Skip to content

Commit 90cfd4e

Browse files
felixfonteinfelixfontein
committed
Add yamllint check for RST code listings.
1 parent 634e6d0 commit 90cfd4e

File tree

5 files changed

+338
-4
lines changed

5 files changed

+338
-4
lines changed

‎.github/workflows/reusable-nox.yml

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -25,6 +25,8 @@ jobs:
2525
python-versions: "3.11"
2626
- session: "checkers(rstcheck)"
2727
python-versions: "3.11"
28+
- session: "checkers(rst-yamllint)"
29+
python-versions: "3.11"
2830
- session: "checkers(docs-build)"
2931
python-versions: "3.11"
3032
- session: "actionlint"

‎tests/checkers/rst-yamllint.json

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,9 @@
1+
{
2+
"extensions": [
3+
".rst",
4+
".txt"
5+
],
6+
"ignore_regexs": [
7+
"^docs/docsite/rst/porting_guides/porting_guide_[0-9]+\\.rst$"
8+
]
9+
}

‎tests/checkers/rst-yamllint.py

Lines changed: 317 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,317 @@
1+
"""Sanity test using rstcheck and sphinx."""
2+
from __future__ import annotations
3+
4+
import io
5+
import os
6+
import re
7+
import sys
8+
import tempfile
9+
import traceback
10+
11+
from yamllint import linter
12+
from yamllint.config import YamlLintConfig
13+
# from yamllint.linter import PROBLEM_LEVELS
14+
15+
from docutils import nodes
16+
from docutils.core import Publisher
17+
from docutils.frontend import Values
18+
from docutils.io import StringInput
19+
from docutils.parsers.rst import Directive, Parser as RstParser
20+
from docutils.parsers.rst.directives import register_directive, unchanged as directive_param_unchanged
21+
from docutils.utils import new_reporter, Reporter, SystemMessage
22+
23+
import sphinx.application
24+
from sphinx.application import Sphinx
25+
26+
27+
YAMLLINT_CONFIG = r"""
28+
extends: default
29+
30+
rules:
31+
line-length:
32+
max: 160
33+
level: warning
34+
document-start:
35+
# present: true
36+
level: warning
37+
document-end:
38+
# present: false
39+
level: warning
40+
truthy:
41+
level: warning
42+
allowed-values:
43+
- 'true'
44+
- 'false'
45+
indentation:
46+
level: warning
47+
spaces: 2
48+
indent-sequences: consistent
49+
key-duplicates:
50+
level: warning
51+
forbid-duplicated-merge-keys: true
52+
trailing-spaces: enable
53+
hyphens:
54+
max-spaces-after: 1
55+
level: warning
56+
empty-lines:
57+
max: 2
58+
max-start: 0
59+
max-end: 0
60+
level: warning
61+
commas:
62+
max-spaces-before: 0
63+
min-spaces-after: 1
64+
max-spaces-after: 1
65+
level: warning
66+
colons:
67+
max-spaces-before: 0
68+
max-spaces-after: 1
69+
level: warning
70+
brackets:
71+
min-spaces-inside: 0
72+
max-spaces-inside: 0
73+
level: warning
74+
braces:
75+
min-spaces-inside: 0
76+
max-spaces-inside: 1
77+
level: warning
78+
octal-values:
79+
forbid-implicit-octal: true
80+
forbid-explicit-octal: true
81+
level: warning
82+
comments:
83+
min-spaces-from-content: 1
84+
level: warning
85+
"""
86+
87+
REPORT_LEVELS = {
88+
# 'warning', # TODO: enable later
89+
'error',
90+
}
91+
92+
93+
class IgnoreDirective(Directive):
94+
has_content = True
95+
96+
def run(self) -> list:
97+
return []
98+
99+
100+
class CodeBlockDirective(Directive):
101+
has_content = True
102+
optional_arguments = 1
103+
104+
# These are all options Sphinx allows for code blocks.
105+
# We need to have them here so that docutils successfully parses this extension.
106+
option_spec = {
107+
'caption': directive_param_unchanged,
108+
'class': directive_param_unchanged,
109+
'dedent': directive_param_unchanged,
110+
'emphasize-lines': directive_param_unchanged,
111+
'name': directive_param_unchanged,
112+
'force': directive_param_unchanged,
113+
'linenos': directive_param_unchanged,
114+
'lineno-start': directive_param_unchanged,
115+
}
116+
117+
def run(self) -> list[nodes.literal_block]:
118+
code = "\n".join(self.content)
119+
literal = nodes.literal_block(code, code)
