Compare commits

...

2 Commits

Author SHA1 Message Date
朱潮
9825a43cad Merge branch 'feature/moshui20260410-file-path-fix' into bot_manager 2026-04-13 18:53:26 +08:00
朱潮
d2bba9e095 update path fix 2026-04-10 17:17:03 +08:00
2 changed files with 90 additions and 0 deletions

View File

@ -22,6 +22,7 @@ from utils.fastapi_utils import detect_provider
from .guideline_middleware import GuidelineMiddleware from .guideline_middleware import GuidelineMiddleware
from .tool_output_length_middleware import ToolOutputLengthMiddleware from .tool_output_length_middleware import ToolOutputLengthMiddleware
from .tool_use_cleanup_middleware import ToolUseCleanupMiddleware from .tool_use_cleanup_middleware import ToolUseCleanupMiddleware
from .filepath_fix_middleware import FilePathFixMiddleware
from utils.settings import ( from utils.settings import (
SUMMARIZATION_MAX_TOKENS, SUMMARIZATION_MAX_TOKENS,
SUMMARIZATION_TOKENS_TO_KEEP, SUMMARIZATION_TOKENS_TO_KEEP,
@ -535,6 +536,7 @@ def create_custom_cli_agent(
deepagent_middleware = [ deepagent_middleware = [
TodoListMiddleware(), TodoListMiddleware(),
FilePathFixMiddleware(), # 修复工具调用参数中 CJK 文件名的多余空格
CustomFilesystemMiddleware(backend=composite_backend), # 使用自定义的 FilesystemMiddleware支持 SKILL.md 完整读取 CustomFilesystemMiddleware(backend=composite_backend), # 使用自定义的 FilesystemMiddleware支持 SKILL.md 完整读取
AnthropicPromptCachingMiddleware(unsupported_model_behavior="ignore"), AnthropicPromptCachingMiddleware(unsupported_model_behavior="ignore"),
PatchToolCallsMiddleware(), PatchToolCallsMiddleware(),

View File

@ -0,0 +1,88 @@
import re
import logging
from langchain.agents.middleware import AgentMiddleware
from langchain.tools.tool_node import ToolCallRequest
logger = logging.getLogger('app')
# 常见文件后缀(小写,用于正则匹配)
_FILE_EXT = (
# 文档
r'txt|pdf|doc|docx|xls|xlsx|ppt|pptx|csv|md|rtf|odt|ods|odp'
r'|jtd|jtt|jaw' # JustSystems Ichitaro
# 图片
r'|jpg|jpeg|png|gif|bmp|svg|webp|ico|tiff|tif'
# 音频
r'|mp3|wav|ogg|flac|aac|m4a'
# 视频
r'|mp4|avi|mov|mkv|wmv|flv|webm'
# 压缩
r'|zip|tar|gz|bz2|rar|7z|tgz'
# 代码/数据
r'|py|js|ts|html|htm|css|json|xml|yaml|yml|sql|sh|bat'
r'|java|c|cpp|h|rb|go|rs|php'
# 其他
r'|log|ini|cfg|conf|toml|env|bin|exe|dmg|iso|db|sqlite|tmp|bak'
)
_EXT_PATTERN = rf'\.(?:{_FILE_EXT})\b'
# 锚点:文件后缀 或 目录分隔符 /(用于匹配路径中间的目录名)
_ANCHOR = rf'(?:{_EXT_PATTERN}|/)'
# 模式1: CJK字符 + 空格 + (以字母数字开头的片段 + 锚点)
# "農作業 2025.jtd" → "農業2025.jtd"
# "ファイル report.txt" → "ファイルreport.txt"
# "/datasets/報告書 2025/file.txt" → "/datasets/報告書2025/file.txt"
_CJK_TO_ALNUM = re.compile(
rf'([\u4e00-\u9fff\u3040-\u309f\u30a0-\u30ff])\s+([a-zA-Z0-9][\w]*{_ANCHOR})'
)
# 模式2: 数字 + 空格 + (以CJK开头的片段 + 锚点)
# "2025 報告書.txt" → "2025報告書.txt"
# "/datasets/2025 報告書/file.txt" → "/datasets/2025報告書/file.txt"
# 注意: 只匹配数字不匹配字母,避免 "cat 報告.txt" 被误修
# 注意: (?<!\.) 排除 "xxx.py ファイル.txt" 这种跨参数的误匹配
_ALNUM_TO_CJK = re.compile(
rf'(?<!\.)(\d+)\s+([\u4e00-\u9fff\u3040-\u309f\u30a0-\u30ff][\w]*{_ANCHOR})'
)
# 模式3: 路径上下文中(/ 后面)的字母 + 空格 + (字母数字开头.后缀)
# "/data/report 2025年度.xlsx" → "/data/report2025年度.xlsx"
# 用 (?<=/) 确保只在路径中生效,避免 "cat report 2025.txt" 被误修
_PATH_LETTER = re.compile(
rf'(?<=/)([a-zA-Z][\w]*)\s+([a-zA-Z0-9][\w]*{_ANCHOR})'
)
def fix_filename_spacing(text: str) -> str:
"""修复文件名/路径中 CJK 与非CJK 交界处的多余空格(循环应用直到无更多匹配)"""
prev = None
while prev != text:
prev = text
text = _CJK_TO_ALNUM.sub(r'\1\2', text)
text = _ALNUM_TO_CJK.sub(r'\1\2', text)
text = _PATH_LETTER.sub(r'\1\2', text)
return text
class FilePathFixMiddleware(AgentMiddleware):
"""修复工具调用参数中 CJK 文件名/路径的多余空格"""
def _fix_tool_call_args(self, request: ToolCallRequest) -> None:
args = request.tool_call.get('args', {})
for key, value in args.items():
if isinstance(value, str):
fixed = fix_filename_spacing(value)
if fixed != value:
logger.info(f"Filename spacing fix: args['{key}'] '{value}' -> '{fixed}'")
args[key] = fixed
def wrap_tool_call(self, request, handler):
self._fix_tool_call_args(request)
return handler(request)
async def awrap_tool_call(self, request, handler):
self._fix_tool_call_args(request)
return await handler(request)