Exemplo n.º 1
0
    def create(self, docs, **kwargs):
        new_guids = []
        provider = get_resource_service('ingest_providers').find_one(source='aapmm', req=None)
        for doc in docs:
            if not doc.get('desk'):
                # if no desk is selected then it is bad request
                raise SuperdeskApiError.badRequestError("Destination desk cannot be empty.")

            archived_doc = self.backend.find_one_raw(doc['guid'], doc['guid'])

            dest_doc = dict(archived_doc)
            new_id = generate_guid(type=GUID_TAG)
            new_guids.append(new_id)
            dest_doc['_id'] = new_id
            generate_unique_id_and_name(dest_doc)

            if provider:
                dest_doc['ingest_provider'] = str(provider[superdesk.config.ID_FIELD])

            dest_doc[config.VERSION] = 1
            send_to(doc=dest_doc, update=None, desk_id=doc.get('desk'), stage_id=doc.get('stage'))
            dest_doc[config.CONTENT_STATE] = doc.get('state', STATE_FETCHED)
            dest_doc[INGEST_ID] = archived_doc['_id']
            dest_doc[FAMILY_ID] = archived_doc['_id']
            remove_unwanted(dest_doc)
            set_original_creator(dest_doc)

            superdesk.get_resource_service(ARCHIVE).post([dest_doc])
            insert_into_versions(dest_doc.get('_id'))

        return new_guids
Exemplo n.º 2
0
    def _create_image_items(self, image_links, text_item):
        """Create a list of picture items that represent the external images
        located on given URLs.

        Each created item's `firstcreated` and `versioncreated` fields are set
        to the same value as the values of these fields in `text_item`.

        :param iterable image_links: list of image URLs
        :param dict text_item: the "main" text item the images are related to

        :return: list of created image items (as dicts)
        """
        image_items = []

        for image_url in image_links:
            img_item = {
                'guid': generate_guid(type=GUID_TAG),
                'type': 'picture',
                'firstcreated': text_item.get('firstcreated'),
                'versioncreated': text_item.get('versioncreated'),
                'renditions': {
                    'baseImage': {
                        'href': image_url
                    }
                }
            }
            image_items.append(img_item)

        return image_items
Exemplo n.º 3
0
    def create(self, docs, **kwargs):
        new_guids = []
        provider = get_resource_service('ingest_providers').find_one(
            source='aapmm', req=None)
        for doc in docs:
            if not doc.get('desk'):
                # if no desk is selected then it is bad request
                raise SuperdeskApiError.badRequestError(
                    "Destination desk cannot be empty.")

            archived_doc = self.backend.find_one_raw(doc['guid'], doc['guid'])

            dest_doc = dict(archived_doc)
            new_id = generate_guid(type=GUID_TAG)
            new_guids.append(new_id)
            dest_doc['_id'] = new_id
            generate_unique_id_and_name(dest_doc)

            if provider:
                dest_doc['ingest_provider'] = str(
                    provider[superdesk.config.ID_FIELD])

            dest_doc[config.VERSION] = 1
            send_to(dest_doc, doc.get('desk'), doc.get('stage'))
            dest_doc[config.CONTENT_STATE] = doc.get('state', STATE_FETCHED)
            dest_doc[INGEST_ID] = archived_doc['_id']
            dest_doc[FAMILY_ID] = archived_doc['_id']
            remove_unwanted(dest_doc)
            set_original_creator(dest_doc)

            superdesk.get_resource_service(ARCHIVE).post([dest_doc])
            insert_into_versions(dest_doc.get('_id'))

        return new_guids
Exemplo n.º 4
0
    def fetch(self, docs, id=None, **kwargs):
        id_of_fetched_items = []

        for doc in docs:
            id_of_item_to_be_fetched = doc.get('_id') if id is None else id

            desk_id = doc.get('desk')
            stage_id = doc.get('stage')

            ingest_service = get_resource_service('ingest')
            ingest_doc = ingest_service.find_one(req=None,
                                                 _id=id_of_item_to_be_fetched)

