def fieldUnique(self, name):
(ord, conf) = self._fmap[name]
if not conf['uniq']:
return True
val = self[name]
if not val:
return True
csum = fieldChecksum(val)
if self.id:
lim = "and fid != :fid"
else:
lim = ""
fids = self.deck.db.list(
"select fid from fsums where csum = ? and fid != ? and mid = ?",
csum, self.id or 0, self.mid)
if not fids:
return True
# grab facts with the same checksums, and see if they're actually
# duplicates
for flds in self.deck.db.list("select flds from facts where id in "+
ids2str(fids)):
fields = splitFields(flds)
if fields[ord] == val:
return False
return True
def flush(self, mod=None):
"If fields or tags have changed, write changes to disk."
assert self.scm == self.col.scm
self._preFlush()
sfld = stripHTMLMedia(self.fields[self.col.models.sortIdx(self._model)])
tags = self.stringTags()
fields = self.joinedFields()
if not mod and self.col.db.scalar(
"select 1 from notes where id = ? and tags = ? and flds = ?", self.id, tags, fields
):
return
csum = fieldChecksum(self.fields[0])
self.mod = mod if mod else intTime()
self.usn = self.col.usn()
res = self.col.db.execute(
"""
insert or replace into notes values (?,?,?,?,?,?,?,?,?,?,?)""",
self.id,
self.guid,
self.mid,
self.mod,
self.usn,
tags,
fields,
sfld,
csum,
self.flags,
self.data,
)
self.col.tags.register(self.tags)
self._postFlush()
def flush(self, mod=None):
assert self.scm == self.col.scm
self._preFlush()
self.mod = mod if mod else intTime()
self.usn = self.col.usn()
sfld = stripHTML(self.fields[self.col.models.sortIdx(self._model)])
tags = self.stringTags()
csum = fieldChecksum(self.fields[0])
res = self.col.db.execute(
"""
insert or replace into notes values (?,?,?,?,?,?,?,?,?,?,?)""",
self.id,
self.guid,
self.mid,
self.mod,
self.usn,
tags,
self.joinedFields(),
sfld,
csum,
self.flags,
self.data,
)
self.col.tags.register(self.tags)
self._postFlush()
def scmhash(self, m):
"Return a hash of the schema, to see if models are compatible."
s = ""
for f in m['flds']:
s += f['name']
for t in m['tmpls']:
s += t['name']
s += t['qfmt']
s += t['afmt']
return fieldChecksum(s)
def updateFieldChecksums(self):
self.deck.db.execute("delete from fsums where fid = ?", self.id)
d = []
for (ord, conf) in self._fmap.values():
if not conf['uniq']:
continue
val = self.fields[ord]
if not val:
continue
d.append((self.id, self.mid, fieldChecksum(val)))
self.deck.db.executemany("insert into fsums values (?, ?, ?)", d)
def isDupe(self, data):
"Takes field, model and returns True if the field is a dupe and False otherwise."
# find any matching csums and compare
csum = fieldChecksum(data["field"])
mid = mw.col.models.byName(data["model"])["id"]
for flds in mw.col.db.list(
"select flds from notes where csum = ? and id != ? and mid = ?",
csum, 0, mid):
if splitFields(flds)[0] == data["field"]:
return True
return False
def updateFieldCache(self, nids):
"Update field checksums and sort cache, after find&replace, etc."
snids = ids2str(nids)
r = []
for (nid, mid, flds) in self._fieldData(snids):
fields = splitFields(flds)
model = self.models.get(mid)
r.append((stripHTML(fields[self.models.sortIdx(model)]),
fieldChecksum(fields[0]),
nid))
# apply, relying on calling code to bump usn+mod
self.db.executemany("update notes set sfld=?, csum=? where id=?", r)
def dupeOrEmpty(self):
"1 if first is empty; 2 if first is a duplicate, False otherwise."
val = self.fields[0]
if not val.strip():
return 1
csum = fieldChecksum(val)
# find any matching csums and compare
for flds in self.col.db.list(
"select flds from notes where csum = ? and id != ? and mid = ?",
csum, self.id or 0, self.mid):
if splitFields(flds)[0] == self.fields[0]:
return 2
return False
def dupeOrEmpty(self):
"1 if first is empty; 2 if first is a duplicate, False otherwise."
