|
1
|
|
|
"""Configuration file handling""" |
|
2
|
|
|
|
|
3
|
|
|
from __future__ import annotations |
|
4
|
|
|
|
|
5
|
|
|
import configparser |
|
6
|
|
|
import os.path |
|
7
|
|
|
from glob import glob |
|
8
|
|
|
|
|
9
|
|
|
try: |
|
10
|
|
|
import tomllib |
|
11
|
|
|
except ImportError: |
|
12
|
|
|
import tomli as tomllib |
|
13
|
|
|
|
|
14
|
|
|
import annif |
|
15
|
|
|
import annif.util |
|
16
|
|
|
from annif.exception import ConfigurationException |
|
17
|
|
|
|
|
18
|
|
|
logger = annif.logger |
|
19
|
|
|
|
|
20
|
|
|
|
|
21
|
|
|
class AnnifConfigCFG: |
|
22
|
|
|
"""Class for reading configuration in CFG/INI format""" |
|
23
|
|
|
|
|
24
|
|
|
def __init__(self, filename: str = None, projstr: str = None) -> None: |
|
25
|
|
|
self._config = configparser.ConfigParser() |
|
26
|
|
|
self._config.optionxform = annif.util.identity |
|
27
|
|
|
if filename is not None: |
|
28
|
|
|
logger.debug(f"Reading configuration file {filename} in CFG format") |
|
29
|
|
|
self._read_config(self._config.read, filename) |
|
30
|
|
|
elif projstr is not None: |
|
31
|
|
|
logger.debug("Reading configuration from a string in CFG format") |
|
32
|
|
|
self._read_config(self._config.read_string, projstr) |
|
33
|
|
|
|
|
34
|
|
|
def _read_config(self, read_method, source): |
|
35
|
|
|
encoding = "utf-8-sig" |
|
36
|
|
|
try: |
|
37
|
|
|
read_method(source, encoding) |
|
38
|
|
|
except ( |
|
39
|
|
|
configparser.DuplicateOptionError, |
|
40
|
|
|
configparser.DuplicateSectionError, |
|
41
|
|
|
) as err: |
|
42
|
|
|
raise ConfigurationException(err.message) |
|
43
|
|
|
|
|
44
|
|
|
@property |
|
45
|
|
|
def project_ids(self) -> list[str]: |
|
46
|
|
|
return self._config.sections() |
|
47
|
|
|
|
|
48
|
|
|
def __getitem__(self, key: str) -> configparser.SectionProxy: |
|
49
|
|
|
return self._config[key] |
|
50
|
|
|
|
|
51
|
|
|
|
|
52
|
|
|
class AnnifConfigTOML: |
|
53
|
|
|
"""Class for reading configuration in TOML format""" |
|
54
|
|
|
|
|
55
|
|
|
def __init__(self, filename: str) -> None: |
|
56
|
|
|
with open(filename, "rb") as projf: |
|
57
|
|
|
try: |
|
58
|
|
|
logger.debug(f"Reading configuration file {filename} in TOML format") |
|
59
|
|
|
self._config = tomllib.load(projf) |
|
60
|
|
|
except tomllib.TOMLDecodeError as err: |
|
61
|
|
|
raise ConfigurationException( |
|
62
|
|
|
f"Parsing TOML file '{filename}' failed: {err}" |
|
63
|
|
|
) |
|
64
|
|
|
|
|
65
|
|
|
@property |
|
66
|
|
|
def project_ids(self): |
|
67
|
|
|
return self._config.keys() |
|
68
|
|
|
|
|
69
|
|
|
def __getitem__(self, key: str) -> dict[str, str]: |
|
70
|
|
|
return self._config[key] |
|
71
|
|
|
|
|
72
|
|
|
|
|
73
|
|
|
class AnnifConfigDirectory: |
|
74
|
|
|
"""Class for reading configuration from directory""" |
|
75
|
|
|
|
|
76
|
|
|
def __init__(self, directory: str) -> None: |
|
77
|
|
|
files = glob(os.path.join(directory, "*.cfg")) |
|
78
|
|
|
files.extend(glob(os.path.join(directory, "*.toml"))) |
|
79
|
|
|
logger.debug(f"Reading configuration files in directory {directory}") |
|
80
|
|
|
|
|
81
|
