1# $Id: __init__.py 10382 2026-07-14 13:32:05Z milde $
2# Author: David Goodger <goodger@python.org>
3# Copyright: This module has been placed in the public domain.
4
5"""
6This package contains Docutils parser modules.
7"""
8
9from __future__ import annotations
10
11__docformat__ = 'reStructuredText'
12
13import importlib
14
15from docutils import Component, frontend, transforms
16
17TYPE_CHECKING = False
18if TYPE_CHECKING:
19 from typing import Final
20
21 from docutils import nodes
22 from docutils.transforms import Transform
23
24
25class Parser(Component):
26 settings_spec = (
27 'Generic Parser Options',
28 None,
29 (('Disable directives that insert the contents of an external file; '
30 'replaced with a "warning" system message.',
31 ['--no-file-insertion'],
32 {'dest': 'file_insertion_enabled', 'action': 'store_false',
33 'default': True, 'validator': frontend.validate_boolean}),
34 ('Enable directives that insert the contents '
35 'of an external file. (default)',
36 ['--file-insertion-enabled'],
37 {'action': 'store_true'}),
38 ('Disable the "raw" directive; '
39 'replaced with a "warning" system message.',
40 ['--no-raw'],
41 {'dest': 'raw_enabled', 'action': 'store_false', 'default': True,
42 'validator': frontend.validate_boolean}),
43 ('Enable the "raw" directive. (default)',
44 ['--raw-enabled'],
45 {'action': 'store_true'}),
46 ('Maximal number of characters in an input line. (default 10 000)',
47 ['--line-length-limit'],
48 {'metavar': '<length>', 'type': 'int', 'default': 10_000,
49 'validator': frontend.validate_nonnegative_int}),
50 ('Keep IDs backwards compatible. (default)',
51 ['--legacy-ids'],
52 {'action': 'store_true', 'default': True,
53 'validator': frontend.validate_boolean}),
54 ('Generate IDs for implicit targets only if required; '
55 'no IDs for external and indirect targets.',
56 ['--lazy-ids'],
57 {'dest': 'legacy_ids', 'action': 'store_false'}),
58 ('Validate the document tree after parsing.',
59 ['--validate'],
60 {'action': 'store_true', 'validator': frontend.validate_boolean}),
61 ('Do not validate the document tree. (default)',
62 ['--no-validation'],
63 {'dest': 'validate', 'action': 'store_false'}),
64 )
65 )
66 component_type: Final = 'parser'
67 config_section: Final = 'parsers'
68
69 def get_transforms(self) -> list[type[Transform]]:
70 return super().get_transforms() + [transforms.universal.Validate]
71
72 def parse(self, inputstring: str, document: nodes.document) -> None:
73 """Override to parse `inputstring` into document tree `document`."""
74 raise NotImplementedError('subclass must override this method')
75
76 def setup_parse(self, inputstring: str, document: nodes.document) -> None:
77 """Initial parse setup. Call at start of `self.parse()`."""
78 self.inputstring = inputstring
79 # provide fallbacks in case the document has only generic settings
80 document.settings.setdefault('file_insertion_enabled', False)
81 document.settings.setdefault('raw_enabled', False)
82 document.settings.setdefault('line_length_limit', 10_000)
83 self.document = document
84 document.reporter.attach_observer(document.note_parse_message)
85
86 def finish_parse(self) -> None:
87 """Finalize parse details. Call at end of `self.parse()`."""
88 self.document.reporter.detach_observer(
89 self.document.note_parse_message)
90
91
92PARSER_ALIASES = { # short names for known parsers
93 'null': 'docutils.parsers.null',
94 # reStructuredText
95 'rst': 'docutils.parsers.rst',
96 'restructuredtext': 'docutils.parsers.rst',
97 'rest': 'docutils.parsers.rst',
98 'restx': 'docutils.parsers.rst',
99 'rtxt': 'docutils.parsers.rst',
100 # Docutils XML
101 'docutils_xml': 'docutils.parsers.docutils_xml',
102 'xml': 'docutils.parsers.docutils_xml',
103 # 3rd-party Markdown parsers
104 'myst': 'myst_parser.docutils_',
105 # 'pycmark': works out of the box (with `legacy_ids`)
106 # dispatcher for 3rd-party Markdown parsers
107 'commonmark': 'docutils.parsers.commonmark_wrapper',
108 'markdown': 'docutils.parsers.commonmark_wrapper',
109 }
110
111
112def get_parser_class(parser_name: str) -> type[Parser]:
113 """Return the Parser class from the `parser_name` module."""
114 name = parser_name.lower()
115
116 try:
117 module = importlib.import_module(PARSER_ALIASES.get(name, name))
118 except ImportError as err:
119 raise ImportError(f'Parser "{parser_name}" not found. {err}') from err
120 return module.Parser