feat: add skill and assistant knowledge bases
This commit is contained in:
1 parent
dd0ee282df
commit
fbd8dec48f
45 files changed
+5262
-21
No files matched your search
@@ -167,6 +167,24 @@ async def convert_file_to_markdown(file_path: Path) -> Path | None:
|
||||
return None
|
||||
|
||||
|
||||
async def extract_file_to_markdown(file_path: Path) -> str | None:
|
||||
"""Extract a supported document to Markdown without creating a sidecar file.
|
||||
|
||||
Knowledge-ingest scanners use this variant because running a read-only
|
||||
distillation task must never add generated ``.md`` files to a live skill.
|
||||
Conversion always runs off the event loop; Office/PDF parsers can be CPU
|
||||
intensive even for deceptively small inputs.
|
||||
"""
|
||||
|
||||
if file_path.suffix.lower() not in CONVERTIBLE_EXTENSIONS:
|
||||
return None
|
||||
try:
|
||||
return await asyncio.to_thread(_do_convert, file_path, _get_pdf_converter())
|
||||
except Exception as exc:
|
||||
logger.error("Failed to extract %s as markdown: %s", file_path.name, exc)
|
||||
return None
|
||||
|
||||
|
||||
# Regex for bold-only lines that look like section headings.
|
||||
# Targets SEC filing structural headings that pymupdf4llm renders as **bold**
|
||||
# rather than # Markdown headings (because they use same font size as body text,
|
||||
|
||||
Reference in new issue
Block a user