Skip to content
Back to search

chandan-digital-local

MarketplaceWorks with Claude CodeGitHub

Video editing and local media tools by Chandan Digital

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

Its scripts were only partly scanned.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/chandandigital-cms/CDVEA with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/chandandigital-cms/CDVEA'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/chandandigital-cms--cdvea/badge.svg)](/listings/chandandigital-cms--cdvea)

HTML

<a href="/listings/chandandigital-cms--cdvea"><img src="/api/listings/chandandigital-cms--cdvea/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

The easiest way: paste one of these prompts into **Claude Code, Codex or Cursor** on your Windows computer. The agent checks and installs Python and FFmpeg, downloads the plugin, runs setup and connects CDVEA to itself, asking you before it installs anything. It follows [AI-INSTALL.md](AI-INSTALL.md).

**English**

Install the CDVEA video editing plugin on this Windows computer.
Follow these instructions exactly: https://raw.githubusercontent.com/chandandigital-cms/CDVEA/main/AI-INSTALL.md
Repository: https://github.com/chandandigital-cms/CDVEA
Ask me before installing any software or changing app settings.

**বাংলা**

এই Windows কম্পিউটারে CDVEA ভিডিও এডিটিং প্লাগইন ইনস্টল করে দাও।
ঠিক এই নির্দেশগুলো মেনে চলো: https://raw.githubusercontent.com/chandandigital-cms/CDVEA/main/AI-INSTALL.md
রিপোজিটরি: https://github.com/chandandigital-cms/CDVEA
কোনো সফটওয়্যার ইনস্টল বা অ্যাপের সেটিং বদলানোর আগে আমাকে জিজ্ঞেস করবে।

Setup downloads about 700 MB and takes 5–15 minutes. In the **Codex app**, the agent finishes by asking you to upload `Downloads\CDVEA.zip` in **New Plugin**, since only you can do that step.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 14 lines
  1. {
  2. "name": "chandan-digital-local",
  3. "owner": {
  4. "name": "Chandan Digital"
  5. },
  6. "plugins": [
  7. {
  8. "name": "cdvea",
  9. "source": "./",
  10. "description": "CDVEA by Chandan Digital: local video editing, editorial planning, motion graphics, captions, sound and delivery."
  11. }
  12. ],
  13. "description": "Video editing and local media tools by Chandan Digital"
  14. }
README.md · README · 244 lines
  1. <div align="center">
  2. <img src="assets/readme/hero.svg" width="100%" alt="CDVEA by Chandan Digital – AI video editing agent for Claude Code and Codex: cut footage, clean sound, add Bengali, Hindi and English captions, motion graphics and platform exports with FFmpeg">
  3. [![Version](https://img.shields.io/badge/version-1.2.1-3bdbc3?style=flat-square)](https://github.com/chandandigital-cms/CDVEA/releases/latest)
  4. [![License: MIT](https://img.shields.io/badge/license-MIT-3bdbc3?style=flat-square)](LICENSE)
  5. [![Windows x64](https://img.shields.io/badge/setup-Windows%2010%2F11%20x64-102a2c?style=flat-square&logo=windows)](SETUP-GUIDE.md)
  6. [![Python 3.10+](https://img.shields.io/badge/python-3.10%2B-102a2c?style=flat-square&logo=python&logoColor=white)](SETUP-GUIDE.md)
  7. [![FFmpeg 6.1+](https://img.shields.io/badge/FFmpeg-6.1%2B-102a2c?style=flat-square&logo=ffmpeg)](SETUP-GUIDE.md)
  8. [![MCP tools](https://img.shields.io/badge/MCP-49%20tools-3bdbc3?style=flat-square)](CAPABILITIES.md)
  9. <a href="https://github.com/chandandigital-cms/CDVEA/releases/latest/download/CDVEA.zip"><img src="https://img.shields.io/badge/Download-CDVEA%20AI%20Plugin%20·%2018%20MB-3bdbc3?style=for-the-badge&logo=windows&logoColor=0a1c1e&labelColor=e8fbf8" alt="Download the CDVEA AI plugin for Codex and Claude Code (CDVEA.zip, 18 MB)"></a>
  10. <a href="SETUP-GUIDE.md"><img src="https://img.shields.io/badge/Setup%20guide-English%20·%20বাংলা-102a2c?style=for-the-badge&labelColor=0a1c1e" alt="CDVEA setup guide in English and Bengali"></a>
  11. <sub>Needs <a href="https://www.python.org/downloads/windows/">Python 3.10+</a> and <a href="https://www.gyan.dev/ffmpeg/builds/">FFmpeg 6.1+</a>. Setup downloads the speech model and the rest automatically.</sub>
  12. **[Install with AI](#-install-with-your-ai-agent)** · **[Quick start](#-quick-start-windows)** · **[Features](#-features)** · **[Demos](#-see-it-work)** · **[Prompts](#-ask-in-english-or-bengali)** · **[FAQ](#-faq)**
  13. </div>
  14. # CDVEA — AI Video Editing Agent for Claude Code & Codex
  15. **CDVEA** (by Chandan Digital) is an open-source **AI video editing plugin** that turns Claude Code or Codex into a local video editor. Describe the edit in plain **English or Bengali**. CDVEA plans it, cuts the footage with **FFmpeg**, cleans and mixes the sound, burns in **Bengali, Hindi or English subtitles**, adds titles and motion graphics, and exports masters for YouTube, Instagram Reels and Shorts.
  16. Everything core runs **on your own computer**. Your footage is not uploaded, and there is no cloud render step.
  17. > 30 skills · 8 specialist agents · 9 slash commands · 49 MCP tools · one-time Windows setup
  18. ## ✨ Why CDVEA
  19. - **Edit by conversation:** "Make a 30-second promo with Bengali captions and a vertical Short" is a complete instruction.
  20. - **Local and private:** Python + FFmpeg on your machine. Optional speech models load only for a transcription job.
  21. - **A full post-production crew:** a director, editor, sound editor, caption editor, motion designer, media analyst, delivery editor and an independent reviewer.
  22. - **Indic captions done properly:** Bengali and Hindi caption review, retiming and font checks, plus English.
  23. - **Reproducible:** edits live in a plan file you can validate, re-render and hand over as an EDL cut list.
