|
1
|
|
# -*- coding: utf8 - *-
|
|
2
|
2
|
"""Cihai core functionality."""
|
|
3
|
2
|
from __future__ import absolute_import, print_function, unicode_literals
|
|
4
|
|
|
|
5
|
2
|
import logging
|
|
6
|
2
|
import os
|
|
7
|
|
|
|
8
|
2
|
import kaptan
|
|
9
|
2
|
from appdirs import AppDirs
|
|
10
|
|
|
|
11
|
2
|
from . import exc, extend
|
|
12
|
2
|
from ._compat import string_types
|
|
13
|
2
|
from .config import expand_config
|
|
14
|
2
|
from .constants import DEFAULT_CONFIG, UNIHAN_CONFIG
|
|
15
|
2
|
from .db import Database
|
|
16
|
2
|
from .utils import import_string, merge_dict
|
|
17
|
|
|
|
18
|
2
|
log = logging.getLogger(__name__)
|
|
19
|
|
|
|
20
|
|
|
|
21
|
2
|
class Cihai(object):
|
|
22
|
|
"""
|
|
23
|
|
Central application object.
|
|
24
|
|
|
|
25
|
|
By default, this automatically adds the UNIHAN dataset.
|
|
26
|
|
|
|
27
|
|
Attributes
|
|
28
|
|
----------
|
|
29
|
|
config : dict
|
|
30
|
|
|
|
31
|
|
Notes
|
|
32
|
|
-----
|
|
33
|
|
Inspired by the early pypa/warehouse applicaton object [1]_.
|
|
34
|
|
|
|
35
|
|
**Configuration templates**
|
|
36
|
|
|
|
37
|
|
The ``config`` :py:class:`dict` parameter supports a basic template system
|
|
38
|
|
for replacing :term:`XDG Base Directory` directory variables, tildes
|
|
39
|
|
and environmentas variables. This is done by passing the option dict
|
|
40
|
|
through :func:`cihai.config.expand_config` during initialization.
|
|
41
|
|
|
|
42
|
|
Examples
|
|
43
|
|
--------
|
|
44
|
|
To use cihai programatically, invoke and install the UNIHAN [2]_ dataset:
|
|
45
|
|
|
|
46
|
|
.. literalinclude:: ../examples/basic_usage.py
|
|
47
|
|
:language: python
|
|
48
|
|
|
|
49
|
|
Above: :attr:`~cihai.data.unihan.bootstrap.is_bootstrapped` can check if the system
|
|
50
|
|
has the database installed.
|
|
51
|
|
|
|
52
|
|
References
|
|
53
|
|
----------
|
|
54
|
|
.. [1] PyPA Warehouse on GitHub. https://github.com/pypa/warehouse.
|
|
55
|
|
Accessed sometime in 2013.
|
|
56
|
|
.. [2] UNICODE HAN DATABASE (UNIHAN) documentation.
|
|
57
|
|
https://www.unicode.org/reports/tr38/. Accessed March 31st, 2018.
