|
1
|
|
|
import json |
|
2
|
|
|
from shlex import split as shlex_split, quote |
|
3
|
|
|
# only in 3.8+ :( |
|
4
|
|
|
# from shlex import join as shlex_join |
|
5
|
|
|
from subprocess import run, PIPE |
|
6
|
|
|
|
|
7
|
|
|
from ocrd_utils import getLogger, parse_json_string_or_file, set_json_key_value_overrides |
|
8
|
|
|
|
|
9
|
|
|
LOG = getLogger('ocrd.wf.step') |
|
10
|
|
|
|
|
11
|
|
|
class OcrdWfStep(): |
|
12
|
|
|
|
|
13
|
|
|
@classmethod |
|
14
|
|
|
def parse(cls, argstr): |
|
15
|
|
|
tokens = shlex_split(argstr) |
|
16
|
|
|
executable = tokens.pop(0) |
|
17
|
|
|
if not executable.startswith('ocrd-'): |
|
18
|
|
|
executable = 'ocrd-%s' % executable |
|
19
|
|
|
input_file_grps = [] |
|
20
|
|
|
output_file_grps = [] |
|
21
|
|
|
parameters = {} |
|
22
|
|
|
while tokens: |
|
23
|
|
|
if tokens[0] == '-I': |
|
24
|
|
|
for grp in tokens[1].split(','): |
|
25
|
|
|
input_file_grps.append(grp) |
|
26
|
|
|
tokens = tokens[2:] |
|
27
|
|
|
elif tokens[0] == '-O': |
|
28
|
|
|
for grp in tokens[1].split(','): |
|
29
|
|
|
output_file_grps.append(grp) |
|
30
|
|
|
tokens = tokens[2:] |
|
31
|
|
|
elif tokens[0] == '-p': |
|
32
|
|
|
parameters = {**parameters, **parse_json_string_or_file(tokens[1])} |
|
33
|
|
|
tokens = tokens[2:] |
|
34
|
|
|
elif tokens[0] == '-P': |
|
35
|
|
|
set_json_key_value_overrides(parameters, tokens[1:3]) |
|
36
|
|
|
tokens = tokens[3:] |
|
37
|
|
|
else: |
|
38
|
|
|
raise Exception("Failed parsing task description '%s' with tokens remaining: '%s'" % (argstr, tokens)) |
|
39
|
|
|
return OcrdWfStep(executable, input_file_grps, output_file_grps, parameters) |
|
40
|
|
|
|
|
41
|
|
|
def __init__(self, executable, input_file_grps, output_file_grps, parameters): |
|
42
|
|
|
self.executable = executable |
|
43
|
|
|
self.input_file_grps = input_file_grps |
|
44
|
|
|
self.output_file_grps = output_file_grps |
|
45
|
|
|
self.parameters = parameters |
|
46
|
|
|
self._ocrd_tool_json = None |
|
47
|
|
|
|
|
48
|
|
|
@property |
|
49
|
|
|
def ocrd_tool_json(self): |
|
50
|
|
|
if self._ocrd_tool_json: |
|
51
|
|
|
return self._ocrd_tool_json |
|
52
|
|
|
result = run([self.executable, '--dump-json'], stdout=PIPE, check=True, universal_newlines=True) |
|
53
|
|
|
self._ocrd_tool_json = json.loads(result.stdout) |
|
54
|
|
|
return self._ocrd_tool_json |
|
55
|
|
|
|
|
56
|
|
|
def __str__(self): |
|
57
|
|
|
ret = [self.executable] |
|
58
|
|
|
if self.input_file_grps: |
|
59
|
|
|
ret += ['-I', ','.join(self.input_file_grps)] |
|
60
|
|
|
if self.output_file_grps: |
|
61
|
|
|
ret += ['-O', ','.join(self.output_file_grps)] |
|
62
|
|
|
for k in self.parameters: |
|
63
|
|
|
ret += ['-P', k, json.dumps(self.parameters[k])] |
|
64
|
|
|
return ' '.join([quote(s) for s in ret]) |
|
65
|
|
|
|
|
66
|
|
|
|
OCR-D /
core