  24. ## 🎬 How it works
  25. <img src="assets/readme/pipeline.svg" width="100%" alt="CDVEA video editing pipeline: footage analysis, cutting, sound mixing, captions, motion graphics and delivery">
  26. 1. **Footage:** probe files, detect scenes, transcribe speech with word timestamps.
  27. 2. **Cut:** trim, join, speed ramps, remove long pauses and filler words, multicam switching.
  28. 3. **Sound:** voice cleanup, music ducking, loudness normalisation (LUFS).
  29. 4. **Captions:** SRT, karaoke and word-by-word captions; Bengali/Hindi retiming.
  30. 5. **Motion:** title cards, lower thirds, logo overlays, HTML and React motion graphics.
  31. 6. **Deliver:** 1080p masters, ProRes/DNxHR, proxies, Reels/Shorts/TikTok versions, QC reports.
  32. ## 🎥 See it work
  33. Every preview below was rendered by CDVEA's own scripts. In each clip, the left half is the input and the right half is the result.
  34. <table>
  35. <tr>
  36. <td align="center"><img src="docs/demos/captions_pop_karaoke.gif" width="260" alt="Karaoke-style pop captions burned into video"><br><sub>Karaoke captions</sub></td>
  37. <td align="center"><img src="docs/demos/lower_third_hindi.gif" width="260" alt="Hindi lower third title overlay"><br><sub>Hindi lower third</sub></td>
  38. <td align="center"><img src="docs/demos/silence_removal.gif" width="260" alt="Automatic silence removal from a talking video"><br><sub>Silence removal</sub></td>
  39. </tr>
  40. <tr>
  41. <td align="center"><img src="docs/demos/color_correct.gif" width="260" alt="Colour correction before and after"><br><sub>Colour correction</sub></td>
  42. <td align="center"><img src="docs/demos/speedramp.gif" width="260" alt="Speed ramp effect"><br><sub>Speed ramp</sub></td>
  43. <td align="center"><img src="docs/demos/multicam_switch.gif" width="260" alt="Multicam angle switching"><br><sub>Multicam switch</sub></td>
  44. </tr>
  45. <tr>
  46. <td align="center"><img src="docs/demos/bgm_ducking.gif" width="260" alt="Background music ducking under dialogue"><br><sub>Music ducking</sub></td>
  47. <td align="center"><img src="docs/demos/greenscreen.gif" width="260" alt="Green screen chroma key removal"><br><sub>Green screen</sub></td>
  48. <td align="center"><img src="docs/demos/stabilize.gif" width="260" alt="Shaky video stabilisation"><br><sub>Stabilisation</sub></td>
  49. </tr>
  50. <tr>
  51. <td align="center"><img src="docs/demos/title_card.gif" width="260" alt="Animated title card"><br><sub>Title card</sub></td>
  52. <td align="center"><img src="docs/demos/broll_cutaway.gif" width="260" alt="B-roll cutaway insert"><br><sub>B-roll cutaway</sub></td>
  53. <td align="center"><img src="docs/demos/export_reels.gif" width="260" alt="Vertical export for Instagram Reels and YouTube Shorts"><br><sub>Reels / Shorts export</sub></td>
  54. </tr>
  55. </table>
  56. <p align="center"><a href="docs/demos.md"><b>Open the full demo gallery (77 examples) →</b></a></p>
  57. ## 💬 Ask in English or Bengali
  58. - "এই footage দিয়ে ৩০ সেকেন্ডের banquet promo বানান। বাংলা captions দিন, সঙ্গে একটি vertical Short।"
  59. - "ভিডিওর লম্বা pause গুলো কেটে দিন আর voice টা পরিষ্কার করুন।"
  60. - "Remove long pauses, keep complete sentences, clean the voice and give me an SRT."
  61. - "Match these shots, add the supplied logo and export a 1080p master."
  62. - "Review this edit for bad cuts, spelling, lip sync and sound problems."
  63. Slash commands in Claude Code: `/cdvea:plan`, `/cdvea:edit`, `/cdvea:audio`, `/cdvea:captions`, `/cdvea:motion`, `/cdvea:shorts`, `/cdvea:review`, `/cdvea:export`, `/cdvea:doctor`.
  64. ## 🧰 Features
  65. | Work | What CDVEA does |
  66. | --- | --- |
  67. | ✂️ Editing | Cut, join, trim, speed ramps, freeze frames, reverse, loops, smart cuts, pause and filler removal |
  68. | 🎞️ Assembly | Reproducible multi-clip timelines from a plan file, multi-track composites, picture-in-picture, B-roll |
  69. | 🎚️ Sound | Voice cleanup, denoise, music ducking, loudness targets, 5.1 downmix, external-mic sync |
  70. | 🔤 Captions | Burned-in subtitles, karaoke and word captions, SRT export, Bengali/Hindi review and retiming |
  71. | 🎨 Look | Colour correction, LUTs, HDR to SDR tone mapping, stabilisation, deinterlace, green screen |
  72. | ✨ Motion | Title cards, lower thirds, logo overlays, HTML (HyperFrames) and React (Remotion) motion graphics |
  73. | 🤖 AI assist | Local transcription with word timestamps (faster-whisper), content-based scene detection |
  74. | 📦 Delivery | 1080p masters, ProRes, DNxHR, alpha video, proxies, Reels/Shorts/TikTok formats, EDL cut lists |
  75. | 🔍 Review | Technical checks, contact sheets, render QC reports, independent reviewer agent |
  76. Some routes need extra software. [CAPABILITIES.md](CAPABILITIES.md) lists exactly what each one requires.
  77. ## 🤖 Install with your AI agent
  78. The easiest way: paste one of these prompts into **Claude Code, Codex or Cursor** on your Windows computer. The agent checks and installs Python and FFmpeg, downloads the plugin, runs setup and connects CDVEA to itself, asking you before it installs anything. It follows [AI-INSTALL.md](AI-INSTALL.md).
