Wordly

Desktop wizard for church media teams: download a Facebook sermon, trim multiple highlights, pick a Bible verse and instrumental by theme, and export a Filmora 14.2.9 .wfp project with separate editable tracks.

Workflow

1. Download — paste a Facebook URL (yt-dlp with 16 parallel fragments)
2. Timestamps — add one or more start/end ranges
3. Preview — scrub each segment
4. Trim & join — FFmpeg trims and concatenates highlights
5. Bible verse — enter a theme; AI suggests verses (offline fallback without API key)
6. Instrumental — AI suggests beds; download the one you pick
7. Layers — review sermon / verse / music tracks
8. Project name
9. Export — writes exports/<name>.wfp for Filmora 14.2.9

Requirements

• Python 3.12+
• FFmpeg / ffprobe on PATH
• Filmora 14.2.9 to open the exported project
• Optional: OPENAI_API_KEY (smarter verse/music suggestions)

Install FFmpeg

• Windows: ffmpeg.org or choco install ffmpeg
• macOS: brew install ffmpeg
• Linux / WSL: sudo apt install ffmpeg

Install

From the project root:

git clone <repo-url> wordly
cd wordly
python3 -m venv .venv

Linux / macOS / WSL:

source .venv/bin/activate
pip install -r requirements.txt

Windows (PowerShell):

.venv\Scripts\Activate.ps1
pip install -r requirements.txt

If activation fails with “running scripts is disabled on this system”, either allow scripts for your user (one time):

Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope CurrentUser

or skip activation and call the venv Python directly (works for install and run):

.\.venv\Scripts\python.exe -m pip install -r requirements.txt

Windows (cmd):

.venv\Scripts\activate.bat
pip install -r requirements.txt

Run

Activate the virtual environment (if you use one), then:

python main.py

Windows without activating:

cd path\to\wordly
.\.venv\Scripts\python.exe main.py

If your system only has python3:

python3 main.py

On first launch, Wordly creates downloads/, clips/, exports/, and temp/ if they are missing.

Optional environment variables

export OPENAI_API_KEY=sk-...          # smarter verse / instrumental suggestions
export WORDLY_OPENAI_MODEL=gpt-4o-mini  # optional model override

Project layout

wordly/
├── main.py
├── models/              # wizard project state
├── services/            # download, trim, AI, Filmora .wfp export
├── ui/                  # wizard_window, preview_player
├── utils/
├── assets/
│   └── filmora_templates/
│       ├── sermon-highlights.wfp   # GUI layout reference (preferred)
│       ├── video.mp4, music.mp3, image.jpg  # media for that template
│       └── reference_media/        # tiny placeholders if bundle media missing
├── downloads/           # sermon sources
├── clips/               # trimmed / joined segments
├── exports/             # one folder per project: .wfp + media/ + verse.txt
└── tests/

Filmora export

Wordly clones assets/filmora_templates/sermon-highlights.wfp (or filmora_14_2_9.wfp) and patches paths, hashes, and timeline trims for your sermon segments, joined highlights, instrumental, and verse. Media paths are written for Windows Filmora (C:\... or \\wsl$\... when running in WSL).

Important

• Use Filmora 14.2.9 (same build as the template).
• In the wizard, click Finish (or Generate .wfp). Wordly writes exports/<project-name>/<project-name>.wfp and copies your joined clip, music, and verse text into exports/<project-name>/media/. Open the .wfp from that folder in Filmora — keep the media/ folder beside it.
• The timeline keeps Filmora’s native layer stack: title (Bible verse), mirrored sermon segments on two video tracks (one muted like your GSM template), joined highlights reel, and instrumental audio.
• Put your working Filmora project in assets/filmora_templates/ as **sermon-highlights.wfp** with **video.mp4**, **music.mp3**, and **image.jpg** in the same folder (paths inside the .wfp should point at those files on this PC).
• After updating the template, run once: python tools/sanitize_filmora_template.py (rewrites leftover Downloads/GSM paths to the bundle media).
• Do not open the template .wfp for editing in Filmora from Wordly’s export step — use Finish and open the copy under exports/<project>/.

To repoint the bundled template at local placeholder media (only if you are not using your own GSM project as the template):

.\.venv\Scripts\python.exe tools\sanitize_filmora_template.py

Check an export for leftover template paths:

.\.venv\Scripts\python.exe tools\scan_wfp_refs.py exports\your-project.wfp
.\.venv\Scripts\python.exe tools\verify_export.py exports\your-project.wfp

A clean export should report timeline JSON: OK and no Facebook_1 / D:/ hits from scan_wfp_refs.py.

Tests

.\.venv\Scripts\python.exe -m unittest discover -s tests -p "test_*.py" -v

On Linux / macOS / WSL with the venv activated:

python -m unittest discover -s tests -p 'test_*.py' -v