|
1
|
|
|
import collections |
|
2
|
|
|
import os |
|
3
|
|
|
import sys |
|
4
|
|
|
|
|
5
|
|
|
import click |
|
6
|
|
|
import click_log |
|
7
|
|
|
from flask import current_app |
|
8
|
|
|
|
|
9
|
|
|
import annif |
|
10
|
|
|
from annif.exception import ConfigurationException |
|
11
|
|
|
from annif.project import Access |
|
12
|
|
|
from annif.suggestion import SuggestionFilter |
|
13
|
|
|
|
|
14
|
|
|
logger = annif.logger |
|
15
|
|
|
|
|
16
|
|
|
|
|
17
|
|
|
def _set_project_config_file_path(ctx, param, value): |
|
18
|
|
|
"""Override the default path or the path given in env by CLI option""" |
|
19
|
|
|
with ctx.obj.load_app().app_context(): |
|
20
|
|
|
if value: |
|
21
|
|
|
current_app.config["PROJECTS_CONFIG_PATH"] = value |
|
22
|
|
|
|
|
23
|
|
|
|
|
24
|
|
|
def common_options(f): |
|
25
|
|
|
"""Decorator to add common options for all CLI commands""" |
|
26
|
|
|
f = click.option( |
|
27
|
|
|
"-p", |
|
28
|
|
|
"--projects", |
|
29
|
|
|
help="Set path to project configuration file or directory", |
|
30
|
|
|
type=click.Path(dir_okay=True, exists=True), |
|
31
|
|
|
callback=_set_project_config_file_path, |
|
32
|
|
|
expose_value=False, |
|
33
|
|
|
is_eager=True, |
|
34
|
|
|
)(f) |
|
35
|
|
|
return click_log.simple_verbosity_option(logger)(f) |
|
36
|
|
|
|
|
37
|
|
|
|
|
38
|
|
|
def backend_param_option(f): |
|
39
|
|
|
"""Decorator to add an option for CLI commands to override BE parameters""" |
|
40
|
|
|
return click.option( |
|
41
|
|
|
"--backend-param", |
|
42
|
|
|
"-b", |
|
43
|
|
|
multiple=True, |
|
44
|
|
|
help="Override backend parameter of the config file. " |
|
45
|
|
|
+ "Syntax: `-b <backend>.<parameter>=<value>`.", |
|
46
|
|
|
)(f) |
|
47
|
|
|
|
|
48
|
|
|
|
|
49
|
|
|
def docs_limit_option(f): |
|
50
|
|
|
"""Decorator to add an option for CLI commands to limit the number of documents to |
|
51
|
|
|
use""" |
|
52
|
|
|
return click.option( |
|
53
|
|
|
"--docs-limit", |
|
54
|
|
|
"-d", |
|
55
|
|
|
default=None, |
|
56
|
|
|
type=click.IntRange(0, None), |
|
57
|
|
|
help="Maximum number of documents to use", |
|
58
|
|
|
)(f) |
|
59
|
|
|
|
|
60
|
|
|
|
|
61
|
|
|
def get_project(project_id): |
|
62
|
|
|
""" |
|
63
|
|
|
Helper function to get a project by ID and bail out if it doesn't exist""" |
|
64
|
|
|
try: |
|
65
|
|
|
return annif.registry.get_project(project_id, min_access=Access.private) |
|
66
|
|
|
except ValueError: |
|
67
|
|
|
click.echo("No projects found with id '{0}'.".format(project_id), err=True) |
|
68
|
|
|
sys.exit(1) |
|
69
|
|
|
|
|
70
|
|
|
|
|
71
|
|
|
def get_vocab(vocab_id): |
|
72
|
|
|
""" |
|
73
|
|
|
Helper function to get a vocabulary by ID and bail out if it doesn't |
|
74
|
|
|
exist""" |
|
75
|
|
|
try: |
|
76
|
|
|
return annif.registry.get_vocab(vocab_id, min_access=Access.private) |
|
77
|
|
|
except ValueError: |
|
78
|
|
|
click.echo(f"No vocabularies found with the id '{vocab_id}'.", err=True) |
|
79
|
|
|
sys.exit(1) |
|
80
|
|
|
|
|
81
|
|
|
|
|
82
|
|
|
def open_documents(paths, subject_index, vocab_lang, docs_limit): |
|
83
|
|
|
"""Helper function to open a document corpus from a list of pathnames, |
|
84
|
|
|
each of which is either a TSV file or a directory of TXT files. For |
|
85
|
|
|
directories with subjects in TSV files, the given vocabulary language |
|
86
|
|
|
will be used to convert subject labels into URIs. The corpus will be |
|
87
|
|
|
returned as an instance of DocumentCorpus or LimitingDocumentCorpus.""" |
|
88
|
|
|
|
|
89
|
|
|
def open_doc_path(path, subject_index): |
|
90
|
|
|
"""open a single path and return it as a DocumentCorpus""" |
|
91
|
|
|
if os.path.isdir(path): |
|
92
|
|
|
return annif.corpus.DocumentDirectory( |
|
93
|
|
|
path, subject_index, vocab_lang, require_subjects=True |
|
94
|
|
|
) |
|
95
|
|
|
