Skip to content

Remove duplicate LLM caption helper from ImageConverter - #2826

Open
Mohamed Emad (TheGeekyM) wants to merge 1 commit into
microsoft:mainfrom
TheGeekyM:feat-expose-llm-captioning-in-the-c
Open

Mohamed Emad (TheGeekyM) wants to merge 1 commit into
microsoft:mainfrom
TheGeekyM:feat-expose-llm-captioning-in-the-c

Conversation

@TheGeekyM

Copy link
Copy Markdown

Summary

ImageConverter now calls the shared llm_caption helper in converters/_llm_caption.py instead of its own private copy, _get_llm_description. PptxConverter already uses that helper. The private method and the imports only it needed are deleted. Behaviour stays the same: errors from the client still propagate, and a None caption still leaves the description out.

Motivation

There were two copies of the image-captioning request code. ImageConverter._get_llm_description and llm_caption in _llm_caption.py were the same line for line. Both used the same default prompt ("Write a detailed caption for this image."), resolved the content type the same way (mimetype, then mimetypes.guess_type on the extension, then application/octet-stream), saved and restored the stream position around the base64 read, and sent the same OpenAI-style message. PptxConverter (_pptx_converter.py:186) already called llm_caption. Only the image converter still had its own copy.

The duplication was already known. tests/test_module_misc.py had a TODO saying the LLM parameter test "may only use one test after the llm caption method duplicate has been removed", with a link to #1254.

Keeping two copies means any change to the request has to be made twice: the default prompt, how content types are handled, the message format, or support for a different client. If someone misses one, images and PPTX slides start getting different captions. This affects anyone who passes llm_client/llm_model to MarkItDown, and it is the main thing standing in the way of the larger goal of unifying the vision-caption code paths.

What changed

  • packages/markitdown/src/markitdown/converters/_image_converter.py
    • Imports llm_caption from ._llm_caption and calls it in convert() with the same arguments as before: stream, stream info, client, model, prompt=kwargs.get("llm_prompt").
    • Deletes _get_llm_description.
    • Removes the base64, mimetypes and Union imports, which only that method used.
  • packages/markitdown/tests/test_image.py
    • test_image_without_llm_is_unchanged now patches image_module.llm_caption instead of ImageConverter._get_llm_description, so it still checks that no caption call happens when no LLM is configured.
    • Adds tests that pin the delegation and the behaviour it has to keep (listed below).
  • packages/markitdown/tests/test_module_misc.py
    • Deletes the TODO in test_markitdown_llm_parameters. The image and PPTX assertions both stay: they exercise two different converters end to end, and each converter decides on its own whether to forward llm_prompt.

How it works

llm_caption is imported as a module-level name in _image_converter.py. That matches how _pptx_converter.py imports it, and tests (and anyone extending the converter) can patch it in one place per converter module: monkeypatch.setattr(image_module, "llm_caption", ...). Existing PPTX tests already do this.

The call site in convert() keeps everything around it the same:

  • The caption is only requested when both llm_client and llm_model are set.
  • A None return is skipped, and anything else is stripped and appended under # Description:.
  • No try/except was added. Before this change, an exception from client.chat.completions.create propagated out of ImageConverter.convert(), and it still does. PptxConverter catches caption failures and falls back to its own handling. I left that difference alone because changing it would change behaviour.

Alternatives I considered:

  • Keep _get_llm_description as a thin wrapper around llm_caption. That would keep a subclass override point, but it is a private method, nothing in the repo overrides it (I checked markitdown and markitdown-ocr), and keeping it would leave two names for one operation.
  • Move the helper onto DocumentConverter or a mixin. That is a bigger API decision, and it belongs with the CLI and vision-caption unification work, not this cleanup.
  • Merge the image and PPTX assertions in test_module_misc.py into one, as the TODO suggested. I did not do this. Sharing the helper does not prove that each converter passes llm_prompt through, and those two assertions are what check it.

Overlap with #1254

Open PR #1254 covers the same duplication, and the TODO removed here links to it. This PR is deliberately narrow: one call site moves to the existing helper, and nothing else changes behaviour. If #1254 lands first, this PR can be closed or rebased down to the extra tests. If this one lands first, #1254 will need a rebase where it touches _image_converter.py and the TODO. Reviewers on both may want to agree which one goes first.

Tests

packages/markitdown/tests/test_image.py

  • test_image_without_llm_is_unchanged (updated): with no LLM configured, llm_caption is not called and the output is the metadata only.
  • test_image_converter_uses_shared_llm_caption: image_module.llm_caption is the same object as _llm_caption.llm_caption and _pptx_converter.llm_caption, and ImageConverter no longer has _get_llm_description.
  • test_image_converter_delegates_to_llm_caption (run with and without llm_prompt): convert() calls llm_caption once with the original stream, stream info, client, model and prompt, strips the returned caption, formats it under # Description:, and does not call the client directly.
  • test_image_converter_skips_description_when_llm_caption_returns_none: a None caption leaves the metadata output as it is, with no description heading.
  • test_image_converter_requires_client_and_model_for_caption (run with only a client, then only a model): llm_caption is not called unless both are set.
  • test_image_converter_default_prompt_via_llm_caption (prompt None, "", " "): the request goes through the real helper and sends the default prompt.
  • test_llm_caption_builds_data_uri_and_restores_stream: checks content-type resolution three ways (explicit mimetype wins over extension, extension-based guess, application/octet-stream fallback). It also checks that the data URI encodes only the bytes from the current position onward, that the stream position is restored afterwards, and that the custom prompt and model are passed through.

packages/markitdown/tests/test_module_misc.py

  • test_markitdown_llm_parameters: only the TODO comment is removed. It still checks that llm_prompt reaches the client for both a JPEG and a PPTX.

To run them:

(cd 'packages/markitdown' && python3 -m unittest discover -s tests -p 'test_image.py') && (cd 'packages/markitdown' && python3 -m unittest discover -s tests -p 'test_module_misc.py')

The full suite and lint checks run through hatch test and pre-commit run --all-files.

Compatibility & risk

  • Public API: no change. ImageConverter.convert() takes the same arguments and gives the same output for the same inputs.
  • Removed private method: ImageConverter._get_llm_description no longer exists. Code that subclassed ImageConverter and overrode it, or patched it in tests, will no longer change captioning. That code should patch markitdown.converters._image_converter.llm_caption instead. Nothing in this repo or in markitdown-ocr relies on the old method.
  • Error handling: unchanged. Exceptions from the client still propagate from ImageConverter, and the stream-read failure path still returns None and is skipped.
  • Performance: none. The same single request is made.
  • Where to look closely: the call site in convert() (_image_converter.py:71) and the overlap with Support Async LLM Client, reuse llm_caption helper #1254 described above.

`ImageConverter` now calls the shared `llm_caption` helper in `converters/_llm_caption.py` instead of its own private copy, `_get_llm_description`. `PptxConverter` already uses that helper. The private method and the imports only it needed are deleted. Behaviour stays the same: errors from the client still propagate, and a `None` caption still leaves the description out.
@TheGeekyM
Mohamed Emad (TheGeekyM) marked this pull request as ready for review November 4, 2026 07:32
@microsoft-github-policy-service

Copy link
Copy Markdown

Mohamed Emad (@TheGeekyM) please read the following Contributor License Agreement(CLA). If you agree with the CLA, please reply with the following information.

@microsoft-github-policy-service agree [company="{your company}"]

Options:

  • (default - no company specified) I have sole ownership of intellectual property rights to my Submissions and I am not making Submissions in the course of work for my employer.
@microsoft-github-policy-service agree
  • (when company given) I am making Submissions in the course of work for my employer (or my employer has intellectual property rights in my Submissions by contract or applicable law). I have permission from my employer to make Submissions and enter into this Agreement on behalf of my employer. By signing below, the defined term “You” includes me and my employer.
@microsoft-github-policy-service agree company="Microsoft"
Contributor License Agreement

Contribution License Agreement

This Contribution License Agreement (“Agreement”) is agreed to by the party signing below (“You”),
and conveys certain license rights to Microsoft Corporation and its affiliates (“Microsoft”) for Your
contributions to Microsoft open source projects. This Agreement is effective as of the latest signature
date below.

