diff --git a/swh/web/browse/utils.py b/swh/web/browse/utils.py index affbeb13..e34102d6 100644 --- a/swh/web/browse/utils.py +++ b/swh/web/browse/utils.py @@ -1,471 +1,495 @@ # Copyright (C) 2017 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import base64 import magic import math import stat import textwrap from django.core.cache import cache from django.utils.safestring import mark_safe from swh.web.common import highlightjs, service from swh.web.common.exc import NotFoundExc from swh.web.common.utils import ( reverse, format_utc_iso_date, parse_timestamp ) def get_directory_entries(sha1_git): """Function that retrieves the content of a SWH directory from the SWH archive. The directories entries are first sorted in lexicographical order. Sub-directories and regular files are then extracted. Args: sha1_git: sha1_git identifier of the directory Returns: A tuple whose first member corresponds to the sub-directories list and second member the regular files list Raises: NotFoundExc if the directory is not found """ cache_entry_id = 'directory_entries_%s' % sha1_git cache_entry = cache.get(cache_entry_id) if cache_entry: return cache_entry entries = list(service.lookup_directory(sha1_git)) entries = sorted(entries, key=lambda e: e['name']) for entry in entries: entry['perms'] = stat.filemode(entry['perms']) dirs = [e for e in entries if e['type'] == 'dir'] files = [e for e in entries if e['type'] == 'file'] cache.set(cache_entry_id, (dirs, files)) return dirs, files def get_mimetype_and_encoding_for_content(content): """Function that returns the mime type and the encoding associated to a content buffer using the magic module under the hood. Args: content (bytes): a content buffer Returns: A tuple (mimetype, encoding), for instance ('text/plain', 'us-ascii'), associated to the provided content. """ magic_result = magic.detect_from_content(content) return magic_result.mime_type, magic_result.encoding def request_content(query_string): """Function that retrieves a SWH content from the SWH archive. Raw bytes content is first retrieved, then the content mime type. If the mime type is not stored in the archive, it will be computed using Python magic module. Args: 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 Returns: A tuple whose first member corresponds to the content raw bytes and second member the content mime type Raises: NotFoundExc if the content is not found """ content_data = service.lookup_content(query_string) content_raw = service.lookup_content_raw(query_string) content_data['raw_data'] = content_raw['data'] filetype = service.lookup_content_filetype(query_string) language = service.lookup_content_language(query_string) license = service.lookup_content_license(query_string) if filetype: mimetype = filetype['mimetype'] encoding = filetype['encoding'] else: mimetype, encoding = \ get_mimetype_and_encoding_for_content(content_data['raw_data']) content_data['mimetype'] = mimetype content_data['encoding'] = encoding # encode textual content to utf-8 if needed if mimetype.startswith('text/') and 'ascii' not in encoding \ and 'utf-8' not in encoding: content_data['raw_data'] = \ content_data['raw_data'].decode(encoding).encode('utf-8') if language: content_data['language'] = language['lang'] else: content_data['language'] = 'not detected' if license: content_data['licenses'] = ', '.join(license['licenses']) else: content_data['licenses'] = 'not detected' return content_data _browsers_supported_image_mimes = set(['image/gif', 'image/png', 'image/jpeg', 'image/bmp', 'image/webp']) def prepare_content_for_display(content_data, mime_type, path): """Function that prepares a content for HTML display. The function tries to associate a programming language to a content in order to perform syntax highlighting client-side using highlightjs. The language is determined using either the content filename or its mime type. If the mime type corresponds to an image format supported by web browsers, the content will be encoded in base64 for displaying the image. Args: content_data (bytes): raw bytes of the content mime_type (string): mime type of the content path (string): path of the content including filename Returns: A dict containing the content bytes (possibly different from the one provided as parameter if it is an image) under the key 'content_data and the corresponding highlightjs language class under the key 'language'. """ language = highlightjs.get_hljs_language_from_filename(path) if not language: language = highlightjs.get_hljs_language_from_mime_type(mime_type) if not language: language = 'nohighlight-swh' elif mime_type.startswith('application/'): mime_type = mime_type.replace('application/', 'text/') if mime_type.startswith('image/'): if mime_type in _browsers_supported_image_mimes: content_data = base64.b64encode(content_data) else: content_data = None return {'content_data': content_data, 'language': language} def get_origin_visits(origin_id): """Function that returns the list of visits for a swh origin. That list is put in cache in order to speedup the navigation in the swh web browse ui. Args: origin_id (int): the id of the swh origin to fetch visits from Returns: A list of dict describing the origin visits:: [{'date': , 'origin': , 'status': <'full' | 'partial'>, 'visit': }, ... ] Raises: NotFoundExc if the origin is not found """ cache_entry_id = 'origin_%s_visits' % origin_id cache_entry = cache.get(cache_entry_id) if cache_entry: return cache_entry origin_visits = [] per_page = service.MAX_LIMIT last_visit = None while 1: visits = list(service.lookup_origin_visits(origin_id, last_visit=last_visit, per_page=per_page)) origin_visits += visits if len(visits) < per_page: break else: if not last_visit: last_visit = per_page else: last_visit += per_page cache.set(cache_entry_id, origin_visits) return origin_visits def get_origin_visit(origin_id, visit_id=None, visit_ts=None): """Function that returns information about a SWH visit for a given origin. The visit is retrieved from the visits list using either: * a visit id * a provided timestamp, in that case, the closest visit from that timestamp is selected Args: origin_id (int): the id of the swh origin to fetch branches from visit_id (int): the id of the origin visit visit_ts (int or str): an ISO date string or Unix timestamp to parse Returns: A dict containing the visit info as described below:: {'origin': 2, 'date': '2017-10-08T11:54:25.582463+00:00', 'metadata': {}, 'visit': 25, 'status': 'full'} """ visits = get_origin_visits(origin_id) if not visit_id and visit_ts: parsed_visit_ts = math.floor(parse_timestamp(visit_ts).timestamp()) for i, visit in enumerate(visits): ts = math.floor(parse_timestamp(visit['date']).timestamp()) if i == 0 and parsed_visit_ts <= ts: return visit elif i == len(visits) - 1: if parsed_visit_ts >= ts: return visit else: next_ts = math.floor( parse_timestamp(visits[i+1]['date']).timestamp()) if parsed_visit_ts >= ts and parsed_visit_ts < next_ts: if (parsed_visit_ts - ts) < (next_ts - parsed_visit_ts): return visit else: return visits[i+1] raise NotFoundExc( 'Visit with timestamp %s for origin with id %s not found!' % (visit_ts, origin_id)) visit = [v for v in visits if v['visit'] == int(visit_id)] if len(visit) == 0: raise NotFoundExc( 'Visit with id %s for origin with id %s not found!' % (visit_id, origin_id)) return visit[0] def get_origin_visit_branches(origin_id, visit_id=None, visit_ts=None): """Function that returns the list of branches associated to a swh origin for a given visit. The visit can be expressed by its id or a timestamp. In the latter case, the closest visit from the provided timestamp will be used. If no visit parameter is provided, it returns the list of branches found for the latest visit. That list is put in cache in order to speedup the navigation in the swh web browse ui. Args: origin_id (int): the id of the swh origin to fetch branches from visit_id (int): the id of the origin visit visit_ts (int or str): an ISO date string or Unix timestamp to parse Returns: A tuple with two members. The first one is a list of dict describing the origin branches for the given visit:: [{'name': , 'revision': , 'directory': }, ... ] The second one is the ISO date string corresponding to the associated SWH visit. Raises: NotFoundExc if the origin or its visit are not found """ visit_info = get_origin_visit(origin_id, visit_id, visit_ts) visit_id = visit_info['visit'] cache_entry_id = 'origin_%s_visit_%s_branches' % (origin_id, visit_id) cache_entry = cache.get(cache_entry_id) if cache_entry: return cache_entry origin_visit_data = service.lookup_origin_visit(origin_id, visit_id) branches = [] revision_ids = [] occurrences = origin_visit_data['occurrences'] for key in sorted(occurrences.keys()): if occurrences[key]['target_type'] == 'revision': branches.append({'name': key, 'revision': occurrences[key]['target']}) revision_ids.append(occurrences[key]['target']) revisions = service.lookup_revision_multiple(revision_ids) branches_to_remove = [] for idx, revision in enumerate(revisions): if revision: branches[idx]['directory'] = revision['directory'] else: branches_to_remove.append(branches[idx]) for b in branches_to_remove: branches.remove(b) cache.set(cache_entry_id, branches) return branches def gen_link(url, link_text): """ Utility function for generating an HTML link to insert in Django templates. Args: url (str): an url link_text (str): the text for the produced link Returns: An HTML link in the form 'link_text' """ link = '%s' % (url, link_text) return mark_safe(link) def gen_person_link(person_id, person_name): """ Utility function for generating a link to a SWH person HTML view to insert in Django templates. Args: person_id (int): a SWH person id person_name (str): the associated person name Returns: An HTML link in the form 'person_name' """ person_url = reverse('browse-person', kwargs={'person_id': person_id}) return gen_link(person_url, person_name) -def gen_revision_link(revision_id, shorten_id=False): +def gen_revision_link(revision_id, shorten_id=False, origin_id=None): """ Utility function for generating a link to a SWH revision HTML view to insert in Django templates. Args: revision_id (int): a SWH revision id shorten_id (boolean): wheter to shorten the revision id to 7 characters for the link text Returns: An HTML link in the form 'revision_id' """ + query_params = None + if origin_id: + query_params = {'origin_id': origin_id} revision_url = reverse('browse-revision', - kwargs={'sha1_git': revision_id}) + kwargs={'sha1_git': revision_id}, + query_params=query_params) if shorten_id: return gen_link(revision_url, revision_id[:7]) else: return gen_link(revision_url, revision_id) -def _format_log_entries(revision_log, per_page): +def gen_origin_link(origin_id, origin_url): + """ + Utility function for generating a link to a SWH origin HTML view + to insert in Django templates. + + Args: + origin_id (int): a SWH origin id + origin_url (str): the url of the origin + + Returns: + An HTML link in the form 'Origin: origin_url' + + """ # noqa + origin_browse_url = reverse('browse-origin', + kwargs={'origin_id': origin_id}) + return gen_link(origin_browse_url, + 'Origin: ' + origin_url) + + +def _format_log_entries(revision_log, per_page, origin_id=None): revision_log_data = [] for i, log in enumerate(revision_log): if i == per_page: break revision_log_data.append( {'author': gen_person_link(log['author']['id'], log['author']['name']), - 'revision': gen_revision_link(log['id'], True), + 'revision': gen_revision_link(log['id'], True, origin_id), 'message': log['message'], 'message_shorten': textwrap.shorten(log['message'], width=80, placeholder='...'), 'date': format_utc_iso_date(log['date']), 'directory': log['directory']}) return revision_log_data def prepare_revision_log_for_display(revision_log, per_page, revs_breadcrumb, - origin_context=False): + origin_context=False, origin_id=None): """ Utility functions that process raw revision log data for HTML display. Its purpose is to: * add links to relevant SWH browse views * format date in human readable format * truncate the message log It also computes the data needed to generate the links for navigating back and forth in the history log. Args: revision_log (list): raw revision log as returned by the SWH web api per_page (int): number of log entries per page revs_breadcrumb (str): breadcrumbs of revisions navigated so far, in the form 'rev1[/rev2/../revN]'. Each revision corresponds to the first one displayed in the HTML view for history log. origin_context (boolean): wheter or not the revision log is browsed from an origin view. """ current_rev = revision_log[0]['id'] next_rev = None prev_rev = None next_revs_breadcrumb = None prev_revs_breadcrumb = None if len(revision_log) == per_page + 1: prev_rev = revision_log[-1]['id'] prev_rev_bc = current_rev if origin_context: prev_rev_bc = prev_rev if revs_breadcrumb: revs = revs_breadcrumb.split('/') next_rev = revs[-1] if len(revs) > 1: next_revs_breadcrumb = '/'.join(revs[:-1]) if len(revision_log) == per_page + 1: prev_revs_breadcrumb = revs_breadcrumb + '/' + prev_rev_bc else: prev_revs_breadcrumb = prev_rev_bc - return {'revision_log_data': _format_log_entries(revision_log, per_page), + return {'revision_log_data': _format_log_entries(revision_log, per_page, + origin_id), 'prev_rev': prev_rev, 'prev_revs_breadcrumb': prev_revs_breadcrumb, 'next_rev': next_rev, 'next_revs_breadcrumb': next_revs_breadcrumb} diff --git a/swh/web/browse/views/origin.py b/swh/web/browse/views/origin.py index 7fb66dc6..042209fb 100644 --- a/swh/web/browse/views/origin.py +++ b/swh/web/browse/views/origin.py @@ -1,673 +1,667 @@ # Copyright (C) 2017 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import dateutil import json from distutils.util import strtobool from django.http import HttpResponse from django.shortcuts import render from django.utils.safestring import mark_safe from django.template.defaultfilters import filesizeformat from swh.web.common import service from swh.web.common.utils import ( gen_path_info, reverse, format_utc_iso_date ) from swh.web.common.exc import NotFoundExc, handle_view_exception from swh.web.browse.utils import ( get_origin_visits, get_origin_visit, get_origin_visit_branches, get_directory_entries, request_content, prepare_content_for_display, gen_link, - prepare_revision_log_for_display + prepare_revision_log_for_display, gen_origin_link ) from swh.web.browse.browseurls import browse_route @browse_route(r'origin/(?P[0-9]+)/', r'origin/(?P[a-z]+)/url/(?P.+)/', view_name='browse-origin') def origin_browse(request, origin_id=None, origin_type=None, origin_url=None): """Django view that produces an HTML display of a swh origin identified by its id or its url. The url scheme that points to it is :http:get:`/browse/origin/(origin_id)/`. Args: request: input django http request origin_id: a swh origin id origin_type: type of origin (git, svn, ...) origin_url: url of the origin (e.g. https://github.com//) Returns: The HMTL rendering for the metadata of the provided origin. """ # noqa try: if origin_id: origin_request_params = { 'id': origin_id, } else: origin_request_params = { 'type': origin_type, 'url': origin_url } origin_info = service.lookup_origin(origin_request_params) origin_id = origin_info['id'] origin_visits = get_origin_visits(origin_id) except Exception as exc: return handle_view_exception(request, exc) origin_info['last swh visit browse url'] = \ reverse('browse-origin-directory', kwargs={'origin_id': origin_id}) origin_visits_data = [] for visit in origin_visits: visit_date = dateutil.parser.parse(visit['date']) visit['date'] = format_utc_iso_date(visit['date']) visit['browse_url'] = reverse('browse-origin-directory', kwargs={'origin_id': origin_id, 'visit_id': visit['visit']}) origin_visits_data.append( {'date': visit_date.timestamp()}) return render(request, 'origin.html', {'heading': 'Origin information', 'top_panel_visible': True, 'top_panel_collapsible': True, 'top_panel_text_left': 'SWH object: Origin', 'top_panel_text_right': 'Url: ' + origin_info['url'], 'swh_object_metadata': origin_info, 'main_panel_visible': True, 'origin_visits_data': origin_visits_data, 'visits': list(reversed(origin_visits)), 'browse_url_base': '/browse/origin/%s/' % origin_id}) def _get_origin_branches_and_url_args(origin_id, visit_id, ts): if not visit_id and ts: branches = get_origin_visit_branches(origin_id, visit_ts=ts) url_args = {'origin_id': origin_id, 'timestamp': ts} else: branches = get_origin_visit_branches(origin_id, visit_id) url_args = {'origin_id': origin_id, 'visit_id': visit_id} return branches, url_args def _raise_exception_if_branch_not_found(origin_id, visit_id, ts, branch, branches): if visit_id: if len(branches) == 0: raise NotFoundExc('Origin with