修复 Gitea 网页端行内公式与加粗不渲染
根因(读 Gitea 解析器 modules/markup/markdown/math/inline_parser.go 源码确认):
行内公式收尾 $ 的后一个字节必须属于 {空格,换行,行尾,'.','!','?',',',';',':',')','$'},
开头 $ 的左邻不能是 ASCII 字母数字或上述标点。中文与全角标点均为多字节字符
(单字节值 >= 0x80),全部落空,于是整对 $...$ 被当作普通文本原样显示。
本地 PDF 走 KaTeX,无此限制,所以只有网页异常。
处理:
- 新增 .workbuddy/tools/md2pdf/fix_gitea_math.py(检测 + 修复,可 --check/--apply)
- 在收尾 $ 与紧跟的中文字符之间补一个半角空格:23 个文件 1889 处
- 顺带修复加粗失效(收尾 ** 被标点前导、被汉字后继,不满足 CommonMark
right-flanking 规则):10 处,星号不再外露
- 新增 rebuild_prob_pdfs.sh,重生成概率论第3/5/6讲全部 12 个 PDF
范围:概率论在学练习 + 每日小练 + 知识点。试卷/模拟卷/多元函数积分学
仍有 1484 处待修(同规则,可随时跑 --scope 之外的 --apply)。
This commit is contained in:
@@ -0,0 +1,224 @@
|
||||
#!/usr/bin/env python3
|
||||
# -*- coding: utf-8 -*-
|
||||
"""
|
||||
fix_gitea_math.py —— 修复 Gitea 网页端 Markdown 公式/加粗不渲染的问题
|
||||
|
||||
【问题一:行内公式不渲染】
|
||||
Gitea 的解析器 modules/markup/markdown/math/inline_parser.go 规则写死:
|
||||
收尾 $ 的后一个字节必须 ∈ { 空格, 换行, 行尾, '.', '!', '?', ',', ';', ':', ')', '$' }
|
||||
开头 $ 的左邻不能是 ASCII 字母/数字或 . ! ? , ; :
|
||||
中文字与全角标点(,。、;())都是多字节字符,单字节值 ≥ 0x80,全部落空,
|
||||
于是整对 $...$ 被当作普通文本原样显示。
|
||||
→ 修法:在收尾 $ 与紧跟的非空白字符之间补一个半角空格。
|
||||
|
||||
【问题二:加粗失效】
|
||||
CommonMark 要求收尾 ** 必须是 right-flanking:
|
||||
不被空白前导,且(不被标点前导 或 被标点前导但被空白/标点后继)。
|
||||
形如 **…"一致性"**的 的写法(收尾 ** 前是标点、后是汉字的)不满足条件,
|
||||
加粗失效,星号会被原样显示(网页与 PDF 皆然)。
|
||||
→ 修法:在收尾 ** 后补一个半角空格。
|
||||
|
||||
用法:
|
||||
python fix_gitea_math.py --check # 全文统计
|
||||
python fix_gitea_math.py --check --scope=prob # 只看概率论在学练习/小练/知识点
|
||||
python fix_gitea_math.py --apply --scope=prob # 就地修复
|
||||
"""
|
||||
import os
|
||||
import re
|
||||
import subprocess
|
||||
import sys
|
||||
import unicodedata
|
||||
|
||||
ROOT = os.path.abspath(os.path.join(os.path.dirname(os.path.abspath(__file__)), '..', '..', '..'))