  79. **English**
  80. ```text
  81. Install the CDVEA video editing plugin on this Windows computer.
  82. Follow these instructions exactly: https://raw.githubusercontent.com/chandandigital-cms/CDVEA/main/AI-INSTALL.md
  83. Repository: https://github.com/chandandigital-cms/CDVEA
  84. Ask me before installing any software or changing app settings.
  85. ```
  86. **বাংলা**
  87. ```text
  88. এই Windows কম্পিউটারে CDVEA ভিডিও এডিটিং প্লাগইন ইনস্টল করে দাও।
  89. ঠিক এই নির্দেশগুলো মেনে চলো: https://raw.githubusercontent.com/chandandigital-cms/CDVEA/main/AI-INSTALL.md
  90. রিপোজিটরি: https://github.com/chandandigital-cms/CDVEA
  91. কোনো সফটওয়্যার ইনস্টল বা অ্যাপের সেটিং বদলানোর আগে আমাকে জিজ্ঞেস করবে।
  92. ```
  93. Setup downloads about 700 MB and takes 5–15 minutes. In the **Codex app**, the agent finishes by asking you to upload `Downloads\CDVEA.zip` in **New Plugin**, since only you can do that step.
  94. ## 🚀 Quick start (Windows)
  95. Prefer to do it yourself? Three steps:
  96. **1. Install the required software** (once):
  97. | Software | Download | Or with winget |
  98. | --- | --- | --- |
  99. | Python 3.10+ (tick **"Add python.exe to PATH"**) | [python.org](https://www.python.org/downloads/windows/) | `winget install --id Python.Python.3.12 -e` |
  100. | FFmpeg 6.1+ with FFprobe | [gyan.dev](https://www.gyan.dev/ffmpeg/builds/) · [ffmpeg.org](https://ffmpeg.org/download.html) | `winget install --id Gyan.FFmpeg -e` |
  101. | Codex or Claude Code | [Codex](https://developers.openai.com/codex/quickstart) · [Claude Code](https://code.claude.com/docs/en/setup) | — |
  102. Close and reopen PowerShell after installing.
  103. **2. Download the plugin and run setup.** Paste into PowerShell:
  104. ```powershell
  105. Invoke-WebRequest https://github.com/chandandigital-cms/CDVEA/releases/latest/download/CDVEA.zip -OutFile "$env:USERPROFILE\Downloads\CDVEA.zip"
  106. Expand-Archive "$env:USERPROFILE\Downloads\CDVEA.zip" "$env:LOCALAPPDATA\CDVEA-Plugin" -Force
  107. powershell -ExecutionPolicy Bypass -File "$env:LOCALAPPDATA\CDVEA-Plugin\CDVEA\setup.ps1" -Hosts none
  108. ```
  109. Setup downloads CDVEA's own Python 3.12, its packages and the [Bengali/Hindi/English speech model](https://huggingface.co/Systran/faster-whisper-small) (about 484 MB, SHA-256 checked), then prints `"status": "ready"`.
  110. **3. Add the plugin.** In Codex, open **New Plugin**, choose **`Downloads\CDVEA.zip`** (don't extract it) and click **Add plugin**. Restart Codex and open a new chat.
  111. For the Claude Code or Codex CLI, run the last line of step 2 again with `-Hosts claude` (or `-Hosts codex`) instead of uploading.
  112. The [setup guide](SETUP-GUIDE.md) covers every step in English and Bengali, plus optional software (Node.js, Blender) and troubleshooting.
  113. ### From source
  114. ```powershell
  115. git clone https://github.com/chandandigital-cms/CDVEA.git
  116. cd CDVEA
  117. powershell -ExecutionPolicy Bypass -File .\setup.ps1
  118. ```
  119. Needs Git, Python and FFmpeg from step 1. This registers CDVEA with the Claude Code and Codex CLIs it finds.
  120. ### Use the CLI directly
  121. ```powershell
  122. python setup.py
  123. python cdvea/run.py doctor
  124. python cdvea/cli.py catalog
  125. python cdvea/cli.py init "D:/VideoJobs/my-edit"
  126. ```
  127. Fill the generated project's clip ranges, then validate and render it. [The plan format](references/cdvea-plan.md) has the exact fields, and [the routing list](references/routing.md) shows every skill.
  128. ## 🗂️ What's inside
  129. | Part | Count | Where |
  130. | --- | --- | --- |
  131. | Skills | 30 | [`skills/`](skills) |
  132. | Specialist agents | 8 | [`agents/`](agents) |
  133. | Slash commands | 9 | [`commands/`](commands) |
  134. | Local MCP server tools | 49 (42 FFmpeg + 6 project + 1 transcription) | [`mcp/`](mcp) |
  135. | Python core and tests | CLI, runner, installer, AI helpers | [`cdvea/`](cdvea) |
  136. **Agents:** director · editor · sound editor · caption editor · motion designer · media analyst · delivery editor · reviewer.
  137. ## ❓ FAQ
  138. <details>
  139. <summary><b>Can Claude Code edit videos?</b></summary>
  140. Yes. With CDVEA installed, Claude Code (or Codex) can run real video edits through local FFmpeg tools and a local MCP server. You describe the result, and it plans, renders and checks the edit.
  141. </details>
  142. <details>
  143. <summary><b>Does CDVEA upload my footage?</b></summary>
  144. No. The core runs locally with Python and FFmpeg. Activation starts no uploads. Hosted services and YouTube publishing work only when you ask for them and set up your own account.
  145. </details>
  146. <details>
  147. <summary><b>Does it support Bengali and Hindi subtitles?</b></summary>
  148. Yes. CDVEA includes an Indic caption skill for Bengali and Hindi caption review, spelling checks, retiming and font coverage, alongside English captions. Local speech-to-text supports these languages; check accuracy on your own sample first.
  149. </details>
  150. <details>
  151. <summary><b>Which operating systems are supported?</b></summary>
  152. The automatic installer supports Windows 10/11 x64. The core FFmpeg tools also work on macOS and Linux, but this release has no automatic installer for them.
  153. </details>
  154. <details>
  155. <summary><b>Is it a replacement for Premiere Pro or DaVinci Resolve?</b></summary>
  156. No. It is an agent workflow for fast, repeatable edits. It can export an EDL cut list for relinking in an editor, but it does not automate native Premiere, Resolve or Final Cut projects.
  157. </details>
  158. ## ⚠️ Tested scope
  159. Version 1.2.1 passes 9 local integration tests on Windows x64: cuts, speed, missing audio, source preservation, subtitle retiming, cut-list export and MCP calls. HyperFrames, Remotion, Blender, hosted generation, YouTube publishing and native editor imports have **not** passed end-to-end tests yet. A fresh install on a clean second computer has not been tested either. [VALIDATION.md](VALIDATION.md) records exactly what ran.
