genanki
allows you to programatically generate decks in Python 3 for Anki, a popular spaced-repetition flashcard
program. Please see below for concepts and usage.
This library and its author(s) are not affiliated/associated with the main Anki project in any way.
git clone [email protected]:zhaisilong/genanki.git
cd genanki
pip install -e .
csv2apkg -h
csv2apkg test_single.csv --mode=single # with single col csv without header
csv2apkg test.csv # csv with question and answer as header
The basic unit in Anki is the Note
, which contains a fact to memorize. Note
s correspond to one or more Card
s.
Here's how you create a Note
:
my_note = genanki.Note(
model=my_model,
fields=['Capital of Argentina', 'Buenos Aires'])
You pass in a Model
, discussed below, and a set of fields
(encoded as HTML).
A Model
defines the fields and cards for a type of Note
. For example:
my_model = genanki.Model(
1607392319,
'Simple Model',
fields=[
{'name': 'Question'},
{'name': 'Answer'},
],
templates=[
{
'name': 'Card 1',
'qfmt': '{{Question}}',
'afmt': '{{FrontSide}}<hr id="answer">{{Answer}}',
},
])
This note-type has two fields and one card. The card displays the Question
field on the front and the Question
and
Answer
fields on the back, separated by a <hr>
. You can also pass a css
argument to Model()
to supply custom
CSS.
You need to pass a model_id
so that Anki can keep track of your model. It's important that you use a unique model_id
for each Model
you define. Use import random; random.randrange(1 << 30, 1 << 31)
to generate a suitable model_id, and hardcode it
into your Model
definition. You can print one at the command line with
python3 -c "import random; print(random.randrange(1 << 30, 1 << 31))"
To import your notes into Anki, you need to add them to a Deck
:
my_deck = genanki.Deck(
2059400110,
'Country Capitals')
my_deck.add_note(my_note)
Once again, you need a unique deck_id
that you should generate once and then hardcode into your .py
file.
Then, create a Package
for your Deck
and write it to a file:
genanki.Package(my_deck).write_to_file('output.apkg')
You can then load output.apkg
into Anki using File -> Import...
To add sounds or images, set the media_files
attribute on your Package
:
my_package = genanki.Package(my_deck)
my_package.media_files = ['sound.mp3', 'images/image.jpg']
media_files
should have the path (relative or absolute) to each file. To use them in notes, first add a field to your model, and reference that field in your template:
my_model = genanki.Model(
1091735104,
'Simple Model with Media',
fields=[
{'name': 'Question'},
{'name': 'Answer'},
{'name': 'MyMedia'}, # ADD THIS
],
templates=[
{
'name': 'Card 1',
'qfmt': '{{Question}}<br>{{MyMedia}}', # AND THIS
'afmt': '{{FrontSide}}<hr id="answer">{{Answer}}',
},
])
Then, set the MyMedia
field on your card to [sound:sound.mp3]
for audio and <img src="image.jpg">
for images.
You cannot put <img src="{MyMedia}">
in the template and image.jpg
in the field. See these sections in the Anki manual for more information: Importing Media and Media & LaTeX.
You should only put the filename (aka basename) and not the full path in the field; <img src="images/image.jpg">
will not work. Media files should have unique filenames.
Note
s have a guid
property that uniquely identifies the note. If you import a new note that has the same GUID as an
existing note, the new note will overwrite the old one (as long as their models have the same fields).
This is an important feature if you want to be able to tweak the design/content of your notes, regenerate your deck, and import the updated version into Anki. Your notes need to have stable GUIDs in order for the new note to replace the existing one.
By default, the GUID is a hash of all the field values. This may not be desirable if, for example, you add a new field with additional info that doesn't change the identity of the note. You can create a custom GUID implementation to hash only the fields that identify the note:
class MyNote(genanki.Note):
@property
def guid(self):
return genanki.guid_for(self.fields[0], self.fields[1])
Anki has a value for each Note
called the sort_field
. Anki uses this value to sort the cards in the Browse
interface. Anki also is happier if you avoid having two notes with the same sort_field
, although this isn't strictly
necessary. By default, the sort_field
is the first field, but you can change it by passing sort_field=
to Note()
or implementing sort_field
as a property in a subclass (similar to guid
).
You can also pass sort_field_index=
to Model()
to change the sort field. 0
means the first field in the Note, 1
means the second, etc.
You can create your template definitions in the YAML format and pass them as a str
to Model()
. You can also do this
for fields.
genanki
supports adding generated notes to the local collection when running inside an Anki 2.1 addon (Anki 2.0
may work but has not been tested). See the .write_to_collection_from_addon() method
.
Due to a mistake, in genanki versions before 0.13.0, builtin_models.CLOZE_MODEL
only had a single field, whereas the real Cloze model that is built into Anki has two fields. If you get a DeprecationWarning
when using CLOZE_MODEL
, simply add another field (it can be an empty string) when creating your Note
, e.g.
my_note = genanki.Note(
model=genanki.CLOZE_MODEL,
fields=['{{c1::Rome}} is the capital of {{c2::Italy}}', ''])
If fields in your notes contain literal <
, >
, or &
characters, you need to HTML-encode them: field data is HTML, not plain text. You can use the html.escape
function.
For example, you should write
fields=['AT&T was originally called', 'Bell Telephone Company']
or
fields=[html.escape(f) for f in ['AT&T was originally called', 'Bell Telephone Company']]
This applies even if the content is LaTeX; for example, you should write
fields=['Piketty calls this the "central contradiction of capitalism".', '[latex]r > g[/latex]']
If your name is Kerrick, you can publish the genanki
package to PyPI by running these commands from the root of the genanki
repo:
rm -rf dist/*
python3 setup.py sdist bdist_wheel
python3 -m twine upload dist/*
Note that this directly uploads to prod PyPI and skips uploading to test PyPI.