|
||||
|
||||
SCOPES = {
|
||||
'prob': [
|
||||
'基于考研的学习/数学一/在学练习/',
|
||||
'基于考研的学习/数学一/每日小练/',
|
||||
'基于考研的学习/数学一/知识点/',
|
||||
],
|
||||
}
|
||||
|
||||
|
||||
def iter_md_files(scope=None):
|
||||
"""只处理 git 已跟踪的 .md(排除 参考仓库),与网页上能看到的内容一致。"""
|
||||
out = subprocess.run(['git', 'ls-files', '-z', '--', '*.md'],
|
||||
cwd=ROOT, capture_output=True, check=True).stdout.decode('utf-8')
|
||||
prefixes = SCOPES.get(scope) if scope else None
|
||||
for rel in out.split('\0'):
|
||||
if not rel or rel.startswith('参考仓库/') or rel.startswith('.workbuddy/'):
|
||||
continue
|
||||
if prefixes and not any(rel.startswith(p) for p in prefixes):
|
||||
continue
|
||||
yield os.path.join(ROOT, rel.replace('/', os.sep))
|
||||
|
||||
|
||||
# ------------------------------------------------------------------ 遮罩
|
||||
|
||||
def mask_code(text):
|
||||
"""遮罩围栏代码块与行内代码,返回等长字符串(被遮位置为 \\x00)。"""
|
||||
n = len(text)
|
||||
mask = [''] * n
|
||||
|
||||
def paint(a, b):
|
||||
for i in range(max(0, a), min(b, n)):
|
||||
mask[i] = '\x00'
|
||||
|
||||
for m in re.finditer(r'(`+)([^\n]*?)\1', text):
|
||||
paint(m.start(), m.end())
|
||||
|
||||
# 围栏代码块:行级状态机(开/闭成对,绝不会误遮到文件末尾)
|
||||
in_fence = False
|
||||
pos = 0
|
||||
for line in text.split('\n'):
|
||||
ln = len(line)
|
||||
if re.match(r'^[ \t]*(```|~~~)', line):
|
||||
in_fence = not in_fence
|
||||
paint(pos, pos + ln)
|
||||
elif in_fence:
|
||||
paint(pos, pos + ln)
|
||||
pos += ln + 1
|
||||
|
||||
return ''.join(c if c else text[i] for i, c in enumerate(mask))
|
||||
|
||||
|
||||
def mask_code_and_math(text):
|
||||
"""在 mask_code 基础上再遮罩公式(用于加粗判定,避免把公式里的 ** 当粗体)。"""
|
||||
base = mask_code(text)
|
||||
mask = list(base)
|
||||
for a, b in find_math_any(text):
|
||||
if base[a] == '\x00':
|
||||
continue
|
||||
for i in range(a, min(b + 1, len(text))):
|
||||
mask[i] = '\x00'
|
||||
return ''.join(mask)
|
||||
|
||||
|
||||
# ------------------------------------------------------------------ 公式
|
||||
|
||||
def find_math_any(text):
|
||||
"""返回 (open_idx, close_idx_inclusive) 列表,绝不跨行。"""
|
||||
spans = []
|
||||
i, n = 0, len(text)
|
||||
while i < n:
|
||||
c = text[i]
|
||||
if c == '\\' and i + 1 < n and text[i + 1] == '$':
|
||||
i += 2
|
||||
continue
|
||||
if c == '$' and i + 1 < n and text[i + 1] == '$':
|
||||
end = text.find('$$', i + 2)
|
||||
if end != -1 and '\n' not in text[i:end]:
|
||||
spans.append((i, end + 1))
|
||||
i = end + 2
|
||||
continue
|
||||
if c == '$':
|
||||
j, close = i + 1, -1
|
||||
while j < n and text[j] != '\n':
|
||||
if text[j] == '\\':
|
||||
j += 2
|
||||
continue
|
||||
if text[j] == '$':
|
||||
close = j
|
||||
break
|
||||
j += 1
|
||||
if close != -1 and close > i + 1:
|
||||
spans.append((i, close))
|
||||
i = close + 1
|
||||
continue
|
||||
i += 1
|
||||
return spans
|
||||
|
||||
|
||||
# Gitea 认可的收尾字符(单字节)
|
||||
END_OK = set('.!?,;:') | {')', ' ', '\n', '$'}
|
||||
# Gitea 不认可的开头左邻(ASCII 字母数字与下列标点)
|
