diff --git a/swh/web/assets/src/bundles/browse/origin-save.js b/swh/web/assets/src/bundles/browse/origin-save.js index b3d392b9..b1795d45 100644 --- a/swh/web/assets/src/bundles/browse/origin-save.js +++ b/swh/web/assets/src/bundles/browse/origin-save.js @@ -1,197 +1,197 @@ /** * 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 */ import {handleFetchError, csrfPost, isGitRepoUrl} from 'utils/functions'; import {validate} from 'validate.js'; let saveRequestsTable; export function initOriginSave() { $(document).ready(() => { $.fn.dataTable.ext.errMode = 'throw'; fetch(Urls.browse_origin_save_types_list()) .then(response => response.json()) .then(data => { for (let originType of data) { $('#swh-input-origin-type').append(``); } }); saveRequestsTable = $('#swh-origin-save-requests').DataTable({ serverSide: true, ajax: Urls.browse_origin_save_requests_list('all'), columns: [ { data: 'save_request_date', name: 'request_date', render: (data, type, row) => { if (type === 'display') { let date = new Date(data); return date.toLocaleString(); } return data; } }, { data: 'origin_type', name: 'origin_type' }, { data: 'origin_url', name: 'origin_url', render: (data, type, row) => { if (type === 'display') { return `${data}`; } return data; } }, { data: 'save_request_status', name: 'status' }, { data: 'save_task_status', - name: 'save_task_status', + name: 'loading_task_status', render: (data, type, row) => { if (data === 'succeed') { let browseOriginUrl = Urls.browse_origin(row.origin_url); if (row.visit_date) { browseOriginUrl += `visit/${row.visit_date}/`; } return `${data}`; } return data; } } ], scrollY: '50vh', scrollCollapse: true, order: [[0, 'desc']] }); $('#swh-origin-save-requests-list-tab').on('shown.bs.tab', () => { saveRequestsTable.draw(); }); let saveRequestAcceptedALert = ``; let saveRequestPendingAlert = ``; let saveRequestRejectedAlert = ``; $('#swh-save-origin-form').submit(event => { event.preventDefault(); event.stopPropagation(); $('.alert').alert('close'); if (event.target.checkValidity()) { $(event.target).removeClass('was-validated'); let originType = $('#swh-input-origin-type').val(); let originUrl = $('#swh-input-origin-url').val(); let addSaveOriginRequestUrl = Urls.browse_origin_save_request(originType, originUrl); let grecaptchaData = {'g-recaptcha-response': grecaptcha.getResponse()}; let headers = { 'Accept': 'application/json', 'Content-Type': 'application/json' }; let body = JSON.stringify(grecaptchaData); csrfPost(addSaveOriginRequestUrl, headers, body) .then(handleFetchError) .then(response => response.json()) .then(data => { if (data.save_request_status === 'accepted') { $('#swh-origin-save-request-status').html(saveRequestAcceptedALert); } else { $('#swh-origin-save-request-status').html(saveRequestPendingAlert); } grecaptcha.reset(); }) .catch(response => { console.log(response); if (response.status === 403) { $('#swh-origin-save-request-status').css('color', 'red'); $('#swh-origin-save-request-status').html(saveRequestRejectedAlert); } grecaptcha.reset(); }); } else { $(event.target).addClass('was-validated'); } }); $('#swh-show-origin-save-requests-list').on('click', (event) => { event.preventDefault(); $('.nav-tabs a[href="#swh-origin-save-requests-list"]').tab('show'); }); $('#swh-input-origin-url').on('input', function(event) { let originUrl = $(this).val().trim(); $(this).val(originUrl); $('#swh-input-origin-type option').each(function() { let val = $(this).val(); if (val && originUrl.includes(val)) { $(this).prop('selected', true); } }); }); }); } export function validateSaveOriginUrl(input) { let validUrl = validate({website: input.value}, { website: { url: { schemes: ['http', 'https', 'svn', 'git'] } } }) === undefined; let originType = $('#swh-input-origin-type').val(); if (originType === 'git' && validUrl) { // additional checks for well known code hosting providers let githubIdx = input.value.indexOf('://github.com'); let gitlabIdx = input.value.indexOf('://gitlab.'); let gitSfIdx = input.value.indexOf('://git.code.sf.net'); let bitbucketIdx = input.value.indexOf('://bitbucket.org'); if (githubIdx !