Updated Metadata Extraction (markdown)

Gramps
2025-08-05 22:03:09 -07:00
parent a25280cc9e
commit e57e5b82fa
2 changed files with 51 additions and 0 deletions

51
Metadata-Extraction.md Normal file

@ -0,0 +1,51 @@
## 🔍 Metadata Extraction
This page describes how the LCS Pipeline reads and interprets per-clip metadata, especially from `notes.json` files used to guide title generation, overlays, and social copy.
---
### 📂 Location
For each clip file, the system expects a corresponding file named:
```
<clipname>.notes.json
```
This file should reside in the same directory as the clip (`hits/`, `misses/`, `outtakes/`, or `montages/`).
---
### 📄 File Format
Each `notes.json` file contains:
| Field | Required? | Description |
|-------------|-----------|-------------|
| `highlight` | ✅ Yes | Short, vivid sentence summarizing the moment |
| `tags` | Optional | List of keywords (used in social media metadata) |
| `gag_name` | Optional | Alternative title, meme reference, or inside joke |
---
### 🔄 Fallback Behavior
- If no `notes.json` is found:
- For `hits/`, `misses/`, or `outtakes/`: the user is prompted to enter one interactively.
- For `montages/`: the system attempts to auto-title from the filename.
---
### 🧪 Example
```json
{
"highlight": "Gramps sticks a 3-elim clutch with only 2 HP left!",
"tags": ["Fortnite", "Clutch", "ZeroBuild"],
"gag_name": "The Wrinkle Reaper"
}
```
---
Next page: [`Metadata Format`](./Metadata-Format)