id %s is empty for visit' ' with id %s! No existing branches were found!' % (origin_id, visit_id)) else: raise NotFoundExc('Branch %s associated to visit with' ' id %s for origin with id %s' ' not found!' % (branch, visit_id, origin_id)) else: if len(branches) == 0: raise NotFoundExc('Origin with id %s is empty for visit' ' with timestamp %s! No existing branches' ' were found!' % (origin_id, ts)) else: raise NotFoundExc('Branch %s associated to visit with' ' timestamp %s for origin with id %s' ' not found!' % (branch, ts, origin_id)) def _get_branch(branches, branch_name): """ Utility function to get a specific branch from an origin 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)] # noqa 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')] # noqa if len(filtered_branches) > 0: return filtered_branches[0] elif len(branches) > 0: return branches[0] return None -def _gen_origin_link(origin_id, origin_url): - origin_browse_url = reverse('browse-origin', - kwargs={'origin_id': origin_id}) - return gen_link(origin_browse_url, - 'Origin: ' + origin_url) - - @browse_route(r'origin/(?P[0-9]+)/directory/', r'origin/(?P[0-9]+)/directory/(?P.+)/', r'origin/(?P[0-9]+)/visit/(?P[0-9]+)/directory/', # noqa r'origin/(?P[0-9]+)/visit/(?P[0-9]+)/directory/(?P.+)/', # noqa r'origin/(?P[0-9]+)/ts/(?P.+)/directory/', # noqa r'origin/(?P[0-9]+)/ts/(?P.+)/directory/(?P.+)/', # noqa view_name='browse-origin-directory') def origin_directory_browse(request, origin_id, visit_id=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_id)/directory/[(path)/]` * :http:get:`/browse/origin/(origin_id)/visit/(visit_id)/directory/[(path)/]` * :http:get:`/browse/origin/(origin_id)/ts/(timestamp)/directory/[(path)/]` Args: request: input django http request origin_id: a swh origin id visit_id: optionnal visit id parameter (the last one will be used by default) timestamp: optionnal visit timestamp parameter (the last one will be used by default) path: optionnal path parameter used to navigate in directories reachable from the origin root one branch: optionnal query parameter that specifies the origin branch from which to retrieve the directory revision: optional query parameter to specify the origin revision from which to retrieve the directory Returns: The HTML rendering for the content of the directory associated to the provided origin and visit. """ # noqa try: if not visit_id and not timestamp: origin_visits = get_origin_visits(origin_id) if not origin_visits: raise NotFoundExc('No SWH visit associated to ' 'origin with id %s' % origin_id) return origin_directory_browse(request, origin_id, origin_visits[-1]['visit'], path=path) origin_info = service.lookup_origin({'id': origin_id}) branches, url_args = \ _get_origin_branches_and_url_args(origin_id, visit_id, timestamp) visit_info = get_origin_visit(origin_id, visit_id, timestamp) for b in branches: branch_url_args = dict(url_args) if path: b['path'] = path branch_url_args['path'] = path b['url'] = reverse('browse-origin-directory', kwargs=branch_url_args, query_params={'branch': b['name']}) revision_id = request.GET.get('revision', 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 else: branch_name = request.GET.get('branch', 'HEAD') branch = _get_branch(branches, branch_name) if branch: branch_name = branch['name'] root_sha1_git = branch['directory'] else: _raise_exception_if_branch_not_found(origin_id, visit_id, timestamp, branch_name, branches) sha1_git = root_sha1_git if path: dir_info = service.lookup_directory_with_path(root_sha1_git, path) sha1_git = dir_info['target'] dirs, files = get_directory_entries(sha1_git) except Exception as exc: return handle_view_exception(request, exc) if revision_id: query_params = {'revision': revision_id} else: query_params = {'branch': branch_name} path_info = gen_path_info(path) breadcrumbs = [] breadcrumbs.append({'name': root_sha1_git[:7], 'url': reverse('browse-origin-directory', 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-origin-directory', 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-origin-directory', kwargs=bc_url_args, query_params=query_params) sum_file_sizes = 0 readme_name = None readme_url = None for f in files: bc_url_args = dict(url_args) bc_url_args['path'] = path + f['name'] f['url'] = reverse('browse-origin-content', kwargs=bc_url_args, query_params=query_params) sum_file_sizes += f['length'] f['length'] = filesizeformat(f['length']) if f['name'].lower().startswith('readme'): readme_name = f['name'] readme_sha1 = f['checksums']['sha1'] readme_url = reverse('browse-content-raw', kwargs={'query_string': readme_sha1}) history_url = reverse('browse-origin-log', kwargs=url_args, query_params=query_params) 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, 'origin id': origin_info['id'], 'origin type': origin_info['type'], 'origin url': origin_info['url'], 'origin visit': format_utc_iso_date(visit_info['date']), 'path': '/' + path} return render(request, 'directory.html', {'heading': 'Directory information', 'top_panel_visible': True, 'top_panel_collapsible': True, 'top_panel_text_left': 'SWH object: Directory', - 'top_panel_text_right': _gen_origin_link( + 'top_panel_text_right': gen_origin_link( origin_id, origin_info['url']), 'swh_object_metadata': dir_metadata, 'main_panel_visible': True, 'dirs': dirs, 'files': files, 'breadcrumbs': breadcrumbs, 'branches': branches, 'branch': branch_name, 'top_right_link': history_url, 'top_right_link_text': mark_safe( '' 'History' ), 'readme_name': readme_name, 'readme_url': readme_url}) @browse_route(r'origin/(?P[0-9]+)/content/(?P.