run-llama/llama_cloud_services
1
0
Fork 0
mirror of https://github.com/run-llama/llama_cloud_services.git synced 2026-07-18 07:44:41 -04:00
Code Issues 352 Packages Projects Releases 10 Wiki Activity

PPTX, DOCS parsing not working with Python SDK but works when I use the UI due to mime type=None #292

New Issue
Closed
opened 2026-02-16 00:17:23 -05:00 by yindo · 1 comment
yindo commented 2026-02-16 00:17:23 -05:00
Owner
Copy Link
Copy Source

Originally created by @ajpanyteam on GitHub (Oct 11, 2024).

(code is below)

Describe the bug

When using llama-index-core==0.11.17 and llama-parse==0.5.7 to upload Powerpoint and WordDoc files from a GCP bucket, it fails, error below.

For reference, in the GCP bucket, the powerpoint MIME type is "application/vnd.openxmlformats-officedocument.presentationml.presentation"

Started parsing the file under job_id 95517e7b-b8d5-48f4-888f-dd49336a9f4c
Error while parsing the file '<bytes/buffer>':
Job ID: 95517e7b-b8d5-48f4-888f-dd49336a9f4c failed with status: ERROR, Error code: _UNKOWN_ERROR, Error message: UNKNOWN_ERROR: PDF_IS_BROKEN

However, using the LlamaParse UI to upload the same PPTX file indicates that no markdown text was present and suggests using the Accurate mode. When Accurate mode is used, markdowns are generated.

Re-running the Python code then retrieves the document, likely due to caching.

Investigation

  • Adding a print statement in base.py on
mime_type = mimetypes.guess_type(file_name)[0]
            print('guessing mime type:', mime_type) **returns None**
  • If I retrieve the file from the bucket, write the file locally and then load_data it successfully processes the document.

guessing mime type: application/vnd.openxmlformats-officedocument.wordprocessingml.document

How can I address this by sending the blob to load_data instead?

Files
Happens with all DOCX and PPTX files

Job ID
If you have it, please provide the ID of the job you ran.
See above.

Client:
Please remove untested options:

  • Frontend (cloud.llamaindex.ai)
  • Python Library
  • Notebook
  • API

My Code

parser = LlamaParse(
    api_key=os.environ["LLAMA_PARSE_API_KEY"],
    result_type="markdown",
)
storage_client = storage.Client() //GCP
bucket_name = os.getenv('GCP_FILE_UPLOAD_BUCKET_LOCAL')
bucket = storage_client.bucket(bucket_name)
blob = bucket.blob(fileId)
filename = blob.name
buffer = blob.download_as_string()
# buffer = blob.download_as_bytes(). --- I tried both options
extra_info = {'file_name': filename}
documents = parser.load_data(buffer ,extra_info=extra_info)
print('documents count:', len(documents))

Additional context

  • The issue is related to MIME type handling differences between the Python library and the UI. The Python library fails to process the file initially but succeeds after the file is processed through the UI, suggesting a potential caching mechanism or MIME type issue.
Originally created by @ajpanyteam on GitHub (Oct 11, 2024). (code is below) **Describe the bug** When using llama-index-core==0.11.17 and llama-parse==0.5.7 to upload Powerpoint and WordDoc files from a GCP bucket, it fails, error below. For reference, in the GCP bucket, the powerpoint MIME type is "application/vnd.openxmlformats-officedocument.presentationml.presentation" ``` Started parsing the file under job_id 95517e7b-b8d5-48f4-888f-dd49336a9f4c Error while parsing the file '<bytes/buffer>': Job ID: 95517e7b-b8d5-48f4-888f-dd49336a9f4c failed with status: ERROR, Error code: _UNKOWN_ERROR, Error message: UNKNOWN_ERROR: PDF_IS_BROKEN ``` However, using the LlamaParse UI to upload the same PPTX file indicates that no markdown text was present and suggests using the Accurate mode. When Accurate mode is used, markdowns are generated. Re-running the Python code then retrieves the document, likely due to caching. **Investigation** - Adding a print statement in base.py on ``` mime_type = mimetypes.guess_type(file_name)[0] print('guessing mime type:', mime_type) **returns None** ``` - If I retrieve the file from the bucket, write the file locally and then `load_data` it successfully processes the document. ` guessing mime type: application/vnd.openxmlformats-officedocument.wordprocessingml.document ` **How can I address this by sending the blob to `load_data` instead?** **Files** Happens with all DOCX and PPTX files **Job ID** If you have it, please provide the ID of the job you ran. See above. **Client:** Please remove untested options: - Frontend (cloud.llamaindex.ai) - Python Library - Notebook - API **My Code** ``` parser = LlamaParse( api_key=os.environ["LLAMA_PARSE_API_KEY"], result_type="markdown", ) storage_client = storage.Client() //GCP bucket_name = os.getenv('GCP_FILE_UPLOAD_BUCKET_LOCAL') bucket = storage_client.bucket(bucket_name) blob = bucket.blob(fileId) filename = blob.name buffer = blob.download_as_string() # buffer = blob.download_as_bytes(). --- I tried both options extra_info = {'file_name': filename} documents = parser.load_data(buffer ,extra_info=extra_info) print('documents count:', len(documents)) ``` **Additional context** - The issue is related to MIME type handling differences between the Python library and the UI. The Python library fails to process the file initially but succeeds after the file is processed through the UI, suggesting a potential caching mechanism or MIME type issue.
yindo added the bug label 2026-02-16 00:17:23 -05:00
yindo closed this issue 2026-02-16 00:17:23 -05:00
yindo commented 2026-02-16 00:17:24 -05:00
Author
Owner
Copy Link
Copy Source

@ajpanyteam commented on GitHub (Oct 11, 2024):

Closing. The filename in my code did not include the file extension, therefore llamaParse did not know what to do with it.

Take this with a pinch of salt...
It looks to me like the magic number is the same for PPTX and DOCX.
.docx files start with PK, which signals a ZIP archive (OpenXML formats are zipped).
thus
.pptx files also start with PK

@ajpanyteam commented on GitHub (Oct 11, 2024): Closing. The filename in my code did not include the file extension, therefore llamaParse did not know what to do with it. _Take this with a pinch of salt..._ It looks to me like the magic number is the same for PPTX and DOCX. .docx files start with PK, which signals a ZIP archive (OpenXML formats are zipped). thus .pptx files also start with PK
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
LlamaCloud
LlamaExtract
LlamaParse
bug
client
complex layout
documentation
duplicate
enhancement
figure
font
header footer
help wanted
ocr
parser
pdf_debug
pdf_debug
pull-request
Mirrored from GitHub Pull Request
 python-lib
question
server_issue
server_issue
ui
No Label bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: run-llama/llama_cloud_services#292
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?