fix(pptx): coerce None shape/notes text to '' before concatenation (#1808)

packages/markitdown/src/markitdown/converters/_pptx_converter.py

@@ -161,10 +161,16 @@ def get_shape_content(shape, **kwargs):
 
                 # Text areas
                 elif shape.has_text_frame:
+                    # python-pptx can yield ``None`` for ``shape.text`` when a
+                    # text frame has a run with no ``<a:t>`` child or when a
+                    # third-party deck stores chart/SmartArt titles as
+                    # ``None``. Coerce to "" so a single bad shape doesn't
+                    # fail the entire conversion (#1808).
+                    shape_text = shape.text or ""
                     if shape == title:
-                        md_content += "# " + shape.text.lstrip() + "\n"
+                        md_content += "# " + shape_text.lstrip() + "\n"
                     else:
-                        md_content += shape.text + "\n"
+                        md_content += shape_text + "\n"
 
                 # Group Shapes
                 if shape.shape_type == pptx.enum.shapes.MSO_SHAPE_TYPE.GROUP:
@@ -194,7 +200,9 @@ def get_shape_content(shape, **kwargs):
                 md_content += "\n\n### Notes:\n"
                 notes_frame = slide.notes_slide.notes_text_frame
                 if notes_frame is not None:
-                    md_content += notes_frame.text
+                    # See note above re: ``shape.text`` returning ``None``;
+                    # the same coercion applies to notes text frames (#1808).
+                    md_content += notes_frame.text or ""
                 md_content = md_content.strip()
 
         return DocumentConverterResult(markdown=md_content.strip())

packages/markitdown/tests/test_pptx_none_text.py

@@ -0,0 +1,113 @@
+"""Regression test for #1808.
+
+``PptxConverter.convert`` performed unguarded ``+`` concatenation on
+``shape.text`` and ``notes_frame.text``. python-pptx can return ``None``
+for either when a text frame has a run with no ``<a:t>`` child, or when
+a third-party deck represents a chart/SmartArt title text as ``None``.
+A single such shape used to fail the entire .pptx conversion with::
+
+    PptxConverter threw TypeError with message:
+      can only concatenate str (not "NoneType") to str
+
+The fix coerces ``shape.text`` / ``notes_frame.text`` to ``""`` before
+concatenation. This pins both call sites.
+"""
+
+from __future__ import annotations
+
+from io import BytesIO
+
+import pytest
+
+pptx = pytest.importorskip("pptx")  # markitdown[pptx] extra
+
+from markitdown import MarkItDown
+
+
+def _build_minimal_pptx_with_text(text: str) -> BytesIO:
+    """Build a minimal one-slide .pptx that contains *text* in the body
+    placeholder. Returns a BytesIO suitable for MarkItDown.convert_stream.
+    """
+    prs = pptx.Presentation()
+    # Layout 1 = "Title and Content"
+    slide = prs.slides.add_slide(prs.slide_layouts[1])
+    title = slide.shapes.title
+    title.text = "Hello"
+    body_placeholder = slide.placeholders[1]
+    body_placeholder.text = text
+    buf = BytesIO()
+    prs.save(buf)
+    buf.seek(0)
+    return buf
+
+
+def test_pptx_with_none_shape_text_does_not_crash(monkeypatch):
+    """Regression for #1808: a shape whose ``.text`` returns ``None`` must
+    not fail the whole conversion."""
+    # Build the deck BEFORE monkey-patching — replacing ``text`` with a
+    # getter-only property breaks the ``placeholder.text = ...`` setter.
+    stream = _build_minimal_pptx_with_text("World")
+
+    pptx_text_module = pptx.text.text
+    original_text = pptx_text_module.TextFrame.text.fget
+
+    def _none_for_world(self):  # type: ignore[no-untyped-def]
+        value = original_text(self)
+        return None if value == "World" else value
+
+    monkeypatch.setattr(
+        pptx_text_module.TextFrame,
+        "text",
+        property(_none_for_world),
+    )
+
+    md = MarkItDown()
+    result = md.convert_stream(stream, file_extension=".pptx")
+
+    # Conversion completes without raising. The deck's body shape resolved
+    # to None and was treated as empty; the title row is still preserved
+    # because it doesn't trip the None branch.
+    assert "# Hello" in result.text_content
+
+
+def test_pptx_with_none_notes_text_does_not_crash(monkeypatch):
+    """Regression for #1808: ``notes_frame.text`` returning ``None`` must
+    not fail conversion either."""
+    # Build a deck and add a notes slide with text we'll monkey-patch to None.
+    prs = pptx.Presentation()
+    slide = prs.slides.add_slide(prs.slide_layouts[1])
+    slide.shapes.title.text = "Hello"
+    notes = slide.notes_slide.notes_text_frame
+    notes.text = "speaker note"
+    buf = BytesIO()
+    prs.save(buf)
+    buf.seek(0)
+
+    # Patch only the notes frame's TextFrame.text accessor to None.
+    pptx_text_module = pptx.text.text
+    original_text = pptx_text_module.TextFrame.text.fget
+
+    def _none_for_speaker_note(self):  # type: ignore[no-untyped-def]
+        value = original_text(self)
+        return None if value == "speaker note" else value
+
+    monkeypatch.setattr(
+        pptx_text_module.TextFrame,
+        "text",
+        property(_none_for_speaker_note),
+    )
+
+    md = MarkItDown()
+    result = md.convert_stream(buf, file_extension=".pptx")
+    # Conversion completes; title still appears.
+    assert "# Hello" in result.text_content
+
+
+def test_pptx_normal_text_still_converts():
+    """Regression guard: the coercion must not change behavior for normal
+    decks where ``shape.text`` is a real string."""
+    stream = _build_minimal_pptx_with_text("body content")
+    md = MarkItDown()
+    result = md.convert_stream(stream, file_extension=".pptx")
+    assert "# Hello" in result.text_content
+    assert "body content" in result.text_content