file_meta.py 文件源码

def __extract_metadata(self, doc, payload):
    filename = os.path.basename(doc.path)
    headers = {
      'Accept': 'application/json',
      'Content-Disposition': 'attachment; filename=%s' % url_quote(filename)
    }

    if doc.meta['Content-Type']:
      headers['Content-Type'] = doc.meta['Content-Type']

    tika_url = self.config.get(helper.TIKA_META)
    connection = self.config[helper.INJECTOR].get_http_connection(tika_url)
    payload.seek(0)
    connection.request('PUT', '/meta', payload.read(), headers)
    payload.seek(0)

    response = connection.getresponse()

    try:
      if response.status >= 400:
        logging.error('tika error %d (%s): %s', response.status,
                      response.reason, doc.path)
        return {}
      response_data = response.read()
    finally:
      response.close()

    try:
      result = json.loads(response_data.decode('utf-8'))
    except (ValueError, UnicodeDecodeError):
      logging.error('invalid response from tika for %s', doc.path)
      result = {}

    return result