  160. The repository and CDVEA.zip contain only the plugin source. They exclude model weights, Python, FFmpeg, virtual environments, media and account credentials. Setup downloads the runtime and model from their official sources, and you install FFmpeg yourself under its own licence.
  161. ## 🙏 Credits and licences
  162. | Project | Included material | Licence |
  163. | --- | --- | --- |
  164. | [kajisho5/ffmpeg-skill](https://github.com/kajisho5/ffmpeg-skill) | Supplied engine, 42 tools, templates and references | MIT |
  165. | [gitethanwoo/video-editing](https://github.com/gitethanwoo/video-editing) | Workbench, analysis, B-roll, product and screen-demo skills | MIT |
  166. | [zaddy99999/video-editing-skills](https://github.com/zaddy99999/video-editing-skills) | Short-form skills, scripts, shared code and brand example | MIT |
  167. | [heygen-com/hyperframes](https://github.com/heygen-com/hyperframes) | Selected HTML motion, CLI, registry and media skills | Apache-2.0; bundled font notices remain |
  168. [THIRD-PARTY-NOTICES.md](THIRD-PARTY-NOTICES.md), [`licenses/`](licenses) and `sources.lock.json` list source revisions and file hashes. Original authors keep credit for their work. Vendor source text does not authorise paid calls, telemetry, promotion or publishing.
  169. <div align="center">
  170. <br>
  171. <img src="assets/cdvea.svg" width="220" alt="CDVEA logo">
  172. <br>
  173. <sub>Made by <b>Chandan Digital</b> · AI video editing for Claude Code &amp; Codex · MIT licence</sub>
  174. </div>
scripts/_common/emoji.py · Bundled script · 512 lines
  1. """Emoji: cluster detection (ZWJ sequences, keycaps, flags, skin tones), the PNG asset lookup,
  2. the per-machine support probe and the overlay filter chain that composites the assets.
  3. Split out of _common.text in the refactor after 1.17.3; every body is byte-identical.
  4. """
  5. from __future__ import annotations
  6. import os
  7. import subprocess
  8. from typing import Any, Dict, List, Tuple, Optional
  9. from _common.decision import escape_filter_path
  10. from _common.emit import die
  11. from _common.runner import ffmpeg_exe
  12. from _common.runner import safe_which as _safe_which
  13. # --------------------------------------------------------------------------- emoji (1.15)
  14. # Emoji are orthogonal to the writing system: "やった 🎉" is Japanese AND emoji. They are detected
  15. # separately from detect_script() so a cue's font resolution is still decided by its letters.
  16. EMOJI_RANGES = (
  17. (0x1F300, 0x1FAFF), # symbols & pictographs, supplemental, extended-A
  18. (0x1F000, 0x1F0FF), # mahjong/domino/playing cards (only 🀄 and 🃏 are emoji)
  19. (0x2600, 0x27BF), # misc symbols + dingbats (Emoji_Presentation only: _is_emoji_char)
  20. (0x2B00, 0x2BFF), # misc symbols and arrows (likewise)
  21. (0xFE0F, 0xFE0F), # VS16 (emoji presentation selector)
  22. (0x1F1E6, 0x1F1FF), # regional indicators (flags)
  23. (0x20E3, 0x20E3), # combining enclosing keycap
  24. (0x1F3FB, 0x1F3FF), # skin-tone modifiers
  25. )
  26. # U+200D ZWJ is deliberately NOT in EMOJI_RANGES: it is ordinary Indic/Persian orthography
  27. # (क्‍ष is ka + virama + ZWJ + ssa) and only becomes emoji glue *between two emoji bases*.
  28. # Characters that never START a cluster: they bind to whatever stands before them.
  29. _EMOJI_TAIL = frozenset({0x200D, 0xFE0F, 0x20E3} | set(range(0x1F3FB, 0x1F400)))
  30. _EMOJI_REGIONAL = range(0x1F1E6, 0x1F200)
  31. _ZWJ = 0x200D
  32. _VS15 = 0xFE0E # text-presentation selector: "draw this as a character, not as an emoji"
  33. _VS16 = 0xFE0F
  34. _KEYCAP = 0x20E3
  35. _KEYCAP_BASES = frozenset("0123456789#*")
  36. def _cps(*spans) -> "frozenset":
  37. out = set()
  38. for span in spans:
  39. lo, hi = span if isinstance(span, tuple) else (span, span)
  40. out.update(range(lo, hi + 1))
  41. return frozenset(out)
  42. # The text-symbol blocks in EMOJI_RANGES (U+2600-27BF, U+2B00-2BFF, U+1F000-1F0FF) are mostly
  43. # *text*: ♪ U+266A, ✓ U+2713 and ★ U+2605 are not emoji at all in Unicode's emoji-data.txt, and
  44. # ♥ U+2665, ✔ U+2714, ☺ U+263A are emoji only when VS16 asks for the picture -- their default
  45. # presentation is the character, which the caption font draws. Counting the whole block made
  46. # SDH music notes and check marks "emoji", and `--emoji none` (what macOS and a Linux box
  47. # without a colour face get under auto) cut them out of the burned text. In those blocks only
  48. # Emoji_Presentation=Yes characters start a cluster on their own; an Emoji=Yes one does when
  49. # VS16 or a skin-tone modifier follows it. Pinned from emoji-data.txt (Unicode 15.1); the
  50. # pictograph planes keep their block rule.