120+
literal["classes"].append("code-block")
121+
literal["ansible-code-language"] = self.arguments[0] if self.arguments else None
122+
literal["ansible-code-block"] = True
123+
literal["ansible-code-lineno"] = self.lineno
124+
return [literal]
125+
126+
127+
class YamlLintVisitor(nodes.SparseNodeVisitor):
128+
def __init__(self, document: nodes.document, path: str, results: list[dict], content: str):
129+
super().__init__(document)
130+
self.__path = path
131+
self.__results = results
132+
self.__content_lines = content.splitlines()
133+
134+
def visit_system_message(self, node: nodes.system_message) -> None:
135+
raise nodes.SkipNode
136+
137+
def visit_error(self, node: nodes.error) -> None:
138+
raise nodes.SkipNode
139+
140+
def visit_literal_block(self, node: nodes.literal_block) -> None:
141+
if "ansible-code-block" not in node.attributes:
142+
if node.attributes["classes"]:
143+
self.__results.append({
144+
'path': self.__path,
145+
'line': node.line or 'unknown',
146+
'col': 0,
147+
'message': f"Warning: found unknown literal block! Could be due to '::'. If not, please report this, this is likely a bug in the checker (could be an unsupported Sphinx directive). Node: {node!r}; attributes: {node.attributes}; content: {node.rawsource!r}",
148+
})
149+
raise nodes.SkipNode
150+
151+
language = node.attributes["ansible-code-language"]
152+
lineno = node.attributes["ansible-code-lineno"]
153+
154+
# Ok, we have to find both the row and the column offset for the actual code content
155+
row_offset = lineno
156+
found_empty_line = False
157+
found_content_lines = False
158+
content_lines = node.rawsource.count('\n') + 1
159+
min_indent = None
160+
for offset, line in enumerate(self.__content_lines[lineno:]):
161+
stripped_line = line.strip()
162+
if not stripped_line:
163+
if not found_empty_line:
164+
row_offset = lineno + offset + 1
165+
found_empty_line = True
166+
elif not found_content_lines:
167+
found_content_lines = True
168+
row_offset = lineno + offset
169+
170+
if found_content_lines and content_lines > 0:
171+
if stripped_line:
172+
indent = len(line) - len(line.lstrip())
173+
if min_indent is None or min_indent > indent:
174+
min_indent = indent
175+
content_lines -= 1
176+
elif not content_lines:
177+
break
178+
179+
min_source_indent = None
180+
for line in node.rawsource.split('\n'):
181+
stripped_line = line.lstrip()
182+
if stripped_line:
183+
indent = len(line) - len(line.lstrip())
184+
if min_source_indent is None or min_source_indent > indent:
185+
min_source_indent = indent
186+
187+
col_offset = max(0, (min_indent or 0) - (min_source_indent or 0))
188+
189+
# Now that we have the offsets, we can actually do some processing...
190+
if language not in {'YAML', 'yaml', 'yaml+jinja', 'YAML+Jinja'}:
191+
if language is None:
192+
self.__results.append({
193+
'path': self.__path,
194+
'line': row_offset + 1,
195+
'col': col_offset + 1,
196+
'message': f"Literal block without language!",
197+
})
198+
return
199+
if language not in {'bash', 'ini', 'console', 'text', 'shell', 'shell-session', 'jinja', 'ansible-output', 'none', 'json', 'python', 'Jinja', 'diff', 'powershell', 'md', 'reStructuredText', 'rst', 'sh', 'Python', 'csharp'}:
200+
self.__results.append({
201+
'path': self.__path,
202+
'line': row_offset + 1,
203+
'col': col_offset + 1,
204+
'message': f"Warning: literal block with disallowed language: {language}. If the language should be allowed, the checker needs to be updated.",
205+
})
206+
raise nodes.SkipNode
207+
208+
# So we have YAML. Let's lint it!
209+
try:
210+
conf = YamlLintConfig(YAMLLINT_CONFIG)
211+
problems = linter.run(io.StringIO(node.rawsource.rstrip() + '\n'), conf, self.__path)
212+
for problem in problems:
213+
if problem.level not in REPORT_LEVELS:
214+
continue
215+
msg = f"{problem.level}: {problem.desc}"
216+
if problem.rule:
217+
msg += f" ({problem.rule})"
218+