val = self.fields[0]
if not val.strip():
return 1
csum = fieldChecksum(val)
# find any matching csums and compare
for flds in self.col.db.list(
"select flds from notes where csum = ? and id != ? and mid = ?",
csum, self.id or 0, self.mid):
if stripHTMLMedia(splitFields(flds)[0]) == stripHTMLMedia(
self.fields[0]):
return 2
return False
def _findDupes(self, val):
# caller must call stripHTMLMedia on passed val
try:
mid, val = val.split(",", 1)
except OSError:
return
csum = fieldChecksum(val)
nids = []
for nid, flds in self.col.db.execute(
"select id, flds from notes where mid=? and csum=?", mid,
csum):
if stripHTMLMedia(splitFields(flds)[0]) == val:
nids.append(nid)
return "n.id in %s" % ids2str(nids)
def _findDupes(self, val):
# caller must call stripHTMLMedia on passed val
try:
mid, val = val.split(",", 1)
except OSError:
return
csum = fieldChecksum(val)
nids = []
for nid, flds in self.col.db.execute(
"select id, flds from notes where mid=? and csum=?",
mid, csum):
if stripHTMLMedia(splitFields(flds)[0]) == val:
nids.append(nid)
return "n.id in %s" % ids2str(nids)
def updateFieldCache(self, nids):
"Update field checksums and sort cache, after find&replace, etc."
snids = ids2str(nids)
r = []
for (nid, mid, flds) in self._fieldData(snids):
fields = splitFields(flds)
model = self.models.get(mid)
if not model:
# note points to invalid model
continue
r.append((stripHTML(fields[self.models.sortIdx(model)]),
fieldChecksum(fields[0]), nid))
# apply, relying on calling code to bump usn+mod
self.db.executemany("update notes set sfld=?, csum=? where id=?", r)
def flush(self, mod=None):
assert self.scm == self.col.scm
self._preFlush()
self.mod = mod if mod else intTime()
self.usn = self.col.usn()
sfld = stripHTML(self.fields[self.col.models.sortIdx(self._model)])
tags = self.stringTags()
csum = fieldChecksum(self.fields[0])
res = self.col.db.execute(
"""
insert or replace into notes values (?,?,?,?,?,?,?,?,?,?,?)""",
self.id, self.guid, self.mid, self.mod, self.usn, tags,
self.joinedFields(), sfld, csum, self.flags, self.data)
self.col.tags.register(self.tags)
self._postFlush()
def updateFieldCache(self, fids, csum=True):
"Update field checksums and sort cache, after find&replace, etc."
sfids = ids2str(fids)
mods = self.models()
r = []
r2 = []
for (fid, mid, flds) in self._fieldData(sfids):
fields = splitFields(flds)
model = mods[mid]
if csum:
for f in model.fields:
if f['uniq'] and fields[f['ord']]:
r.append((fid, mid, fieldChecksum(fields[f['ord']])))
r2.append((stripHTML(fields[model.sortIdx()]), fid))
if csum:
self.db.execute("delete from fsums where fid in "+sfids)
self.db.executemany("insert into fsums values (?,?,?)", r)
self.db.executemany("update facts set sfld = ? where id = ?", r2)
def flush(self, mod=None):
"If fields or tags have changed, write changes to disk."
assert self.scm == self.col.scm
self._preFlush()
sfld = stripHTMLMedia(self.fields[self.col.models.sortIdx(
self._model)])
tags = self.stringTags()
fields = self.joinedFields()
if not mod and self.col.db.scalar(
"select 1 from notes where id = ? and tags = ? and flds = ?",
self.id, tags, fields):
return
csum = fieldChecksum(self.fields[0])
self.mod = mod if mod else intTime()
self.usn = self.col.usn()
res = self.col.db.execute(
"""
insert or replace into notes values (?,?,?,?,?,?,?,?,?,?,?)""", self.id,
self.guid, self.mid, self.mod, self.usn, tags, fields, sfld, csum,
self.flags, self.data)
self.col.tags.register(self.tags)
self._postFlush()
def dupeOrEmptyWithOrds(self):
"""
Returns a tuple. The contents of each element are as follows:
1) 1 if first is empty; 2 if first is a duplicate, False otherwise.
