/origin.py +++ b/swh/web/browse/views/origin.py @@ -1,235 +1,237 @@ # Copyright (C) 2017-2018 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 json from distutils.util import strtobool from django.http import HttpResponse from django.shortcuts import render, redirect from swh.web.common import service from swh.web.common.utils import ( reverse, format_utc_iso_date, parse_timestamp, get_origin_visits ) from swh.web.common.exc import handle_view_exception from swh.web.browse.utils import ( get_origin_info ) from swh.web.browse.browseurls import browse_route from .utils.snapshot_context import ( browse_snapshot_directory, browse_snapshot_content, browse_snapshot_log, browse_snapshot_branches, browse_snapshot_releases ) @browse_route(r'origin/(?P[a-z]+)/url/(?P.+)/visit/(?P.+)/directory/', # noqa r'origin/(?P[a-z]+)/url/(?P.+)/visit/(?P.+)/directory/(?P.+)/', # noqa r'origin/(?P[a-z]+)/url/(?P.+)/directory/', # noqa r'origin/(?P[a-z]+)/url/(?P.+)/directory/(?P.+)/', # noqa r'origin/(?P.+)/visit/(?P.+)/directory/', # noqa r'origin/(?P.+)/visit/(?P.+)/directory/(?P.+)/', # noqa r'origin/(?P.+)/directory/', # noqa r'origin/(?P.+)/directory/(?P.+)/', # noqa view_name='browse-origin-directory') def origin_directory_browse(request, origin_url, origin_type=None, timestamp=None, path=None): """Django view for browsing the content of a SWH directory associated to an origin for a given visit. The url scheme that points to it is the following: * :http:get:`/browse/origin/[(origin_type)/url/](origin_url)/directory/[(path)/]` * :http:get:`/browse/origin/[(origin_type)/url/](origin_url)/visit/(timestamp)/directory/[(path)/]` """ # noqa return browse_snapshot_directory( request, origin_type=origin_type, origin_url=origin_url, timestamp=timestamp, path=path) @browse_route(r'origin/(?P[a-z]+)/url/(?P.+)/visit/(?P.+)/content/(?P.+)/', # noqa r'origin/(?P[a-z]+)/url/(?P.+)/content/(?P.+)/', # noqa r'origin/(?P.+)/visit/(?P.+)/content/(?P.+)/', # noqa r'origin/(?P.+)/content/(?P.+)/', # noqa view_name='browse-origin-content') def origin_content_browse(request, origin_url, origin_type=None, path=None, timestamp=None): """Django view that produces an HTML display of a SWH content associated to an origin for a given visit. The url scheme that points to it is the following: * :http:get:`/browse/origin/[(origin_type)/url/](origin_url)/content/(path)/` * :http:get:`/browse/origin/[(origin_type)/url/](origin_url)/visit/(timestamp)/content/(path)/` """ # noqa return browse_snapshot_content(request, origin_type=origin_type, origin_url=origin_url, timestamp=timestamp, path=path) PER_PAGE = 20 @browse_route(r'origin/(?P[a-z]+)/url/(?P.+)/visit/(?P.+)/log/', # noqa r'origin/(?P[a-z]+)/url/(?P.+)/log/', r'origin/(?P.+)/visit/(?P.+)/log/', # noqa r'origin/(?P.+)/log/', view_name='browse-origin-log') def origin_log_browse(request, origin_url, origin_type=None, timestamp=None): """Django view that produces an HTML display of revisions history (aka the commit log) associated to a SWH origin. The url scheme that points to it is the following: * :http:get:`/browse/origin/[(origin_type)/url/](origin_url)/log/` * :http:get:`/browse/origin/[(origin_type)/url/](origin_url)/visit/(timestamp)/log/` """ # noqa return browse_snapshot_log(request, origin_type=origin_type, origin_url=origin_url, timestamp=timestamp) @browse_route(r'origin/(?P[a-z]+)/url/(?P.+)/visit/(?P.+)/branches/', # noqa r'origin/(?P[a-z]+)/url/(?P.+)/branches/', # noqa r'origin/(?P.+)/visit/(?P.+)/branches/', # noqa r'origin/(?P.+)/branches/', # noqa view_name='browse-origin-branches') def origin_branches_browse(request, origin_url, origin_type=None, timestamp=None): """Django view that produces an HTML display of the list of branches associated to an origin for a given visit. The url scheme that points to it is the following: * :http:get:`/browse/origin/[(origin_type)/url/](origin_url)/branches/` * :http:get:`/browse/origin/[(origin_type)/url/](origin_url)/visit/(timestamp)/branches/` """ # noqa return browse_snapshot_branches(request, origin_type=origin_type, origin_url=origin_url, timestamp=timestamp) @browse_route(r'origin/(?P[a-z]+)/url/(?P.+)/visit/(?P.+)/releases/', # noqa r'origin/(?P[a-z]+)/url/(?P.+)/releases/', # noqa r'origin/(?P.+)/visit/(?P.+)/releases/', # noqa r'origin/(?P.+)/releases/', # noqa view_name='browse-origin-releases') def origin_releases_browse(request, origin_url, origin_type=None, timestamp=None): """Django view that produces an HTML display of the list of releases associated to an origin for a given visit. The url scheme that points to it is the following: * :http:get:`/browse/origin/[(origin_type)/url/](origin_url)/releases/` * :http:get:`/browse/origin/[(origin_type)/url/](origin_url)/visit/(timestamp)/releases/` """ # noqa return browse_snapshot_releases(request, origin_type=origin_type, origin_url=origin_url, timestamp=timestamp) @browse_route(r'origin/(?P[a-z]+)/url/(?P.