            if not ingest_doc:
                raise SuperdeskApiError.notFoundError(
                    'Fail to found ingest item with _id: %s' %
                    id_of_item_to_be_fetched)

            if not is_workflow_state_transition_valid(
                    'fetch_from_ingest', ingest_doc[config.CONTENT_STATE]):
                raise InvalidStateTransitionError()

            if doc.get('macro'):  # there is a macro so transform it
                ingest_doc = get_resource_service('macros').execute_macro(
                    ingest_doc, doc.get('macro'))

            archived = utcnow()
            ingest_service.patch(id_of_item_to_be_fetched,
                                 {'archived': archived})

            dest_doc = dict(ingest_doc)
            new_id = generate_guid(type=GUID_TAG)
            id_of_fetched_items.append(new_id)
            dest_doc['_id'] = new_id
            dest_doc['guid'] = new_id
            dest_doc['destination_groups'] = doc.get('destination_groups')
            generate_unique_id_and_name(dest_doc)

            dest_doc[config.VERSION] = 1
            send_to(dest_doc, desk_id, stage_id)
            dest_doc[config.CONTENT_STATE] = doc.get('state', STATE_FETCHED)
            dest_doc[INGEST_ID] = dest_doc[FAMILY_ID] = ingest_doc['_id']

            remove_unwanted(dest_doc)
            set_original_creator(dest_doc)
            self.__fetch_items_in_package(dest_doc, desk_id, stage_id,
                                          doc.get('state', STATE_FETCHED),
                                          doc.get('destination_groups'))

            get_resource_service(ARCHIVE).post([dest_doc])
            insert_into_versions(doc=dest_doc)
            build_custom_hateoas(custom_hateoas, dest_doc)
            doc.update(dest_doc)

        if kwargs.get('notify', True):
            push_notification('item:fetch', fetched=1)

        return id_of_fetched_items
Exemplo n.º 5
0
def ingest_item(item, provider, rule_set=None, routing_scheme=None):
    try:
        item.setdefault(superdesk.config.ID_FIELD, generate_guid(type=GUID_NEWSML))
        item[FAMILY_ID] = item[superdesk.config.ID_FIELD]
        providers[provider.get("type")].provider = provider

        item["ingest_provider"] = str(provider[superdesk.config.ID_FIELD])
        item.setdefault("source", provider.get("source", ""))
        set_default_state(item, STATE_INGESTED)
        item["expiry"] = get_expiry_date(
            provider.get("content_expiry", INGEST_EXPIRY_MINUTES), item.get("versioncreated")
        )

        if "anpa-category" in item:
            process_anpa_category(item, provider)

        if "subject" in item:
            process_iptc_codes(item, provider)

        apply_rule_set(item, provider, rule_set)

        ingest_service = superdesk.get_resource_service("ingest")

        if item.get("ingest_provider_sequence") is None:
            ingest_service.set_ingest_provider_sequence(item, provider)

        rend = item.get("renditions", {})
        if rend:
            baseImageRend = rend.get("baseImage") or next(iter(rend.values()))
            if baseImageRend:
                href = providers[provider.get("type")].prepare_href(baseImageRend["href"])
                update_renditions(item, href)

        old_item = ingest_service.find_one(guid=item[GUID_FIELD], req=None)

        if old_item:
            # In case we already have the item, preserve the _id
            item[superdesk.config.ID_FIELD] = old_item[superdesk.config.ID_FIELD]
            ingest_service.put_in_mongo(item[superdesk.config.ID_FIELD], item)
        else:
            try:
                ingest_service.post_in_mongo([item])
            except HTTPException as e:
                logger.error("Exception while persisting item in ingest collection", e)

        if routing_scheme:
            routed = ingest_service.find_one(_id=item[superdesk.config.ID_FIELD], req=None)
            superdesk.get_resource_service("routing_schemes").apply_routing_scheme(routed, provider, routing_scheme)
    except Exception as ex:
        logger.exception(ex)
        try:
            superdesk.app.sentry.captureException()
        except:
            pass
        return False
    return True
Exemplo n.º 6
0
    def fetch(self, docs, id=None, **kwargs):
        id_of_fetched_items = []

