mspowerpoint_backend: handle_groups/handle_shapes crash on shapes with unrecognized shape_type

[pateltejas]

Bug

_walk_linear() in mspowerpoint_backend.py crashes when processing PPTX files containing <p:sp> elements that python-pptx cannot classify.
handle_groups() and handle_shapes() access shape.shape_type without catching NotImplementedError.
python-pptx's Shape.shape_type only recognizes 4 subtypes (placeholder, freeform, autoshape, textbox). For any <p:sp> with an empty <p:spPr> — no geometry definition, no placeholder, no txBox attribute — it raises:

NotImplementedError("Shape instance of unrecognized shape type")

This kills the entire PPTX conversion, even though the shape's text, tables, etc. would be extractable through the rest of the pipeline (which doesn't use shape_type).

Traceback

File "docling/backend/mspowerpoint_backend.py", line 684, in handle_shapes                                                            
   handle_groups(shape, parent_slide, slide_ind, doc, slide_size)
File "docling/backend/mspowerpoint_backend.py", line 711, in handle_groups                                                            
   if shape.shape_type == MSO_SHAPE_TYPE.GROUP:                                                                                      
      ^^^^^^^^^^^^^^^^                                                                                                               
File "pptx/shapes/autoshape.py", line 325, in shape_type                                                                              
   raise NotImplementedError("Shape instance of unrecognized shape type")    

Steps to reproduce

from lxml import etree                                                                                                                
from pptx import Presentation                                                                                                         
from docling.document_converter import DocumentConverter                                                                              
from pathlib import Path                                                                                                              
                                                                                                                                      
# Create a PPTX with a shape python-pptx can't classify                                                                               
prs = Presentation()                                                                                                                  
slide = prs.slides.add_slide(prs.slide_layouts[6])                                                                                    
                                                                                                                                      
# <p:sp> with empty <p:spPr> — no geometry, no placeholder, no txBox                                                                  
sp_xml = """                                                                                                                          
<p:sp xmlns:p="http://schemas.openxmlformats.org/presentationml/2006/main"                                                            
      xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main"                                                                 
      xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships">                                                  
  <p:nvSpPr>                                                                                                                          
    <p:cNvPr id="100" name="BadShape"/>                                                                                               
    <p:cNvSpPr/>                                                                                                                      
    <p:nvPr/>                                                                                                                         
  </p:nvSpPr>                                                                                                                         
  <p:spPr/>                                                                                                                           
  <p:txBody>                                                                                                                          
    <a:bodyPr/>                                                                                                                       
    <a:p><a:r><a:t>Text in an unrecognized shape</a:t></a:r></a:p>                                                                    
  </p:txBody>                                                                                                                         
</p:sp>                                                                                                                               
"""                                                                                                                                   
slide._element.spTree.append(etree.fromstring(sp_xml))                                                                                
prs.save("/tmp/bad_shape.pptx")                                                                                                       
                                                                                                                                      
# This crashes                                                                                                                        
converter = DocumentConverter()                                                                                                       
converter.convert(Path("/tmp/bad_shape.pptx"))                                                                                        

Docling version

2.84.0

Python version

3.11.7

[dosubot]

Hi @pateltejas! I'm Dosu and I’m helping the docling team.

Good find — the root cause is confirmed. There are two unguarded shape.shape_type accesses in mspowerpoint_backend.py:

1. handle_groups (line 719) — shape.shape_type == MSO_SHAPE_TYPE.GROUP [1]
2. handle_shapes (line 696) — shape.shape_type == MSO_SHAPE_TYPE.PICTURE [2]

Neither has a try/except around the property access, so python-pptx's NotImplementedError propagates and kills the conversion.

A similar class of issue was fixed in PR #2914, where ValueError("no embedded image") from picture shapes was crashing conversions. The fix there was to catch the exception, log a warning, and continue — the same pattern would apply here:

# In handle_groups:
def handle_groups(shape, parent_slide, slide_ind, doc, slide_size):
    try:
        is_group = shape.shape_type == MSO_SHAPE_TYPE.GROUP
    except NotImplementedError:
        _log.warning("Skipping shape with unrecognized shape_type")
        return

    if is_group:
        for groupedshape in shape.shapes:
            handle_shapes(groupedshape, parent_slide, slide_ind, doc, slide_size)

# In handle_shapes (around line 696):
try:
    is_picture = shape.shape_type == MSO_SHAPE_TYPE.PICTURE
except NotImplementedError:
    _log.warning("Skipping shape with unrecognized shape_type")
    is_picture = False

if is_picture:
    self._handle_pictures(shape, parent_slide, slide_ind, doc, slide_size)

This ensures the unrecognized shape is skipped gracefully while the rest of the slide (including text/tables in other shapes) still gets processed.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}