For legacy documentation, please check the Open Library Archives. Most Bot code now lives at https://github.com/internetarchive/openlibrary-bots#openlibrary-bots
Are you frustrated with the number of metadata mistakes on OpenLibrary editions, works, authors, etc? Do you notice the same problem happening over and over? Are you seeing a lot of duplicate Authors which should be merged or fake books which should be deleted? Are there books missing from OpenLibrary that you'd like to import from other sources? These are jobs for
bot accounts are used to make http POSTs to work, author, edition, subject, etc., pages in order to correct metadata or create/import new records (e.g. books, authors, bookcovers) in bulk.
Applying for a
Follow these 2 steps to apply for a privileged
- First, register a new OpenLibrary account which meets the following guidelines: (a) Your
bot accountshould be different from your personal OpenLibrary account and (b) the username of your new account should end with the word "Bot" (e.g.
ImportBot). These conventions allow us to monitor / filter through
bots to identify only those changes made by
bot accounts. (Since bots tend to repeat the same small operation with high frequency, they would overwhelm the list if shown alongside edits by humans.)
- Open a github issue and ask a site
admin(e.g. @mekarpeles or @hornc as of 2019) to set your account to have
botprivileges and to add your account to the
Getting Started: Rules
Question Great, I now have a bot account, I can start fixing thousands of problems I've noticed, right?
Answer Hold on! After you have been granted bot access, please do not run a bot script to change metadata in bulk (for more than 100 records) until it has first been reviewed by charles (@hornc on github). The right process is to create a new directory for your bot within the https://github.com/internetarchive/openlibrary-bots repository, to open a PR, and to add charles (@hornc) or myself (@mekarpeles) as the reviewer.
Question I have a metadata question about how my bot should work, a question about using the openlibrary-client, or I need a code review for my bot. Who do I ask?
Answer @hornc (@charles on slack) is our lead on metadata and is a great person to answer questions about the
openlibrary-client and writing/registering
bots to fix metadata or add new books to our catalog
Question I have written a bot, am I ready to run it?
Answer If you have written a bot script and think it's ready to be run, the correct procedure is to fork the https://github.com/internetarchive/openlibrary-bots repository, create a new branch for your bot, add a directory in the project named according to the action your bot will perform, and then submit a PR (pull request) for review. Please don't run bulk modifications until @horn or @mekarpeles have reviewed and approved your script or your bot privileges may be revoked.
Question My bot reads metadata from a source file and then modifies records on Open Library -- should these source data files be saved somewhere?
Answer Yes please -- If you are writing a bot to add new or updated metadata to Open Library from a file(s), those files should also be committed with your script to https://github.com/internetarchive/openlibrary-bots
Instead of making POSTs to API endpoints directly using a
bot account, the OpenLibrary community has created an official python client library called
openlibrary-client which streamlines the process of making metadata updates and writing
Bots used to be written using the http://github.com/internetarchive/openlibrary/blob/master/openlibrary/api.py library, but this has been deprecated in favor of the much easier (and safer)
Bot account Etiquette
bot accounts should not be used directly to make 1-off changes, e.g. from the command line. Changes instead should be coded into
bot scripts (i.e.
bots) and checked in to the openlibrary-bots repository. This way we can look back in time and see/figure out what exact changes were made (so we can debug if something goes wrong or if we have to modify or update and re-apply the logic).
The following are legacy examples which used
api.py (which is deprecated) instead of
openlibrary-client, but they give an idea of what OpenLibrary
bots are, what they do, how they are written, and how they are used. Please refer to the openlibrary-client documentation for more modern examples:
Ideas for new bots