A web bot to crawl websites and scrape images.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.



This bot (image scraper) crawls a given url(s) and downloads all the images.


  • Supported platforms: Linux / Windows / Python 2.7.
  • Maintains a database of all downloaded images to avoid duplicate downloads.
  • Optionally, it can scrape only under a particular url, e.g. scraping http://website.com/albums/new with this option will only download from new album.
  • Filters urls by regex.
  • Filters images by minimum size.
  • Scrapes through javascript popup links (limited support).
  • Live monitor window for displaying images as they are scraped.
  • Asynchronous i/o design using scrapy and twisted.


crawl command:

  • Scrape images:

    imagebot crawl http://website.com
    imagebot crawl http://website.com,http://otherwebsite.com
  • Options for crawl command:

    -d, --domains

    Scrape images while allowing images to be downloaded from other domain(s) (add multiple domains with comma separated list). The domain in the start url(s) is(are) allowed by default.

    imagebot crawl http://website.com -d otherwebsite.com,anotherwebsite.com

    -is, --images-store

    Specify image store location. Default: ~/Pictures/crawled/[jobname]

    imagebot crawl http://website.com -is /home/images

    -s, --min-size

    Specify minimum size of image to be downloaded (width x height).

    imagebot crawl http://website.com -s 300x300

    -u, --stay-under

    Stay under the start url. Only those urls that have the start url as prefix will be crawled. Useful, for example, to crawl an album or a subsection on a website.

    imagebot crawl http://website.com/albums/new -u

    -m, --monitor

    Launch monitor window for displaying images as they are scraped.

    imagebot crawl http://website.com -m

    -a, --user-agent

    Set user-agent string. Default: imagebot. It is recommended to change it to identify your bot as a matter of responsible crawling.

    imagebot crawl http://website.com -a "my_imagebot(http://mysite.com)"

    -r, --url-regex

    Specify regex for urls. Only those urls matching the regex will be crawled. It does not apply to start url(s).

    imagebot crawl http://website.com -r .*gallery.*

    -dl, --depth-limit

    Specify depth limit for crawling. Use value of 0 to scrape only on start url(s).

    imagebot crawl http://website.com -dl 2


    A list of well known cdn's is included and enabled by default for image downloads. Use this option to disable it.

    -at, --auto-throttle

    Enable auto throttle feature of scrapy. (details in scrapy docs).

    -j, --jobname

    Specify a job name. This will be used to store image meta data in the database. By default, domain name of the start url is used as the job name.

    -nc, --no-cache

    Disable http caching.

    -l, --log-level

    Specify log level. Supported levels: info, silent, critical, error, debug, warning. Default: error.

    imagebot crawl http://website.com -l debug

    -h, --help

    Get help on crawl command options.

clear command:

  • This command is useful for various kinds of cleanup.

  • Options for clear command:


    Clear http cache.


    Remove image metadata for a job from the database.

    imagebot clear --db website.com


    Multiple copies of same image may be downloaded due to different urls. Use this option to delete duplicate images for a job.

    imagebot clear --duplicate-images website.com

    -h, --help

    Get help on clear command options.


  1. pywin32 (http://sourceforge.net/projects/pywin32/)

    Needed on Windows.

  2. python-gi (Python GObject Introspection API)

    Needed on Linux for gtk UI. (Optional). If not found, python built-in Tkinter will be used. On Ubuntu: apt-get install python-gi

  3. scrapy (web crawling framework)

    It will be automatically installed by pip.

  4. Pillow (Python Imaging Library)

    It will be automatically installed by pip.