+)/visits/', r'origin/(?P.+)/visits/', view_name='browse-origin-visits') def origin_visits_browse(request, origin_url, origin_type=None): """Django view that produces an HTML display of visits reporting for a swh origin identified by its id or its url. The url that points to it is :http:get:`/browse/origin/[(origin_type)/url/](origin_url)/visits/`. """ # noqa try: origin_info = get_origin_info(origin_url, origin_type) origin_visits = get_origin_visits(origin_info) except Exception as exc: return handle_view_exception(request, exc) origin_info['last swh visit browse url'] = \ reverse('browse-origin-directory', kwargs={'origin_url': origin_url, 'origin_type': origin_type}) for i, visit in enumerate(origin_visits): url_date = format_utc_iso_date(visit['date'], '%Y-%m-%dT%H:%M:%SZ') visit['fmt_date'] = format_utc_iso_date(visit['date']) query_params = {} 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['browse_url'] = reverse('browse-origin-directory', kwargs={'origin_type': origin_type, 'origin_url': origin_url, 'timestamp': url_date}, query_params=query_params) if not snapshot: visit['snapshot'] = '' visit['date'] = parse_timestamp(visit['date']).timestamp() + heading = 'Origin visits - %s' % origin_url + return render(request, 'origin-visits.html', - {'heading': 'Origin visits', + {'heading': heading, 'swh_object_name': 'Visits', 'swh_object_metadata': origin_info, 'origin_visits': origin_visits, 'origin_info': origin_info, 'browse_url_base': '/browse/origin/%s/url/%s/' % (origin_type, origin_url), 'vault_cooking': None, 'show_actions_menu': False}) @browse_route(r'origin/search/(?P.+)/', view_name='browse-origin-search') def _origin_search(request, url_pattern): """Internal browse endpoint to search for origins whose urls contain a provided string pattern or match a provided regular expression. The search is performed in a case insensitive way. """ offset = int(request.GET.get('offset', '0')) limit = int(request.GET.get('limit', '50')) regexp = request.GET.get('regexp', 'false') results = service.search_origin(url_pattern, offset, limit, bool(strtobool(regexp))) results = json.dumps(list(results), sort_keys=True, indent=4, separators=(',', ': ')) return HttpResponse(results, content_type='application/json') @browse_route(r'origin/(?P[0-9]+)/latest_snapshot/', view_name='browse-origin-latest-snapshot') def _origin_latest_snapshot(request, origin_id): """ Internal browse endpoint used to check if an origin has already been visited by Software Heritage and has at least one full visit. """ result = service.lookup_latest_origin_snapshot(origin_id, allowed_statuses=['full']) result = json.dumps(result, sort_keys=True, indent=4, separators=(',', ': ')) return HttpResponse(result, content_type='application/json') @browse_route(r'origin/(?P[a-z]+)/url/(?P.+)/', r'origin/(?P.+)/', view_name='browse-origin') def origin_browse(request, origin_url, origin_type=None): """Django view that redirects to the display of the latest archived snapshot for a given software origin. """ # noqa last_snapshot_url = reverse('browse-origin-directory', kwargs={'origin_type': origin_type, 'origin_url': origin_url}) return redirect(last_snapshot_url) diff --git a/swh/web/browse/views/person.py b/swh/web/browse/views/person.py index ff2ad32c..e093a1cc 100644 --- a/swh/web/browse/views/person.py +++ b/swh/web/browse/views/person.py @@ -1,45 +1,52 @@ # Copyright (C) 2017-2018 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.shortcuts import render from swh.web.common import service from swh.web.common.exc import handle_view_exception from swh.web.browse.browseurls import browse_route from swh.web.browse.utils import get_snapshot_context @browse_route(r'person/(?P[0-9]+)/', view_name='browse-person') def person_browse(request, person_id): """ Django view that produces an HTML display of a swh person identified by its id. The url that points to it is :http:get:`/browse/person/(person_id)/`. """ try: 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) snapshot_id = request.GET.get('snapshot_id', None) if origin_url: snapshot_context = get_snapshot_context(None, origin_type, origin_url) elif snapshot_id: snapshot_context = get_snapshot_context(snapshot_id) person = service.lookup_person(person_id) except Exception as exc: return handle_view_exception(request, exc) + heading = 'Person - %s' % person['fullname'] + if snapshot_context: + context_found = 'snapshot: %s' % snapshot_context['snapshot_id'] + if origin_url: + context_found = 'origin: %s' % origin_url + heading += ' - %s' % context_found + return render(request, 'person.html', - {'heading': 'Person', + {'heading': heading, 'swh_object_name': 'Person', 'swh_object_metadata': person, 'snapshot_context': snapshot_context, 'vault_cooking': None, 'show_actions_menu': False}) diff --git a/swh/web/browse/views/release.py b/swh/web/browse/views/release.py