Esempio n. 1
0
def main(task, num_docs):
    config(task)
    workspace = os.environ['WORKDIR']
    image_src = '/tmp/jina/celeb/lfw/**/*.jpg'
    if task == 'index':
        if os.path.exists(workspace):
            print(
                f'\n +---------------------------------------------------------------------------------+ \
                    \n |                                   ������                                        | \
                    \n | The directory {workspace} already exists. Please remove it before indexing again. | \
                    \n |                                   ������                                        | \
                    \n +---------------------------------------------------------------------------------+'
            )
        f = Flow().load_config('flow-index.yml')
        with f:
            f.index_files(image_src,
                          batch_size=8,
                          read_mode='rb',
                          size=num_docs)
    elif task == 'query':
        f = Flow().load_config('flow-query.yml')
        f.use_rest_gateway()
        with f:
            f.block()
    elif task == 'dryrun':
        f = Flow.load_config('flow-query.yml')
        with f:
            pass
    else:
        raise NotImplementedError(
            f'unknown task: {task}. A valid task is either `index` or `query` or `dryrun`.'
        )
Esempio n. 2
0
def main(task, num_docs):
    config()
    data_path = os.path.join(os.environ['TMP_DATA_DIR'], 'jpg')
    if task == 'index':
        workspace = os.environ['WORKDIR']
        if os.path.exists(workspace):
            print(f'\n +---------------------------------------------------------------------------------+ \
                    \n |                                   ������                                        | \
                    \n | The directory {workspace} already exists. Please remove it before indexing again. | \
                    \n |                                   ������                                        | \
                    \n +---------------------------------------------------------------------------------+')
        f = Flow().load_config('flow-index.yml')
        with f:
            f.index_files(f'{data_path}/*.jpg', size=num_docs, read_mode='rb', batch_size=2)
    elif task == 'query':
        f = Flow().load_config('flow-query.yml')
        f.use_rest_gateway()
        with f:
            f.block()
    elif task == 'dryrun':
        f = Flow.load_config('flow-query.yml')
        with f:
            pass
    else:
        raise NotImplementedError(
            f'unknown task: {task}. A valid task is either `index` or `query` or `dryrun`.')
Esempio n. 3
0
def index_documents():
    f = Flow().load_config(INDEX_FLOW_FILE_PATH)

    with f:
        f.index_files(os.environ['JINA_DATA'],
                      batch_size=1,
                      read_mode='rb',
                      size=NUM_DOCS)
Esempio n. 4
0
def index_documents():
    f = Flow().load_config('flows/index.yml')

    with f:
        f.index_files(os.environ['JINA_DATA'],
                      request_size=64,
                      read_mode='rb',
                      size=NUM_DOCS)
def index_documents():
    f = Flow().load_config('flow-index.yml')

    with f:
        f.index_files(GIF_BLOB,
                      request_size=REQ_SIZE,
                      read_mode='rb',
                      skip_dry_run=True)
Esempio n. 6
0
def main(task, num_docs):
    config()
    data_path = os.path.join(os.environ['JINA_DATA_DIR'], 'jpg/*.jpg')
    if task == 'index':
        f = Flow().load_config('flow-index.yml')
        with f:
            f.index_files(data_path, batch_size=64, read_mode='rb', size=num_docs)
    elif task == 'query':
        f = Flow().load_config('flow-query.yml')
        with f:
            f.block()
    else:
        raise NotImplementedError(f'unknown task: {task}. A valid task is either `index` or `query`.')
Esempio n. 7
0
def main(task, num_docs):
    os.environ['TMP_WORKSPACE'] = get_random_ws(os.environ['TMP_DATA_DIR'])
    data_path = os.path.join(os.environ['TMP_DATA_DIR'], 'jpg')
    if task == 'index':
        f = Flow().load_config('flow-index.yml')
        with f:
            f.index_files(f'{data_path}/*.jpg', size=num_docs, read_mode='rb', batch_size=2)
    elif task == 'query':
        f = Flow().load_config('flow-query.yml')
        f.use_rest_gateway()
        with f:
            f.block()
    else:
        raise NotImplementedError(f'unknown task: {task}. A valid task is either `index` or `query`.')
Esempio n. 8
0
def main(task, num_docs):
    config()
    data_path = os.path.join(os.environ['TMP_DATA_DIR'], 'audio')
    if task == 'index':
        f = Flow().load_config('./flows/flow-index.yml')
        with f:
            f.index_files(f'{data_path}/*.wav', size=num_docs, batch_size=2)
    elif task == 'query':
        f = Flow().load_config('./flows/flow-query.yml')
        f.use_rest_gateway()
        with f:
            f.block()
    elif task == 'dryrun':
        f = Flow.load_config('./flows/flow-query.yml')
        with f:
            pass
    else:
        raise NotImplementedError(
            f'unknown task: {task}. A valid task is either `index` or `query` or `dryrun`.'
        )
Esempio n. 9
0
def main(task, num_docs):
    config(task)
    image_src = '/tmp/jina/celeb/lfw/**/*.jpg'
    if task == 'index':
        f = Flow().load_config('flow-index.yml')
        with f:
            f.index_files(image_src,
                          batch_size=8,
                          read_mode='rb',
                          size=num_docs)
    elif task == 'query':
        f = Flow().load_config('flow-query.yml')
        f.use_rest_gateway()
        with f:
            f.block()
    elif task == 'dryrun':
        f = Flow.load_config('flow-query.yml')
        with f:
            pass
    else:
        raise NotImplementedError(
            f'unknown task: {task}. A valid task is either `index` or `query` or `dryrun`.'
        )
Esempio n. 10
0
@click.command()
@click.option('--task', '-t')
@click.option('--num_docs', '-n', default=50)
def main(task, num_docs):
    config()
    data_path = os.path.join(os.environ['WORKDIR'], 'audio')
        if os.path.exists(data_path):
            print(f'\n +---------------------------------------------------------------------------------+ \
                    \n |                                   ������                                        | \
                    \n | The directory {data_path} already exists. Please remove it before indexing again. | \
                    \n |                                   ������                                        | \
                    \n +---------------------------------------------------------------------------------+')
    if task == 'index':
        f = Flow().load_config('./flows/flow-index.yml')
        with f:
            f.index_files(f'{data_path}/*.wav', size=num_docs, batch_size=2)
    elif task == 'query':
        f = Flow().load_config('./flows/flow-query.yml')
        f.use_rest_gateway()
        with f:
            f.block()
    elif task == 'dryrun':
        f = Flow.load_config('./flows/flow-query.yml')
        with f:
            pass
    else:
        raise NotImplementedError(
            f'unknown task: {task}. A valid task is either `index` or `query` or `dryrun`.')


if __name__ == '__main__':
Esempio n. 11
0
def index():
    f = Flow().add(yaml_path='indexer.yml')
    with f:
        f.index_files(sys.argv[2])
Esempio n. 12
0
def index_documents():
    f = Flow().load_config('flows/index.yml')
    with f:
        f.index_files('tests/data/*.wav', batch_size=2, size=NUM_DOCS)
Esempio n. 13
0
def index():
    f = Flow().add(uses='indexer.yml')
    with f:
        f.index_files(sys.argv[2])