2) For a duplicate (2), this returns the list of ordinals that make up the key.
Otherwise this is None.
"""
val = self.fields[0]
if not val.strip():
return 1, None
csum = fieldChecksum(val)
# find any matching csums and compare
for flds in self.col.db.list(
"select flds from notes where csum = ? and id != ? and mid = ?",
csum, self.id or 0, self.mid):
model = self.model()
field_ords = [0]
for fld in model["flds"]:
if fld["ord"] == 0:
continue
elif fld["name"].endswith(KEY_SUFFIX):
field_ords.append(fld["ord"])
all_fields_equal = True
fields_split = splitFields(flds)
for field_ord in field_ords:
if stripHTMLMedia(fields_split[field_ord]) != stripHTMLMedia(
self.fields[field_ord]):
all_fields_equal = False
if all_fields_equal:
return 2, field_ords
return False, None
def updateCompleteDeck(self, data):
self.startEditing()
did = self.decks().id(data["deck"])
self.decks().flush()
model_manager = self.collection().models
for _, card in data["cards"].items():
self.database().execute(
"replace into cards (id, nid, did, ord, type, queue, due, ivl, factor, reps, lapses, left, "
"mod, usn, odue, odid, flags, data) "
"values (" + "?," * (12 + 6 - 1) + "?)",
card["id"], card["nid"], did, card["ord"], card["type"], card["queue"], card["due"],
card["ivl"], card["factor"], card["reps"], card["lapses"], card["left"],
intTime(), -1, 0, 0, 0, 0
)
note = data["notes"][str(card["nid"])]
tags = self.collection().tags.join(self.collection().tags.canonify(note["tags"]))
self.database().execute(
"replace into notes(id, mid, tags, flds,"
"guid, mod, usn, flags, data, sfld, csum) values (" + "?," * (4 + 7 - 1) + "?)",
note["id"], note["mid"], tags, joinFields(note["fields"]),
guid64(), intTime(), -1, 0, 0, "", fieldChecksum(note["fields"][0])
)
model = data["models"][str(note["mid"])]
if not model_manager.get(model["id"]):
model_o = model_manager.new(model["name"])
for field_name in model["fields"]:
field = model_manager.newField(field_name)
model_manager.addField(model_o, field)
for template_name in model["templateNames"]:
template = model_manager.newTemplate(template_name)
model_manager.addTemplate(model_o, template)
model_o["id"] = model["id"]
model_manager.update(model_o)
model_manager.flush()
self.stopEditing()
def updateCompleteDeck(self, data):
self.startEditing()
did = self.decks().id(data['deck'])
self.decks().flush()
model_manager = self.collection().models
for _, card in data['cards'].items():
self.database().execute(
'replace into cards (id, nid, did, ord, type, queue, due, ivl, factor, reps, lapses, left, '
'mod, usn, odue, odid, flags, data) '
'values (' + '?,' * (12 + 6 - 1) + '?)',
card['id'], card['nid'], did, card['ord'], card['type'], card['queue'], card['due'],
card['ivl'], card['factor'], card['reps'], card['lapses'], card['left'],
intTime(), -1, 0, 0, 0, 0
)
note = data['notes'][str(card['nid'])]
tags = self.collection().tags.join(self.collection().tags.canonify(note['tags']))
self.database().execute(
'replace into notes(id, mid, tags, flds,'
'guid, mod, usn, flags, data, sfld, csum) values (' + '?,' * (4 + 7 - 1) + '?)',
note['id'], note['mid'], tags, joinFields(note['fields']),
guid64(), intTime(), -1, 0, 0, '', fieldChecksum(note['fields'][0])
)
model = data['models'][str(note['mid'])]
if not model_manager.get(model['id']):
model_o = model_manager.new(model['name'])
for field_name in model['fields']:
field = model_manager.newField(field_name)
model_manager.addField(model_o, field)
for template_name in model['templateNames']:
template = model_manager.newTemplate(template_name)
model_manager.addTemplate(model_o, template)
model_o['id'] = model['id']
model_manager.update(model_o)
model_manager.flush()
self.stopEditing()
def scmhash(self, m):
"Return a hash of the schema, to see if models are compatible."
