phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-30 19:20:01 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

add check for interleavedContent

Browse source 
Signed-off-by: Kevin <kpostlet@redhat.com>
This commit is contained in:
Kevin 2025-04-16 15:47:22 -04:00
parent 2976b5d992
commit 36de927fd6
1 changed files with 16 additions and 17 deletions
Download patch file Download diff file Expand all files Collapse all files

11
llama_stack/providers/utils/memory/vector_store.py
View file

@ -118,27 +118,26 @@ async def content_from_doc(doc: RAGDocument) -> str:
if isinstance(doc.content, URL): if isinstance(doc.content, URL):
if doc.content.uri.startswith("data:"): if doc.content.uri.startswith("data:"):
return content_from_data(doc.content.uri) return content_from_data(doc.content.uri)
else:
async with httpx.AsyncClient() as client: async with httpx.AsyncClient() as client:
r = await client.get(doc.content.uri) r = await client.get(doc.content.uri)
if doc.mime_type == "application/pdf": if doc.mime_type == "application/pdf":
return parse_pdf(r.content) return parse_pdf(r.content)
else:
return r.text return r.text
elif isinstance(doc.content, str):
pattern = re.compile("^(https?://|file://|data:)") pattern = re.compile("^(https?://|file://|data:)")
if pattern.match(doc.content): if pattern.match(doc.content):
if doc.content.startswith("data:"): if doc.content.startswith("data:"):
return content_from_data(doc.content) return content_from_data(doc.content)
else:
async with httpx.AsyncClient() as client: async with httpx.AsyncClient() as client:
r = await client.get(doc.content) r = await client.get(doc.content)
if doc.mime_type == "application/pdf": if doc.mime_type == "application/pdf":
return parse_pdf(r.content) return parse_pdf(r.content)
else:
return r.text return r.text
return doc.content
elif isinstance(doc.content, InterleavedContent):
return interleaved_content_as_str(doc.content) return interleaved_content_as_str(doc.content)
else:
raise ValueError(f"{type(doc)} not supported document.")
def make_overlapped_chunks(document_id: str, text: str, window_len: int, overlap_len: int) -> List[Chunk]: def make_overlapped_chunks(document_id: str, text: str, window_len: int, overlap_len: int) -> List[Chunk]: