Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.
Python agent-contextagent-memoryagent-pluginsagentic-ragcontext-databasedsh-pluginknowledge-basememorymemory-managementself-evolvingvector-database
fix(pdf): offload local conversion to thread (#2220)
xiejianqiao07dfde379402
diff --git a/openviking/parse/parsers/pdf.py b/openviking/parse/parsers/pdf.pyindex 00b31bea7..e1f3b1c63 100644--- a/openviking/parse/parsers/pdf.py+++ b/openviking/parse/parsers/pdf.py@@ -12,6 +12,7 @@ This design simplifies PDF handling by delegating structure analysisto the MarkdownParser after conversion."""+import asyncioimport reimport timefrom collections import Counter, defaultdict@@ -209,25 +210,16 @@ class PDFParser(BaseParser):async def _convert_local(self, pdf_path: Path, storage=None, resource_name: Optional[str] = None) -> tuple[str, Dict[str, Any]]:- """- Convert PDF to Markdown using pdfplumber.-- When the PDF contains bookmarks/outlines, these are extracted and- injected as markdown headings at the appropriate page positions.- This allows MarkdownParser to build a hierarchical directory tree- instead of producing flat numbered files.-- Args:- pdf_path: Path to PDF file- storage: Optional StoragePath for saving images- resource_name: Resource name for organizing saved images+ # pdfplumber / pdfminer 的解析与图片/表格提取通常是 CPU/IO 密集且为同步实现,+ # 放到线程池中执行,避免阻塞事件循环。+ return await asyncio.to_thread(self._convert_local_sync, pdf_path, storage, resource_name)- Returns:- Tuple of (markdown_content, metadata)+ def _convert_local_sync(+ self, pdf_path: Path, storage=None, resource_name: Optional[str] = None+ ) -> tuple[str, Dict[str, Any]]:+ """同步版:用 pdfplumber 将 PDF 转 Markdown。- Raises:- ImportError: If pdfplumber not installed- Exception: If conversion fails+ 该方法会在 :meth:`_convert_local` 中通过 asyncio.to_thread 调用。"""pdfplumber = lazy_import("pdfplumber")