s = ""
for f in m['flds']:
s += f['name']
return fieldChecksum(s)
def updateNotes(allDb):
t_0, now, db = time.time(), intTime(), mw.col.db
TAG = mw.col.tags # type: TagManager
ds, nid2mmi = [], {}
mw.progress.start(label='Updating data', immediate=True)
fidDb = allDb.fidDb(recalc=True)
loc_db = allDb.locDb(recalc=False) # type: Dict[Location, Set[Morpheme]]
# read tag names
compTag, vocabTag, freshTag, notReadyTag, alreadyKnownTag, priorityTag, tooShortTag, tooLongTag, frequencyTag = tagNames = cfg(
'Tag_Comprehension'), cfg('Tag_Vocab'), cfg('Tag_Fresh'), cfg(
'Tag_NotReady'), cfg('Tag_AlreadyKnown'), cfg('Tag_Priority'), cfg(
'Tag_TooShort'), cfg('Tag_TooLong'), cfg('Tag_Frequency')
TAG.register(tagNames)
badLengthTag = cfg('Tag_BadLength')
# handle secondary databases
mw.progress.update(label='Creating seen/known/mature from all.db')
seenDb = filterDbByMat(allDb, cfg('threshold_seen'))
knownDb = filterDbByMat(allDb, cfg('threshold_known'))
matureDb = filterDbByMat(allDb, cfg('threshold_mature'))
mw.progress.update(label='Loading priority.db')
priorityDb = MorphDb(cfg('path_priority'), ignoreErrors=True).db
mw.progress.update(label='Loading frequency.txt')
frequencyListPath = cfg('path_frequency')
frequency_map = {}
frequency_has_morphemes = False
try:
with io.open(frequencyListPath, encoding='utf-8-sig') as csvfile:
csvreader = csv.reader(csvfile, delimiter="\t")
rows = [row for row in csvreader]
if rows[0][0] == "#study_plan_frequency":
frequency_has_morphemes = True
frequency_map = dict(
zip([
Morpheme(row[0], row[1], row[2], row[3], row[4],
row[5]) for row in rows[1:]
], itertools.count(0)))
else:
frequency_map = dict(
zip([row[0] for row in rows], itertools.count(0)))
except FileNotFoundError:
pass
frequencyListLength = len(frequency_map)
# prefetch cfg for fields
field_focus_morph = cfg('Field_FocusMorph')
field_unknown_count = cfg('Field_UnknownMorphCount')
field_unmature_count = cfg('Field_UnmatureMorphCount')
field_morph_man_index = cfg('Field_MorphManIndex')
field_unknowns = cfg('Field_Unknowns')
field_unmatures = cfg('Field_Unmatures')
field_unknown_freq = cfg('Field_UnknownFreq')
field_focus_morph_pos = cfg("Field_FocusMorphPos")
skip_comprehension_cards = cfg('Option_SkipComprehensionCards')
skip_fresh_cards = cfg('Option_SkipFreshVocabCards')
# Find all morphs that changed maturity and the notes that refer to them.
last_maturities = allDb.meta.get('last_maturities', {})
new_maturities = {}
refresh_notes = set()
# Recompute everything if preferences changed.
last_preferences = allDb.meta.get('last_preferences', {})
if not last_preferences == get_preferences():
print("Preferences changed. Updating all notes...")
last_updated = 0
else:
last_updated = allDb.meta.get('last_updated', 0)