+)/', r'origin/(?P[0-9]+)/visit/(?P[0-9]+)/content/(?P.+)/', # noqa r'origin/(?P[0-9]+)/ts/(?P.+)/content/(?P.+)/', # noqa view_name='browse-origin-content') def origin_content_display(request, origin_id, path, visit_id=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_id)/content/(path)/` * :http:get:`/browse/origin/(origin_id)/visit/(visit_id)/content/(path)/` * :http:get:`/browse/origin/(origin_id)/ts/(timestamp)/content/(path)/` Args: request: input django http request origin_id: id of a swh origin path: path of the content relative to the origin root directory visit_id: optionnal visit id parameter (the last one will be used by default) timestamp: optionnal visit timestamp parameter (the last one will be used by default) branch: optionnal query parameter that specifies the origin branch from which to retrieve the content revision: optional query parameter to specify the origin revision from which to retrieve the content Returns: The HTML rendering of the requested content associated to the provided origin and visit. """ # noqa try: if not visit_id and not timestamp: origin_visits = get_origin_visits(origin_id) if not origin_visits: raise NotFoundExc('No SWH visit associated to ' 'origin with id %s' % origin_id) return origin_content_display(request, origin_id, path, origin_visits[-1]['visit']) origin_info = service.lookup_origin({'id': origin_id}) branches, url_args = \ _get_origin_branches_and_url_args(origin_id, visit_id, timestamp) visit_info = get_origin_visit(origin_id, visit_id, timestamp) for b in branches: bc_url_args = dict(url_args) bc_url_args['path'] = path b['url'] = reverse('browse-origin-content', kwargs=bc_url_args, query_params={'branch': b['name']}) revision_id = request.GET.get('revision', 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 else: branch_name = request.GET.get('branch', 'HEAD') branch = _get_branch(branches, branch_name) if branch: branch_name = branch['name'] root_sha1_git = branch['directory'] else: _raise_exception_if_branch_not_found(origin_id, visit_id, timestamp, branch_name, branches) 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) if revision_id: query_params = {'revision': revision_id} else: query_params = {'branch': branch_name} content_display_data = prepare_content_for_display( content_data['raw_data'], content_data['mimetype'], path) filename = None path_info = None breadcrumbs = [] split_path = path.split('/') filename = split_path[-1] path = path.replace(filename, '') path_info = gen_path_info(path) breadcrumbs.append({'name': root_sha1_git[:7], 'url': reverse('browse-origin-directory', 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-origin-directory', kwargs=bc_url_args, query_params=query_params)}) breadcrumbs.append({'name': filename, 'url': None}) content_raw_url = reverse('browse-content-raw', kwargs={'query_string': query_string}, query_params={'filename': filename}) 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'], 'origin id': origin_info['id'], 'origin type': origin_info['type'], 'origin url': origin_info['url'], 'origin visit': format_utc_iso_date(visit_info['date']), 'path': '/' + path, 'filename': filename } return render(request, 'content.html', {'heading': 'Content information', 'top_panel_visible': True, 'top_panel_collapsible': True, 'top_panel_text_left': 'SWH object: Content', - 'top_panel_text_right': _gen_origin_link( + 'top_panel_text_right': gen_origin_link( origin_id, origin_info['url']), 'swh_object_metadata': content_metadata, 'main_panel_visible': True, 'content': content_display_data['content_data'], 'mimetype': content_data['mimetype'], 'language': content_display_data['language'], 'breadcrumbs': breadcrumbs, 'branches': branches, 'branch': branch_name, 'top_right_link': content_raw_url, 'top_right_link_text': mark_safe( 'Raw File') }) def _gen_directory_link(url_args, revision, link_text): directory_url = reverse('browse-origin-directory', kwargs=url_args, query_params={'revision': revision}) return gen_link(directory_url, link_text) NB_LOG_ENTRIES = 20 @browse_route(r'origin/(?P[0-9]+)/log/', r'origin/(?P[0-9]+)/visit/(?P[0-9]+)/log/', # noqa r'origin/(?P[0-9]+)/ts/(?P.+)/log/', view_name='browse-origin-log') def origin_log_browse(request, origin_id, visit_id=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_id)/log/` * :http:get:`/browse/origin/(origin_id)/visit/(visit_id)/log/` * :http:get:`/browse/origin/(origin_id)/ts/(timestamp)/log/` Args: request: input django http request origin_id: id of a swh origin visit_id: optionnal visit id parameter (the last one will be used by default) timestamp: optionnal visit timestamp parameter (the last one will be used by default) revs_breadcrumb: query parameter used internally to store the navigation breadcrumbs (i.e. the list of descendant revisions visited so far). per_page: optionnal query parameter used to specify the number of log entries per page branch: optionnal query parameter that specifies the origin branch from which to retrieve the content revision: optional query parameter to specify the origin revision from which to retrieve the directory Returns: The HTML rendering of revisions history for a given SWH visit. """ # noqa try: if not visit_id and not timestamp: origin_visits = get_origin_visits(origin_id) if not origin_visits: raise NotFoundExc('No SWH visit associated to ' 'origin with id %s' % origin_id) return origin_log_browse(request, origin_id, origin_visits[-1]['visit']) branches, url_args = \ _get_origin_branches_and_url_args(origin_id, visit_id, timestamp) visit_info = get_origin_visit(origin_id, visit_id, timestamp) for b in branches: b['url'] = reverse('browse-origin-log', kwargs=url_args, query_params={'branch': b['name']}) revision_id = request.GET.get('revision', None) revs_breadcrumb = request.GET.get('revs_breadcrumb', None) branch_name = request.GET.get('branch', 'HEAD') if revision_id: revision = service.lookup_revision(revision_id) branches.append({'name': revision_id, 'revision': revision_id, 'directory': revision['directory']}) revision = revision_id branch_name = revision_id elif revs_breadcrumb: revs = revs_breadcrumb.split('/') revision = revs[-1] else: branch = _get_branch(branches, branch_name) if branch: branch_name = branch['name'] revision = branch['revision'] else: _raise_exception_if_branch_not_found(origin_id, visit_id, timestamp, branch_name, branches) per_page = int(request.GET.get('per_page', NB_LOG_ENTRIES)) revision_log = service.lookup_revision_log(revision, limit=per_page+1) revision_log = list(revision_log) except Exception as exc: return handle_view_exception(request, exc) revision_log_display_data = prepare_revision_log_for_display( - revision_log, per_page, revs_breadcrumb, origin_context=True) + revision_log, per_page, revs_breadcrumb, origin_context=True, + origin_id=origin_id) 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-origin-log', kwargs=url_args, query_params={'revs_breadcrumb': prev_revs_breadcrumb, 'per_page': per_page, 'branch': branch_name}) 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-origin-log', kwargs=url_args, query_params={'revs_breadcrumb': next_revs_breadcrumb, 'per_page': per_page, 'branch': branch_name}) revision_log_data = revision_log_display_data['revision_log_data'] for i, log in enumerate(revision_log_data): log['directory'] = _gen_directory_link(url_args, revision_log[i]['id'], 'Tree') origin_info = service.lookup_origin({'id': origin_id}) revision_metadata = { 'origin id': origin_info['id'], 'origin type': origin_info['type'], 'origin url': origin_info['url'], 'origin visit': format_utc_iso_date(visit_info['date']) } return render(request, 'revision-log.html', {'heading': 'Revision history information', 'top_panel_visible': True, 'top_panel_collapsible': True, 'top_panel_text_left': 'SWH object: Revision history', - 'top_panel_text_right': _gen_origin_link( + 'top_panel_text_right': gen_origin_link( origin_id, origin_info['url']), 'swh_object_metadata': revision_metadata, 'main_panel_visible': True, 'revision_log': revision_log_data, 'next_log_url': next_log_url, 'prev_log_url': prev_log_url, 'breadcrumbs': None, 'branches': branches, 'branch': branch_name, 'top_right_link': None, 'top_right_link_text': None, 'include_top_navigation': True, 'no_origin_context': False}) @browse_route(r'origin/search/(?P.+)/', view_name='browse-origin-search') def origin_search(request, url_pattern): """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') diff --git a/swh/web/browse/views/revision.py b/swh/web/browse/views/revision.py index 563ec8bd..fed62cb5 100644 --- a/swh/web/browse/views/revision.py +++ b/swh/web/browse/views/revision.py @@ -1,166 +1,206 @@ # Copyright (C) 2017 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import json 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_link, gen_person_link, gen_revision_link, - prepare_revision_log_for_display + prepare_revision_log_for_display, gen_origin_link ) -def _gen_directory_link(sha1_git, link_text): +def _gen_directory_link(sha1_git, link_text=None): directory_url = reverse('browse-directory', kwargs={'sha1_git': sha1_git}) + if not link_text: + link_text = directory_url return gen_link(directory_url, link_text) -def _gen_revision_log_link(revision_id): - revision_log_url = reverse('browse-revision-log', - kwargs={'sha1_git': revision_id}) - return gen_link(revision_log_url, revision_log_url) +def _gen_origin_directory_link(origin_id, revision_id): + directory_url = reverse('browse-origin-directory', + kwargs={'origin_id': origin_id}, + query_params={'revision': revision_id}) + return gen_link(directory_url, directory_url) + + +def _gen_revision_log_link(revision_id, origin_id=None, link_text=None): + if origin_id: + revision_log_url = reverse('browse-origin-log', + kwargs={'origin_id': origin_id}, + query_params={'revision': revision_id}) + else: + revision_log_url = reverse('browse-revision-log', + kwargs={'sha1_git': revision_id}) + if not link_text: + link_text = revision_log_url + return gen_link(revision_log_url, link_text) @browse_route(r'revision/(?P[0-9a-f]+)/', view_name='browse-revision') def revision_browse(request, sha1_git): """ 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)/`. Args: request: input django http request sha1_git: a SWH revision id Returns: The HMTL rendering for the metadata of the provided revision. """ try: revision = service.lookup_revision(sha1_git) + origin_info = None + origin_id = request.GET.get('origin_id', None) + origin_type = request.GET.get('origin_type', None) + origin_url = request.GET.get('origin_url', None) + if origin_id: + origin_info = service.lookup_origin({'id': origin_id}) + elif origin_type and origin_url: + origin_info = service.lookup_origin({'type': origin_type, + 'url': origin_url}) except Exception as exc: return handle_view_exception(request, exc) revision_data = {} revision_data['author'] = gen_person_link( revision['author']['id'], revision['author']['name']) revision_data['committer'] = gen_person_link( revision['committer']['id'], revision['committer']['name']) revision_data['committer date'] = format_utc_iso_date( revision['committer_date']) revision_data['date'] = format_utc_iso_date(revision['date']) - revision_data['directory'] = _gen_directory_link(revision['directory'], - revision['directory']) - revision_data['history