self.__results.append({
219+
'path': self.__path,
220+
'line': row_offset + problem.line,
221+
'col': col_offset + problem.column,
222+
'message': msg,
223+
})
224+
except Exception as exc:
225+
self.__results.append({
226+
'path': self.__path,
227+
'line': row_offset + 1,
228+
'col': col_offset + 1,
229+
'message': f"Internal error while linting YAML: exception {type(exc)}: {str(exc).replace("\n", " / ")}; traceback: {traceback.format_exc()!r}",
230+
})
231+
232+
raise nodes.SkipNode
233+
234+
235+
def main():
236+
paths = sys.argv[1:] or sys.stdin.read().splitlines()
237+
results = []
238+
239+
for directive in (
240+
'code',
241+
'code-block',
242+
'sourcecode',
243+
):
244+
register_directive(directive, CodeBlockDirective)
245+
246+
# The following docutils directives should better be ignored:
247+
for directive in (
248+
'parsed-literal',
249+
):
250+
register_directive(directive, IgnoreDirective)
251+
252+
# TODO: should we handle the 'literalinclude' directive? maybe check file directly if right extension?
253+
# (https://www.sphinx-doc.org/en/master/usage/restructuredtext/directives.html#directive-literalinclude)
254+
255+
docs_root = os.path.normpath(os.path.join(os.path.dirname(__file__), '..', '..', 'docs', 'docsite', 'rst'))
256+
257+
parser = RstParser()
258+
for path in paths:
259+
with open(path, 'rt', encoding='utf-8') as f:
260+
content = f.read()
261+
262+
# We create a Publisher only to have a mechanism which gives us the settings object.
263+
# Doing this more explicit is a bad idea since the classes used are deprecated and will
264+
# eventually get replaced. Publisher.get_settings() looks like a stable enough API that
265+
# we can 'just use'.
266+
publisher = Publisher(source_class=StringInput)
267+
publisher.set_components('standalone', 'restructuredtext', 'pseudoxml')
268+
override = {
269+
"root_prefix": docs_root,
270+
"input_encoding": "utf-8",
271+
"file_insertion_enabled": False,
272+
"raw_enabled": False,
273+
"_disable_config": True,
274+
"report_level": Reporter.ERROR_LEVEL,
275+
"warning_stream": io.StringIO(),
276+
}
277+
publisher.process_programmatic_settings(None, override, None)
278+
publisher.set_source(content, path)
279+
280+
# Parse the document
281+
try:
282+
doc = publisher.reader.read(publisher.source, publisher.parser, publisher.settings)
283+
except SystemMessage as exc:
284+
results.append({
285+
'path': path,
286+
'line': 0,
287+
'col': 0,
288+
'message': f'Cannot parse document: {str(exc).replace("\n", " / ")}'
289+
})
290+
continue
291+
except Exception as exc:
292+
results.append({
293+
'path': path,
294+
'line': 0,
295+
'col': 0,
296+
'message': f'Cannot parse document, unexpected error {type(exc)}: {str(exc).replace("\n", " / ")}; traceback: {traceback.format_exc()!r}'
297+
})
298+
continue
299+
300+
# Process the document
301+
try:
302+
visitor = YamlLintVisitor(doc, path, results, content)
303+
doc.walk(visitor)
304+
except Exception as exc:
305+
results.append({
306+
'path': path,
307+
'line': 0,
308+
'col': 0,
309+
'message': f'Cannot process document: {type(exc)} {str(exc).replace("\n", " / ")}; traceback: {traceback.format_exc()!r}'
310+
})
311+
312+
for result in sorted(results, key=lambda result: (result['path'], result['line'], result['col'], result['message'])):
313+
print('{path}:{line}:{col}: {message}'.format(**result))
314+
315+
316+
if __name__ == '__main__':
317+
main()

‎tests/requirements.in

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -11,6 +11,7 @@ sphinx-notfound-page # extension used for the custom 404 page (cowsay)
1111
sphinx-ansible-theme # extension used for the custom docs theme
1212
sphinx-rtd-theme
1313
rstcheck
14+
yamllint
1415
sphinx-copybutton
1516
jinja2 # used by hacking/build_library/build_ansible/command_plugins/generate_man.py and dump_keywords.py
1617
pyyaml # used by ansible-core

0 commit comments

Comments
 (0)