# Todo: Remove this forced 0 once we add checks for other changes like new frequency.txt files.
last_updated = 0
# If we're updating everything anyway, clear the notes set.
if last_updated > 0:
for m, locs in allDb.db.items():
maturity_bits = 0
if seenDb.matches(m):
maturity_bits |= 1
if knownDb.matches(m):
maturity_bits |= 2
if matureDb.matches(m):
maturity_bits |= 4
new_maturities[m] = maturity_bits
if last_maturities.get(m, -1) != maturity_bits:
for loc in locs:
if isinstance(loc, AnkiDeck):
refresh_notes.add(loc.noteId)
included_types, include_all = getModifyEnabledModels()
included_mids = [
m['id'] for m in mw.col.models.all()
if include_all or m['name'] in included_types
]
query = '''
select id, mid, flds, guid, tags from notes
WHERE mid IN ({0}) and ( mod > {2} or id in ({1}) )
'''.format(','.join([str(m) for m in included_mids]),
','.join([str(id) for id in refresh_notes]), last_updated)
query_results = db.execute(query)
N_notes = len(query_results)
mw.progress.finish()
mw.progress.start(label='Updating notes', max=N_notes, immediate=True)
for i, (nid, mid, flds, guid, tags) in enumerate(query_results):
ts = TAG.split(tags)
if i % 500 == 0:
mw.progress.update(value=i)
C = partial(cfg, model_id=mid)
notecfg = getFilterByMidAndTags(mid, ts)
if notecfg is None or not notecfg['Modify']:
continue
# Get all morphemes for note
morphemes = set()
for fieldName in notecfg['Fields']:
try:
loc = fidDb[(nid, guid, fieldName)]
morphemes.update(loc_db[loc])
except KeyError:
continue
proper_nouns_known = cfg('Option_ProperNounsAlreadyKnown')
# Determine un-seen/known/mature and i+N
unseens, unknowns, unmatures, new_knowns = set(), set(), set(), set()
for morpheme in morphemes:
if proper_nouns_known and morpheme.isProperNoun():
continue
if not seenDb.matches(morpheme):
unseens.add(morpheme)
if not knownDb.matches(morpheme):
unknowns.add(morpheme)
if not matureDb.matches(morpheme):
unmatures.add(morpheme)
if knownDb.matches(morpheme):
new_knowns.add(morpheme)
# Determine MMI - Morph Man Index
N, N_s, N_k, N_m = len(morphemes), len(unseens), len(unknowns), len(
unmatures)
# Bail early for lite update
if N_k > 2 and C('only update k+2 and below'):
continue
# add bonus for morphs in priority.db and frequency.txt
frequencyBonus = C('frequency.txt bonus')
isPriority = False
isFrequency = False
focusMorph = None
F_k = 0
usefulness = 0
for focusMorph in unknowns:
F_k += allDb.frequency(focusMorph)
if focusMorph in priorityDb:
isPriority = True
usefulness += C('priority.db weight')
if frequency_has_morphemes:
focusMorphIndex = frequency_map.get(focusMorph, -1)
else:
focusMorphIndex = frequency_map.get(focusMorph.base, -1)
if focusMorphIndex >= 0:
isFrequency = True
# The bigger this number, the lower mmi becomes
usefulness += int(
round(frequencyBonus *
(1 - focusMorphIndex / frequencyListLength)))
# average frequency of unknowns (ie. how common the word is within your collection)
F_k_avg = F_k // N_k if N_k > 0 else F_k
usefulness += F_k_avg
# add bonus for studying recent learned knowns (reinforce)
for morpheme in new_knowns:
locs = knownDb.getMatchingLocs(morpheme)
if locs:
ivl = min(1, max(loc.maturity for loc in locs))
# TODO: maybe average this so it doesnt favor long sentences
usefulness += C('reinforce new vocab weight') // ivl
if any(morpheme.pos == '動詞'
for morpheme in unknowns): # FIXME: this isn't working???
