|
1
|
|
|
import sys |
|
2
|
|
|
|
|
3
|
|
|
import click |
|
4
|
|
|
from json import loads |
|
5
|
|
|
import codecs |
|
6
|
|
|
|
|
7
|
|
|
from ocrd import Resolver, Workspace |
|
8
|
|
|
from ocrd.task_sequence import ProcessorTask, validate_tasks |
|
9
|
|
|
|
|
10
|
|
|
from ocrd_utils import ( |
|
11
|
|
|
parse_json_string_or_file |
|
12
|
|
|
) |
|
13
|
|
|
from ocrd_validators import ( |
|
14
|
|
|
OcrdToolValidator, |
|
15
|
|
|
OcrdZipValidator, |
|
16
|
|
|
PageValidator, |
|
17
|
|
|
ParameterValidator, |
|
18
|
|
|
WorkspaceValidator, |
|
19
|
|
|
) |
|
20
|
|
|
|
|
21
|
|
|
def _inform_of_result(report): |
|
22
|
|
|
if not report.is_valid: |
|
23
|
|
|
print(report.to_xml()) |
|
24
|
|
|
sys.exit(1) |
|
25
|
|
|
|
|
26
|
|
|
|
|
27
|
|
|
@click.group("validate") |
|
28
|
|
|
def validate_cli(): |
|
29
|
|
|
""" |
|
30
|
|
|
All the validation in one CLI |
|
31
|
|
|
""" |
|
32
|
|
|
|
|
33
|
|
|
@validate_cli.command('tool-json') |
|
34
|
|
|
@click.argument('ocrd_tool', required=False, nargs=1) |
|
35
|
|
|
def validate_ocrd_tool(ocrd_tool): |
|
36
|
|
|
''' |
|
37
|
|
|
Validate OCRD_TOOL as an ocrd-tool.json file. |
|
38
|
|
|
''' |
|
39
|
|
|
if not ocrd_tool: |
|
40
|
|
|
ocrd_tool = 'ocrd-tool.json' |
|
41
|
|
|
with codecs.open(ocrd_tool, encoding='utf-8') as f: |
|
42
|
|
|
ocrd_tool = loads(f.read()) |
|
43
|
|
|
_inform_of_result(OcrdToolValidator.validate(ocrd_tool)) |
|
44
|
|
|
|
|
45
|
|
|
@validate_cli.command('parameters') |
|
46
|
|
|
@click.argument('ocrd_tool') |
|
47
|
|
|
@click.argument('executable') |
|
48
|
|
|
@click.argument('param_json') |
|
49
|
|
|
def validate_parameters(ocrd_tool, executable, param_json): |
|
50
|
|
|
''' |
|
51
|
|
|
Validate PARAM_JSON against parameter definition of EXECUTABLE in OCRD_TOOL |
|
52
|
|
|
''' |
|
53
|
|
|
with codecs.open(ocrd_tool, encoding='utf-8') as f: |
|
54
|
|
|
ocrd_tool = loads(f.read()) |
|
55
|
|
|
_inform_of_result(ParameterValidator(ocrd_tool['tools'][executable]).validate(parse_json_string_or_file(param_json))) |
|
56
|
|
|
|
|
57
|
|
|
@validate_cli.command('page') |
|
58
|
|
|
@click.argument('page', required=True, nargs=1) |
|
59
|
|
|
@click.option('--page-textequiv-consistency', help="How strict to check PAGE multi-level textequiv consistency", type=click.Choice(['strict', 'lax', 'fix', 'off']), default='strict') |
|
60
|
|
|
@click.option('--page-textequiv-strategy', help="Strategy to determine the correct textequiv", type=click.Choice(['index1']), default='index1') |
|
61
|
|
|
@click.option('--check-baseline', help="Whether Baseline must be fully within TextLine/Coords", is_flag=True, default=False) |
|
62
|
|
|
@click.option('--check-coords', help="Whether *Region/TextLine/Word/Glyph must each be fully contained within Border/*Region/TextLine/Word, resp.", is_flag=True, default=False) |
|
63
|
|
|
def validate_page(page, **kwargs): |
|
64
|
|
|
''' |
|
65
|
|
|
Validate PAGE against OCR-D conventions |
|
66
|
|
|
''' |
|
67
|
|
|
_inform_of_result(PageValidator.validate(filename=page, **kwargs)) |