        for doc in docs:
            id_of_item_to_be_fetched = doc.get('_id') if id is None else id

            desk_id = doc.get('desk')
            stage_id = doc.get('stage')

            ingest_service = get_resource_service('ingest')
            ingest_doc = ingest_service.find_one(req=None, _id=id_of_item_to_be_fetched)

            if not ingest_doc:
                raise SuperdeskApiError.notFoundError('Fail to found ingest item with _id: %s' %
                                                      id_of_item_to_be_fetched)

            if not is_workflow_state_transition_valid('fetch_from_ingest', ingest_doc[config.CONTENT_STATE]):
                raise InvalidStateTransitionError()

            if doc.get('macro'):  # there is a macro so transform it
                ingest_doc = get_resource_service('macros').execute_macro(ingest_doc, doc.get('macro'))

            archived = utcnow()
            ingest_service.patch(id_of_item_to_be_fetched, {'archived': archived})

            dest_doc = dict(ingest_doc)
            new_id = generate_guid(type=GUID_TAG)
            id_of_fetched_items.append(new_id)
            dest_doc['_id'] = new_id
            dest_doc['guid'] = new_id
            dest_doc['destination_groups'] = doc.get('destination_groups')
            generate_unique_id_and_name(dest_doc)

            dest_doc[config.VERSION] = 1
            send_to(dest_doc, desk_id, stage_id)
            dest_doc[config.CONTENT_STATE] = doc.get('state', STATE_FETCHED)
            dest_doc[INGEST_ID] = dest_doc[FAMILY_ID] = ingest_doc['_id']

            remove_unwanted(dest_doc)
            set_original_creator(dest_doc)
            self.__fetch_items_in_package(dest_doc, desk_id, stage_id,
                                          doc.get('state', STATE_FETCHED),
                                          doc.get('destination_groups'))

            get_resource_service(ARCHIVE).post([dest_doc])
            insert_into_versions(doc=dest_doc)
            build_custom_hateoas(custom_hateoas, dest_doc)
            doc.update(dest_doc)

        if kwargs.get('notify', True):
            push_notification('item:fetch', fetched=1)

        return id_of_fetched_items
Exemplo n.º 7
0
 def on_create(self, docs):
     for doc in docs:
         update_dates_for(doc)
         doc['original_creator'] = str(get_user().get('_id'))
         doc['guid'] = generate_guid(type=GUID_TAG)
         # set the blog_preferences by merging given preferences with global_prefs
         global_prefs = get_resource_service('global_preferences').get_global_prefs()
         prefs = global_prefs.copy()
         prefs.update(doc.get('blog_preferences', {}))
         doc['blog_preferences'] = prefs
         # save a snapshot of the theme in the `theme` field
         if 'theme' in prefs:
             doc['theme'] = self.get_theme_snapshot(prefs['theme'])
Exemplo n.º 8
0
    def _create_package(self, text_item, image_items):
        """Create a new content package from given content items.

        The package's `main` group contains only the references to given items,
        not the items themselves. In the list of references, the reference to
        the text item preceeds the references to image items.

        Package's `firstcreated` and `versioncreated` fields are set to values
        of these fields in `text_item`.

        :param dict text_item: item representing the text content
        :param list image_items: list of items (dicts) representing the images
            related to the text content
        :return: the created content package
        :rtype: dict
        """
        package = {
            'type': 'composite',
            'guid': generate_guid(type=GUID_TAG),
            'firstcreated': text_item['firstcreated'],
            'versioncreated': text_item['versioncreated'],
            'groups': [
                {
                    'id': 'root',
                    'role': 'grpRole:NEP',
                    'refs': [{'idRef': 'main'}],
                }, {
                    'id': 'main',
                    'role': 'main',
                    'refs': [],
                }
            ]
        }

        item_references = package['groups'][1]['refs']
        item_references.append({'residRef': text_item['guid']})

        for image in image_items:
            item_references.append({'residRef': image['guid']})

