- should never skip recording graves, for the sake of merging
- 1.0 upgrade will fail on decks that have the same fact creation date. need
to work around this in the future
The approach of using incrementing id numbers works for syncing if we assume
the server is canonical and all other clients rewrite their ids as necessary,
but upon reflection it is not sufficient for merging decks in general, as we
have no way of knowing whether objects with the same id are actually the same
or not. So we need some way of uniquely identifying the object.
One approach would be to go back to Anki 1.0's random 64bit numbers, but as
outlined in a previous commit such large numbers can't be handled easy in some
languages like Javascript, and they tend to be fragmented on disk which
impacts performance. It's much better if we can keep content added at the same
time in the same place on disk, so that operations like syncing which are mainly
interested in newly added content can run faster.
Another approach is to add a separate column containing the unique id, which
is what Mnemosyne 2.0 will be doing. Unfortunately it means adding an index
for that column, leading to slower inserts and larger deck files. And if the
current sequential ids are kept, a bunch of code needs to be kept to ensure ids
don't conflict when merging.
To address the above, the plan is to use a millisecond timestamp as the id.
This ensures disk order reflects creation order, allows us to merge the id and
crt columns, avoids the need for a separate index, and saves us from worrying
about rewriting ids. There is of course a small chance that the objects to be
merged were created at exactly the same time, but this is extremely unlikely.
This commit changes models. Other objects will follow.
the initial plan was to zero the creation time and leave the cards/facts there
until we have a chance to garbage collect them on a schema change, but such an
approach won't work with deck subscriptions
- use negative numbers to denote second intervals
- record the rev ivl when leaving lrn queue
- improve revlog upgrade
- don't truncate precision when recording time taken
reps should now be equal to the number of entries in the revlog, and only
exists so that we can order by review count in the browser efficiently
streak is no longer necessary as we have a learn queue now
Previously cloze deletions were handled by copying the contents of one field
into another and applying transforms to it. This had a number of problems:
- after you add a card, you can't undo the cloze deletion
- if you spot a mistake, you have to edit it twice (or more if you have more
than one cloze for a sentence)
- making multiple clozes requires copying & pasting the sentence multiple
times
- this also lead to much bigger decks if the sentences being cloze-deleted are
large
- related clozes can't be spaced apart as siblings
To address these issues, we introduce the idea of cloze tags in the card
template and fields. If the template has the text:
{{cloze:1:field}}
And a field has the following contents:
{{c1::hello}}
Then the template will automatically replace that part of the text with either
occluded text, or a highlighted answer. All other clozes in the field are
displayed normally.
At the same time, we add support for text: into the template library, instead
of manually creating text: fields in the dict for every field.
Finally, add a forecast routine to get the due counts for the next week, which
is used in the GUI.
The media table was originally introduced when Anki hashed media filenames,
and needed a way to remember the original filename. It also helped with:
1) getting a quick list of all media used in the deck, or the media added
since the last sync, for mobile clients
2) merging identical files with different names
But had some drawbacks:
- every operation that modifies templates, models or facts meant generating
the q/a and checking if any new media had appeared
- each entry is about 70 bytes, and some decks have 100k+ media files
So we remove the media table. We address 1) by being more intelligent about
media downloads on the mobile platform. We ask the user after a full sync if
they want to look for missing media, and they can choose not to if they know
they haven't added any. And on a partial sync, we can scan the contents of the
incoming facts for media references, and download any references we find. This
also avoids all the issues people had with media not downloading because it
was in their media folder but not in the media database.
For 2), when copying media to the media folder, if we have a duplicate
filename, we check if that file has the same md5, and avoid copying if so.
This won't merge identical content that has separate names, but instances
where users need that are rare.
The 'entry due' is the due time of a failed card before it enters the learning
queue. When the card graduates or is removed, it has its old due time
restored. We could pull this from the revlog, but it's cheaper to do it this
way.
A lot of the old checks in fixIntegrity() are no longer relevant, and some of
the others may no longer be required. They can be added back in as the need
arises.
We want to ensure that we never recycle ids from deleted cards. We could do
this with an autoincrement column in sqlite, but it's cheaper for us to handle
the ids ourselves, as the deck object is always in memory.
- remove revlog.py and move code into scheduler
- add a routine to log a learn repetition
- rename flags to type and set type=0 for learn mode
- add to unit test
Because the cards table is small now, loading it in and doing a table scan
isn't necessary. The facts table is bigger now, so we still need that index.