return annif.corpus.DocumentFile(path, subject_index) |
|
96
|
|
|
|
|
97
|
|
|
if len(paths) == 0: |
|
98
|
|
|
logger.warning("Reading empty file") |
|
99
|
|
|
docs = open_doc_path(os.path.devnull, subject_index) |
|
100
|
|
|
elif len(paths) == 1: |
|
101
|
|
|
docs = open_doc_path(paths[0], subject_index) |
|
102
|
|
|
else: |
|
103
|
|
|
corpora = [open_doc_path(path, subject_index) for path in paths] |
|
104
|
|
|
docs = annif.corpus.CombinedCorpus(corpora) |
|
105
|
|
|
if docs_limit is not None: |
|
106
|
|
|
docs = annif.corpus.LimitingDocumentCorpus(docs, docs_limit) |
|
107
|
|
|
return docs |
|
108
|
|
|
|
|
109
|
|
|
|
|
110
|
|
|
def open_text_documents(paths, docs_limit): |
|
111
|
|
|
""" |
|
112
|
|
|
Helper function to read text documents from the given file paths. Returns a |
|
113
|
|
|
DocumentList object with Documents having no subjects. If a path is "-", the |
|
114
|
|
|
document text is read from standard input. The maximum number of documents to read |
|
115
|
|
|
is set by docs_limit parameter. |
|
116
|
|
|
""" |
|
117
|
|
|
|
|
118
|
|
|
def _docs(paths): |
|
119
|
|
|
for path in paths: |
|
120
|
|
|
if path == "-": |
|
121
|
|
|
doc = annif.corpus.Document(text=sys.stdin.read(), subject_set=None) |
|
122
|
|
|
else: |
|
123
|
|
|
with open(path, errors="replace", encoding="utf-8-sig") as docfile: |
|
124
|
|
|
doc = annif.corpus.Document(text=docfile.read(), subject_set=None) |
|
125
|
|
|
yield doc |
|
126
|
|
|
|
|
127
|
|
|
return annif.corpus.DocumentList(_docs(paths[:docs_limit])) |
|
128
|
|
|
|
|
129
|
|
|
|
|
130
|
|
|
def show_hits(hits, project, lang, file=None): |
|
131
|
|
|
""" |
|
132
|
|
|
Print subject suggestions to the console or a file. The suggestions are displayed as |
|
133
|
|
|
a table, with one row per hit. Each row contains the URI, label, possible notation, |
|
134
|
|
|
and score of the suggestion. The label is given in the specified language. |
|
135
|
|
|
""" |
|
136
|
|
|
for hit in hits.as_list(): |
|
137
|
|
|
subj = project.subjects[hit.subject_id] |
|
138
|
|
|
line = "<{}>\t{}\t{}".format( |
|
139
|
|
|
subj.uri, |
|
140
|
|
|
"\t".join(filter(None, (subj.labels[lang], subj.notation))), |
|
141
|
|
|
hit.score, |
|
142
|
|
|
) |
|
143
|
|
|
click.echo(line, file=file) |
|
144
|
|
|
|
|
145
|
|
|
|
|
146
|
|
|
def parse_backend_params(backend_param, project): |
|
147
|
|
|
"""Parse a list of backend parameters given with the --backend-param |
|
148
|
|
|
option into a nested dict structure""" |
|
149
|
|
|
backend_params = collections.defaultdict(dict) |
|
150
|
|
|
for beparam in backend_param: |
|
151
|
|
|
backend, param = beparam.split(".", 1) |
|
152
|
|
|
key, val = param.split("=", 1) |
|
153
|
|
|
_validate_backend_params(backend, beparam, project) |
|
154
|
|
|
backend_params[backend][key] = val |
|
155
|
|
|
return backend_params |
|
156
|
|
|
|
|
157
|
|
|
|
|
158
|
|
|
def _validate_backend_params(backend, beparam, project): |
|
159
|
|
|
if backend != project.config["backend"]: |
|
160
|
|
|
raise ConfigurationException( |
|
161
|
|
|
'The backend {} in CLI option "-b {}" not matching the project' |
|
162
|
|
|
" backend {}.".format(backend, beparam, project.config["backend"]) |
|
163
|
|
|
) |
|
164
|
|
|
|
|
165
|
|
|
|
|
166
|
|
|
def generate_filter_batches(subjects, filter_batch_max_limit): |
|
167
|
|
|
import annif.eval |
|
168
|
|
|
|
|
169
|
|
|
filter_batches = {} |
|
170
|
|
|
for limit in range(1, filter_batch_max_limit + 1): |
|
171
|
|
|
for threshold in [i * 0.05 for i in range(20)]: |
|
172
|
|
|
hit_filter = SuggestionFilter(subjects, limit, threshold) |
|
173
|
|
|
batch = annif.eval.EvaluationBatch(subjects) |
|
174
|
|
|
filter_batches[(limit, threshold)] = (hit_filter, batch) |
|
175
|
|
|
return filter_batches |
|
176
|
|
|
|
NatLibFi /
Annif
