mirror of
https://github.com/piabrota/mu.git
synced 2026-11-07 06:15:59 -10:00
7508c777f3
* md-same-text: also compare blocks as the plan tools find them (base-ci follow-up) Co-Authored-By: Claude Sonnet 5.5 <[email protected]> * plan-generate.py: a blank line before the Files list, for MD032 (base-ci follow-up) Co-Authored-By: Claude Sonnet 5.5 <[email protected]> * md-same-text: allow only the empty -> text info change in plan blocks (base-ci follow-up) Co-Authored-By: Claude Sonnet 5.5 <[email protected]> --------- Co-authored-by: Bruno Guilhermo de Barros Valério <[email protected]> Co-authored-by: Claude Sonnet 5.5 <[email protected]>
75 lines
3.6 KiB
Python
Executable File
75 lines
3.6 KiB
Python
Executable File
#!/usr/bin/env python3
|
|
# Compare two versions of one Markdown file: print "same", "lintoff", "code", "links" or "text". Used by
|
|
# md-same-text.sh.
|
|
# md-same-text.py OLD NEW
|
|
# Checked in this order, and the first that differs is printed:
|
|
# "lintoff": NEW has more markdownlint-disable (-next-line, -line) or markdownlint-configure-file
|
|
# comments than OLD. Turning a rule off is not a formatting fix.
|
|
# "code": the contents of fenced and indented code blocks differ by even one byte, or the blocks the plan
|
|
# tools find (plan-blocks.py's regex, which CommonMark can read differently) differ.
|
|
# "links": the link and image targets differ (href of links, src of images, href= and src= in raw HTML).
|
|
# An autolink (<https://...>) is not counted: its target is its text.
|
|
# "text": the rendered text differs (tags stripped, entities decoded, whitespace collapsed). Raw HTML
|
|
# stays in the text as its source, so `<issue>` and <issue> are the same text.
|
|
# Run it as `python3 -I`, so nothing is imported from the current folder; md-same-text.sh does that
|
|
# through `uvx --with markdown-it-py==3.0.0`. It only reads the two paths.
|
|
import html
|
|
import re
|
|
import sys
|
|
|
|
from markdown_it import MarkdownIt
|
|
|
|
|
|
# Copied from plan-blocks.py (its `block`): the plan tools extract code with this, not with CommonMark. They
|
|
# differ, for example, on a diff context line of one space and three backticks, which CommonMark takes as a
|
|
# closing fence. Compared as (info, body) pairs; see plan_same.
|
|
PLAN_BLOCK = re.compile(r'^(`{3,4})(\w*)\n(.*?)\n\1$', re.M | re.S)
|
|
|
|
|
|
def walk(tokens):
|
|
for t in tokens:
|
|
yield t
|
|
yield from walk(t.children or [])
|
|
|
|
|
|
def parts(path):
|
|
with open(path, encoding='utf-8', newline='') as f:
|
|
src = f.read()
|
|
md = MarkdownIt('commonmark').enable('table')
|
|
# Raw HTML is rendered escaped, so stripping the tags keeps its source as text.
|
|
for rule in ('html_inline', 'html_block'):
|
|
md.add_render_rule(rule, lambda self, tokens, idx, options, env: html.escape(tokens[idx].content))
|
|
text = html.unescape(re.sub(r'<[^>]*>', '', md.render(src)))
|
|
toks = list(walk(md.parse(src)))
|
|
code = [t.content for t in toks if t.type in ('fence', 'code_block')]
|
|
plan = [(m.group(2), m.group(3)) for m in PLAN_BLOCK.finditer(src)]
|
|
links = []
|
|
for t in toks:
|
|
# An autolink (<https://...>) shows its target as its text, which the text check compares.
|
|
if t.type == 'link_open' and t.markup != 'autolink':
|
|
links.append(t.attrGet('href'))
|
|
elif t.type == 'image':
|
|
links.append(t.attrGet('src'))
|
|
elif t.type in ('html_block', 'html_inline'):
|
|
links += re.findall(r'\b(?:href|src)\s*=\s*("[^"]*"|\'[^\']*\'|[^\s>]+)', t.content)
|
|
lintoff = len(re.findall(r'<!--\s*markdownlint-(?:disable|configure-file)', src))
|
|
return ' '.join(text.split()), code, links, lintoff, plan
|
|
|
|
|
|
if len(sys.argv) != 3:
|
|
sys.exit('usage: md-same-text.py OLD NEW')
|
|
old, new = parts(sys.argv[1]), parts(sys.argv[2])
|
|
|
|
|
|
def plan_same(a, b):
|
|
# The (info, body) pairs match byte for byte; the one allowed change is empty -> `text` (markdownlint MD040).
|
|
# Other infos matter to the plan tools: bash and output blocks are never applied.
|
|
return len(a) == len(b) and all(x[1] == y[1] and (x[0] == y[0] or (x[0], y[0]) == ('', 'text')) for x, y in zip(a, b))
|
|
|
|
|
|
if new[3] > old[3]:
|
|
print('lintoff')
|
|
else:
|
|
diff = {'code': old[1] != new[1] or not plan_same(old[4], new[4]), 'links': old[2] != new[2], 'text': old[0] != new[0]}
|
|
print(next((n for n, d in diff.items() if d), 'same'))
|