== -1 && githubIdx <= 5) { validUrl = isGitRepoUrl(input.value, 'github.com'); } else if (gitlabIdx !== -1 && gitlabIdx <= 5) { let startIdx = gitlabIdx + 3; let idx = input.value.indexOf('/', startIdx); if (idx !== -1) { let gitlabDomain = input.value.substr(startIdx, idx - startIdx); // GitLab repo url needs to be suffixed by '.git' in order to be successfully loaded validUrl = isGitRepoUrl(input.value, gitlabDomain) && input.value.endsWith('.git'); } else { validUrl = false; } } else if (gitSfIdx !== -1 && gitSfIdx <= 5) { validUrl = isGitRepoUrl(input.value, 'git.code.sf.net/p'); } else if (bitbucketIdx !== -1 && bitbucketIdx <= 5) { validUrl = isGitRepoUrl(input.value, 'bitbucket.org'); } } if (validUrl) { input.setCustomValidity(''); } else { input.setCustomValidity('The origin url is not valid or does not reference a code repository'); } } diff --git a/swh/web/browse/views/origin_save.py b/swh/web/browse/views/origin_save.py index 46e32ddc..2f70e713 100644 --- a/swh/web/browse/views/origin_save.py +++ b/swh/web/browse/views/origin_save.py @@ -1,91 +1,86 @@ # 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 import json from django.core.paginator import Paginator from django.http import HttpResponse, HttpResponseForbidden from django.views.decorators.http import require_POST from swh.web.browse.browseurls import browse_route from swh.web.common.exc import ForbiddenExc from swh.web.common.models import SaveOriginRequest from swh.web.common.utils import is_recaptcha_valid from swh.web.common.origin_save import ( create_save_origin_request, get_savable_origin_types, get_save_origin_requests_from_queryset ) @browse_route(r'origin/save/(?P.+)/url/(?P.+)/', view_name='browse-origin-save-request') @require_POST def _browse_origin_save_request(request, origin_type, origin_url): body_unicode = request.body.decode('utf-8') body = json.loads(body_unicode) if is_recaptcha_valid(request, body['g-recaptcha-response']): try: response = json.dumps(create_save_origin_request(origin_type, origin_url), separators=(',', ': ')) return HttpResponse(response, content_type='application/json') except ForbiddenExc as exc: return HttpResponseForbidden(str(exc)) else: return HttpResponseForbidden('The reCAPTCHA could not be validated !') @browse_route(r'origin/save/types/list/', view_name='browse-origin-save-types-list') def _browse_origin_save_types_list(request): origin_types = json.dumps(get_savable_origin_types(), separators=(',', ': ')) return HttpResponse(origin_types, content_type='application/json') @browse_route(r'origin/save/requests/list/(?P.+)/', view_name='browse-origin-save-requests-list') def _browse_origin_save_requests_list(request, status): if status != 'all': save_requests = SaveOriginRequest.objects.filter(status=status) else: save_requests = SaveOriginRequest.objects.all() table_data = {} table_data['recordsTotal'] = save_requests.count() table_data['draw'] = int(request.GET['draw']) search_value = request.GET['search[value]'] column_order = request.GET['order[0][column]'] field_order = request.GET['columns[%s][name]' % column_order] order_dir = request.GET['order[0][dir]'] if order_dir == 'desc': field_order = '-' + field_order - if 'save_task_status' not in field_order: - save_requests = save_requests.order_by(field_order) + save_requests = save_requests.order_by(field_order) length = int(request.GET['length']) page = int(request.GET['start']) / length + 1 save_requests = get_save_origin_requests_from_queryset(save_requests) if search_value: save_requests = \ [sr for sr in save_requests if search_value.lower() in sr['save_request_status'].lower() or search_value.lower() in sr['save_task_status'].lower() or search_value.lower() in