|
|
58
|
|
"""
|
|
59
|
|
|
|
60
|
|
#: :py:class:`dict` of default config, can be monkey-patched during tests
|
|
61
|
2
|
default_config = DEFAULT_CONFIG
|
|
62
|
|
|
|
63
|
2
|
def __init__(self, config=None, unihan=True):
|
|
64
|
|
"""
|
|
65
|
|
Parameters
|
|
66
|
|
----------
|
|
67
|
|
config : dict, optional
|
|
68
|
|
unihan : boolean, optional
|
|
69
|
|
Bootstrap the core UNIHAN dataset (recommended)
|
|
70
|
|
"""
|
|
71
|
2
|
if config is None:
|
|
72
|
2
|
config = {}
|
|
73
|
|
|
|
74
|
|
# Merges custom configuration settings on top of defaults
|
|
75
|
|
#: Configuration dictionary
|
|
76
|
2
|
self.config = merge_dict(self.default_config, config)
|
|
77
|
|
|
|
78
|
2
|
if unihan:
|
|
79
|
2
|
self.config = merge_dict(UNIHAN_CONFIG, self.config)
|
|
80
|
|
|
|
81
|
|
#: XDG App directory locations
|
|
82
|
2
|
dirs = AppDirs("cihai", "cihai team") # appname # app author
|
|
83
|
|
|
|
84
|
|
#: Expand template variables
|
|
85
|
2
|
expand_config(self.config, dirs)
|
|
86
|
|
|
|
87
|
2
|
if not os.path.exists(dirs.user_data_dir):
|
|
88
|
2
|
os.makedirs(dirs.user_data_dir)
|
|
89
|
|
|
|
90
|
|
#: :class:`cihai.db.Database` : Database instance
|
|
91
|
2
|
self.sql = Database(self.config)
|
|
92
|
|
|
|
93
|
2
|
self.bootstrap()
|
|
94
|
|
|
|
95
|
2
|
def bootstrap(self):
|
|
96
|
2
|
for namespace, class_string in self.config.get('datasets', {}).items():
|
|
97
|
2
|
self.add_dataset(class_string, namespace)
|
|
98
|
|
|
|
99
|
2
|
for dataset, plugins in self.config.get('plugins', {}).items():
|
|
100
|
0
|
for namespace, class_string in plugins.items():
|
|
101
|
0
|
getattr(self, dataset).add_plugin(class_string, namespace)
|
|
102
|
|
|
|
103
|
2
|
def add_dataset(self, _cls, namespace):
|
|
104
|
2
|
if isinstance(_cls, string_types):
|
|
105
|
2
|
_cls = import_string(_cls)
|
|
106
|
|
|
|
107
|
2
|
setattr(self, namespace, _cls())
|
|
108
|
2
|
dataset = getattr(self, namespace)
|
|
109
|
|
|
|
110
|
2
|
if isinstance(dataset, extend.SQLAlchemyMixin):
|
|
111
|
2
|
dataset.sql = self.sql
|
|
112
|
|
|
|
113
|
2
|
@classmethod
|
|
114
|
2
|
def from_file(cls, config_path=None, *args, **kwargs):
|
|
115
|
|
"""
|
|
116
|
|
Create a Cihai instance from a JSON or YAML config.
|
|
117
|
|
|
|
118
|
|
Parameters
|
|
119
|
|
----------
|
|
120
|
|
config_path : str, optional
|
|
121
|
|
path to custom config file
|
|
122
|
|
|
|
123
|
|
Returns
|
|
124
|
|
-------
|
|
125
|
|
:class:`Cihai` :
|
|
126
|
|
application object
|
|
127
|
|
"""
|
|
128
|
|
|
|
129
|
2
|
config_reader = kaptan.Kaptan()
|
|
130
|
|
|
|
131
|
2
|
config = {}
|
|
132
|
|
|
|
133
|
2
|
if config_path:
|
|
134
|
2
|
if not os.path.exists(config_path):
|
|
135
|
0
|
raise exc.CihaiException(
|
|
136
|
|
'{0} does not exist.'.format(os.path.abspath(config_path))
|
|
137
|
|
)
|
|
138
|
2
|
if not any(
|
|
139
|
|
config_path.endswith(ext) for ext in ('json', 'yml', 'yaml', 'ini')
|
|
140
|
|
):
|
|
141
|
0
|
raise exc.CihaiException(
|
|
142
|
|
'{0} does not have a yaml,yml,json,ini extend.'.format(
|
|
143
|
|
os.path.abspath(config_path)
|
|
144
|
|
)
|
|
145
|
|
)
|
|
146
|
|
else:
|
|
147
|
2
|
custom_config = config_reader.import_config(config_path).get()
|
|
148
|
2
|
config = merge_dict(config, custom_config)
|
|
149
|
|
|
|
150
|
2
|
return cls(config)
|
tony