  51. _EMOJI_PRESENTATION_SYMBOLS = _cps(
  52. (0x2614, 0x2615), (0x2648, 0x2653), 0x267F, 0x2693, 0x26A1, (0x26AA, 0x26AB),
  53. (0x26BD, 0x26BE), (0x26C4, 0x26C5), 0x26CE, 0x26D4, 0x26EA, (0x26F2, 0x26F3), 0x26F5,
  54. 0x26FA, 0x26FD, 0x2705, (0x270A, 0x270B), 0x2728, 0x274C, 0x274E, (0x2753, 0x2755), 0x2757,
  55. (0x2795, 0x2797), 0x27B0, 0x27BF, (0x2B1B, 0x2B1C), 0x2B50, 0x2B55, 0x1F004, 0x1F0CF)
  56. _EMOJI_TEXT_DEFAULT_SYMBOLS = _cps(
  57. (0x2600, 0x2604), 0x260E, 0x2611, 0x2618, 0x261D, 0x2620, (0x2622, 0x2623), 0x2626, 0x262A,
  58. (0x262E, 0x262F), (0x2638, 0x263A), 0x2640, 0x2642, (0x265F, 0x2660), 0x2663,
  59. (0x2665, 0x2666), 0x2668, 0x267B, 0x267E, (0x2692, 0x2697), 0x2699, (0x269B, 0x269C),
  60. 0x26A0, 0x26A7, (0x26B0, 0x26B1), 0x26C8, 0x26CF, 0x26D1, 0x26D3, 0x26E9, (0x26F0, 0x26F1),
  61. 0x26F4, (0x26F7, 0x26F9), 0x2702, (0x2708, 0x2709), (0x270C, 0x270D), 0x270F, 0x2712, 0x2714,
  62. 0x2716, 0x271D, 0x2721, (0x2733, 0x2734), 0x2744, 0x2747, (0x2763, 0x2764), 0x27A1,
  63. (0x2B05, 0x2B07))
  64. _SYMBOL_BLOCKS = ((0x2600, 0x27BF), (0x2B00, 0x2BFF), (0x1F000, 0x1F0FF))
  65. def _in_symbol_block(cp: int) -> bool:
  66. return any(lo <= cp <= hi for lo, hi in _SYMBOL_BLOCKS)
  67. def _is_emoji_char(ch: str) -> bool:
  68. """Is this character drawn as an emoji with nothing after it? In the text-symbol blocks only
  69. the Emoji_Presentation characters are; ♪ ✓ ★ ♥ are text (see _EMOJI_PRESENTATION_SYMBOLS)."""
  70. cp = ord(ch)
  71. if _in_symbol_block(cp):
  72. return cp in _EMOJI_PRESENTATION_SYMBOLS
  73. return any(lo <= cp <= hi for lo, hi in EMOJI_RANGES)
  74. def _can_be_emoji(ch: str) -> bool:
  75. """Emoji=Yes: an emoji by default, or a text-default symbol VS16 can turn into one (♥ ✔)."""
  76. return _is_emoji_char(ch) or ord(ch) in _EMOJI_TEXT_DEFAULT_SYMBOLS
  77. def _is_emoji_base(ch: str) -> bool:
  78. """Can this character START an emoji cluster? Pictographs and regional indicators can;
  79. the joiners and modifiers (ZWJ, VS16, keycap, skin tone) never can -- they only bind to an
  80. emoji base that already stands before them. Without this, a ZWJ or a VS16 sitting after an
  81. ordinary letter turned that letter into "an emoji" and the PNG route replaced it with a gap.
  82. A text-default symbol (♥ ✔ ☝) is a base only with the VS16 or modifier that follows it:
  83. emoji_clusters() checks that, this function answers for the character alone."""
  84. return ord(ch) not in _EMOJI_TAIL and _is_emoji_char(ch)
  85. def _starts_emoji(text: str, i: int) -> bool:
  86. """Does an emoji cluster start at text[i]? A default-emoji base does; a text-default symbol
  87. does only when VS16 or a skin-tone modifier follows it (♥️, ☝🏽), the sequences Unicode
  88. defines as emoji; ♥ alone, and ♪ ✓ ★ always, are characters the font draws."""
  89. ch = text[i]
  90. if _is_emoji_base(ch):
  91. return True
  92. if ord(ch) in _EMOJI_TEXT_DEFAULT_SYMBOLS and i + 1 < len(text):
  93. nxt = ord(text[i + 1])
  94. return nxt == _VS16 or 0x1F3FB <= nxt <= 0x1F3FF
  95. return False
  96. def emoji_clusters(text: str) -> "List[Tuple[int, str]]":
  97. """(index in `text`, cluster) for every emoji in it, ZWJ sequences, VS16, keycaps, flag pairs
  98. and skin-tone modifiers kept together -- 👩‍💻 is one cluster, not three, and 1️⃣ starts at the
  99. digit even though the digit is not itself an emoji character.
  100. A cluster can only START at an emoji base (a pictograph, a regional indicator) or at a keycap
  101. base (`0-9 # *`) that is actually followed by U+20E3. A ZWJ is glue *inside* a cluster, never
  102. a starter and never a tail on its own: `क्‍ष` (Hindi ka + virama + ZWJ + ssa) and `abc‍def`
  103. contain no emoji. A base explicitly marked with U+FE0E (VS15, text presentation) is likewise
  104. not an emoji -- the author asked for the character, not the picture. A text-default symbol
  105. (♥ ✔ ☝) starts one only with the VS16 or skin-tone modifier after it, and a symbol that is
  106. not emoji at all (♪ ✓ ★) never does: the caption font draws those.
  107. """
  108. out: "List[Tuple[int, str]]" = []
  109. i = 0
  110. n = len(text or "")
  111. while i < n:
  112. ch = text[i]
  113. start = i
  114. if _starts_emoji(text, i):
  115. j = i + 1
  116. if j < n and ord(text[j]) == _VS15: # text presentation requested: not an emoji
  117. i = j + 1
  118. continue
  119. elif ch in _KEYCAP_BASES:
  120. j = i + 1
  121. if j < n and ord(text[j]) == _VS16:
  122. j += 1
  123. if not (j < n and ord(text[j]) == _KEYCAP):
  124. i += 1
  125. continue
  126. j += 1
  127. else:
  128. i += 1
  129. continue
  130. # extend: modifiers bind rightwards, a ZWJ only when a real emoji base follows it
  131. while j < n:
  132. cp = ord(text[j])
  133. if cp in (_VS16, _KEYCAP) or 0x1F3FB <= cp <= 0x1F3FF:
  134. j += 1
  135. continue
  136. if cp == _ZWJ and j + 1 < n and ord(text[j + 1]) not in _EMOJI_TAIL \
  137. and _can_be_emoji(text[j + 1]):
  138. # inside a ZWJ sequence a text-default symbol is emoji glue too: 🏃‍♀ ends in ♀
  139. j += 2
  140. continue
  141. if (j == start + 1 and ord(ch) in _EMOJI_REGIONAL and cp in _EMOJI_REGIONAL):
  142. j += 1
  143. continue
  144. break
  145. out.append((start, text[start:j]))
  146. i = j
  147. return out
  148. def has_emoji(text: str) -> bool:
  149. return bool(emoji_clusters(text or ""))
  150. def emoji_codepoint_name(cluster: str) -> str:
  151. """The asset filename stem for a cluster: lowercase hex code points joined by '-', the
  152. Twemoji convention (1f389, 1f469-200d-1f4bb, 1f1ef-1f1f5); emoji_asset_for() also tries Noto's."""
  153. return "-".join(f"{ord(c):x}" for c in cluster)
  154. def _emoji_name_candidates(cluster: str) -> "List[str]":
  155. """Asset stems to try, most specific first: exact, without VS16, without skin tone, the ZWJ
  156. sequence reduced to its first code point, the bare base."""