sr['origin_type'].lower() or search_value.lower() in sr['origin_url'].lower()] - if field_order == 'save_task_status': - save_requests = sorted(save_requests, - key=lambda sr: sr['save_task_status']) - if order_dir == 'desc': - save_requests = reversed(save_requests) + table_data['recordsFiltered'] = len(save_requests) paginator = Paginator(save_requests, length) table_data['data'] = paginator.page(page).object_list table_data_json = json.dumps(table_data, separators=(',', ': ')) return HttpResponse(table_data_json, content_type='application/json') diff --git a/swh/web/common/origin_save.py b/swh/web/common/origin_save.py index c0a7208d..0434e3fe 100644 --- a/swh/web/common/origin_save.py +++ b/swh/web/common/origin_save.py @@ -1,350 +1,354 @@ # 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 from bisect import bisect_right from django.core.exceptions import ObjectDoesNotExist from django.core.exceptions import ValidationError from django.core.validators import URLValidator from swh.web import config from swh.web.common import service from swh.web.common.exc import BadInputExc, ForbiddenExc from swh.web.common.models import ( SaveUnauthorizedOrigin, SaveAuthorizedOrigin, SaveOriginRequest, SAVE_REQUEST_ACCEPTED, SAVE_REQUEST_REJECTED, SAVE_REQUEST_PENDING, SAVE_TASK_NOT_YET_SCHEDULED, SAVE_TASK_SCHEDULED, SAVE_TASK_SUCCEED, SAVE_TASK_FAILED ) from swh.web.common.utils import get_origin_visits, parse_timestamp from swh.scheduler.utils import create_oneshot_task_dict scheduler = config.scheduler() def get_origin_save_authorized_urls(): """ Get the list of origin url prefixes authorized to be immediately loaded into the archive (whitelist). Returns: list: The list of authorized origin url prefix """ return [origin.url for origin in SaveAuthorizedOrigin.objects.all()] def get_origin_save_unauthorized_urls(): """ Get the list of origin url prefixes forbidden to be loaded into the archive (blacklist). Returns: list: the list of unauthorized origin url prefix """ return [origin.url for origin in SaveUnauthorizedOrigin.objects.all()] def can_save_origin(origin_url): """ Check if a software origin can be saved into the archive. Based on the origin url, the save request will be either: * immediately accepted if the url is whitelisted * rejected if the url is blacklisted * put in pending state for manual review otherwise Args: origin_url (str): the software origin url to check Returns: str: the origin save request status, either *accepted*, *rejected* or *pending* """ # origin url may be blacklisted for url_prefix in get_origin_save_unauthorized_urls(): if origin_url.startswith(url_prefix): return SAVE_REQUEST_REJECTED # if the origin url is in the white list, it can be immediately saved for url_prefix in get_origin_save_authorized_urls(): if origin_url.startswith(url_prefix): return SAVE_REQUEST_ACCEPTED # otherwise, the origin url needs to be manually verified return SAVE_REQUEST_PENDING # map origin type to scheduler task # TODO: do not hardcode the task name here # TODO: unlock hg and svn loading once the scheduler # loading tasks are available in production _origin_type_task = { 'git': 'origin-update-git', # 'hg': 'origin-load-hg', # 'svn': 'origin-load-svn' } # map scheduler task status to origin save status _save_task_status = { 'next_run_not_scheduled': SAVE_TASK_NOT_YET_SCHEDULED, 'next_run_scheduled': SAVE_TASK_SCHEDULED, 'completed': SAVE_TASK_SUCCEED, 'disabled': SAVE_TASK_FAILED } def get_savable_origin_types(): return sorted(list(_origin_type_task.keys())) def _check_origin_type_savable(origin_type): """ Get the list of software origin types that can be loaded through a save request. Returns: list: the list of savable origin types """ allowed_origin_types = ', '.join(get_savable_origin_types()) if origin_type not in _origin_type_task: raise BadInputExc('Origin of type %s can not be saved! ' 'Allowed types are the following: %s' % (origin_type, allowed_origin_types)) _validate_url = URLValidator(schemes=['http', 'https', 'svn', 'git']) def _check_origin_url_valid(origin_url): try: _validate_url(origin_url) except ValidationError: raise BadInputExc('The provided origin url (%s) is not valid!' % origin_url) def _get_visit_date_for_save_request(save_request): visit_date = None try: origin = {'type': save_request.origin_type, 'url': save_request.origin_url} origin_info = service.lookup_origin(origin) origin_visits = get_origin_visits(origin_info) visit_dates = [parse_timestamp(v['date']) for v in origin_visits] i = bisect_right(visit_dates, save_request.request_date) if i != len(visit_dates): - save_request.visit_date = visit_dates[i] - save_request.save() visit_date = visit_dates[i] except Exception: pass return visit_date def _save_request_dict(save_request, task=None): + must_save = False visit_date = save_request.visit_date if task: save_task_status = _save_task_status[task['status']] if save_task_status in (SAVE_TASK_FAILED, SAVE_TASK_SUCCEED) \ - and not visit_date: + and not visit_date: visit_date = _get_visit_date_for_save_request(save_request) + save_request.visit_date = visit_date + must_save = True # Ensure last origin visit is available in database # before reporting the task execution as successful if save_task_status == SAVE_TASK_SUCCEED and not visit_date: save_task_status = SAVE_TASK_SCHEDULED - save_request.loading_task_status = save_task_status - save_request.save() + if save_request.loading_task_status != save_task_status: + save_request.loading_task_status = save_task_status + must_save = True + if must_save: + save_request.save() else: save_task_status = save_request.loading_task_status return {'origin_type': save_request.origin_type, 'origin_url': save_request.origin_url, 'save_request_date': save_request.request_date.isoformat(), 'save_request_status': save_request.status, 'save_task_status': save_task_status, 'visit_date': visit_date.isoformat() if visit_date else None} def create_save_origin_request(origin_type, origin_url): """ Create a loading task to save a software origin into the archive. This function aims to create a software origin loading task trough the use of the swh-scheduler component. First, some checks are performed to see if the origin type and url are valid but also if the the save request can be accepted. If those checks passed, the loading task is then created. Otherwise, the save request is put in pending or rejected state. All the submitted save requests are logged into the swh-web database to keep track of them. Args: origin_type (str): the type of origin to save (*git*, *hg*, *svn*, ...) origin_url (str): the url of the origin to save Raises: BadInputExc: the origin type or url is invalid ForbiddenExc: the provided origin url is blacklisted Returns: dict: A dict describing the save request with the following keys: * **origin_type**: the type of the origin to save * **origin_url**: the url of the origin * **save_request_date**: the date the request was submitted * **save_request_status**: the request status, either *accepted*, *rejected* or *pending* * **save_task_status**: the origin loading task status, either *not created*, *not yet scheduled*, *scheduled*, *succeed* or *failed* """ _check_origin_type_savable(origin_type) _check_origin_url_valid(origin_url) save_request_status = can_save_origin(origin_url) task = None # if the origin save request is accepted, create a scheduler # task to load it into the archive if save_request_status == SAVE_REQUEST_ACCEPTED: # create a task with high priority kwargs = {'priority': 'high'} # set task parameters according to the origin type if origin_type == 'git': kwargs['repo_url'] = origin_url elif origin_type == 'hg': kwargs['origin_url'] = origin_url elif origin_type == 'svn': kwargs['origin_url'] = origin_url kwargs['svn_url'] = origin_url sor = None # get list of previously sumitted save requests