fix: process alphaModFix for images

[VladaHarbour]

No description provided.

[linear-code]

SD-3243

[github-actions]

The ecma-spec MCP tools are installed but every call is being rejected with "haven't granted it yet" — the permission prompt isn't resolving in this session, so I couldn't run live schema lookups. I'll flag that, and give you the review based on the ECMA-376 DrawingML schema for this element (which is unambiguous here).

---

Status: PASS

The change parses a:alphaModFix out of the a:blip element and stores its amt attribute. That's all spec-correct:

• a:alphaModFix is a real DrawingML element (CT_AlphaModulateFixedEffect, namespace …/drawingml/2006/main) and is a legal child of a:blip (CT_Blip), alongside the other blip effects like a:lum, a:grayscl, a:biLevel, etc. The handler reads it via findChildByLocalName(blip?.elements, 'alphaModFix'), which matches where the spec places it. ✓ — https://ooxml.dev/spec?q=alphaModFix
• amt is the correct (and only) attribute on that element. Its type is ST_PositiveFixedPercentage, range 0..100000 where 100000 = 100%, and it's optional with a default of 100000 (fully opaque). ✓
• Default handling is right. Because the spec default is 100% opaque, resolveImageOpacity returning null when the computed opacity is 1 (and the converter omitting alphaModFix when absent) correctly treats "no alphaModFix" and "amt=100000" as no-op. No incorrect default baked in. ✓
• No missing required attributes — amt is optional, and extractAlphaModFix correctly returns undefined when it's absent or non-finite rather than fabricating a value.
• Unit math matches the spec range — 9000 / 100000 = 0.09, and the painter clamps to [0, 100000] before dividing, consistent with ST_PositiveFixedPercentage.

No non-existent elements/attributes, no spec violations in the two changed files (encode-image-node-helpers.js and its test). The accompanying contract/adapter/painter changes carry the same {amt} shape through faithfully.

One non-spec note (not a compliance issue): the converter doesn't clamp amt to the 0..100000 range — a malformed file with amt="250000" would pass through, but the painter clamps at paint time, so output stays valid. Fine as-is.

If you want the live MCP verification on record, re-run once the ecma-spec tool permissions are granted and I'll confirm CT_AlphaModulateFixedEffect/amt against the parsed XSD directly.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7e5765f120

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[luccas-harbour]

hey @VladaHarbour! it's all looking good here. Claude/codex found a few things that I left as inline comments.
I noticed that there's a unit test that's failing but it doesn't seem to be related to your changes, perhaps merging the latest changes from main into this branch might fix it?

also, not sure if you did this already but it would be good to upload the test document to the corpus (with pnpm corpus:upload) to lock this fix.

[luccas-harbour]

LGTM