fix: knowledge upload file types have single source of truth#878
Conversation
philnash
force-pushed
the
philnash/upload-file-types
branch
from
b068583 to
745fce2
Compare
| ### Supported file types | ||
|
|
||
| OpenRAG supports ingesting the following document types: | ||
|
|
||
| * .doc | ||
| * .docx | ||
| * .pptx | ||
| * .ppt | ||
| * .xlsx | ||
| * .xls | ||
| * .csv | ||
| * .txt | ||
| * .md | ||
| * .html | ||
| * .htm | ||
| * .rtf | ||
| * .odt | ||
| * .asciidoc | ||
| * .adoc | ||
| * .png | ||
| * .jpg | ||
| * .jpeg | ||
| * .gif | ||
| * .bmp | ||
| * .tiff | ||
| * .webp | ||
|
|
There was a problem hiding this comment.
I wouldn't include these in the docs for two reasons:
- They can see the list in the app, and it is more relevant at the "point of performance" when they are uploading files.
- Someone will have to remember to edit this list any time SUPPORTED_FILE_TYPES changes, so it could become misaligned easily.
There was a problem hiding this comment.
I don't believe this full list is displayed in the app anywhere and the troubleshooting page specifically says to "Make sure you ingest only supported file types" without otherwise listing what those are.
As for having to keep the documentation in line with the application, that's true for all features. The logical conclusion to that argument is that we don't have docs at all.
There was a problem hiding this comment.
oh I apologize. I thought this PR added the list to the UI as well.
There was a problem hiding this comment.
I am working on another PR to reorganize some of the knowledge content. It will probably conflict with this PR. I will add this list to my PR so it doesn't get lost.
There was a problem hiding this comment.
We could add a comment in the code pointing to the list in the docs as a reminder to update the list.
| "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet": [".xlsx"], | ||
| "text/csv": [".csv"], | ||
| "text/plain": [".txt"], | ||
| "text/markdown": [".md"], |
There was a problem hiding this comment.
Is .mdx supported?
| "text/markdown": [".md"], | |
| "text/markdown": [".md", ".mdx"], |
There was a problem hiding this comment.
Not according to the Docling Serve component in Langflow.
Co-authored-by: April I. Murphy <[email protected]>
4 participants
Fixes #877.
Uses the const
SUPPORTED_EXTENSIONSfrom the knowledge upload page to make the uploaded file types consistent across all parts of the app.Added the list of supported file types to the docs too. Gave it a header and linked from the troubleshooting section.