When adding facts, you can now pass in a group id which the GUI should support
editing. Templates will have an optional group id which overrides the provided
id, so users can automatically put certain card types in a different group (or
all of them, if desired). Greying out the group box in the GUI in that case
would be a good idea.
The model now has a css column, and when it's flushed, it generates the css
for the fields and templates. This means we don't have to generate the CSS on
deck load anymore.
The hex cache has also been removed. Javascript couldn't handle big ints, but
since ints are small numbers now, we no longer need a cache to efficiently
convert an id to hex.
- convert checksums to int
- add bulk update & update on upgrade
- add indices pending performance testing. The fsum table & indices add about
2MB to a deck with 50k unique fields
the field cache (fsums table) also needs to store the model id to preserve the
old behaviour of limiting duplicate checks to a given model, and to ensure
we're actually comparing against the same fields
removed the dingsbums and wcu importers; will accept them back if the authors
port them to the new codebase.
Since Anki first moved to an SQL backend, it has stored fields in a fields
table, with one field per line. This is a natural layout in a relational
database, and it had some nice properties. It meant we could retrieve an
individual field of a fact, which we used for limiting searches to a
particular field, for sorting, and for determining if a field was unique, by
adding an index on the field value.
The index was very expensive, so as part of the early work towards 2.0 I added
a checksum field instead, and added an index to that. This was a lot cheaper
than storing the entire value twice for the purpose of fast searches, but it
only partly solved the problem. We still needed an index on factId so that we
could retrieve a given fact's fields quickly. For simple models this was
fairly cheap, but as the number of fields grows the table grows very big. 25k
facts with 30 fields each and the fields table has grown to 750k entries. This
makes the factId index and checksum index really expensive - with the q/a
cache removed, about 30% of the deck in such a situation.
Equally problematic was sorting on those fields. Short of adding another
expensive index, a sort involves a table scan of the entire table.
We solve these problems by moving all fields into the facts table. For this to
work, we need to address some issues:
Sorting: we'll add an option to the model to specify the sort field. When
facts are modified, that field is written to a separate sort column. It can be
HTML stripped, and possibly truncated to a maximum number of letters. This
means that switching sort to a different field involves an expensive rewrite
of the sort column, but people tend to leave their sort field set to the same
value, and we don't need to clear the field if the user switches temporarily
to a non-field sort like due order. And it has the nice properties of allowing
different models to be sorted on different columns at the same time, and
makes it impossible for models to be hidden because the user has sorted on a
field which doesn't appear in some models.
Searching for words with embedded HTML: 1.2 introduced a HTML-stripped cache
of the fields content, which both sped up searches (since we didn't have to
search the possibly large fields table), and meant we could find "bob" in
"b<b>ob</b>" quickly. The ability to quickly search for words peppered with
HTML was nice, but it meant doubling the cost of storing text in many cases,
and meant after any edit more data has to be written to the DB. Instead, we'll
do it on the fly. On this i7 computer, stripping HTML from all fields takes
1-2.6 seconds on 25-50k decks. We could possibly skip the stripping for people
who don't require it - the number of people who bold parts of words is
actually pretty small.
Duplicate detection: one option would be to fetch all fields when the add
cards dialog or editor are opened. But this will be expensive on mobile
devices. Instead, we'll create a separate table of (fid, csum), with an index
on both columns. When we edit a fact, we delete all the existing checksums for
that fact, and add checksums for any fields that must be checked as unique. We
could optionally skip the index on csum - some benchmarking is required.
As for the new table layout, creating separate columns for each field won't
scale. Instead, we store the fields in a single column, separated by an ascii
record separator. We split on that character when extracting from
the database, and join on it when writing to the DB.
Searching on a particular field in the browser will be accomplished by finding
all facts that match, and then unpacking to see if the relevant field matched.
Tags have been moved back to a separate column. Now that fields are on the
facts table, there is no need to pack them in as a field simply to avoid
another table hit.
Anki used random 64bit IDs for cards, facts and fields. This had some nice
properties:
- merging data in syncs and imports was simply a matter of copying each way,
as conflicts were astronomically unlikely
- it made it easy to identify identical cards and prevent them from being
reimported
But there were some negatives too:
- they're more expensive to store
- javascript can't handle numbers > 2**53, which means AnkiMobile, iAnki and
so on have to treat the ids as strings, which is slow
- simply copying data in a sync or import can lead to corruption, as while a
duplicate id indicates the data was originally the same, it may have
diverged. A more intelligent approach is necessary.
