def parse(self,
option: str,
url_or_path: str,
server_endpoint: str = None,
verbose: int = 0,
tika_server_jar: str = None,
response_mime_type: str = 'application/json',
services: dict = None,
raw_response: bool = False,
extra_headers: Dict[str, str] = None) -> Dict:
"""
The method is called from parse_file_on_server to parse the file
calling Tika as a server.
:param option: command line options to send to Tika's server
:param url_or_path: local path (or URL) to the file being parsed
:param server_endpoint: Tika server's URL
:param verbose: make Tika produse verbose log
:param tika_server_jar: path to Tika's JAR file
:param response_mime_type: response format (application/json) for plain text + metadata in JSON format
:param services:
:param raw_response: get raw response from Tika (text + metadata + warnings), False by default
:param extra_headers: extra request header
:return: dictionary with "content" (text) and "metadata" (another dictionary) keys
"""
services = services if services else \
{'meta': '/meta', 'text': '/tika', 'all': '/rmeta/text'}
tika_server_jar = tika_server_jar if tika_server_jar else self.tika_jar_path
server_endpoint = server_endpoint if server_endpoint else self.server_endpoint
path, file_type = getRemoteFile(url_or_path, self.tika_files_path)
service = services.get(option, services['all'])
if service == '/tika':
response_mime_type = 'text/plain'
content_path = self.make_content_disposition_header(path)
headers = {
'Accept': response_mime_type,
'Content-Disposition': content_path
}
if extra_headers:
headers = {**headers, **extra_headers}
status, response = callServer('put',
server_endpoint,
service,
open(path, 'rb'),
headers,
verbose,
tika_server_jar,
rawResponse=raw_response)
if file_type == 'remote':
os.unlink(path)
return _parse((status, response))
def parse(self,
option: str,
url_or_path: str,
server_endpoint: str = None,
verbose: int = 0,
tika_server_jar: str = None,
response_mime_type: str = 'application/json',
services: dict = None,
raw_response: bool = False,
extra_headers: Dict[str, str] = None) -> Dict:
services = services if services else \
{'meta': '/meta', 'text': '/tika', 'all': '/rmeta/text'}
tika_server_jar = tika_server_jar if tika_server_jar else self.tika_jar_path
server_endpoint = server_endpoint if server_endpoint else self.server_endpoint
path, file_type = getRemoteFile(url_or_path, self.tika_files_path)
service = services.get(option, services['all'])
if service == '/tika':
response_mime_type = 'text/plain'
content_path = self.make_content_disposition_header(path)
headers = {
'Accept': response_mime_type,
'Content-Disposition': content_path
}
if extra_headers:
headers = {**headers, **extra_headers}
status, response = callServer('put',
server_endpoint,
service,
open(path, 'rb'),
headers,
verbose,
tika_server_jar,
rawResponse=raw_response)
if file_type == 'remote':
os.unlink(path)
return _parse((status, response))