index 959ce78f..1b0b7a33 100644 --- a/swh/web/browse/views/release.py +++ b/swh/web/browse/views/release.py @@ -1,157 +1,164 @@ # Copyright (C) 2017-2018 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.shortcuts import render from django.utils.safestring import mark_safe from swh.web.common import service from swh.web.common.utils import ( reverse, format_utc_iso_date ) from swh.web.common.exc import handle_view_exception from swh.web.browse.browseurls import browse_route from swh.web.browse.utils import ( gen_person_link, gen_revision_link, get_snapshot_context, gen_link, gen_snapshot_link, get_swh_persistent_ids ) @browse_route(r'release/(?P[0-9a-f]+)/', view_name='browse-release') def release_browse(request, sha1_git): """ Django view that produces an HTML display of a SWH 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: snapshot_context = get_snapshot_context(snapshot_id, origin_type, origin_url, timestamp, visit_id) 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 = {} release_data['author'] = \ gen_person_link(release['author']['id'], release['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', kwargs={'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', kwargs={'sha1_git': release['target']}) release_data['target'] = gen_link(directory_url, release['target']) elif release['target_type'] == 'release': release_url = \ reverse('browse-release', kwargs={'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 release_target_link = 'Target: ' if release['target_type'] == 'revision': release_target_link += '' # noqa 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 else: release_target_link += release['target_type'] release_target_link += ' ' + release_data['target'] if snapshot_context: release_data['snapshot id'] = snapshot_context['snapshot_id'] if origin_info: release_url = reverse('browse-release', kwargs={'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) release_note_header = 'None' if len(release_note_lines) > 0: release_note_header = release_note_lines[0] + 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, 'release.html', - {'heading': 'Release', + {'heading': heading, 'swh_object_name': 'Release', 'swh_object_metadata': release_data, 'release_name': release['name'], 'release_note_header': release_note_header, 'release_note_body': '\n'.join(release_note_lines[1:]), 'release_target_link': mark_safe(release_target_link), 'snapshot_context': snapshot_context, 'show_actions_menu': True, 'breadcrumbs': None, 'vault_cooking': vault_cooking, 'top_right_link': None, 'top_right_link_text': None, 'swh_ids': swh_ids}) diff --git a/swh/web/browse/views/revision.py b/swh/web/browse/views/revision.py index 54ac3380..7a4c8b44 100644 --- a/swh/web/browse/views/revision.py +++ b/swh/web/browse/views/revision.py @@ -1,489 +1,498 @@ # Copyright (C) 2017-2018 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 django.http import HttpResponse from django.shortcuts import render, redirect from django.template.defaultfilters import filesizeformat from django.utils.safestring import mark_safe from swh.web.common import service from swh.web.common.utils import ( reverse, format_utc_iso_date, gen_path_info ) from swh.web.common.exc import handle_view_exception from swh.web.browse.browseurls import browse_route from swh.web.browse.utils import ( gen_link, gen_person_link, gen_revision_link, prepare_revision_log_for_display, get_snapshot_context, gen_snapshot_directory_link, get_revision_log_url, get_directory_entries, gen_directory_link, request_content, prepare_content_for_display, content_display_max_size, gen_snapshot_link, get_readme_to_display, get_swh_persistent_ids ) def _gen_content_url(revision, query_string, path, snapshot_context): if snapshot_context: url_args = snapshot_context['url_args'] url_args['path'] = path query_params = snapshot_context['query_params'] query_params['revision'] = revision['id'] content_url = reverse('browse-origin-content', kwargs=url_args, query_params=query_params) else: content_path = '%s/%s' % (revision['directory'], path) content_url = reverse('browse-content', kwargs={'query_string': query_string}, query_params={'path': content_path}) return content_url 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 # TODO: put in conf _max_displayed_file_diffs = 1000 def _gen_revision_changes_list(revision, changes, snapshot_context): """ 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', kwargs=url_args, query_params=query_params) hasher.update(diff_id.encode('utf-8')) diff_id = hasher.hexdigest() change['id'] = diff_id panel_diff_link = '#panel_' + 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, panel_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, panel_diff_link, change['to_path'])) elif change['type'] == 'delete': parent = service.