feat: support caller media in text processing

[gabewillen]

Summary

• Add Cortext::ProcessText(text, source, Media, retention) while preserving canonical text encoding/processing.
• Expose C status and JSON wrappers: cortext_process_text_with_media, cortext_process_text_with_media_json, and cortext_process_text_with_media_json_with_options.
• Verify caller media persistence, opaque binary hydration, C error handling, and legacy behavior.

Exact API contract

• ProcessText(text, source, Media, retention) uses text as the canonical embedding/processing input.
• For non-empty Media, bytes are copied and stored with Media.mimetype; custom MIME payloads are opaque binary and hydrate unchanged.
• Media{} / NULL / size == 0 omits the stored payload for media overloads. Legacy calls without Media retain canonical payload behavior.
• C status wrapper returns 0/1/2 for success/invalid parameters/internal errors. JSON wrappers return Context JSON and support retention/include-embedding options.

Validation

• cmake --build build-local-clang -j2 passed.
• build-local-clang/tests/cortext_tests "[cortext][media][text]": 2 cases, 19 assertions passed.
• build-local-clang/tests/cortext_tests "[capi]": 4 cases, 128 assertions passed.
• build-local-clang/tests/cortext_tests "[media]": 5 cases, 57 assertions passed.
• Full ctest --test-dir build-local-clang --output-on-failure: 676 passed, 31 unrelated existing failures, 5 skipped (reported, not waived).
• EMSCRIPTEN=/opt/homebrew/Cellar/emscripten/4.0.18/libexec emcmake cmake ... configure succeeded in /tmp/cortext-wasm-proof and generated CMakeFiles/cortext_wasm.dir/link.txt linker flags include _cortext_process_text_with_media_json and _cortext_process_text_with_media_json_with_options.
• git diff --check passed.

The feature commit contains only the six intended tracked files; the separate required Emscripten export commit adds the two linker-list entries documented above. .claude/ and scripts/metis/ remain untracked and untouched. Do not merge this PR yet.

[Copilot]

Pull request overview

Adds first-class support for caller-provided media payloads on text processing, extending the existing “media overrides stored payload” behavior to the text path and exposing it through both the C++ and C APIs.

Changes:

• Add Cortext::ProcessText(..., const Media&, Retention) to allow storing an opaque caller payload (with MIME type) while keeping canonical text as the embedding input.
• Add C API wrappers cortext_process_text_with_media* (status + JSON) with media validation consistent with other *_with_media entry points.
• Expand tests and docs to cover persistence semantics (default text payload vs supplied payload vs empty media payload).

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
tests/cortext.test.cpp	Adds coverage for text-with-media persistence behavior and C API entry points.
src/cortext.cpp	Implements the new text-with-media overload and updates payload surface validation rules for text + custom MIME payloads.
src/capi.cpp	Exposes the new text-with-media processing entry points in the C API (status + JSON).
README.md	Documents the storage vs embedding behavior for Media overloads (including empty media semantics).
include/cortext/cortext.hpp	Documents and declares the new C++ ProcessText overload taking Media.
include/cortext/capi.h	Documents and declares new C API functions for processing text with media payloads.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.