|
1
|
|
|
"""Registry that keeps track of Annif projects""" |
|
2
|
|
|
|
|
3
|
|
|
import collections |
|
4
|
|
|
import re |
|
5
|
|
|
|
|
6
|
|
|
from flask import current_app |
|
7
|
|
|
|
|
8
|
|
|
import annif |
|
9
|
|
|
from annif.config import parse_config |
|
10
|
|
|
from annif.exception import ConfigurationException |
|
11
|
|
|
from annif.project import Access, AnnifProject |
|
12
|
|
|
from annif.util import parse_args |
|
13
|
|
|
from annif.vocab import AnnifVocabulary |
|
14
|
|
|
|
|
15
|
|
|
logger = annif.logger |
|
16
|
|
|
|
|
17
|
|
|
|
|
18
|
|
|
class AnnifRegistry: |
|
19
|
|
|
"""Class that keeps track of the Annif projects and vocabularies""" |
|
20
|
|
|
|
|
21
|
|
|
# Note: The individual projects and vocabularies are stored in shared |
|
22
|
|
|
# static variables, keyed by the "registry ID" which is unique to the |
|
23
|
|
|
# registry instance. This is done to make it possible to serialize |
|
24
|
|
|
# AnnifRegistry instances without including the potentially huge objects |
|
25
|
|
|
# (which contain backends with large models, vocabularies with lots of |
|
26
|
|
|
# concepts etc). Serialized AnnifRegistry instances can then be passed |
|
27
|
|
|
# between processes when using the multiprocessing module. |
|
28
|
|
|
_projects = {} |
|
29
|
|
|
_vocabs = {} |
|
30
|
|
|
|
|
31
|
|
|
def __init__(self, projects_config_path, datadir, init_projects): |
|
32
|
|
|
self._rid = id(self) |
|
33
|
|
|
self._projects_config_path = projects_config_path |
|
34
|
|
|
self._datadir = datadir |
|
35
|
|
|
self._init_vars() |
|
36
|
|
|
if init_projects: |
|
37
|
|
|
for project in self._projects[self._rid].values(): |
|
38
|
|
|
project.initialize() |
|
39
|
|
|
|
|
40
|
|
|
def _init_vars(self): |
|
41
|
|
|
# initialize the static variables, if necessary |
|
42
|
|
|
if self._rid not in self._projects: |
|
43
|
|
|
self._projects[self._rid] = self._create_projects() |
|
44
|
|
|
self._vocabs[self._rid] = {} |
|
45
|
|
|
|
|
46
|
|
|
def _create_projects(self): |
|
47
|
|
|
# parse the configuration |
|
48
|
|
|
config = parse_config(self._projects_config_path) |
|
49
|
|
|
|
|
50
|
|
|
# handle the case where the config file doesn't exist |
|
51
|
|
|
if config is None: |
|
52
|
|
|
return {} |
|
53
|
|
|
|
|
54
|
|
|
# create AnnifProject objects from the configuration file |
|
55
|
|
|
projects = collections.OrderedDict() |
|
56
|
|
|
for project_id in config.project_ids: |
|
57
|
|
|
projects[project_id] = AnnifProject( |
|
58
|
|
|
project_id, config[project_id], self._datadir, self |
|
59
|
|
|
) |
|
60
|
|
|
return projects |
|
61
|
|
|
|
|
62
|
|
|
def get_projects(self, min_access=Access.private): |
|
63
|
|
|
"""Return the available projects as a dict of project_id -> |
|
64
|
|
|
AnnifProject. The min_access parameter may be used to set the minimum |
|
65
|
|
|
access level required for the returned projects.""" |
|
66
|
|
|
|
|
67
|
|
|
self._init_vars() |
|
68
|
|
|
return { |
|
69
|
|
|
project_id: project |
|
70
|
|
|
for project_id, project in self._projects[self._rid].items() |
|
71
|
|
|
if project.access >= min_access |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
def get_project(self, project_id, min_access=Access.private): |
|
75
|
|
|
"""return the definition of a single Project by project_id""" |
|
76
|
|
|
|
|
77
|
|
|
projects = self.get_projects(min_access) |