usefulness += C('verb bonus')
usefulness = 99999 - min(99999, usefulness)
# difference from optimal length range (too little context vs long sentence)
lenDiffRaw = min(N - C('min good sentence length'),
max(0, N - C('max good sentence length')))
lenDiff = min(9, abs(lenDiffRaw))
# Fill in various fields/tags on the note based on cfg
fs = splitFields(flds)
# clear any 'special' tags, the appropriate will be set in the next few lines
ts = [
t for t in ts
if t not in (notReadyTag, compTag, vocabTag, freshTag)
]
# determine card type
if N_m == 0: # sentence comprehension card, m+0
ts.append(compTag)
if skip_comprehension_cards:
usefulness += 1000000 # Add a penalty to put these cards at the end of the queue
elif N_k == 1: # new vocab card, k+1
ts.append(vocabTag)
setField(mid, fs, field_focus_morph, focusMorph.base)
setField(mid, fs, field_focus_morph_pos, focusMorph.pos)
elif N_k > 1: # M+1+ and K+2+
ts.append(notReadyTag)
elif N_m == 1: # we have k+0, and m+1, so this card does not introduce a new vocabulary -> card for newly learned morpheme
ts.append(freshTag)
if skip_fresh_cards:
usefulness += 1000000 # Add a penalty to put these cards at the end of the queue
focusMorph = next(iter(unmatures))
setField(mid, fs, field_focus_morph, focusMorph.base)
setField(mid, fs, field_focus_morph_pos, focusMorph.pos)
else: # only case left: we have k+0, but m+2 or higher, so this card does not introduce a new vocabulary -> card for newly learned morpheme
ts.append(freshTag)
if skip_fresh_cards:
usefulness += 1000000 # Add a penalty to put these cards at the end of the queue
# calculate mmi
mmi = 100000 * N_k + 1000 * lenDiff + int(round(usefulness))
if C('set due based on mmi'):
nid2mmi[nid] = mmi
# set type agnostic fields
setField(mid, fs, field_unknown_count, '%d' % N_k)
setField(mid, fs, field_unmature_count, '%d' % N_m)
setField(mid, fs, field_morph_man_index, '%d' % mmi)
setField(mid, fs, field_unknowns, ', '.join(u.base for u in unknowns))
setField(mid, fs, field_unmatures,
', '.join(u.base for u in unmatures))
setField(mid, fs, field_unknown_freq, '%d' % F_k_avg)
# remove deprecated tag
if badLengthTag is not None and badLengthTag in ts:
ts.remove(badLengthTag)
# other tags
if priorityTag in ts:
ts.remove(priorityTag)
if isPriority:
ts.append(priorityTag)
if frequencyTag in ts:
ts.remove(frequencyTag)
if isFrequency:
ts.append(frequencyTag)
if tooShortTag in ts:
ts.remove(tooShortTag)
if lenDiffRaw < 0:
ts.append(tooShortTag)
if tooLongTag in ts:
ts.remove(tooLongTag)
if lenDiffRaw > 0:
ts.append(tooLongTag)
# remove unnecessary tags
if not cfg('Option_SetNotRequiredTags'):
unnecessary = [priorityTag, tooShortTag, tooLongTag]
ts = [tag for tag in ts if tag not in unnecessary]
# update sql db
tags_ = TAG.join(TAG.canonify(ts))
flds_ = joinFields(fs)
if flds != flds_ or tags != tags_: # only update notes that have changed
csum = fieldChecksum(fs[0])
sfld = stripHTML(fs[getSortFieldIndex(mid)])
ds.append((tags_, flds_, sfld, csum, now, mw.col.usn(), nid))
mw.progress.update(label='Updating anki database...')
mw.col.db.executemany(
'update notes set tags=?, flds=?, sfld=?, csum=?, mod=?, usn=? where id=?',
ds)
# Now reorder new cards based on MMI
mw.progress.update(label='Updating new card ordering...')
ds = []
# "type = 0": new cards
# "type = 1": learning cards [is supposed to be learning: in my case no learning card had this type]
# "type = 2": review cards
for (cid, nid,
due) in db.execute('select id, nid, due from cards where type = 0'):
if nid in nid2mmi: # owise it was disabled
due_ = nid2mmi[nid]
if due != due_: # only update cards that have changed
ds.append((due_, now, mw.col.usn(), cid))
mw.col.db.executemany('update cards set due=?, mod=?, usn=? where id=?',
ds)
mw.reset()
allDb.meta['last_preferences'] = get_preferences()
allDb.meta['last_maturities'] = new_maturities
allDb.meta['last_updated'] = int(time.time() + 0.5)
printf('Updated %d notes in %f sec' % (N_notes, time.time() - t_0))
if cfg('saveDbs'):
mw.progress.update(label='Saving all/seen/known/mature dbs')
