youpaishidifu/msgbot
1
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: 行合并把EasyOCR正确切分的消息错误合并\n\n根因：EasyOCR输出'没有个西瓜'+ '关于我们'是两条独立消息，\n但'没有个西瓜'不以_SENT_END结尾，被合并成'没有个西瓜，关于我们'。\n\n修复：新增'下一行保护'——若当前行后面是短消息(<12字)或包含\n昵称前缀(@/用户名开头)，不合并，保留EasyOCR正确切分。\n\n场景: '没有个西瓜'(6字)→下一行'关于我们'(4字<12)→不合并\n场景: '@C418minecraft'→不合并(@开头)\n场景: '你好，'+'世界'→正常合并

Browse Source
This commit is contained in:
youpaishidifu
2026-08-19 21:12:19 +08:00
parent ab8286b6fa
commit beb6812223
1 changed files with 12 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
monitor.py
+12 -1
View File
@@ -3400,10 +3400,21 @@ class MessageMonitor:
# 它是OCR对@提及的误读残片，会被短句碎片拦截；并入会把上行用户消息污染。 # 它是OCR对@提及的误读残片，会被短句碎片拦截；并入会把上行用户消息污染。
_cjk_in_rl = sum(1 for c in rl if '\u4e00' <= c <= '\u9fff') _cjk_in_rl = sum(1 for c in rl if '\u4e00' <= c <= '\u9fff')
_at_noise_line = ('@' in rl) and not re.search(r'[?？]', rl) and _cjk_in_rl <= 4 _at_noise_line = ('@' in rl) and not re.search(r'[?？]', rl) and _cjk_in_rl <= 4
# ⭐ 2026-08-19: 下一条是短消息/用户名前缀时不合并
# EasyOCR常把相邻用户消息切成多行（如"没有个西瓜"+"关于我们"），
# 若下一行是短句(<12字)或包含昵称前缀(/@/括号)，不应并入上行。
_next_is_short_or_name = False
if i + 1 < len(ocr_lines):
_next_line = ocr_lines[i + 1].strip()
_next_cjk = sum(1 for c in _next_line if '\u4e00' <= c <= '\u9fff')
if len(_next_line) < 12 or '@' in _next_line \
or re.match(r'^[\u4e00-\u9fff]{2,8}[@，,]', _next_line) \
or re.match(r'^[/@（(]', _next_line):
_next_is_short_or_name = True
if merged_lines and not _SENT_END.search(merged_lines[-1]) \ if merged_lines and not _SENT_END.search(merged_lines[-1]) \
and not _TS_START.match(rl) and not _TS_START.match(merged_lines[-1]) \ and not _TS_START.match(rl) and not _TS_START.match(merged_lines[-1]) \
and not _is_bot_reply_line and not _is_single_excl and not _prev_is_single_excl \ and not _is_bot_reply_line and not _is_single_excl and not _prev_is_single_excl \
and not _at_noise_line: and not _at_noise_line and not _next_is_short_or_name:
merged_lines[-1] = merged_lines[-1] + '，' + rl merged_lines[-1] = merged_lines[-1] + '，' + rl
else: else:
merged_lines.append(rl) merged_lines.append(rl)