|
78
|
|
|
try: |
|
79
|
|
|
return projects[project_id] |
|
80
|
|
|
except KeyError: |
|
81
|
|
|
raise ValueError("No such project {}".format(project_id)) |
|
82
|
|
|
|
|
83
|
|
|
def get_vocab(self, vocab_spec, default_language): |
|
84
|
|
|
"""Return an (AnnifVocabulary, language) pair corresponding to the |
|
85
|
|
|
vocab_spec. If no language information is specified, use the given |
|
86
|
|
|
default language.""" |
|
87
|
|
|
|
|
88
|
|
|
match = re.match(r"([\w-]+)(\((.*)\))?$", vocab_spec) |
|
89
|
|
|
if match is None: |
|
90
|
|
|
raise ValueError(f"Invalid vocabulary specification: {vocab_spec}") |
|
91
|
|
|
vocab_id = match.group(1) |
|
92
|
|
|
posargs, kwargs = parse_args(match.group(3)) |
|
93
|
|
|
language = posargs[0] if posargs else default_language |
|
94
|
|
|
vocab_key = (vocab_id, language) |
|
95
|
|
|
|
|
96
|
|
|
self._init_vars() |
|
97
|
|
|
if vocab_key not in self._vocabs[self._rid]: |
|
98
|
|
|
self._vocabs[self._rid][vocab_key] = AnnifVocabulary( |
|
99
|
|
|
vocab_id, self._datadir |
|
100
|
|
|
) |
|
101
|
|
|
return self._vocabs[self._rid][vocab_key], language |
|
102
|
|
|
|
|
103
|
|
|
|
|
104
|
|
|
def initialize_projects(app): |
|
105
|
|
|
projects_config_path = app.config["PROJECTS_CONFIG_PATH"] |
|
106
|
|
|
datadir = app.config["DATADIR"] |
|
107
|
|
|
init_projects = app.config["INITIALIZE_PROJECTS"] |
|
108
|
|
|
app.annif_registry = AnnifRegistry(projects_config_path, datadir, init_projects) |
|
109
|
|
|
|
|
110
|
|
|
|
|
111
|
|
|
def get_projects(min_access=Access.private): |
|
112
|
|
|
"""Return the available projects as a dict of project_id -> |
|
113
|
|
|
AnnifProject. The min_access parameter may be used to set the minimum |
|
114
|
|
|
access level required for the returned projects.""" |
|
115
|
|
|
if not hasattr(current_app, "annif_registry"): |
|
116
|
|
|
initialize_projects(current_app) |
|
117
|
|
|
|
|
118
|
|
|
return current_app.annif_registry.get_projects(min_access) |
|
119
|
|
|
|
|
120
|
|
|
|
|
121
|
|
|
def get_project(project_id, min_access=Access.private): |
|
122
|
|
|
"""return the definition of a single Project by project_id""" |
|
123
|
|
|
|
|
124
|
|
|
projects = get_projects(min_access) |
|
125
|
|
|
try: |
|
126
|
|
|
return projects[project_id] |
|
127
|
|
|
except KeyError: |
|
128
|
|
|
raise ValueError(f"No such project '{project_id}'") |
|
129
|
|
|
|
|
130
|
|
|
|
|
131
|
|
|
def get_vocabs(min_access=Access.private): |
|
132
|
|
|
"""Return the available vocabularies as a dict of vocab_id -> |
|
133
|
|
|
AnnifVocabulary. The min_access parameter may be used to set the minimum |
|
134
|
|
|
access level required for the returned vocabularies.""" |
|
135
|
|
|
|
|
136
|
|
|
vocabs = {} |
|
137
|
|
|
for proj in get_projects(min_access).values(): |
|
138
|
|
|
try: |
|
139
|
|
|
vocabs[proj.vocab.vocab_id] = proj.vocab |
|
140
|
|
|
except ConfigurationException: |
|
141
|
|
|
pass |
|
142
|
|
|
|
|
143
|
|
|
return vocabs |
|
144
|
|
|
|
|
145
|
|
|
|
|
146
|
|
|
def get_vocab(vocab_id, min_access=Access.private): |
|
147
|
|
|
"""return a single AnnifVocabulary by vocabulary id""" |
|
148
|
|
|
|
|
149
|
|
|
vocabs = get_vocabs(min_access) |
|
150
|
|
|
try: |
|
151
|
|
|
return vocabs[vocab_id] |
|
152
|
|
|
except KeyError: |
|
153
|
|
|
raise ValueError(f"No such vocabulary '{vocab_id}'") |
|
154
|
|
|
|
NatLibFi /
Annif
