Esempio n. 1
0
def ingest(path, dataset, languages=None):
    """Queue a set of files for ingest."""
    context = {'languages': languages}
    conn = get_redis()
    job = Job.create(conn, dataset)
    stage = job.get_stage(Stage.INGEST)
    manager = Manager(stage, context)
    path = ensure_path(path)
    if path is not None:
        if path.is_file():
            entity = manager.make_entity('Document')
            checksum = manager.store(path)
            entity.set('contentHash', checksum)
            entity.make_id(checksum)
            entity.set('fileName', path.name)
            manager.queue_entity(entity)
        if path.is_dir():
            DirectoryIngestor.crawl(manager, path)
    manager.close()
Esempio n. 2
0
File: core.py Progetto: we1l1n/aleph
def get_cache():
    if not hasattr(settings, '_cache') or settings._cache is None:
        settings._cache = Cache(get_redis(), prefix=settings.APP_NAME)
    return settings._cache
Esempio n. 3
0
def connect_redis():
    if settings.TESTING:
        return get_fakeredis()
    return get_redis()
Esempio n. 4
0
def killthekitten():
    """Completely kill redis contents."""
    conn = get_redis()
    conn.flushall()
Esempio n. 5
0
def cancel(dataset):
    """Delete scheduled tasks for given dataset"""
    conn = get_redis()
    Dataset(conn, dataset).cancel()
Esempio n. 6
0
def ingest(path, dataset, languages=None):
    """Queue a set of files for ingest."""
    conn = get_redis()
    db = get_dataset(dataset, OP_INGEST)
    _ingest_path(db, conn, dataset, path, languages=languages)
Esempio n. 7
0
File: core.py Progetto: pudo/aleph
def get_cache():
    if not hasattr(settings, '_cache') or settings._cache is None:
        settings._cache = Cache(get_redis(), prefix=settings.APP_NAME)
    return settings._cache
Esempio n. 8
0
File: cli.py Progetto: x0rzkov/aleph
def ingest(path, dataset, languages=None):
    """Queue a set of files for ingest."""
    conn = get_redis()
    _ingest_path(conn, dataset, path, languages=languages)
Esempio n. 9
0
 def setUp(self):
     self.conn = get_redis()
     self.dataset = "my-dataset"
     self.operation = "OP_FOO"
     self.job = Job.create(self.conn, self.dataset)