ao3 0.1.2

A Python API for scraping AO3 (the Archive of Our Own)

This Python package provides a scripted interface to some of the data on AO3 (the Archive of Our Own).

It is not an official API.

Motivation

I want to be able to write Python scripts that use data from AO3.

An official API for AO3 data has been on the roadmap for a couple of years. Until that appears, I’ve cobbled together my own page-scraping code that does the job. It’s a bit messy and fragile, but it seems to work most of the time.

If/when we get the proper API, I’d drop this in a heartbeat and do it properly.

Installation

Install using pip:

$ pip install ao3

I’m trying to support Python 2.7, Python 3.3+ and PyPy.

Usage

Create an API instance:

>>> from ao3 import AO3
>>> api = AO3()

Looking up information about a work

Getting a work:

>>> work = api.work(id='258626')

The id is the numeric portion of the URL. For example, the work ID of https://archiveofourown.org/works/258626 is 258626.

Get a URL:

>>> work.url
'https://archiveofourown.org/works/258626'

You can then look up a number of attributes, similar to the Stats panel at the top of a page. Here’s the full set you can look up:

>>> work.title
'The Morning After'

>>> work.author
'ambyr'

>>> work.summary
"<p>Delicious just can't understand why it's the shy, quiet ones who get all the girls.</p>"

>>> work.rating
['Teen And Up Audiences']

>>> work.warnings
[]

(An empty list is synonymous with “No Archive Warnings”, so that it’s a falsey value.)

>>> work.category
['F/M']

>>> work.fandoms
['Anthropomorfic - Fandom']

>>> work.relationship
['Pinboard/Fandom']

>>> work.characters
['Pinboard', 'Delicious - Character', 'Diigo - Character']

>>> work.additional_tags
['crackfic', 'Meta', 'so very not my usual thing']

>>> work.language
'English'

>>> work.published
datetime.date(2011, 9, 29)

>>> work.words
605

>>> work.comments
122

>>> work.kudos
1238

>>> for name in work.kudos_left_by:
...     print(name)
...
winterbelles
AnonEhouse
SailAweigh
# and so on

>>> work.bookmarks
99

>>> work.hits
43037

There’s also a method for dumping all the information about a work into JSON, for easy export/passing into other places:

>>> work.json()
'{"rating": ["Teen And Up Audiences"], "fandoms": ["Anthropomorfic - Fandom"], "characters": ["Pinboard", "Delicious - Character", "Diigo - Character"], "language": "English", "additional_tags": ["crackfic", "Meta", "so very not my usual thing"], "warnings": [], "id": "258626", "stats": {"hits": 43037, "words": 605, "bookmarks": 99, "comments": 122, "published": "2011-09-29", "kudos": 1238}, "author": "ambyr", "category": ["F/M"], "title": "The Morning After", "relationship": ["Pinboard/Fandom"], "summary": "<p>Delicious just can\'t understand why it\'s the shy, quiet ones who get all the girls.</p>"}'

Looking up your account

If you have an account on AO3, you can log in to access pages that aren’t available to the public:

>>> api.login('username', 'password')

Currently there’s only one thing you can do with this: if you have Viewing History enabled, you can get a list of work IDs from that history, like so:

>>> for work_id in api.user.reading_history():
...     print(work_id)
...
'123'
'456'
'789'
# and so on

You can enable Viewing History in the settings pane.

One interesting side effect of this is that you can use it to get a list of works where you’ve left kudos:

from ao3 import AO3
from ao3.works import RestrictedWork

api = AO3()
api.login('username', 'password')

for work_id in api.user.reading_history():
    try:
        work = api.work(id=work_id)
    except RestrictedWork:
        continue
    print(work.url + '... ', end='')
    if api.user.username in work.kudos_left_by:
        print('yes')
    else:
        print('no')

Warning: this is very slow. It has to go back and load a page for everything you’ve ever read. Don’t use this if you’re on a connection with limited bandwidth.

This doesn’t include “restricted” works – works that require you to be a logged-in user to see them.

(The reading page tells you when you last read something. If you cached the results, and then subsequent runs only rechecked fics you’d read since the last run, you could make this quite efficient. Exercise for the reader.)

License

The project is licensed under the MIT license.