Skip to content

chore: Add Lua sample scripts and sync fixes with Python variants#141

Open
hanatyan128 wants to merge 10 commits intodevfrom
chore-add-lua-sample-codes
Open

chore: Add Lua sample scripts and sync fixes with Python variants#141
hanatyan128 wants to merge 10 commits intodevfrom
chore-add-lua-sample-codes

Conversation

@hanatyan128
Copy link
Copy Markdown
Member

@hanatyan128 hanatyan128 commented Apr 12, 2026

Summary

  • Add Lua sample scripts for recording-filename-from-text and replay-buffer-filename-from-text, mirroring the existing Python variants.
  • Enhance API.md / API_ja.md with a "Threading and Call Context" section, an "Obtaining the Filter Source" walkthrough, and notes on osi_branch_output_get_filter_list (private source exclusion, obs_module_post_load() timing, snapshot semantics).
  • Sync robustness fixes between the Lua and Python variants so both share: override_cleared flag, enhanced sanitize_filename (control chars / trailing dots/spaces / Windows reserved names), UTF-8/UTF-16 BOM handling with 4KB read limit, old-filter override clearing on selection change, filter ID validation with proc return-value logging, and a defensive timer_remove in script_load.

Test plan

  • Load each Lua sample script in OBS, verify the filename override updates as the text source changes
  • Load each Python sample script in OBS, verify the same behavior for parity
  • Switch the selected filter in the script properties and confirm the previous filter's override is cleared
  • Point the text source at a file with a UTF-8 BOM and confirm it is stripped; point it at a UTF-16 file and confirm the warning is logged
  • Unload/reload the script and confirm no duplicate timers via timer_remove + timer_add
  • Verify the API.md / API_ja.md rendering looks correct (new sections, notes)

🤖 Generated with Claude Code

@hanatyan128 hanatyan128 mentioned this pull request Apr 12, 2026
Open
@hanatyan128 hanatyan128 added this to the 1.1.0 milestone Apr 12, 2026
@hanatyan128 hanatyan128 self-assigned this Apr 12, 2026
@tonywingmah
Copy link
Copy Markdown

tonywingmah commented Apr 13, 2026

I have tested the script and it seems to work very well.

  • Load each Lua sample script in OBS, verify the filename override updates as the text source changes
  • Load each Python sample script in OBS, verify the same behavior for parity
  • Switch the selected filter in the script properties and confirm the previous filter's override is cleared

My works flow is
4 x Google Sheets > Companion to scroll athletes > 4 x OBS Text Box > 4x Branch Output Lua Rename script > Record / Stop toggle.

I have modified the script for 4 simultaneous streams to be easier for volunteers.

@hanatyan128 hanatyan128 force-pushed the chore-add-lua-sample-codes branch 9 times, most recently from f9a2238 to 90cd399 Compare April 26, 2026 04:04
@hanatyan128 hanatyan128 force-pushed the chore-add-lua-sample-codes branch from 90cd399 to 288b272 Compare May 4, 2026 04:44
hanatyan128 and others added 10 commits May 4, 2026 13:48
@hanatyan128 @claude
chore: Add Lua sample scripts and enhance API docs
583f6b6 
Add Lua variants of recording-filename-from-text and
replay-buffer-filename-from-text sample scripts. Both include:
- Robust filename sanitization (control chars, trailing dots/spaces,
  Windows reserved names)
- UTF-8 BOM stripping and UTF-16 detection
- 4 KB read size limit for safety
- Filter id verification before proc calls
- Proper cleanup on filter selection change and script reload

Enhance API.md and API_ja.md with:
- Threading and call context requirements
- Filter source acquisition guide
- osi_branch_output_get_filter_list notes (post_load timing,
  private source exclusion)
- Sample scripts section covering both Python and Lua variants