  1. Definitions.
    “Code” means the computer software code, whether in human-readable or machine-executable form,
    that is delivered by You to Microsoft under this Agreement.
    “Project” means any of the projects owned or managed by Microsoft and offered under a license
    approved by the Open Source Initiative (www.opensource.org).
    “Submit” is the act of uploading, submitting, transmitting, or distributing code or other content to any
    Project, including but not limited to communication on electronic mailing lists, source code control
    systems, and issue tracking systems that are managed by, or on behalf of, the Project for the purpose of
    discussing and improving that Project, but excluding communication that is conspicuously marked or
    otherwise designated in writing by You as “Not a Submission.”
    “Submission” means the Code and any other copyrightable material Submitted by You, including any
    associated comments and documentation.
  2. Your Submission. You must agree to the terms of this Agreement before making a Submission to any
    Project. This Agreement covers any and all Submissions that You, now or in the future (except as
    described in Section 4 below), Submit to any Project.
  3. Originality of Work. You represent that each of Your Submissions is entirely Your original work.
    Should You wish to Submit materials that are not Your original work, You may Submit them separately
    to the Project if You (a) retain all copyright and license information that was in the materials as You
    received them, (b) in the description accompanying Your Submission, include the phrase “Submission
    containing materials of a third party:” followed by the names of the third party and any licenses or other
    restrictions of which You are aware, and (c) follow any other instructions in the Project’s written
    guidelines concerning Submissions.
  4. Your Employer. References to “employer” in this Agreement include Your employer or anyone else
    for whom You are acting in making Your Submission, e.g. as a contractor, vendor, or agent. If Your
    Submission is made in the course of Your work for an employer or Your employer has intellectual
    property rights in Your Submission by contract or applicable law, You must secure permission from Your
    employer to make the Submission before signing this Agreement. In that case, the term “You” in this
    Agreement will refer to You and the employer collectively. If You change employers in the future and
    desire to Submit additional Submissions for the new employer, then You agree to sign a new Agreement
    and secure permission from the new employer before Submitting those Submissions.
  5. Licenses.
  • Copyright License. You grant Microsoft, and those who receive the Submission directly or
    indirectly from Microsoft, a perpetual, worldwide, non-exclusive, royalty-free, irrevocable license in the
    Submission to reproduce, prepare derivative works of, publicly display, publicly perform, and distribute
    the Submission and such derivative works, and to sublicense any or all of the foregoing rights to third
    parties.
  • Patent License. You grant Microsoft, and those who receive the Submission directly or
    indirectly from Microsoft, a perpetual, worldwide, non-exclusive, royalty-free, irrevocable license under
    Your patent claims that are necessarily infringed by the Submission or the combination of the
    Submission with the Project to which it was Submitted to make, have made, use, offer to sell, sell and
    import or otherwise dispose of the Submission alone or with the Project.
  • Other Rights Reserved. Each party reserves all rights not expressly granted in this Agreement.
    No additional licenses or rights whatsoever (including, without limitation, any implied licenses) are
    granted by implication, exhaustion, estoppel or otherwise.
  1. Representations and Warranties. You represent that You are legally entitled to grant the above
    licenses. You represent that each of Your Submissions is entirely Your original work (except as You may
    have disclosed under Section 3). You represent that You have secured permission from Your employer to
    make the Submission in cases where Your Submission is made in the course of Your work for Your
    employer or Your employer has intellectual property rights in Your Submission by contract or applicable
    law. If You are signing this Agreement on behalf of Your employer, You represent and warrant that You
    have the necessary authority to bind the listed employer to the obligations contained in this Agreement.
    You are not expected to provide support for Your Submission, unless You choose to do so. UNLESS
    REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING, AND EXCEPT FOR THE WARRANTIES
    EXPRESSLY STATED IN SECTIONS 3, 4, AND 6, THE SUBMISSION PROVIDED UNDER THIS AGREEMENT IS
    PROVIDED WITHOUT WARRANTY OF ANY KIND, INCLUDING, BUT NOT LIMITED TO, ANY WARRANTY OF
    NONINFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
  2. Notice to Microsoft. You agree to notify Microsoft in writing of any facts or circumstances of which
    You later become aware that would make Your representations in this Agreement inaccurate in any
    respect.
  3. Information about Submissions. You agree that contributions to Projects and information about
    contributions may be maintained indefinitely and disclosed publicly, including Your name and other
    information that You submit with Your Submission.
  4. Governing Law/Jurisdiction. This Agreement is governed by the laws of the State of Washington, and
    the parties consent to exclusive jurisdiction and venue in the federal courts sitting in King County,
    Washington, unless no federal subject matter jurisdiction exists, in which case the parties consent to
    exclusive jurisdiction and venue in the Superior Court of King County, Washington. The parties waive all
    defenses of lack of personal jurisdiction and forum non-conveniens.
  5. Entire Agreement/Assignment. This Agreement is the entire agreement between the parties, and
    supersedes any and all prior agreements, understandings or communications, written or oral, between
    the parties relating to the subject matter hereof. This Agreement may be assigned by Microsoft.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant