ghibli-gallery-crawler
Image downloader in Go (originally written in Python and JavaScript in 2020-08-31).
Install
go build -o ghibli-gallery-crawler .
Or run without building: go run . <command>.
Usage
Two commands, one per behavior carried over from the original scripts.
gallery — numbered ghibli.jp galleries
Downloads <film>001.jpg … <film>050.jpg for each film, into one directory per film:
# All 8 default films into ./marnie, ./kaguyahime, ...
ghibli-gallery-crawler gallery
# Pick films, image count, and output location
ghibli-gallery-crawler gallery -films ponyo,chihiro -count 50 -out ./images
| Flag | Default | Meaning |
|---|---|---|
-films |
the 8 films below | comma-separated film slugs |
-count |
50 |
highest image number tried per film |
-out |
. |
where the per-film directories are created |
-base |
https://www.ghibli.jp/gallery/ |
base gallery URL |
-concurrency |
8 |
parallel downloads |
-timeout |
30s |
per-request timeout |
-user-agent |
tool identifier | User-Agent header to send |
Default films: marnie, kaguyahime, kazetachinu, kokurikozaka, karigurashi, ponyo, ged, chihiro.
Image numbers are a fixed range, so films with fewer than -count images are normal — those are reported as missing, not as failures.
scrape — every image on a page
Parses a page's HTML, collects every <img src>, resolves relative URLs, and downloads them:
# Saves into ./en.wikipedia.org (the URL's host)
ghibli-gallery-crawler scrape https://en.wikipedia.org/wiki/Studio_Ghibli
# Explicit output directory
ghibli-gallery-crawler scrape https://example.com -path ./images
Flags: -path (default: the URL's host), plus -concurrency, -timeout, and -user-agent as above.
Query strings are stripped from image URLs so that names like /hsts-pixel.gif?c=3.2.5 produce clean file names. data: URIs and duplicates are skipped.
Output behavior
Images are named after the last path segment of their URL. Existing files with the same name are overwritten. Missing directories are created. Non-200 responses are never written to disk, so error pages cannot land in your output as .jpg files.
Note that ghibli.jp serves gallery images to any client but returns 403 for its HTML pages unless the request looks like a browser, so gallery works there while scrape does not.
License
Apache-2.0