|
1
|
|
|
"""Registry that keeps track of Annif projects""" |
|
2
|
|
|
|
|
3
|
|
|
import collections |
|
4
|
|
|
import configparser |
|
5
|
|
|
import os.path |
|
6
|
|
|
from flask import current_app |
|
7
|
|
|
import tomli |
|
8
|
|
|
import annif |
|
9
|
|
|
import annif.util |
|
10
|
|
|
from annif.exception import ConfigurationException |
|
11
|
|
|
from annif.project import Access, AnnifProject |
|
12
|
|
|
|
|
13
|
|
|
logger = annif.logger |
|
14
|
|
|
|
|
15
|
|
|
|
|
16
|
|
View Code Duplication |
class AnnifRegistry: |
|
|
|
|
|
|
17
|
|
|
"""Class that keeps track of the Annif projects""" |
|
18
|
|
|
|
|
19
|
|
|
# Note: The individual projects are stored in a shared static variable, |
|
20
|
|
|
# keyed by the "registry ID" which is unique to the registry instance. |
|
21
|
|
|
# This is done to make it possible to serialize AnnifRegistry instances |
|
22
|
|
|
# without including the potentially huge project objects (which contain |
|
23
|
|
|
# backends with large models, vocabularies with lots of concepts etc). |
|
24
|
|
|
# Serialized AnnifRegistry instances can then be passed between |
|
25
|
|
|
# processes when using the multiprocessing module. |
|
26
|
|
|
_projects = {} |
|
27
|
|
|
|
|
28
|
|
|
def __init__(self, projects_file, datadir, init_projects): |
|
29
|
|
|
self._rid = id(self) |
|
30
|
|
|
self._projects[self._rid] = \ |
|
31
|
|
|
self._create_projects(projects_file, datadir) |
|
32
|
|
|
if init_projects: |
|
33
|
|
|
for project in self._projects[self._rid].values(): |
|
34
|
|
|
project.initialize() |
|
35
|
|
|
|
|
36
|
|
|
def _create_projects(self, projects_file, datadir): |
|
37
|
|
|
if projects_file: |
|
38
|
|
|
if not os.path.exists(projects_file): |
|
39
|
|
|
logger.warning( |
|
40
|
|
|
f'Project configuration file "{projects_file}" is ' + |
|
41
|
|
|
'missing. Please provide one. ' + |
|
42
|
|
|
'You can set the path to the project configuration ' + |
|
43
|
|
|
'file using the ANNIF_PROJECTS environment ' + |
|
44
|
|
|
'variable or the command-line option "--projects".') |
|
45
|
|
|
return {} |
|
46
|
|
|
else: |
|
47
|
|
|
if os.path.exists('projects.cfg'): |
|
48
|
|
|
projects_file = 'projects.cfg' |
|
49
|
|
|
elif os.path.exists('projects.toml'): |
|
50
|
|
|
projects_file = 'projects.toml' |
|
51
|
|
|
else: |
|
52
|
|
|
logger.warning( |
|
53
|
|
|
'Could not find project configuration file ' + |
|
54
|
|
|
'"projects.cfg" or "projects.toml". ' + |
|
55
|
|
|
'You can set the path to the project configuration ' + |
|
56
|
|
|
'file using the ANNIF_PROJECTS environment ' + |
|
57
|
|
|
'variable or the command-line option "--projects".') |
|
58
|
|
|
return {} |
|
59
|
|
|
|
|
60
|
|
|
if projects_file.endswith('.toml'): # TOML format |
|
61
|
|
|
with open(projects_file, "rb") as projf: |
|
62
|
|
|
try: |
|
63
|
|
|
config = tomli.load(projf) |
|
64
|
|
|
except tomli.TOMLDecodeError as err: |
|
65
|
|
|
raise ConfigurationException(err) |
|
66
|
|
|
project_ids = config.keys() |
|
67
|
|
|
else: # classic INI style format |
|
68
|
|
|
config = configparser.ConfigParser() |
|
69
|
|
|
config.optionxform = annif.util.identity |
|
70
|
|
|
with open(projects_file, encoding='utf-8-sig') as projf: |
|
71
|
|
|
try: |
|
72
|
|
|
config.read_file(projf) |
|
73
|
|
|
except (configparser.DuplicateOptionError, |
|
74
|
|
|
configparser.DuplicateSectionError) as err: |
|
75
|
|
|
raise ConfigurationException(err) |
|
76
|
|
|
project_ids = config.sections() |
|
77
|
|
|
|
|
78
|
|
|
# create AnnifProject objects from the configuration file |
|
79
|
|
|
projects = collections.OrderedDict() |
|
80
|
|
|
for project_id in project_ids: |
|
81
|
|
|
projects[project_id] = AnnifProject(project_id, |
|
82
|
|
|
config[project_id], |
|
83
|
|
|
datadir, |
|
84
|
|
|
self) |
|
85
|
|
|
return projects |
|
86
|
|
|
|
|
87
|
|
|
def get_projects(self, min_access=Access.private): |
|
88
|
|
|
"""Return the available projects as a dict of project_id -> |
|
89
|
|
|
AnnifProject. The min_access parameter may be used to set the minimum |
|
90
|
|
|
access level required for the returned projects.""" |
|
91
|
|
|
|
|
92
|
|
|
return {project_id: project |
|
93
|
|
|
for project_id, project in self._projects[self._rid].items() |
|
94
|
|
|
if project.access >= min_access} |
|
95
|
|
|
|
|
96
|
|
|
def get_project(self, project_id, min_access=Access.private): |
|
97
|
|
|
"""return the definition of a single Project by project_id""" |
|
98
|
|
|
|
|
99
|
|
|
projects = self.get_projects(min_access) |
|
100
|
|
|
try: |
|
101
|
|
|
return projects[project_id] |
|
102
|
|
|
except KeyError: |
|
103
|
|
|
raise ValueError("No such project {}".format(project_id)) |
|
104
|
|
|
|
|
105
|
|
|
|
|
106
|
|
|
def initialize_projects(app): |
|
107
|
|
|
projects_file = app.config['PROJECTS_FILE'] |
|
108
|
|
|
datadir = app.config['DATADIR'] |
|
109
|
|
|
init_projects = app.config['INITIALIZE_PROJECTS'] |
|
110
|
|
|
app.annif_registry = AnnifRegistry(projects_file, datadir, init_projects) |
|
111
|
|
|
|
|
112
|
|
|
|
|
113
|
|
|
def get_projects(min_access=Access.private): |
|
114
|
|
|
"""Return the available projects as a dict of project_id -> |
|
115
|
|
|
AnnifProject. The min_access parameter may be used to set the minimum |
|
116
|
|
|
access level required for the returned projects.""" |
|
117
|
|
|
if not hasattr(current_app, 'annif_registry'): |
|
118
|
|
|
initialize_projects(current_app) |
|
119
|
|
|
|
|
120
|
|
|
return current_app.annif_registry.get_projects(min_access) |
|
121
|
|
|
|
|
122
|
|
|
|
|
123
|
|
|
def get_project(project_id, min_access=Access.private): |
|
124
|
|
|
"""return the definition of a single Project by project_id""" |
|
125
|
|
|
|
|
126
|
|
|
projects = get_projects(min_access) |
|
127
|
|
|
try: |
|
128
|
|
|
return projects[project_id] |
|
129
|
|
|
except KeyError: |
|
130
|
|
|
raise ValueError("No such project {}".format(project_id)) |
|
131
|
|
|
|
NatLibFi /
Annif
Loading history...