late commit to rename .txt to .rst
[docutils.git] / docutils / docutils / parsers / __init__.py
blobd9fc25554897ca60c782b581bc4dfce64a4ffc8e
1 # $Id$
2 # Author: David Goodger <goodger@python.org>
3 # Copyright: This module has been placed in the public domain.
5 """
6 This package contains Docutils parser modules.
7 """
9 from __future__ import annotations
11 __docformat__ = 'reStructuredText'
13 import importlib
14 from typing import TYPE_CHECKING, overload
16 from docutils import Component, frontend, transforms
18 if TYPE_CHECKING:
19 from typing import Final, Literal
21 from docutils import nodes
22 from docutils.parsers import (
23 commonmark_wrapper,
24 docutils_xml,
25 null,
26 rst,
27 recommonmark_wrapper,
29 from docutils.transforms import Transform
31 from myst_parser import docutils_ as myst_wrapper
34 class Parser(Component):
35 settings_spec = (
36 'Generic Parser Options',
37 None,
38 (('Disable directives that insert the contents of an external file; '
39 'replaced with a "warning" system message.',
40 ['--no-file-insertion'],
41 {'action': 'store_false', 'default': True,
42 'dest': 'file_insertion_enabled',
43 'validator': frontend.validate_boolean}),
44 ('Enable directives that insert the contents '
45 'of an external file. (default)',
46 ['--file-insertion-enabled'],
47 {'action': 'store_true'}),
48 ('Disable the "raw" directive; '
49 'replaced with a "warning" system message.',
50 ['--no-raw'],
51 {'action': 'store_false', 'default': True, 'dest': 'raw_enabled',
52 'validator': frontend.validate_boolean}),
53 ('Enable the "raw" directive. (default)',
54 ['--raw-enabled'],
55 {'action': 'store_true'}),
56 ('Maximal number of characters in an input line. Default 10 000.',
57 ['--line-length-limit'],
58 {'metavar': '<length>', 'type': 'int', 'default': 10_000,
59 'validator': frontend.validate_nonnegative_int}),
60 ('Validate the document tree after parsing.',
61 ['--validate'],
62 {'action': 'store_true',
63 'validator': frontend.validate_boolean}),
64 ('Do not validate the document tree. (default)',
65 ['--no-validation'],
66 {'action': 'store_false', 'dest': 'validate'}),
69 component_type: Final = 'parser'
70 config_section: Final = 'parsers'
72 def get_transforms(self) -> list[type[Transform]]:
73 return super().get_transforms() + [transforms.universal.Validate]
75 def parse(self, inputstring: str, document: nodes.document) -> None:
76 """Override to parse `inputstring` into document tree `document`."""
77 raise NotImplementedError('subclass must override this method')
79 def setup_parse(self, inputstring: str, document: nodes.document) -> None:
80 """Initial parse setup. Call at start of `self.parse()`."""
81 self.inputstring = inputstring
82 # provide fallbacks in case the document has only generic settings
83 document.settings.setdefault('file_insertion_enabled', False)
84 document.settings.setdefault('raw_enabled', False)
85 document.settings.setdefault('line_length_limit', 10_000)
86 self.document = document
87 document.reporter.attach_observer(document.note_parse_message)
89 def finish_parse(self) -> None:
90 """Finalize parse details. Call at end of `self.parse()`."""
91 self.document.reporter.detach_observer(
92 self.document.note_parse_message)
95 @overload
96 def get_parser_class(parser_name: Literal['null']) -> type[null.Parser]:
97 ...
100 @overload
101 def get_parser_class(
102 parser_name: Literal['rst', 'restructuredtext']
103 ) -> type[rst.Parser]:
107 @overload
108 def get_parser_class(
109 parser_name: Literal['xml', 'docutils_xml']
110 ) -> type[docutils_xml.Parser]:
114 @overload
115 def get_parser_class(
116 parser_name: Literal['recommonmark']
117 ) -> type[recommonmark_wrapper.Parser]:
121 @overload
122 def get_parser_class(
123 parser_name: Literal['myst']
124 ) -> type[myst_wrapper.Parser]:
128 @overload
129 def get_parser_class(
130 parser_name: Literal['commonmark', 'markdown']
131 ) -> type[commonmark_wrapper.Parser]:
135 @overload
136 def get_parser_class(parser_name: str) -> type[Parser]:
140 def get_parser_class(parser_name: str) -> type[Parser]:
141 """Return the Parser class from the `parser_name` module."""
142 name = parser_name.lower()
144 # short names for known parsers
145 if name == 'null':
146 from docutils.parsers import null
147 return null.Parser
148 if name in {'rst', 'restructuredtext', 'rest', 'restx', 'rtxt'}:
149 from docutils.parsers import rst
150 return rst.Parser
151 if name in {'docutils_xml', 'xml'}:
152 from docutils.parsers import docutils_xml
153 return docutils_xml.Parser
155 try:
156 # 3rd-party Markdown parsers
157 # (pycmark works out of the box)
158 if name == 'recommonmark':
159 from docutils.parsers import recommonmark_wrapper
160 return recommonmark_wrapper.Parser
161 if name == 'myst':
162 from myst_parser import docutils_ as myst_wrapper
163 return myst_wrapper.Parser
165 # dispatcher for 3rd-party Markdown parsers
166 if name in {'commonmark', 'markdown'}:
167 from docutils.parsers import commonmark_wrapper
168 return commonmark_wrapper.Parser
170 # fallback to importing a fully-qualified name
171 module = importlib.import_module(name)
172 except ImportError as err:
173 raise ImportError(f'Parser "{parser_name}" not found. {err}')
174 else:
175 return module.Parser