# Batch runner for Generic Plex Salvage

Keep `batch.sh` in the same directory as:

```text
scan.sh
recover.sh
salvage.py
```

The batch runner reads one folder per line and calls the existing generic
scanner or recovery script for each item.

## Absolute paths

Example `priority.txt`:

```text
# Family priority recovery list
/Media/TV/SEAL Team
/Media/TV/Doctor Who
/Media/Movies/Alien
/Media/Movies/Aliens
```

Run:

```bash
./batch.sh scan priority.txt
./batch.sh recover priority.txt
```

## Relative paths with a common base

Example `priority.txt`:

```text
TV/SEAL Team
TV/Doctor Who
Movies/Alien
Movies/Aliens
```

Run:

```bash
./batch.sh scan priority.txt --base '/Media'
./batch.sh recover priority.txt --base '/Media'
```

## Behaviour

- blank lines are ignored;
- lines beginning with `#` are comments;
- Windows CRLF text files are accepted;
- leading/trailing whitespace is removed;
- duplicate resolved paths are skipped;
- spaces in folder names are handled;
- folders are processed sequentially;
- one failed or metadata-dead folder does not stop later items;
- the batch exits nonzero after finishing if any requested folder failed.

The underlying `scan.sh` / `recover.sh` safety rules are unchanged. Recovery
still processes only files classified `SAFE`.

## Logging

When `/mnt/recovery` is writable, batch logs and summaries are written there:

```text
batch-scan-priority-YYYYMMDD-HHMMSS.log
batch-scan-priority-YYYYMMDD-HHMMSS-summary.txt
batch-recover-priority-YYYYMMDD-HHMMSS.log
batch-recover-priority-YYYYMMDD-HHMMSS-summary.txt
```

Each source folder still gets its own normal manifest, scan summary, recovery
TSV, and recovery summary from the generic salvage scripts.

## Recommended workflow

Scan the whole priority list first:

```bash
./batch.sh scan priority.txt --base '/Media'
```

Review the resulting per-folder summaries and available recovery-disk space,
then recover the same list:

```bash
./batch.sh recover priority.txt --base '/Media'
```

Recovery is restartable. Existing destination files whose sizes already match
their source inode are skipped by `recover.sh`.