feat: export 64-bit file_size_bigint at publish

[rtibblesbot]

Summary

Channels containing a file larger than 2.1 GB export with a wrong or missing size, because publish writes each file's size only into the legacy 32-bit LocalFile.file_size column, which cannot represent it. Publish now also writes a 64-bit LocalFile.file_size_bigint carrying the true size, and clamps the legacy 32-bit file_size (on both LocalFile and File) to NULL when the size overflows. Current Kolibri reads the real size from the bigint column; older Kolibri keeps reading the legacy column and simply sees NULL for those oversized files. No content-schema version bump — VERSION_6 is already advertised.

References

Closes #5987. Ships with #6040 (included_presets) in the same VERSION_6 hotfix. Kolibri reader: learningequality/kolibri#14881. Studio-side bigint storage: #5974.

Reviewer guidance

• contentcuration/contentcuration/utils/publish.py:67 — this change deliberately adds no version bump; confirm the content-schema VERSION stays 6 and MIN_SCHEMA_VERSION stays "1" so pre-VERSION_6 Kolibri still imports these databases.
• contentcuration/contentcuration/utils/publish.py:704 — File.file_size now takes the clamped legacy_size; confirm the File.file_size == LocalFile.file_size invariant still holds for oversized files (both NULL).
• contentcuration/kolibri_content/migrations/0025_localfile_file_size_bigint.py and contentcuration/kolibri_public/migrations/0010_localfile_file_size_bigint.py — confirm both mirrors migrate the column and chain off the merged included_presets migrations.

AI usage

Used Claude Code to implement this against a pre-approved plan, following the sibling included_presets export (#6040) for the migration and publish patterns. Verified with the test_exportchannel suite, including two new tests covering the ≤2.1 GB (both columns match) and >2.1 GB (bigint real, legacy NULL) cases and the existing file_size invariant regression.

---

@rtibblesbot's comments are generated by an LLM, and should be evaluated accordingly

How was this generated?

• Ran pre-flight CI checks (lint, format, tests) and verified all pass
• Rebased onto the target branch and resolved any conflicts
• Reorganized commit history into clean, logical commits
• Audited the diff to ensure only issue-relevant files are changed
• Built PR body from the repository's PR template with evidence blocks

🟡 Waiting for feedback Last updated: 2026-07-13 01:37 UTC

[rtibbles]

Changes as expected.