||||
BAD_OPEN_PREV = set('abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789.!?,;:')
|
||||
|
||||
|
||||
def is_punct(ch):
|
||||
return unicodedata.category(ch)[0] in ('P', 'S')
|
||||
|
||||
|
||||
def fix_math(text, mask):
|
||||
inserts = []
|
||||
for start, end in find_math_any(text):
|
||||
if mask[start] == '\x00':
|
||||
continue
|
||||
if start > 0 and mask[start - 1] != '\x00' and text[start - 1] in BAD_OPEN_PREV:
|
||||
inserts.append((start, '公式前'))
|
||||
pos = end + 1
|
||||
if pos < len(text) and mask[pos] != '\x00' and text[pos] not in END_OK:
|
||||
inserts.append((pos, '公式后'))
|
||||
return inserts
|
||||
|
||||
|
||||
# ------------------------------------------------------------------ 加粗
|
||||
|
||||
def fix_bold(text, mask):
|
||||
"""收尾 ** 被标点前导、被非空白非标点后继 → 右侧补空格,使其成为 right-flanking。"""
|
||||
stars = [m.start() for m in re.finditer(r'\*\*', text) if mask[m.start()] != '\x00']
|
||||
inserts = []
|
||||
is_open = True
|
||||
for idx in stars:
|
||||
if is_open:
|
||||
is_open = False
|
||||
continue
|
||||
is_open = True
|
||||
prev = text[idx - 1] if idx > 0 else ''
|
||||
nxt = text[idx + 2] if idx + 2 < len(text) else ''
|
||||
if not prev or not nxt:
|
||||
continue
|
||||
if mask[idx - 1] == '\x00' or mask[idx + 2] == '\x00':
|
||||
continue
|
||||
if is_punct(prev) and not nxt.isspace() and not is_punct(nxt):
|
||||
inserts.append((idx + 2, '加粗后'))
|
||||
return inserts
|
||||
|
||||
|
||||
# ------------------------------------------------------------------ 主流程
|
||||
|
||||
def fix_text(text):
|
||||
mc = mask_code(text)
|
||||
mcm = mask_code_and_math(text)
|
||||
inserts = fix_bold(text, mcm) + fix_math(text, mc)
|
||||
if not inserts:
|
||||
return text, []
|
||||
seen = set()
|
||||
uniq = []
|
||||
for pos, kind in inserts:
|
||||
if (pos, kind) in seen:
|
||||
continue
|
||||
seen.add((pos, kind))
|
||||
uniq.append((pos, kind))
|
||||
for pos, _ in sorted(uniq, reverse=True):
|
||||
text = text[:pos] + ' ' + text[pos:]
|
||||
return text, sorted(uniq, key=lambda x: x[0])
|
||||
|
||||
|
||||
def main():
|
||||
apply = '--apply' in sys.argv
|
||||
scope = None
|
||||
for a in sys.argv:
|
||||
if a.startswith('--scope='):
|
||||
scope = a.split('=', 1)[1]
|
||||
|
||||
total_files = total_hits = 0
|
||||
report = []
|
||||
for path in sorted(iter_md_files(scope)):
|
||||
raw = open(path, encoding='utf-8').read()
|
||||
new, changes = fix_text(raw)
|
||||
if changes:
|
||||
total_files += 1
|
||||
total_hits += len(changes)
|
||||
rel = os.path.relpath(path, ROOT)
|
||||
report.append((rel, len(changes)))
|
||||
if apply:
|
||||
open(path, 'w', encoding='utf-8', newline='').write(new)
|
||||
print(f'== {"已修复" if apply else "待修复(仅统计)"} ==')
|
||||
for rel, cnt in report:
|
||||
print(f' {cnt:4d} {rel}')
|
||||
print(f'--- 共 {total_files} 个文件,{total_hits} 处需要修 ---')
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
Reference in New Issue
Block a user