log'] = _gen_revision_log_link(sha1_git) + if origin_info: + revision_data['directory'] = _gen_origin_directory_link( + origin_info['id'], sha1_git) + revision_data['history log'] = _gen_revision_log_link( + sha1_git, origin_info['id']) + else: + revision_data['directory'] = _gen_directory_link(revision['directory']) + revision_data['history log'] = _gen_revision_log_link(sha1_git) revision_data['id'] = sha1_git revision_data['merge'] = revision['merge'] revision_data['message'] = revision['message'] revision_data['metadata'] = json.dumps(revision['metadata'], sort_keys=True, indent=4, separators=(',', ': ')) + if origin_info: + 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']) + top_panel_text_right = gen_origin_link(origin_info['id'], + origin_info['url']) + else: + top_panel_text_right = 'Sha1 git: ' + sha1_git + parents = '' for p in revision['parents']: - parent_link = gen_revision_link(p) + parent_link = gen_revision_link(p, origin_id=origin_id) parents += parent_link + '
' revision_data['parents'] = mark_safe(parents) revision_data['synthetic'] = revision['synthetic'] revision_data['type'] = revision['type'] return render(request, 'revision.html', {'heading': 'Revision information', 'top_panel_visible': False, 'top_panel_collapsible': False, 'top_panel_text_left': 'SWH object: Revision', - 'top_panel_text_right': 'Sha1 git: ' + sha1_git, + 'top_panel_text_right': top_panel_text_right, 'swh_object_metadata': None, 'main_panel_visible': True, 'revision': revision_data}) 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/`. Args: request: input django http request sha1_git: a SWH revision id Returns: The HMTL rendering of the revision history 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'], 'Tree') return render(request, 'revision-log.html', {'heading': 'Revision history information', 'top_panel_visible': False, 'top_panel_collapsible': False, 'top_panel_text_left': 'SWH object: Revision history', 'top_panel_text_right': 'Sha1 git: ' + sha1_git, 'swh_object_metadata': None, 'main_panel_visible': True, 'revision_log': revision_log_data, 'next_log_url': next_log_url, 'prev_log_url': prev_log_url, 'breadcrumbs': None, 'branches': None, 'branch': None, 'top_right_link': None, 'top_right_link_text': None, 'include_top_navigation': False, 'no_origin_context': True}) diff --git a/swh/web/templates/revision.html b/swh/web/templates/revision.html index f19c477a..8081a5cd 100644 --- a/swh/web/templates/revision.html +++ b/swh/web/templates/revision.html @@ -1,24 +1,24 @@ {% extends "browse.html" %} {% block swh-browse-main-panel-content %}
-
+

{{ top_panel_text_left }}

-
+

{{ top_panel_text_right }}

{% for key, val in revision.items|dictsort:"0.lower" %} {% endfor %}
{{ key }} 
{{ val }}
{% endblock %} \ No newline at end of file diff --git a/swh/web/tests/browse/views/test_revision.py b/swh/web/tests/browse/views/test_revision.py index 430548b0..8cdaa6af 100644 --- a/swh/web/tests/browse/views/test_revision.py +++ b/swh/web/tests/browse/views/test_revision.py @@ -1,200 +1,251 @@ # Copyright (C) 2017 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information from unittest.mock import patch from nose.tools import istest from django.test import TestCase from django.utils.html import escape from swh.web.common.exc import NotFoundExc from swh.web.common.utils import reverse, format_utc_iso_date from swh.web.tests.testbase import SWHWebTestBase from .data.revision_test_data import ( revision_id_test, revision_metadata_test, revision_history_log_test ) class SwhBrowseRevisionTest(SWHWebTestBase, TestCase): @patch('swh.web.browse.views.revision.service') @istest def revision_browse(self, mock_service): mock_service.lookup_revision.return_value = revision_metadata_test url = reverse('browse-revision', kwargs={'sha1_git': revision_id_test}) author_id = revision_metadata_test['author']['id'] author_name = revision_metadata_test['author']['name'] committer_id = revision_metadata_test['committer']['id'] committer_name = revision_metadata_test['committer']['name'] dir_id = revision_metadata_test['directory'] author_url = reverse('browse-person', kwargs={'person_id': author_id}) committer_url = reverse('browse-person', kwargs={'person_id': committer_id}) directory_url = reverse('browse-directory', kwargs={'sha1_git': dir_id}) history_url = reverse('browse-revision-log', kwargs={'sha1_git': revision_id_test}) resp = self.client.get(url) self.assertEquals(resp.status_code, 200) self.assertTemplateUsed('revision.html') self.assertContains(resp, '%s' % (author_url, author_name)) self.assertContains(resp, '%s' % (committer_url, committer_name)) self.assertContains(resp, '%s' % - (directory_url, dir_id)) + (directory_url, directory_url)) self.assertContains(resp, '%s' % (history_url, history_url)) for parent in revision_metadata_test['parents']: parent_url = reverse('browse-revision', kwargs={'sha1_git': parent}) self.assertContains(resp, '%s' % (parent_url, parent)) author_date = revision_metadata_test['date'] committer_date = revision_metadata_test['committer_date'] message = revision_metadata_test['message'] self.assertContains(resp, format_utc_iso_date(author_date)) self.assertContains(resp, format_utc_iso_date(committer_date)) self.assertContains(resp, message) + origin_info = { + 'id': '7416001', + 'type': 'git', + 'url': 'https://github.com/webpack/webpack' + } + + mock_service.lookup_origin.return_value = origin_info + + origin_directory_url = reverse('browse-origin-directory', + kwargs={'origin_id': origin_info['id']}, + query_params={'revision': revision_id_test}) # noqa + + origin_revision_log_url = reverse('browse-origin-log', + kwargs={'origin_id': origin_info['id']}, # noqa + query_params={'revision': revision_id_test}) # noqa + + url = reverse('browse-revision', + kwargs={'sha1_git': revision_id_test}, + query_params={'origin_id': origin_info['id']}) # noqa + + resp = self.client.get(url) + + self.assertContains(resp, '%s' % + (origin_directory_url, origin_directory_url)) + + self.assertContains(resp, '%s' % + (origin_revision_log_url, origin_revision_log_url)) # noqa + + self.assertContains(resp, '%s' % + (origin_info['url'], origin_info['url'])) + + for parent in revision_metadata_test['parents']: + parent_url = reverse('browse-revision', + kwargs={'sha1_git': parent}, + query_params={'origin_id': origin_info['id']}) + self.assertContains(resp, '%s' % + (parent_url, parent)) + @patch('swh.web.browse.views.revision.service') @istest def revision_log_browse(self, mock_service): per_page = 10 mock_service.lookup_revision_log.return_value = \ revision_history_log_test[:per_page+1] url = reverse('browse-revision-log', kwargs={'sha1_git': revision_id_test}, query_params={'per_page': per_page}) resp = self.client.get(url) prev_rev = revision_history_log_test[per_page]['id'] next_page_url = reverse('browse-revision-log', kwargs={'sha1_git': prev_rev}, query_params={'revs_breadcrumb': revision_id_test, # noqa 'per_page': per_page}) self.assertEquals(resp.status_code, 200) self.assertTemplateUsed('revision-log.html') self.assertContains(resp, '', count=per_page) self.assertContains(resp, '
  • Newer
    • ') self.assertContains(resp, '
  • Older
    • ' % escape(next_page_url)) for log in revision_history_log_test[:per_page]: author_url = reverse('browse-person', kwargs={'person_id': log['author']['id']}) revision_url = reverse('browse-revision', kwargs={'sha1_git': log['id']}) directory_url = reverse('browse-directory', kwargs={'sha1_git': log['directory']}) self.assertContains(resp, '%s' % (author_url, log['author']['name'])) self.assertContains(resp, '%s' % (revision_url, log['id'][:7])) self.assertContains(resp, '%s' % (directory_url, 'Tree')) mock_service.lookup_revision_log.return_value = \ revision_history_log_test[per_page:2*per_page+1] resp = self.client.get(next_page_url) prev_prev_rev = revision_history_log_test[2*per_page]['id'] prev_page_url = reverse('browse-revision-log', kwargs={'sha1_git': revision_id_test}, query_params={'per_page': per_page}) next_page_url = reverse('browse-revision-log', kwargs={'sha1_git': prev_prev_rev}, query_params={'revs_breadcrumb': revision_id_test + '/' + prev_rev, # noqa 'per_page': per_page}) self.assertEquals(resp.status_code, 200) self.assertTemplateUsed('revision-log.html') self.assertContains(resp, '', count=per_page) self.assertContains(resp, '
  • Newer
    • ' % escape(prev_page_url)) self.assertContains(resp, '
  • Older
    • ' % escape(next_page_url)) mock_service.lookup_revision_log.return_value = \ revision_history_log_test[2*per_page:3*per_page+1] resp = self.client.get(next_page_url) prev_prev_prev_rev = revision_history_log_test[3*per_page]['id'] prev_page_url = reverse('browse-revision-log', kwargs={'sha1_git': prev_rev}, query_params={'revs_breadcrumb': revision_id_test, # noqa 'per_page': per_page}) next_page_url = reverse('browse-revision-log', kwargs={'sha1_git': prev_prev_prev_rev}, query_params={'revs_breadcrumb': revision_id_test + '/' + prev_rev + '/' + prev_prev_rev, # noqa 'per_page': per_page}) self.assertEquals(resp.status_code, 200) self.assertTemplateUsed('revision-log.html') self.assertContains(resp, '', count=per_page) self.assertContains(resp, '
  • Newer
    • ' % escape(prev_page_url)) self.assertContains(resp, '
  • Older
    • ' % escape(next_page_url)) mock_service.lookup_revision_log.return_value = \ revision_history_log_test[3*per_page:3*per_page+per_page//2] resp = self.client.get(next_page_url) prev_page_url = reverse('browse-revision-log', kwargs={'sha1_git': prev_prev_rev}, query_params={'revs_breadcrumb': revision_id_test + '/' + prev_rev, # noqa 'per_page': per_page}) self.assertEquals(resp.status_code, 200) self.assertTemplateUsed('revision-log.html') self.assertContains(resp, '', count=per_page//2) self.assertContains(resp, '
  • Older
    • ') self.assertContains(resp, '
  • Newer
    • ' % escape(prev_page_url)) @patch('swh.web.browse.views.revision.service') @istest def revision_request_errors(self, mock_service): mock_service.lookup_revision.side_effect = \ NotFoundExc('Revision not found') url = reverse('browse-revision', kwargs={'sha1_git': revision_id_test}) resp = self.client.get(url) self.assertEquals(resp.status_code, 404) self.assertTemplateUsed('error.html') self.assertContains(resp, 'Revision not found', status_code=404) mock_service.lookup_revision_log.side_effect = \ NotFoundExc('Revision not found') url = reverse('browse-revision-log', kwargs={'sha1_git': revision_id_test}) resp = self.client.get(url) self.assertEquals(resp.status_code, 404) self.assertTemplateUsed('error.html') self.assertContains(resp, 'Revision not found', status_code=404) + + url = reverse('browse-revision', + kwargs={'sha1_git': revision_id_test}, + query_params={'origin_id': '54'}) + + mock_service.lookup_revision.side_effect = None + mock_service.lookup_origin.side_effect = \ + NotFoundExc('Origin not found') + + resp = self.client.get(url) + self.assertEquals(resp.status_code, 404) + self.assertTemplateUsed('error.html') + self.assertContains(resp, 'Origin not found', status_code=404)