Update README TIPS sections to mention both Python and Lua variants.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@hanatyan128 @claude
chore: Sync sample script fixes between Lua and Python variants
bc42f09 
Align recording/replay-buffer filename-from-text sample scripts so the
Lua and Python variants share the same robustness fixes: version note
corrected to v1.0.9+, override_cleared flag to suppress redundant proc
calls, enhanced sanitize_filename (control chars, trailing dots/spaces,
Windows reserved names), BOM handling with 4KB read limit, old-filter
override clearing on selection change, filter ID validation with proc
return-value logging, and defensive timer_remove in script_load.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@hanatyan128 @claude
fix: Code review fixes for sample scripts and related proc handlers
a969e61 
- onGetFilterList: dispatch to UI thread via QMetaObject::invokeMethod,
  with same-thread direct-call fallback to avoid BlockingQueuedConnection
  self-deadlock; null-out statusDock on module unload
- Sample scripts: reset last_text (and throttle state) on clear_override,
  warn on empty file path, extend Windows reserved-name check to cover
  base names with extensions (e.g. CON.txt)
- API docs: revise threading notes and add call-context guidance for
  osi_branch_output_get_filter_list

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@hanatyan128 @claude
fix: Additional review fixes for multi-byte safety and statusDock lif…
6f4eda3 
…etime

Round 1-3 review findings across sample scripts and related native code:
- Lua sanitize_filename now normalizes NBSP/U+3000/ZWSP/U+2028/U+2029/BOM
- Document Windows ANSI code page limitation of Lua io.open
- statusDock promoted to std::atomic and unpublished before dock teardown
- Clarify callback safety and hotkey context in API docs
- Document lock ordering and proc_handler lifetime in native code

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@hanatyan128 @claude
refactor: Simplify UTF-8 handling in Lua filename scripts
60d7345 
Replace the per-byte-sequence gsub chain in sanitize_filename() with a
utf8.codes() walk driven by a strip/space lookup. Also folds the en/em
space family (U+2000–200A) for parity with the Python variant.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@hanatyan128
fix: address review findings
9564842 
- Lua: replace Lua 5.3 utf8 library with LuaJIT-compatible manual
  UTF-8 decoder/encoder (utf8_codepoints, cp_to_utf8), removing
  dead code and misleading pcall warning
- C++: change JSON fallback from empty string to '{}' for safer
  downstream parsing
- Python: wrap calldata_create/calldata_free in try/finally to
  prevent leaks on exception
- Python: add MAX_SANITIZED_LENGTH=128 to sanitize_filename to
  avoid exceeding Windows MAX_PATH
@hanatyan128
Fix script filename handling and atomic statusDock usage
691c6f8 
Lua/Python scripts:
- Truncate sanitized filenames to 200 bytes with UTF-8 boundary respect
- Call clear_override() when text source is deselected in script_update()
- Validate empty base_format to prevent ambiguous empty overrides (Lua)
- Add group source limitation note to script_description()

C++:
- Use statusDock.store() instead of implicit operator= for consistency
- Fix comment: 'acquire/release semantics' -> 'sequential consistency'
- Clarify Q_DECLARE_METATYPE comment as qRegisterMetaType safety net
@hanatyan128 @claude
refactor: Compress verbose comments and API documentation
d951c67 
Trim review-round artifacts where long justification comments were
added to defend design choices. Replace prose with concise FIXMEs
where the root-cause fix is deferred. Drop internal-implementation
details from API.md / API_ja.md that callers do not need.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@hanatyan128
fix: Restrict statusDock to internal linkage
dc52bfa
@hanatyan128
fix: Correct filename script reliability in Lua and Python variants
40ec4c4 
- Defer last_text update to proc call success to prevent failed calls
  from being silently swallowed on subsequent ticks
- Fix 4-branch new_format logic in Python to match Lua (trailing space,
  unintended clear when base_format is empty)
- Add UTF-8 validation for non-UTF-8 file content (CP932 etc.)
- Add cp_to_utf8 surrogate-half guard
- Detect and warn on file truncation at MAX_READ_SIZE
- Unify sanitize_filename step order with Python variant
- Move MAX_FILENAME_BYTES to module-level constants
- Log warning on JSON parse failure
- Remove redundant state resets after clear_override()
@hanatyan128 hanatyan128 force-pushed the chore-add-lua-sample-codes branch from 288b272 to 40ec4c4 Compare May 4, 2026 16:59
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

@hanatyan128 hanatyan128

Labels

None yet

Projects

None yet

Milestone

1.1.0

Development

Successfully merging this pull request may close these issues.

2 participants

@hanatyan128 @tonywingmah