  157. cps = [ord(c) for c in cluster]
  158. names = [emoji_codepoint_name(cluster)]
  159. def add(seq):
  160. name = "-".join(f"{c:x}" for c in seq)
  161. if name and name not in names:
  162. names.append(name)
  163. add([c for c in cps if c != 0xFE0F])
  164. add([c for c in cps if c != 0xFE0F and not (0x1F3FB <= c <= 0x1F3FF)])
  165. if 0x200D in cps:
  166. add([cps[0]])
  167. add([cps[0]])
  168. return names
  169. def emoji_asset_for(cluster: str, assets_dir: "Optional[str]") -> "Optional[str]":
  170. """The PNG for `cluster` under `assets_dir`, or None when nothing matches."""
  171. if not assets_dir or not os.path.isdir(assets_dir):
  172. return None
  173. for name in _emoji_name_candidates(cluster):
  174. # Twemoji's name (1f469-200d-1f4bb.png, 31-20e3.png), then Noto Emoji's, whose code points
  175. # are at least four hex digits (emoji_u1f469_200d_1f4bb.png, emoji_u0031_20e3.png): the hint
  176. # names noto-emoji/png/128, whose files were never found (#404)
  177. for stem in (name, "emoji_u" + "_".join(cp.zfill(4) for cp in name.split("-"))):
  178. for ext in (".png", ".PNG"):
  179. candidate = os.path.join(assets_dir, stem + ext)
  180. if os.path.isfile(candidate):
  181. return candidate
  182. return None
  183. def emoji_mono_on_colour_note(support: "Dict[str, Any]") -> str:
  184. """--emoji mono on a build whose libass draws the emoji face in colour: nothing here makes that
  185. glyph monochrome (mono and color build the same text), so the result says color (#403: the
  186. warning said "no colour path" beside a detail saying it renders in colour)."""
  187. return ("--emoji mono: libass draws the emoji face in colour on this build and nothing here makes it "
  188. "monochrome, so the emoji are drawn in colour (mode: color). " + str(support.get("detail") or ""))
  189. def emoji_missing_asset_note(missing: "List[str]", support: "Dict[str, Any]") -> str:
  190. """What happens to the clusters --emoji-assets has no PNG for: they stay in the text for libass,
  191. which is said as measured (libass_glyph), not as "drawn by the text font" (#403)."""
  192. names = ", ".join(missing)
  193. if support.get("libass_glyph") is False:
  194. return (f"no PNG in the assets directory for {names} -- those clusters stay in the text, and libass draws no "
  195. "emoji glyph on this build: an empty box at best")
  196. return (f"no PNG in the assets directory for {names} -- those clusters stay in the text for libass to draw "
  197. "with an installed face" + ("" if support.get("libass_glyph") else " (whether one has the glyph was not measured)"))
  198. EMOJI_ASSET_HINT = (
  199. "point --emoji-assets at a directory of PNGs named by code point (1f389.png): "
  200. "twemoji/assets/72x72 (Twemoji, CC-BY 4.0) or noto-emoji/png/128 (Noto Emoji, OFL/Apache-2.0) "
  201. "are the two people already have. The skill has no network at runtime, so the assets must "
  202. "already exist on this machine -- nothing is ever downloaded")
  203. _EMOJI_COLOR_FAMILIES = ("Noto Color Emoji", "Apple Color Emoji", "Segoe UI Emoji")
  204. def _fc_list_path(line: str) -> str:
  205. """The file from one `fc-list ... file` line.
  206. fontconfig prints `<file>: <families>`. The separator is a colon followed by a space.
  207. A Windows drive letter is `C:` followed by a backslash, so splitting on the first colon
  208. yields `C` and the font is reported missing.
  209. """
  210. if ": " in line:
  211. return line.partition(": ")[0].strip()
  212. return line.strip().rstrip(":")
  213. _EMOJI_SUPPORT_CACHE: "Dict[Tuple[Optional[str], bool], Dict[str, Any]]" = {}
  214. def _emoji_color_font() -> "Tuple[Optional[str], Optional[str], bool]":
  215. """(family, file, fontconfig_answered) for the first installed colour emoji family."""
  216. exe = _safe_which("fc-list")
  217. if not exe:
  218. return None, None, False
  219. for family in _EMOJI_COLOR_FAMILIES:
  220. try:
  221. proc = subprocess.run([exe, f":family={family}", "file"], stdout=subprocess.PIPE,
  222. stderr=subprocess.DEVNULL, text=True, encoding="utf-8", errors="replace", timeout=10)
  223. except (subprocess.TimeoutExpired, OSError):
  224. return None, None, False
  225. if proc.returncode != 0:
  226. return None, None, False
  227. for line in proc.stdout.splitlines():
  228. path = _fc_list_path(line)
  229. if path and os.path.exists(path):
  230. return family, path, True
  231. return None, None, True
  232. def _libass_color_probe() -> "Optional[bool]":
  233. """Does THIS ffmpeg render an emoji in colour through libass? Answered by a render, never by
  234. the font listing: Noto Color Emoji installs happily on builds whose freetype/libass has no
  235. colour-bitmap path at all, and those render a monochrome outline instead (measured). ~80 ms.