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, panel_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, panel_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') def _revision_diff(request, sha1_git): """ Browse internal endpoint to compute revision diff """ try: revision = service.lookup_revision(sha1_git) 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) if origin_url: snapshot_context = get_snapshot_context(None, origin_type, origin_url, timestamp, visit_id) except Exception as exc: return handle_view_exception(request, exc) changes = service.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 } diff_data_json = json.dumps(diff_data, separators=(',', ': ')) return HttpResponse(diff_data_json, content_type='application/json') NB_LOG_ENTRIES = 20 @browse_route(r'revision/(?P[0-9a-f]+)/log/', view_name='browse-revision-log') def revision_log_browse(request, sha1_git): """ Django view that produces an HTML display of the history log for a SWH revision identified by its id. The url that points to it is :http:get:`/browse/revision/(sha1_git)/log/`. """ # noqa try: per_page = int(request.GET.get('per_page', NB_LOG_ENTRIES)) revision_log = service.lookup_revision_log(sha1_git, limit=per_page+1) revision_log = list(revision_log) except Exception as exc: return handle_view_exception(request, exc) revs_breadcrumb = request.GET.get('revs_breadcrumb', None) revision_log_display_data = prepare_revision_log_for_display( revision_log, per_page, revs_breadcrumb) prev_rev = revision_log_display_data['prev_rev'] prev_revs_breadcrumb = revision_log_display_data['prev_revs_breadcrumb'] prev_log_url = None if prev_rev: prev_log_url = \ reverse('browse-revision-log', kwargs={'sha1_git': prev_rev}, query_params={'revs_breadcrumb': prev_revs_breadcrumb, 'per_page': per_page}) next_rev = revision_log_display_data['next_rev'] next_revs_breadcrumb = revision_log_display_data['next_revs_breadcrumb'] next_log_url = None if next_rev: next_log_url = \ reverse('browse-revision-log', kwargs={'sha1_git': next_rev}, query_params={'revs_breadcrumb': next_revs_breadcrumb, 'per_page': per_page}) revision_log_data = revision_log_display_data['revision_log_data'] for log in revision_log_data: log['directory'] = gen_directory_link( log['directory'], link_text='Browse files', link_attrs={'class': 'btn btn-default btn-sm', 'role': 'button'}) return render(request, 'revision-log.html', {'heading': 'Revision history', 'swh_object_name': 'Revision history', 'swh_object_metadata': None, 'revision_log': revision_log_data, 'next_log_url': next_log_url, 'prev_log_url': prev_log_url, 'breadcrumbs': None, 'top_right_link': None, 'top_right_link_text': None, 'snapshot_context': None, 'vault_cooking': None, 'show_actions_menu': True, 'swh_ids': None}) @browse_route(r'revision/(?P[0-9a-f]+)/', r'revision/(?P[0-9a-f]+)/(?P.+)/', view_name='browse-revision') def revision_browse(request, sha1_git, extra_path=None): """ Django view that produces an HTML display of a SWH 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', kwargs={'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: snapshot_context = get_snapshot_context(None, origin_type, origin_url, timestamp, visit_id) origin_info = snapshot_context['origin_info'] snapshot_id = snapshot_context['snapshot_id'] elif snapshot_id: snapshot_context = get_snapshot_context(snapshot_id) if path: path_info = \ service.lookup_directory_with_path(revision['directory'], path) if path_info['type'] == 'dir': dir_id = path_info['target'] else: query_string = 'sha1_git:' + path_info['target'] content_data = request_content(query_string) 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 = {} revision_data['author'] = \ gen_person_link(revision['author']['id'], revision['author']['name'], snapshot_context) 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'] = 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 + '
' revision_data['parents'] = mark_safe(parents) revision_data['synthetic'] = revision['synthetic'] revision_data['type'] = revision['type'] message_lines = revision['message'].split('\n') parents_links = '%s parent%s ' % \ (len(revision['parents']), '' if len(revision['parents']) == 1 else 's') parents_links += ' ' for p in revision['parents']: parent_link = gen_revision_link(p, shorten_id=True, snapshot_context=snapshot_context) parents_links += parent_link if p != revision['parents'][-1]: parents_links += ' + ' 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', kwargs={'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', kwargs={'sha1_git': sha1_git}, query_params=query_params)}) vault_cooking = { 'directory_context': False, 'directory_id': None, 'revision_context': True, 'revision_id': sha1_git } content = None content_size = None mimetype = None language = None readme_name = None readme_url = None readme_html = None readmes = {} 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'] query_params = {} if path: query_params['filename'] = path_info[-1]['name'] top_right_link = reverse('browse-content-raw', kwargs={'query_string': query_string}, query_params=query_params) top_right_link_text = mark_safe( 'Raw File') else: for d in dirs: query_params['path'] = path + d['name'] d['url'] = reverse('browse-revision', kwargs={'sha1_git': sha1_git}, query_params=query_params) for f in files: query_params['path'] = path + f['name'] f['url'] = reverse('browse-revision', kwargs={'sha1_git': sha1_git}, query_params=query_params) 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 = get_revision_log_url(sha1_git, snapshot_context) top_right_link_text = mark_safe( '' 'History') vault_cooking['directory_context'] = True vault_cooking['directory_id'] = dir_id diff_revision_url = reverse('diff-revision', kwargs={'sha1_git': sha1_git}, query_params={'origin_type': origin_type, 'origin': origin_url, 'timestamp': timestamp, 'visit_id': visit_id}) swh_objects = [{'type': 'revision', 'id': sha1_git}] 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, 'revision.html', - {'heading': 'Revision', + {'heading': heading, 'swh_object_name': 'Revision', 'swh_object_metadata': revision_data, 'message_header': message_lines[0], 'message_body': '\n'.join(message_lines[1:]), 'parents_links': mark_safe(parents_links), '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, 'top_right_link_text': top_right_link_text, 'vault_cooking': vault_cooking, 'diff_revision_url': diff_revision_url, 'show_actions_menu': True, 'swh_ids': swh_ids}) diff --git a/swh/web/browse/views/utils/snapshot_context.py b/swh/web/browse/views/utils/snapshot_context.py index 1223c6bc..c0890ef5 100644 --- a/swh/web/browse/views/utils/snapshot_context.py +++ b/swh/web/browse/views/utils/snapshot_context.py @@ -1,781 +1,813 @@ # Copyright (C) 2018 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 implementing Django views for browsing the SWH archive # in a snapshot context. # Its purpose is to factorize code for the views reachable from the # /origin/.* and /snapshot/.* endpoints. from django.shortcuts import render, redirect from django.utils.safestring import mark_safe from django.template.defaultfilters import filesizeformat from swh.web.browse.utils import ( get_snapshot_context, get_directory_entries, gen_directory_link, gen_revision_link, request_content, gen_content_link, prepare_content_for_display, content_display_max_size, prepare_revision_log_for_display, gen_snapshot_directory_link, gen_revision_log_link, gen_link, get_readme_to_display, get_swh_persistent_ids ) from swh.web.common import service from swh.web.common.exc import ( handle_view_exception, NotFoundExc ) from swh.web.common.utils import ( reverse, gen_path_info, format_utc_iso_date ) def _get_branch(branches, branch_name): """ Utility function to get a specific branch from a branches list. Its purpose is to get the default HEAD branch as some SWH origin (e.g those with svn type) does not have it. In that latter case, check if there is a master branch instead and returns it. """ filtered_branches = \ [b for b in branches if b['name'].endswith(branch_name)] if len(filtered_branches) > 0: return filtered_branches[0] elif branch_name == 'HEAD': filtered_branches = \ [b for b in branches if b['name'].endswith('master')] if len(filtered_branches) > 0: return filtered_branches[0] elif len(branches) > 0: return branches[0] return None def _get_release(releases, release_name): """ Utility function to get a specific release from a releases list. Returns None if the release can not be found in the list. """ filtered_releases = \ [r for r in releases if r['name'] == release_name] if len(filtered_releases) > 0: return filtered_releases[0] else: return None def _branch_not_found(branch_type, branch, branches, snapshot_id=None, origin_info=None, timestamp=None, visit_id=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' else: branch_type = 'Release' branch_type_plural = 'releases' if snapshot_id and len(branches) == 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 len(branches) == 0: msg = 'Origin with type %s and url %s' \ ' for visit with id %s has an empty list' \ ' of %s!' % (origin_info['type'], origin_info['url'], visit_id, branch_type_plural) elif visit_id: msg = '%s %s associated to visit with' \ ' id %s for origin with type %s and url %s' \ ' not found!' % (branch_type, branch, visit_id, origin_info['type'], origin_info['url']) elif len(branches) == 0: msg = 'Origin with type %s and url %s' \ ' for visit with timestamp %s has an empty list' \ ' of %s!' % (origin_info['type'], origin_info['url'], timestamp, branch_type_plural) else: msg = '%s %s associated to visit with' \ ' timestamp %s for origin with type %s' \ ' and url %s not found!' % (branch_type, branch, timestamp, origin_info['type'], origin_info['url']) raise NotFoundExc(msg) def _process_snapshot_request(request, snapshot_id=None, origin_type=None, origin_url=None, timestamp=None, path=None, browse_context='directory'): """ Utility function to perform common input request processing for snapshot context views. """ visit_id = request.GET.get('visit_id', None) snapshot_context = get_snapshot_context(snapshot_id, origin_type, origin_url, timestamp, visit_id) swh_type = snapshot_context['swh_type'] origin_info = snapshot_context['origin_info'] branches = snapshot_context['branches'] releases = snapshot_context['releases'] url_args = snapshot_context['url_args'] query_params = snapshot_context['query_params'] browse_view_name = 'browse-' + swh_type + '-' + browse_context for b in branches: branch_url_args = dict(url_args) branch_query_params = dict(query_params) branch_query_params['branch'] = b['name'] if path: b['path'] = path branch_url_args['path'] = path b['url'] = reverse(browse_view_name, kwargs=branch_url_args, query_params=branch_query_params) for r in releases: release_url_args = dict(url_args) release_query_params = dict(query_params) release_query_params['release'] = r['name'] if path: r['path'] = path release_url_args['path'] = path r['url'] = reverse(browse_view_name, kwargs=release_url_args, query_params=release_query_params) root_sha1_git = None revision_id = request.GET.get('revision', None) release_name = request.GET.get('release', None) branch_name = None if revision_id: revision = service.lookup_revision(revision_id) root_sha1_git = revision['directory'] branches.append({'name': revision_id, 'revision': revision_id, 'directory': root_sha1_git, 'url': None}) branch_name = revision_id query_params['revision'] = revision_id elif release_name: release = _get_release(releases, release_name) if release: root_sha1_git = release['directory'] revision_id = release['target'] query_params['release'] = release_name else: _branch_not_found("release", release_name, releases, snapshot_id, origin_info, timestamp, visit_id) else: branch_name = request.GET.get('branch', None) if branch_name: query_params['branch'] = branch_name branch = _get_branch(branches, branch_name or 'HEAD') if branch: branch_name = branch['name'] root_sha1_git = branch['directory'] revision_id = branch['revision'] else: _branch_not_found("branch", branch_name, branches, snapshot_id, origin_info, timestamp, visit_id) snapshot_context['query_params'] = query_params snapshot_context['root_sha1_git'] = root_sha1_git snapshot_context['revision_id'] = revision_id snapshot_context['branch'] = branch_name snapshot_context['release'] = release_name return snapshot_context 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 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', kwargs={'query_string': dir_info['checksums']['sha1']}) return redirect(file_raw_url) sha1_git = dir_info['target'] 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 = [] breadcrumbs.append({'name': root_sha1_git[:7], 'url': reverse(browse_view_name, kwargs=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, kwargs=bc_url_args, query_params=query_params)}) path = '' if path is None else (path + '/') for d in dirs: bc_url_args = dict(url_args) bc_url_args['path'] = path + d['name'] d['url'] = reverse(browse_view_name, kwargs=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, kwargs=bc_url_args, query_params=query_params) 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 = reverse(browse_view_name, kwargs=url_args, query_params=query_params) sum_file_sizes = filesizeformat(sum_file_sizes) 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': len(files), 'number of subdirectories': len(dirs), 'sum of regular file sizes': sum_file_sizes, 'path': '/' + 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', kwargs={'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}] 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, 'directory.html', - {'heading': 'Directory', + {'heading': heading, 'swh_object_name': 'Directory', 'swh_object_metadata': dir_metadata, 'dirs': dirs, 'files': files, 'breadcrumbs': breadcrumbs, 'top_right_link': history_url, 'top_right_link_text': mark_safe( '' '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'] 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) 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 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'] 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)]) breadcrumbs.append({'name': root_sha1_git[:7], 'url': reverse(browse_view_name, kwargs=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, kwargs=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 = reverse('browse-content-raw', kwargs={'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, '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'], 'path': '/' + path[:-len(filename)], 'filename': filename, 'revision id': revision_id, 'revision': browse_rev_link, 'snapshot id': snapshot_id } 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', kwargs={'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 cnt_sha1_git = content_data['checksums']['sha1_git'] swh_objects = [{'type': 'content', 'id': cnt_sha1_git}, {'type': 'revision', 'id': revision_id}, {'type': 'snapshot', 'id': snapshot_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, 'content.html', - {'heading': 'Content', + {'heading': heading, 'swh_object_name': 'Content', 'swh_object_metadata': content_metadata, 'content': content, 'content_size': content_data['length'], 'max_content_size': content_display_max_size, 'mimetype': content_data['mimetype'], 'language': language, 'breadcrumbs': breadcrumbs, 'top_right_link': content_raw_url, 'top_right_link_text': mark_safe( 'Raw File'), 'snapshot_context': snapshot_context, 'vault_cooking': None, 'show_actions_menu': True, 'swh_ids': swh_ids}) PER_PAGE = 20 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'] current_rev = revision_id per_page = int(request.GET.get('per_page', PER_PAGE)) revs_breadcrumb = request.GET.get('revs_breadcrumb', None) if revs_breadcrumb: current_rev = revs_breadcrumb.split('/')[-1] revision_log = service.lookup_revision_log(current_rev, limit=per_page+1) revision_log = list(revision_log) 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 revision_log_display_data = prepare_revision_log_for_display( revision_log, per_page, revs_breadcrumb, snapshot_context) browse_view_name = 'browse-' + swh_type + '-log' prev_rev = revision_log_display_data['prev_rev'] prev_revs_breadcrumb = revision_log_display_data['prev_revs_breadcrumb'] prev_log_url = None query_params['revs_breadcrumb'] = prev_revs_breadcrumb if prev_rev: prev_log_url = \ reverse(browse_view_name, kwargs=url_args, query_params=query_params) next_rev = revision_log_display_data['next_rev'] next_revs_breadcrumb = revision_log_display_data['next_revs_breadcrumb'] next_log_url = None query_params['revs_breadcrumb'] = next_revs_breadcrumb if next_rev: next_log_url = \ reverse(browse_view_name, kwargs=url_args, query_params=query_params) revision_log_data = revision_log_display_data['revision_log_data'] for i, log in enumerate(revision_log_data): params = { 'revision': revision_log[i]['id'], } if 'visit_id' in query_params: params['visit_id'] = query_params['visit_id'] log['directory'] = gen_snapshot_directory_link( snapshot_context, revision_log[i]['id'], link_text='Browse files', link_attrs={'class': 'btn btn-default btn-sm', 'role': 'button'}) 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', kwargs={'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}] 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, 'revision-log.html', - {'heading': 'Revision history', + {'heading': heading, 'swh_object_name': 'Revision history', 'swh_object_metadata': revision_metadata, 'revision_log': revision_log_data, 'next_log_url': next_log_url, 'prev_log_url': prev_log_url, 'breadcrumbs': None, 'top_right_link': None, 'top_right_link_text': None, 'snapshot_context': snapshot_context, 'vault_cooking': None, 'show_actions_menu': True, 'swh_ids': swh_ids}) def browse_snapshot_branches(request, snapshot_id=None, origin_type=None, origin_url=None, timestamp=None): """ Django view implementation for browsing a list of branches in a snapshot context. """ try: snapshot_context = _process_snapshot_request(request, snapshot_id, origin_type, origin_url, timestamp) except Exception as exc: return handle_view_exception(request, exc) branches_offset = int(request.GET.get('branches_offset', 0)) swh_type = snapshot_context['swh_type'] origin_info = snapshot_context['origin_info'] url_args = snapshot_context['url_args'] query_params = snapshot_context['query_params'] browse_view_name = 'browse-' + swh_type + '-directory' branches = snapshot_context['branches'] displayed_branches = \ branches[branches_offset:branches_offset+PER_PAGE] for branch in displayed_branches: if snapshot_id: revision_url = reverse('browse-revision', kwargs={'sha1_git': branch['revision']}, query_params={'snapshot_id': snapshot_id}) else: revision_url = reverse('browse-revision', kwargs={'sha1_git': branch['revision']}, query_params={'origin_type': origin_info['type'], # noqa 'origin': origin_info['url']}) # noqa query_params['branch'] = branch['name'] directory_url = reverse(browse_view_name, kwargs=url_args, query_params=query_params) del query_params['branch'] branch['revision_url'] = revision_url branch['directory_url'] = directory_url browse_view_name = 'browse-' + swh_type + '-branches' prev_branches_url = None next_branches_url = None next_offset = branches_offset + PER_PAGE prev_offset = branches_offset - PER_PAGE if next_offset < len(branches): query_params['branches_offset'] = next_offset next_branches_url = reverse(browse_view_name, kwargs=url_args, query_params=query_params) query_params['branches_offset'] = None if prev_offset >= 0: if prev_offset != 0: query_params['branches_offset'] = prev_offset