current_sors = \ list(SaveOriginRequest.objects.filter(origin_type=origin_type, origin_url=origin_url)) can_create_task = False # if no save requests previously submitted, create the scheduler task if not current_sors: can_create_task = True else: # get the latest submitted save request sor = current_sors[0] # if it was in pending state, we need to create the scheduler task # and update the save request info in the database if sor.status == SAVE_REQUEST_PENDING: can_create_task = True # a task has already been created to load the origin elif sor.loading_task_id != -1: # get the scheduler task and its status tasks = scheduler.get_tasks([sor.loading_task_id]) task = tasks[0] if tasks else None task_status = _save_request_dict(sor, task)['save_task_status'] # create a new scheduler task only if the previous one has been # already executed if task_status == SAVE_TASK_FAILED or \ task_status == SAVE_TASK_SUCCEED: can_create_task = True sor = None else: can_create_task = False if can_create_task: # effectively create the scheduler task task_dict = create_oneshot_task_dict( _origin_type_task[origin_type], **kwargs) task = scheduler.create_tasks([task_dict])[0] # pending save request has been accepted if sor: sor.status = SAVE_REQUEST_ACCEPTED sor.loading_task_id = task['id'] sor.save() else: sor = SaveOriginRequest.objects.create(origin_type=origin_type, origin_url=origin_url, status=save_request_status, # noqa loading_task_id=task['id']) # noqa # save request must be manually reviewed for acceptation elif save_request_status == SAVE_REQUEST_PENDING: # check if there is already such a save request already submitted, # no need to add it to the database in that case try: sor = SaveOriginRequest.objects.get(origin_type=origin_type, origin_url=origin_url, status=save_request_status) # if not add it to the database except ObjectDoesNotExist: sor = SaveOriginRequest.objects.create(origin_type=origin_type, origin_url=origin_url, status=save_request_status) # origin can not be saved as its url is blacklisted, # log the request to the database anyway else: sor = SaveOriginRequest.objects.create(origin_type=origin_type, origin_url=origin_url, status=save_request_status) if save_request_status == SAVE_REQUEST_REJECTED: raise ForbiddenExc('The origin url is blacklisted and will not be ' 'loaded into the archive.') return _save_request_dict(sor, task) def get_save_origin_requests_from_queryset(requests_queryset): """ Get all save requests from a SaveOriginRequest queryset. Args: requests_queryset (django.db.models.QuerySet): input SaveOriginRequest queryset Returns: list: A list of save origin requests dict as described in :func:`swh.web.common.origin_save.create_save_origin_request` """ + task_ids = [] + for sor in requests_queryset: + task_ids.append(sor.loading_task_id) + tasks = scheduler.get_tasks(task_ids) + tasks = {task['id']: task for task in tasks} requests = [] for sor in requests_queryset: - task = None - # save task has not been created, no need to query - # the scheduler API - if sor.loading_task_id != -1: - tasks = scheduler.get_tasks([sor.loading_task_id]) - task = tasks[0] if tasks else None - requests.append(_save_request_dict(sor, task)) + sr_dict = _save_request_dict(sor, tasks.get(sor.loading_task_id, None)) + requests.append(sr_dict) return requests def get_save_origin_requests(origin_type, origin_url): """ Get all save requests for a given software origin. Args: origin_type (str): the type of the origin origin_url (str): the url of the origin Raises: BadInputExc: the origin type or url is invalid Returns: list: A list of save origin requests dict as described in :func:`swh.web.common.origin_save.create_save_origin_request` """ _check_origin_type_savable(origin_type) _check_origin_url_valid(origin_url) sors = SaveOriginRequest.objects.filter(origin_type=origin_type, origin_url=origin_url) return get_save_origin_requests_from_queryset(sors)