  236. None means the probe could not be run (no ffmpeg, a failure) -- unknown, not false."""
  237. exe = ffmpeg_exe("ffmpeg")[0]
  238. if not exe:
  239. return None
  240. import tempfile
  241. with tempfile.TemporaryDirectory() as td:
  242. srt = os.path.join(td, "e.srt")
  243. with open(srt, "w", encoding="utf-8") as fh:
  244. fh.write("1\n00:00:00,000 --> 00:00:01,000\n\U0001F389\n")
  245. try:
  246. proc = subprocess.run(
  247. [exe, "-hide_banner", "-loglevel", "error", "-f", "lavfi",
  248. "-i", "color=c=black:s=64x64:d=0.04",
  249. "-vf", "subtitles=" + escape_filter_path(srt, bind=False), "-frames:v", "1",
  250. "-f", "rawvideo", "-pix_fmt", "rgb24", "-"],
  251. stdout=subprocess.PIPE, stderr=subprocess.DEVNULL, timeout=20)
  252. except (subprocess.TimeoutExpired, OSError):
  253. return None
  254. if proc.returncode != 0 or len(proc.stdout) < 64 * 64 * 3:
  255. return None
  256. data = proc.stdout
  257. for i in range(0, 64 * 64 * 3, 3):
  258. r, g, b = data[i], data[i + 1], data[i + 2]
  259. if max(r, g, b) - min(r, g, b) > 40:
  260. return True
  261. return False
  262. def _libass_render(exe: str, char: str, fonts_dir: "Optional[str]" = None) -> "Optional[bytes]":
  263. """One 96x96 rgb24 frame of `char` drawn through libass (subtitles=), or None when it failed."""
  264. import tempfile
  265. with tempfile.TemporaryDirectory() as td:
  266. srt = os.path.join(td, "g.srt")
  267. with open(srt, "w", encoding="utf-8") as fh:
  268. fh.write(f"1\n00:00:00,000 --> 00:00:01,000\n{char}\n")
  269. try:
  270. proc = subprocess.run(
  271. [exe, "-hide_banner", "-loglevel", "error", "-f", "lavfi",
  272. "-i", "color=c=black:s=96x96:d=0.04",
  273. "-vf", "subtitles=" + escape_filter_path(srt, bind=False) + ":force_style='FontSize=64,Alignment=5'"
  274. + (":fontsdir=" + escape_filter_path(fonts_dir, bind=False) if fonts_dir else ""),
  275. "-frames:v", "1", "-f", "rawvideo", "-pix_fmt", "rgb24", "-"],
  276. stdout=subprocess.PIPE, stderr=subprocess.DEVNULL, timeout=20)
  277. except (subprocess.TimeoutExpired, OSError):
  278. return None
  279. if proc.returncode != 0 or len(proc.stdout) < 96 * 96 * 3:
  280. return None
  281. return proc.stdout
  282. def _libass_glyph_probe(fonts_dir: "Optional[str]" = None) -> "Optional[bool]":
  283. """Does libass draw an emoji GLYPH at all, or only what it draws for a character no face has?
  284. "No colour pixels" (_libass_color_probe False) covers both a monochrome glyph and the empty
  285. .notdef box: Apple Color Emoji (an sbix bitmap face) gives libass no usable glyph, and the box
  286. is white (#395). So the emoji is compared with U+10FFFD, a code point no font carries: the same
  287. frame, or nothing lit, is no glyph. A caller's --fonts-dir is searched too, as the burn searches it.
  288. None: the probe could not run."""
  289. exe = ffmpeg_exe("ffmpeg")[0]
  290. if not exe:
  291. return None
  292. emoji, nothing = _libass_render(exe, "\U0001F389", fonts_dir), _libass_render(exe, "\U0010FFFD", fonts_dir)
  293. if emoji is None or nothing is None:
  294. return None
  295. lit = any(emoji[i] > 60 for i in range(0, len(emoji)))
  296. return bool(lit and emoji != nothing)
  297. def emoji_support(assets: "Optional[str]" = None, probe: bool = True, fonts_dir: "Optional[str]" = None) -> "Dict[str, Any]":
  298. """What this machine can actually do with emoji, cached per process.
  299. `mode` is `color` when a render probe proves libass draws colour, else `png` when an assets
  300. directory resolves, else `mono` when some installed face has a glyph at all, else `none`.
  301. An installed colour emoji font proves nothing on its own -- that is why `libass_color` comes
  302. from a render (see references/gotchas.md#emoji). `probe=False` (`contract --json --static`, and every
  303. static/JSON-only path) skips the render entirely and leaves `libass_color` unknown.
  304. """
  305. key = (assets or None, bool(probe), fonts_dir or None)
  306. if key in _EMOJI_SUPPORT_CACHE:
  307. return dict(_EMOJI_SUPPORT_CACHE[key])
  308. family, file, fc_answered = _emoji_color_font()
  309. libass_color = _libass_color_probe() if probe else None
  310. # a colour render is a glyph; without one, ask whether libass draws one at all (#395)
  311. libass_glyph = True if libass_color else (_libass_glyph_probe(fonts_dir) if probe and libass_color is False else None)
  312. assets_dir = assets if (assets and os.path.isdir(assets)) else None
  313. if libass_color:
  314. mode = "color"
  315. elif assets_dir:
  316. mode = "png"
  317. elif family and libass_glyph is False:
  318. mode = "none"
  319. elif family or libass_glyph:
  320. # a face libass draws the emoji from, colour family listed or not (A64, #457: a glyph from
  321. # a monochrome face was answered `none`, against this docstring and libass_glyph: true)
  322. mode = "mono"
  323. elif not fc_answered:
  324. # No fontconfig to ask (a static ffmpeg build, a bare container): the PNG path needs none,
  325. # so the honest answer is png-or-none, never "none because fc-list is missing".