        return package
    def ingest_items_for(self, desk, no_of_stories, skip_index):
        desk_id = desk['_id']
        stage_id = desk['incoming_stage']

        bucket_size = min(100, no_of_stories)

        no_of_buckets = len(range(0, no_of_stories, bucket_size))

        for x in range(0, no_of_buckets):
            skip = x * bucket_size * skip_index
            self.logger.info('Page : {}, skip: {}'.format(x + 1, skip))
            cursor = get_resource_service('text_archive').get_from_mongo(None, {})
            cursor.skip(skip)
            cursor.limit(bucket_size)
            items = list(cursor)
            self.logger.info('Inserting {} items'.format(len(items)))
            archive_items = []

            for item in items:
                dest_doc = dict(item)
                new_id = generate_guid(type=GUID_TAG)
                dest_doc['_id'] = new_id
                dest_doc['guid'] = new_id
                generate_unique_id_and_name(dest_doc)

                dest_doc[app.config['VERSION']] = 1
                dest_doc['state'] = 'fetched'
                user_id = desk.get('members', [{'user': None}])[0].get('user')
                dest_doc['original_creator'] = user_id
                dest_doc['version_creator'] = user_id
                send_to(dest_doc, desk_id=desk_id, stage_id=stage_id, user_id=user_id)
                dest_doc[FAMILY_ID] = item['_id']

                remove_unwanted(dest_doc)
                archive_items.append(dest_doc)

            get_resource_service(ARCHIVE).post(archive_items)
            for item in archive_items:
                insert_into_versions(id_=item['_id'])
Exemplo n.º 10
0
 def on_create(self, docs):
     for doc in docs:
         update_dates_for(doc)
         doc['original_creator'] = str(get_user().get('_id'))
         doc['guid'] = generate_guid(type=GUID_TAG)
Exemplo n.º 11
0
def ingest_item(item, provider, rule_set=None, routing_scheme=None):
    try:
        item.setdefault(superdesk.config.ID_FIELD,
                        generate_guid(type=GUID_NEWSML))
        item[FAMILY_ID] = item[superdesk.config.ID_FIELD]
        providers[provider.get('type')].provider = provider

        item['ingest_provider'] = str(provider[superdesk.config.ID_FIELD])
        item.setdefault('source', provider.get('source', ''))
        set_default_state(item, STATE_INGESTED)
        item['expiry'] = get_expiry_date(
            provider.get('content_expiry', INGEST_EXPIRY_MINUTES),
            item.get('versioncreated'))

        if 'anpa-category' in item:
            process_anpa_category(item, provider)

        if 'subject' in item:
            process_iptc_codes(item, provider)

        apply_rule_set(item, provider, rule_set)

        ingest_service = superdesk.get_resource_service('ingest')

        if item.get('ingest_provider_sequence') is None:
            ingest_service.set_ingest_provider_sequence(item, provider)

        rend = item.get('renditions', {})
        if rend:
            baseImageRend = rend.get('baseImage') or next(iter(rend.values()))
            if baseImageRend:
                href = providers[provider.get('type')].prepare_href(
                    baseImageRend['href'])
                update_renditions(item, href)

        old_item = ingest_service.find_one(guid=item[GUID_FIELD], req=None)

        if old_item:
            # In case we already have the item, preserve the _id
            item[superdesk.config.ID_FIELD] = old_item[
                superdesk.config.ID_FIELD]
            ingest_service.put_in_mongo(item[superdesk.config.ID_FIELD], item)
        else:
            try:
                ingest_service.post_in_mongo([item])
            except HTTPException as e:
                logger.error(
                    "Exception while persisting item in ingest collection", e)

        if routing_scheme:
            routed = ingest_service.find_one(
                _id=item[superdesk.config.ID_FIELD], req=None)
            superdesk.get_resource_service(
                'routing_schemes').apply_routing_scheme(
                    routed, provider, routing_scheme)
    except Exception as ex:
        logger.exception(ex)
        try:
            superdesk.app.sentry.captureException()
        except:
            pass
        return False
    return True