|
68
|
|
|
|
|
69
|
|
|
# @validate_cli.command('zip') |
|
70
|
|
|
# @click.argument('src', type=click.Path(dir_okay=True, readable=True, resolve_path=True), required=True) |
|
71
|
|
|
# @click.option('-Z', '--skip-unzip', help="Treat SRC as a directory not a ZIP", is_flag=True, default=False) |
|
72
|
|
|
# @click.option('-B', '--skip-bag', help="Whether to skip all checks of manifests and files", is_flag=True, default=False) |
|
73
|
|
|
# @click.option('-C', '--skip-checksums', help="Whether to omit checksum checks but still check basic BagIt conformance", is_flag=True, default=False) |
|
74
|
|
|
# @click.option('-D', '--skip-delete', help="Whether to skip deleting the unpacked OCRD-ZIP dir after valdiation", is_flag=True, default=False) |
|
75
|
|
|
# @click.option('-j', '--processes', help="Number of parallel processes", type=int, default=1) |
|
76
|
|
|
# def validate(src, **kwargs): |
|
77
|
|
|
# """ |
|
78
|
|
|
# Validate OCRD-ZIP |
|
79
|
|
|
|
|
80
|
|
|
# SRC must exist an be an OCRD-ZIP, either a ZIP file or a directory. |
|
81
|
|
|
# """ |
|
82
|
|
|
# _inform_of_result(OcrdZipValidator(Resolver(), src).validate(**kwargs)) |
|
83
|
|
|
|
|
84
|
|
|
# @validate_cli.command('workspace') |
|
85
|
|
|
# @click.option('-a', '--download', is_flag=True, help="Download all files") |
|
86
|
|
|
# @click.option('-s', '--skip', help="Tests to skip", default=[], multiple=True, type=click.Choice(['imagefilename', 'dimension', 'mets_unique_identifier', 'mets_file_group_names', 'mets_files', 'pixel_density', 'page', 'url'])) |
|
87
|
|
|
# @click.option('--page-textequiv-consistency', '--page-strictness', help="How strict to check PAGE multi-level textequiv consistency", type=click.Choice(['strict', 'lax', 'fix', 'off']), default='strict') |
|
88
|
|
|
# @click.option('--page-coordinate-consistency', help="How fierce to check PAGE multi-level coordinate consistency", type=click.Choice(['poly', 'baseline', 'both', 'off']), default='poly') |
|
89
|
|
|
# @click.argument('mets_url') |
|
90
|
|
|
# def validate_workspace(mets_url, **kwargs): |
|
91
|
|
|
# ''' |
|
92
|
|
|
# Validate a workspace |
|
93
|
|
|
# ''' |
|
94
|
|
|
# _inform_of_result(WorkspaceValidator.validate(Resolver(), mets_url, **kwargs)) |
|
95
|
|
|
|
|
96
|
|
|
@validate_cli.command('tasks') |
|
97
|
|
|
@click.option('--workspace', nargs=1, required=False, help='Workspace these tasks are to be run. If omitted, only validate syntax') |
|
98
|
|
|
@click.argument('tasks', nargs=-1, required=True) |
|
99
|
|
|
def validate_process(tasks, workspace): |
|
100
|
|
|
''' |
|
101
|
|
|
Validate a sequence of tasks passable to 'ocrd process' |
|
102
|
|
|
''' |
|
103
|
|
|
if workspace: |
|
104
|
|
|
_inform_of_result(validate_tasks([ProcessorTask.parse(t) for t in tasks], Workspace(Resolver(), directory=workspace))) |
|
105
|
|
|
else: |
|
106
|
|
|
for t in [ProcessorTask.parse(t) for t in tasks]: |
|
107
|
|
|
_inform_of_result(t.validate()) |
|
108
|
|
|
|
OCR-D /
core