  326. mode = "none"
  327. else:
  328. mode = "none"
  329. if not fc_answered:
  330. detail = "no fontconfig on this machine; the PNG overlay path needs none"
  331. elif libass_color:
  332. detail = f"{family or 'an installed face'} renders in colour through libass on this ffmpeg"
  333. elif family and libass_glyph is False:
  334. detail = (f"{family} installed but libass draws no emoji glyph from it on this build (the render "
  335. "matches a code point no font has: an empty box or nothing)")
  336. elif family and libass_color is False:
  337. detail = f"{family} installed but libass renders it monochrome on this build"
  338. elif family and libass_color is None:
  339. detail = f"{family} installed; the colour render probe was not run"
  340. elif assets_dir:
  341. detail = "no colour emoji family installed; using the PNG assets directory"
  342. elif libass_glyph:
  343. detail = "no colour emoji family installed; libass draws the emoji monochrome from an installed face"
  344. else:
  345. detail = "no colour emoji family installed and no --emoji-assets directory"
  346. result = {"mode": mode, "color_font": family, "color_font_file": file,
  347. "libass_color": libass_color, "libass_glyph": libass_glyph, "assets": assets_dir,
  348. "detail": detail, "fix": EMOJI_ASSET_HINT}
  349. _EMOJI_SUPPORT_CACHE[key] = result
  350. return dict(result)
  351. def resolve_emoji_assets(flag: "Optional[str]" = None, project: "Optional[str]" = None,
  352. brand: "Optional[dict]" = None) -> "Optional[str]":
  353. """--emoji-assets DIR, else the project key, else brand.json, else FFMPEG_SKILL_EMOJI_ASSETS.
  354. A directory that was named but does not exist is a failed job, never a silent downgrade."""
  355. brand = brand or {}
  356. styles = (brand.get("styles") or {}).get("caption") or {}
  357. for value, where in ((flag, "--emoji-assets"), (project, "the project's text.emoji_assets"),
  358. (styles.get("emoji_assets"), "brand.json styles.caption.emoji_assets"),
  359. (brand.get("emoji_assets"), "brand.json emoji_assets"),
  360. (os.environ.get("FFMPEG_SKILL_EMOJI_ASSETS"), "FFMPEG_SKILL_EMOJI_ASSETS")):
  361. if not value:
  362. continue
  363. if not os.path.isdir(str(value)):
  364. die(f"{where}: {value} is not a readable directory -- {EMOJI_ASSET_HINT}", kind="input")
  365. return str(value)
  366. return None
  367. def emoji_filter_chain(plan, base_label, out_label, first_input=1, to_yuv="", base_chain="", overlay_opts=""):
  368. """(chains, inputs) that composite the planned PNGs on top of `base_label`.
  369. `inputs` is a list of argv fragments, each ending in the asset path, to be appended to the
  370. ffmpeg command in order (an overlay that fades needs `-loop 1` on its input so the still has
  371. a timeline the fade filter can move along; one that does not is a plain `-i`).
  372. `to_yuv` ends each PNG's chain: the caller's rgb_chain_to_bt709(meta), so the RGBA asset is
  373. converted to the BT.709 the output is tagged with instead of by overlay's BT.601 default;
  374. `base_chain` (label_chain(meta, meta): an untagged source's BT.709 label) goes in front of
  375. the first overlay, so 7.1+ does not convert the two pictures to one another.
  376. `overlay_opts` ends each overlay's options: the caller's overlay_format(meta), so a 10-bit
  377. picture is not composited at overlay's 8-bit default.
  378. """
  379. overlays = plan.get("overlays") or []
  380. if not overlays:
  381. return [], []
  382. # Group by everything that makes two uses of the same PNG a different STREAM: the fade is
  383. # expressed in the cue's own timeline, so two cues cannot share one faded input.
  384. def _key(o):
  385. fades = (round(float(o.get("fade_in") or 0.0), 3), round(float(o.get("fade_out") or 0.0), 3))
  386. window = (round(float(o["start"]), 3), round(float(o["end"]), 3)) if any(fades) else (None, None)
  387. return (o["asset"], o["box"]) + fades + window
  388. groups: "List[Tuple]" = []
  389. for o in overlays:
  390. if _key(o) not in groups:
  391. groups.append(_key(o))
  392. chains: List[str] = []
  393. inputs: "List[List[str]]" = []
  394. pads: "Dict[Tuple, List[str]]" = {}
  395. for k, key in enumerate(groups):
  396. asset, box, fin, fout, gstart, gend = key
  397. uses = [o for o in overlays if _key(o) == key]
  398. idx = first_input + k
  399. labels = [f"e{k}_{j}" for j in range(len(uses))]
  400. chain = f"[{idx}:v]format=rgba,scale={box}:{box}"
  401. if fin or fout:
  402. # -loop 1 gives the still an advancing timeline on the SAME clock as the main video,
  403. # so the fade times below are the cue's own seconds. The emoji then appears and
  404. # leaves with the text instead of popping in against a fading line.
  405. # -t bounds the loop at the cue's end: an unbounded looped still never EOFs and the
  406. # whole encode hangs (overlay keeps pulling from it after the main video is done).
  407. inputs.append(["-loop", "1", "-t", f"{gend:.3f}", "-i", asset])
  408. if fin:
  409. chain += f",fade=t=in:st={gstart:.3f}:d={fin:.3f}:alpha=1"
  410. if fout:
  411. chain += f",fade=t=out:st={max(gstart, gend - fout):.3f}:d={fout:.3f}:alpha=1"
  412. else:
  413. inputs.append(["-i", asset])
  414. chain += to_yuv
  415. if len(labels) > 1:
  416. chain += f",split={len(labels)}"
  417. chains.append(chain + "".join(f"[{l}]" for l in labels))
  418. pads[key] = labels
  419. cur = base_label
  420. if base_chain:
  421. chains.append(f"[{cur}]{base_chain}null[ebase]")
  422. cur = "ebase"
  423. remaining = {key: list(v) for key, v in pads.items()}
  424. for j, o in enumerate(overlays):
  425. label = remaining[_key(o)].pop(0)
  426. nxt = out_label if j == len(overlays) - 1 else f"eov{j}"
  427. x = o["x"]
  428. x = f"'{x}'" if isinstance(x, str) else x
  429. # No eof_action=pass here: a PNG input is a SINGLE frame at pts 0, and eof_action=pass
  430. # switches off overlay's default "hold the last frame of the secondary input", so the
  431. # asset would be composited on frame 0 only and vanish for the rest of the cue (that is
  432. # exactly what shipped first). eof_action=repeat (the default) holds the still for the
  433. # whole timeline; enable= is what confines it to the cue's window.
  434. chains.append(f"[{cur}][{label}]overlay=x={x}:y={o['y']}:"
  435. f"enable='between(t,{o['start']:.3f},{o['end']:.3f})'{overlay_opts}[{nxt}]")
  436. cur = nxt
  437. return chains, inputs

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.