|
|
self._config = dict() |
|
82
|
|
|
for file in sorted(files): |
|
83
|
|
|
source_config = parse_config(file) |
|
84
|
|
|
for proj_id in source_config.project_ids: |
|
85
|
|
|
self._check_duplicate_project_ids(proj_id, file) |
|
86
|
|
|
self._config[proj_id] = source_config[proj_id] |
|
87
|
|
|
|
|
88
|
|
|
def _check_duplicate_project_ids(self, proj_id: str, file: str) -> None: |
|
89
|
|
|
if proj_id in self._config: |
|
90
|
|
|
# Error message resembles configparser's DuplicateSection message |
|
91
|
|
|
raise ConfigurationException( |
|
92
|
|
|
f'While reading from "{file}": project ID "{proj_id}" already ' |
|
93
|
|
|
"exists in another configuration file in the directory." |
|
94
|
|
|
) |
|
95
|
|
|
|
|
96
|
|
|
@property |
|
97
|
|
|
def project_ids(self): |
|
98
|
|
|
return self._config.keys() |
|
99
|
|
|
|
|
100
|
|
|
def __getitem__(self, key: str) -> dict[str, str] | configparser.SectionProxy: |
|
101
|
|
|
return self._config[key] |
|
102
|
|
|
|
|
103
|
|
|
|
|
104
|
|
|
def check_config(projects_config_path: str) -> str | None: |
|
105
|
|
|
if os.path.exists(projects_config_path): |
|
106
|
|
|
return projects_config_path |
|
107
|
|
|
else: |
|
108
|
|
|
logger.warning( |
|
109
|
|
|
"Project configuration file or directory " |
|
110
|
|
|
+ f'"{projects_config_path}" is missing. Please provide one. ' |
|
111
|
|
|
+ "You can set the path to the project configuration " |
|
112
|
|
|
+ "using the ANNIF_PROJECTS environment " |
|
113
|
|
|
+ 'variable or the command-line option "--projects".' |
|
114
|
|
|
) |
|
115
|
|
|
return None |
|
116
|
|
|
|
|
117
|
|
|
|
|
118
|
|
|
def find_config() -> str | None: |
|
119
|
|
|
for path in ("projects.cfg", "projects.toml", "projects.d"): |
|
120
|
|
|
if os.path.exists(path): |
|
121
|
|
|
return path |
|
122
|
|
|
|
|
123
|
|
|
logger.warning( |
|
124
|
|
|
"Could not find project configuration " |
|
125
|
|
|
+ '"projects.cfg", "projects.toml" or "projects.d". ' |
|
126
|
|
|
+ "You can set the path to the project configuration " |
|
127
|
|
|
+ "using the ANNIF_PROJECTS environment " |
|
128
|
|
|
+ 'variable or the command-line option "--projects".' |
|
129
|
|
|
) |
|
130
|
|
|
return None |
|
131
|
|
|
|
|
132
|
|
|
|
|
133
|
|
|
def parse_config( |
|
134
|
|
|
projects_config_path: str, |
|
135
|
|
|
) -> AnnifConfigDirectory | AnnifConfigCFG | AnnifConfigTOML | None: |
|
136
|
|
|
if projects_config_path: |
|
137
|
|
|
projects_config_path = check_config(projects_config_path) |
|
138
|
|
|
else: |
|
139
|
|
|
projects_config_path = find_config() |
|
140
|
|
|
|
|
141
|
|
|
if not projects_config_path: # not found |
|
142
|
|
|
return None |
|
143
|
|
|
|
|
144
|
|
|
if os.path.isdir(projects_config_path): |
|
145
|
|
|
return AnnifConfigDirectory(projects_config_path) |
|
146
|
|
|
elif projects_config_path.endswith(".toml"): # TOML format |
|
147
|
|
|
return AnnifConfigTOML(projects_config_path) |
|
148
|
|
|
else: # classic CFG/INI style format |
|
149
|
|
|
return AnnifConfigCFG(projects_config_path) |
|
150
|
|
|
|
NatLibFi /
Annif
