|
1
|
|
|
import click |
|
2
|
|
|
from http.server import BaseHTTPRequestHandler, HTTPServer |
|
3
|
|
|
from typing import Optional |
|
4
|
|
|
|
|
5
|
|
|
from ocrd.decorators import parameter_option |
|
6
|
|
|
from ocrd_network import Client |
|
7
|
|
|
from ocrd_utils import DEFAULT_METS_BASENAME |
|
8
|
|
|
|
|
9
|
|
|
|
|
10
|
|
|
STOP_WAITING_CALLBACK = False |
|
11
|
|
|
|
|
12
|
|
|
|
|
13
|
|
|
class ClientCallbackHandler(BaseHTTPRequestHandler): |
|
14
|
|
|
""" |
|
15
|
|
|
A simple callback handler for the network client to be invoked when the Processing Worker |
|
16
|
|
|
sends requests to the `callback_url` set in the processing request submitted to the Processing Server. |
|
17
|
|
|
""" |
|
18
|
|
|
|
|
19
|
|
|
def do_POST(self): |
|
20
|
|
|
self.send_response(200) |
|
21
|
|
|
self.send_header("Content-Type", "text/plain") |
|
22
|
|
|
self.end_headers() |
|
23
|
|
|
self.wfile.write("finished".encode("utf-8")) |
|
24
|
|
|
len = int(self.headers.get("Content-Length", 0)) |
|
25
|
|
|
data = self.rfile.read(len).decode("utf-8") |
|
26
|
|
|
# TODO: how should the callback-content be handled/printed |
|
27
|
|
|
print(f"Processor finished: {data}") |
|
28
|
|
|
global STOP_WAITING_CALLBACK |
|
29
|
|
|
STOP_WAITING_CALLBACK = True |
|
30
|
|
|
|
|
31
|
|
|
|
|
32
|
|
|
class ClientCallbackServer(HTTPServer): |
|
33
|
|
|
""" |
|
34
|
|
|
A simple http-server that listens for callbacks from the Processing Server/Worker. |
|
35
|
|
|
""" |
|
36
|
|
|
def __init__(self): |
|
37
|
|
|
super().__init__(server_address=("0.0.0.0", 0), RequestHandlerClass=ClientCallbackHandler) |
|
38
|
|
|
self.callback_url = f"http://{self.server_address[0]}:{self.server_address[1]}" |
|
39
|
|
|
|
|
40
|
|
|
|
|
41
|
|
|
@click.group('client') |
|
42
|
|
|
def client_cli(): |
|
43
|
|
|
""" |
|
44
|
|
|
A client for interacting with the network modules. |
|
45
|
|
|
The client CLI mimics the WebAPI endpoints |
|
46
|
|
|
""" |
|
47
|
|
|
pass |
|
48
|
|
|
|
|
49
|
|
|
|
|
50
|
|
|
@client_cli.group('discovery') |
|
51
|
|
|
def discovery_cli(): |
|
52
|
|
|
""" |
|
53
|
|
|
The discovery endpoint of the WebAPI |
|
54
|
|
|
""" |
|
55
|
|
|
pass |
|
56
|
|
|
|
|
57
|
|
|
|
|
58
|
|
|
@client_cli.group('processing') |
|
59
|
|
|
def processing_cli(): |
|
60
|
|
|
""" |
|
61
|
|
|
The processing endpoint of the WebAPI |
|
62
|
|
|
""" |
|
63
|
|
|
pass |
|
64
|
|
|
|
|
65
|
|
|
|
|
66
|
|
|
@processing_cli.command('processor') |
|
67
|
|
|
@click.argument('processor_name', required=True, type=click.STRING) |
|
68
|
|
|
@click.option('--address') |
|
69
|
|
|
@click.option('-m', '--mets', required=True, default=DEFAULT_METS_BASENAME) |
|
70
|
|
|
@click.option('-I', '--input-file-grp', default='OCR-D-INPUT') |
|
71
|
|
|
@click.option('-O', '--output-file-grp', default='OCR-D-OUTPUT') |
|
72
|
|
|
@click.option('-g', '--page-id') |
|
73
|
|
|
@parameter_option |
|
74
|
|
|
@click.option('--result-queue-name') |
|
75
|
|
|
@click.option('--callback-url') |
|
76
|
|
|
@click.option('--agent-type', default='worker') |
|
77
|
|
|
def send_processing_request( |
|
78
|
|
|
address: Optional[str], |
|
79
|
|
|
processor_name: str, |
|
80
|
|
|
mets: str, |
|
81
|
|
|
input_file_grp: str, |
|
82
|
|
|
output_file_grp: Optional[str], |
|
83
|
|
|
page_id: Optional[str], |
|
84
|
|
|
parameter: Optional[dict], |
|
85
|
|
|
result_queue_name: Optional[str], |
|
86
|
|
|
callback_url: Optional[str], |
|
87
|
|
|
# TODO: This is temporally available to toggle |
|
88
|
|
|
# between the ProcessingWorker/ProcessorServer |
|
89
|
|
|
agent_type: Optional[str] |
|
90
|
|
|
): |
|
91
|
|
|
callback_server = ClientCallbackServer() |
|
92
|
|
|
req_params = { |
|
93
|
|
|
"path_to_mets": mets, |
|
94
|
|
|
"description": "OCR-D Network client request", |
|
95
|
|
|
"input_file_grps": input_file_grp.split(','), |
|
96
|
|
|
"parameters": parameter if parameter else {}, |
|
97
|
|
|
"agent_type": agent_type, |
|
98
|
|
|
"callback_url": callback_server.callback_url |
|
99
|
|
|
} |
|
100
|
|
|
if output_file_grp: |
|
101
|
|
|
req_params["output_file_grps"] = output_file_grp.split(',') |
|
102
|
|
|
if page_id: |
|
103
|
|
|
req_params["page_id"] = page_id |
|
104
|
|
|
if result_queue_name: |
|
105
|
|
|
req_params["result_queue_name"] = result_queue_name |
|
106
|
|
|
if callback_url: |
|
107
|
|
|
req_params["callback_url"] = callback_url |
|
108
|
|
|
|
|
109
|
|
|
client = Client(server_addr_processing=address) |
|
110
|
|
|
response = client.send_processing_request(processor_name=processor_name, req_params=req_params) |
|
111
|
|
|
processing_job_id = response.get('job_id', None) |
|
112
|
|
|
print(f"Processing job id: {processing_job_id}") |
|
113
|
|
|
while not STOP_WAITING_CALLBACK: |
|
114
|
|
|
callback_server.handle_request() |
|
115
|
|
|
|
|
116
|
|
|
|
|
117
|
|
|
@client_cli.group('workflow') |
|
118
|
|
|
def workflow_cli(): |
|
119
|
|
|
""" |
|
120
|
|
|
The workflow endpoint of the WebAPI |
|
121
|
|
|
""" |
|
122
|
|
|
pass |
|
123
|
|
|
|
|
124
|
|
|
|
|
125
|
|
|
@client_cli.group('workspace') |
|
126
|
|
|
def workspace_cli(): |
|
127
|
|
|
""" |
|
128
|
|
|
The workspace endpoint of the WebAPI |
|
129
|
|
|
""" |
|
130
|
|
|
pass |
|
131
|
|
|
|
OCR-D /
core