- sqlite was sorting the fields table based on the id, which meant the fields
were spread across the table, and costly to fetch
So instead, we'll move to incremental ids. In the case of model changes we'll
declare that a schema change and force a full sync to avoid having to deal
with conflicts, and in the case of cards and facts, we'll need to update the
ids on one end to merge. Identical cards can be detected by checking to see if
their id is the same and their creation time is the same.
Creation time has been added back to cards and facts because it's necessary
for sync conflict merging. That means facts.pos is not required.
The graves table has been removed. It's not necessary for schema related
changes, and dead cards/facts can be represented as a card with queue=-4 and
created=0. Because we will record schema modification time and can ensure a
full sync propagates to all endpoints, it means we can remove the dead
cards/facts on schema change.
Tags have been removed from the facts table and are represented as a field
with ord=-1 and fmid=0. Combined with the locality improvement for fields, it
means that fetching fields is not much more expensive than using the q/a
cache.
Because of the above, removing the q/a cache is a possibility now. The q and a
columns on cards has been dropped. It will still be necessary to render the
q/a on fact add/edit, since we need to record media references. It would be
nice to avoid this in the future. Perhaps one way would be the ability to
assign a type to fields, like "image", "audio", or "latex". LaTeX needs
special consider anyway, as it was being rendered into the q/a cache.
- make sure we're actually stripping text in the field cache
- make sure a default group is added on upgrade
- make sure old style field references are upgrade
SQLAlchemy is a great tool, but it wasn't a great fit for Anki:
- We often had to drop down to raw SQL for performance reasons.
- The DB cursors and results were wrapped, which incurred a
sizable performance hit due to introspection. Operations like fetching 50k
records from a hot cache were taking more than twice as long to complete.
- We take advantage of sqlite-specific features, so SQL language abstraction
is useless to us.
- The anki schema is quite small, so manually saving and loading objects is
not a big burden.
In the process of porting to DBAPI, I've refactored the database schema:
- App configuration data that we don't need in joins or bulk updates has been
moved into JSON objects. This simplifies serializing, and means we won't
need DB schema changes to store extra options in the future. This change
obsoletes the deckVars table.
- Renamed tables:
-- fieldModels -> fields
-- cardModels -> templates
-- fields -> fdata
- a number of attribute names have been shortened
Classes like Card, Fact & Model remain. They maintain a reference to the deck.
To write their state to the DB, call .flush().
Objects no longer have their modification time manually updated. Instead, the
modification time is updated when they are flushed. This also applies to the
deck.
Decks will now save on close, because various operations that were done at
deck load will be moved into deck close instead. Operations like undoing
buried card are cheap on a hot cache, but expensive on startup.
Programmatically you can call .close(save=False) to avoid a save and a
modification bump. This will be useful for generating due counts.
Because of the new saving behaviour, the save and save as options will be
removed from the GUI in the future.
The q/a cache and field cache generating has been centralized. Facts will
automatically rebuild the cache on flush; models can do so with
model.updateCache().
Media handling has also been reworked. It has moved into a MediaRegistry
object, which the deck holds. Refcounting has been dropped - it meant we had
to compare old and new value every time facts or models were changed, and
existed for the sole purpose of not showing errors on a missing media
download. Instead we just media.registerText(q+a) when it's updated. The
download function will be expanded to ask the user if they want to continue
after a certain number of files have failed to download, which should be an
adequate alternative. And we now add the file into the media DB when it's
copied to th emedia directory, not when the card is commited. This fixes
duplicates a user would get if they added the same media to a card twice
without adding the card.
The old DeckStorage object had its upgrade code split in a previous commit;
the opening and upgrading code has been merged back together, and put in a
separate storage.py file. The correct way to open a deck now is import anki; d
= anki.Deck(path).
deck.getCard() -> deck.sched.getCard()
same with answerCard
deck.getCard(id) returns a Card object now.
And the DB wrapper has had a few changes:
- sql statements are a more standard DBAPI:
- statement() -> execute()
- statements() -> executemany()
- called like execute(sql, 1, 2, 3) or execute(sql, a=1, b=2, c=3)
- column0 -> list