prev_branches_url = reverse(browse_view_name, kwargs=url_args, query_params=query_params) + heading = 'Branches - ' + if origin_info: + heading += 'origin: %s' % origin_info['url'] + else: + heading += 'snapshot: %s' % snapshot_id + return render(request, 'branches.html', - {'heading': 'Origin branches', + {'heading': heading, 'swh_object_name': 'Branches', 'swh_object_metadata': {}, 'top_right_link': None, 'top_right_link_text': None, 'displayed_branches': displayed_branches, 'prev_branches_url': prev_branches_url, 'next_branches_url': next_branches_url, 'snapshot_context': snapshot_context}) def browse_snapshot_releases(request, snapshot_id=None, origin_type=None, origin_url=None, timestamp=None): """ Django view implementation for browsing a list of releases in a snapshot context. """ try: snapshot_context = _process_snapshot_request(request, snapshot_id, origin_type, origin_url, timestamp) except Exception as exc: return handle_view_exception(request, exc) releases_offset = int(request.GET.get('releases_offset', 0)) swh_type = snapshot_context['swh_type'] origin_info = snapshot_context['origin_info'] url_args = snapshot_context['url_args'] query_params = snapshot_context['query_params'] releases = snapshot_context['releases'] displayed_releases = \ releases[releases_offset:releases_offset+PER_PAGE] for release in displayed_releases: if snapshot_id: release_url = reverse('browse-release', kwargs={'sha1_git': release['id']}, query_params={'snapshot_id': snapshot_id}) else: release_url = reverse('browse-release', kwargs={'sha1_git': release['id']}, query_params={'origin_type': origin_info['type'], # noqa 'origin': origin_info['url']}) # noqa query_params['release'] = release['name'] del query_params['release'] release['release_url'] = release_url browse_view_name = 'browse-' + swh_type + '-releases' prev_releases_url = None next_releases_url = None next_offset = releases_offset + PER_PAGE prev_offset = releases_offset - PER_PAGE if next_offset < len(releases): query_params['releases_offset'] = next_offset next_releases_url = reverse(browse_view_name, kwargs=url_args, query_params=query_params) query_params['releases_offset'] = None if prev_offset >= 0: if prev_offset != 0: query_params['releases_offset'] = prev_offset prev_releases_url = reverse(browse_view_name, kwargs=url_args, query_params=query_params) + heading = 'Releases - ' + if origin_info: + heading += 'origin: %s' % origin_info['url'] + else: + heading += 'snapshot: %s' % snapshot_id + return render(request, 'releases.html', - {'heading': 'Origin releases', + {'heading': heading, 'top_panel_visible': False, 'top_panel_collapsible': False, 'swh_object_name': 'Releases', 'swh_object_metadata': {}, 'top_right_link': None, 'top_right_link_text': None, 'displayed_releases': displayed_releases, 'prev_releases_url': prev_releases_url, 'next_releases_url': next_releases_url, 'snapshot_context': snapshot_context, 'vault_cooking': None, 'show_actions_menu': False}) diff --git a/swh/web/templates/homepage.html b/swh/web/templates/homepage.html index 130b3fb7..4af383a8 100644 --- a/swh/web/templates/homepage.html +++ b/swh/web/templates/homepage.html @@ -1,124 +1,124 @@ {% extends "layout.html" %} {% comment %} Copyright (C) 2017-2018 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 title %}The Software Heritage archive{% endblock %} +{% block title %}Welcome to the Software Heritage archive{% endblock %} {% block navbar-content %}

Welcome to the Software Heritage archive

{% endblock %} {% block content %}

Overview

The long term goal of the Software Heritage initiative is to collect all publicly available software in source code form together with its development history, replicate it massively to ensure its preservation, and share it with everyone who needs it. The Software Heritage archive is growing over time as we crawl new source code from software projects and development forges. We will incrementally release archive search and browse functionalities — as of now you can check whether source code you care about is already present in the archive or not.

Content

A significant amount of source code has already been ingested in the Software Heritage archive. It currently includes:

  • public repositories from GitHub
  • source packages from the Debian distribution
  • public repositories from the former Gitorious code hosting service
  • public repositories from the former Google Code project hosting service
  • releases from the GNU project (as of August 2015)

Size

As of today the archive already contains and keeps safe for you the following amount of objects:

Source files
Directories
Commits
Authors
Projects
Releases

Access

Browse the Software Heritage archive
Browse

Explore the archive using a dedicated web interface.
Software Heritage Web API
Web API

Query the archive programmatically using a RESTful API.

{% endblock %}