fix: truncate uppercase data image URIs

[Lucas Ma (pony-maggie)]

Summary

• make image data URI truncation case-insensitive
• keep preserving full data URIs when keep_data_uris=True
• add regression coverage for uppercase DATA:image sources

Reproduction

Before this change, converting HTML with an image source like DATA:image/png;base64,AAAA emitted the full payload even though keep_data_uris defaults to False. Lowercase data:image sources were truncated, so uppercase schemes bypassed the intended behavior.

Tests

• /tmp/markitdown-venv/bin/python -m pytest packages/markitdown/tests/test_module_misc.py::test_uppercase_data_image_uri_is_truncated_by_default packages/markitdown/tests/test_module_misc.py::test_data_uris -q
• /tmp/markitdown-venv/bin/python -m pytest packages/markitdown/tests/test_module_vectors.py::test_convert_keep_data_uris packages/markitdown/tests/test_module_vectors.py::test_convert_stream_keep_data_uris -q
• git diff --check

[Lucas Ma (pony-maggie)]

@microsoft-github-policy-service agree

[Copilot]

🟡 Changes recommended

Lowercasing the entire potentially large URI introduces avoidable time and memory overhead.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

Makes data-URI image truncation case-insensitive while preserving keep_data_uris=True.

Changes:

• Detects uppercase and mixed-case data: schemes.
• Adds regression coverage for truncation and preservation.

File summaries

File	Description
_markdownify.py	Adds case-insensitive scheme detection.
test_module_misc.py	Tests uppercase data-image URIs.

Review details

• Files reviewed: 2/2 changed files
• Comments generated: 1
• Review effort level: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.