# hitomi-downloader

A small Go command-line downloader for `hitomi.moe` galleries.

The tool fetches gallery metadata from `https://hitomi.moe/reader/info/<id>`,
writes it to `info.json`, and downloads all gallery images with concurrent
workers.

## Requirements

- Go 1.26 or newer
- Network access to `hitomi.moe`

## Build

```sh
go build -o hitomi-downloader .
```

## Usage

Download by gallery URL:

```sh
go run . -url https://hitomi.moe/g/123456
```

Download by gallery ID:

```sh
go run . -url 123456
```

The gallery identifier can also be provided as a positional argument:

```sh
go run . 123456
```

After building:

```sh
./hitomi-downloader -url 123456 -out downloads
```

## Options

```text
-url string
    Hitomi gallery URL or ID.

-out string
    Output directory. Defaults to "downloads".

-workers int
    Number of concurrent download workers. Defaults to a value based on CPU
    count, with a minimum of 2 and a maximum of 8.

-timeout int
    Per-request timeout in seconds. Defaults to 30.

-skip-existing bool
    Skip existing non-empty files. Defaults to true.
    Use -skip-existing=false to force re-downloads.

-metadata-only bool
    Fetch metadata and write info.json without downloading images.
```

## Output

Files are written under:

```text
<out>/<gallery-id>_<sanitized-title>/
```

Each gallery directory contains:

```text
info.json
001.<ext>
002.<ext>
003.<ext>
...
```

Images are first downloaded to a temporary `.part` file and then renamed into
place when the download completes successfully.

## Examples

Fetch metadata only:

```sh
go run . -url 123456 -metadata-only
```

Use more workers and a longer timeout:

```sh
go run . -url 123456 -workers 12 -timeout 60
```

Force re-downloading existing files:

```sh
go run . -url 123456 -skip-existing=false
```

## Notes

Use this tool responsibly and follow the terms and rules of the site you are
accessing. Very high worker counts can increase load on the remote server and
may make downloads less reliable.