allDb.save(cfg('path_all'))
seenDb.save(cfg('path_seen'))
knownDb.save(cfg('path_known'))
matureDb.save(cfg('path_mature'))
printf('Updated %d notes + saved dbs in %f sec' %
(N_notes, time.time() - t_0))
mw.progress.finish()
return knownDb
def updateNotes( allDb ):
t_0, now, db, TAG = time.time(), intTime(), mw.col.db, mw.col.tags
ds, nid2mmi = [], {}
N_notes = db.scalar( 'select count() from notes' )
mw.progress.start( label='Updating data', max=N_notes, immediate=True )
fidDb = allDb.fidDb()
locDb = allDb.locDb( recalc=False ) # fidDb() already forces locDb recalc
# read tag names
compTag, vocabTag, freshTag, notReadyTag, alreadyKnownTag, priorityTag, tooShortTag, tooLongTag = tagNames = jcfg('Tag_Comprehension'), jcfg('Tag_Vocab'), jcfg('Tag_Fresh'), jcfg('Tag_NotReady'), jcfg('Tag_AlreadyKnown'), jcfg('Tag_Priority'), jcfg('Tag_TooShort'), jcfg('Tag_TooLong')
TAG.register( tagNames )
badLengthTag = jcfg2().get('Tag_BadLength')
# handle secondary databases
mw.progress.update( label='Creating seen/known/mature from all.db' )
seenDb = filterDbByMat( allDb, cfg1('threshold_seen') )
knownDb = filterDbByMat( allDb, cfg1('threshold_known') )
matureDb = filterDbByMat( allDb, cfg1('threshold_mature') )
mw.progress.update( label='Loading priority.db' )
priorityDb = MorphDb( cfg1('path_priority'), ignoreErrors=True ).db
if cfg1('saveDbs'):
mw.progress.update( label='Saving seen/known/mature dbs' )
seenDb.save( cfg1('path_seen') )
knownDb.save( cfg1('path_known') )
matureDb.save( cfg1('path_mature') )
mw.progress.update( label='Updating notes' )
for i,( nid, mid, flds, guid, tags ) in enumerate( db.execute( 'select id, mid, flds, guid, tags from notes' ) ):
if i % 500 == 0: mw.progress.update( value=i )
C = partial( cfg, mid, None )
note = mw.col.getNote(nid)
notecfg = getFilter(note)
if notecfg is None or not notecfg['Modify']: continue
# Get all morphemes for note
morphemes = set()
for fieldName in notecfg['Fields']:
try:
loc = fidDb[ ( nid, guid, fieldName ) ]
morphemes.update( locDb[ loc ] )
except KeyError: continue
# Determine un-seen/known/mature and i+N
unseens, unknowns, unmatures, newKnowns = set(), set(), set(), set()
for morpheme in morphemes:
if morpheme not in seenDb.db: unseens.add( morpheme )
if morpheme not in knownDb.db: unknowns.add( morpheme )
if morpheme not in matureDb.db: unmatures.add( morpheme )
if morpheme not in matureDb.db and morpheme in knownDb.db:
newKnowns.add( morpheme )
# Determine MMI - Morph Man Index
N, N_s, N_k, N_m = len( morphemes ), len( unseens ), len( unknowns ), len( unmatures )
# Bail early for lite update
if N_k > 2 and C('only update k+2 and below'): continue
# average frequency of unknowns (ie. how common the word is within your collection)
F_k = 0
for focusMorph in unknowns: # focusMorph used outside loop
F_k += allDb.frequency(focusMorph)
F_k_avg = F_k // N_k if N_k > 0 else F_k
usefulness = F_k_avg
# add bonus for morphs in priority.db
isPriority = False
for focusMorph in unknowns:
if focusMorph in priorityDb:
isPriority = True
usefulness += C('priority.db weight')
# add bonus for studying recent learned knowns (reinforce)
for morpheme in newKnowns:
locs = allDb.db[ morpheme ]
if locs:
ivl = min( 1, max( loc.maturity for loc in locs ) )
usefulness += C('reinforce new vocab weight') // ivl #TODO: maybe average this so it doesnt favor long sentences
if any( morpheme.pos == u'動詞' for morpheme in unknowns ): #FIXME: this isn't working???
usefulness += C('verb bonus')
usefulness = 999 - min( 999, usefulness )
# difference from optimal length range (too little context vs long sentence)
lenDiffRaw = min(N - C('min good sentence length'),
max(0, N - C('max good sentence length')))
lenDiff = min(9, abs(lenDiffRaw))
# calculate mmi
mmi = 10000*N_k + 1000*lenDiff + usefulness
if C('set due based on mmi'):
nid2mmi[ nid ] = mmi
# Fill in various fields/tags on the note based on cfg
ts, fs = TAG.split( tags ), splitFields( flds )
# clear any 'special' tags, the appropriate will be set in the next few lines
ts = [ t for t in ts if t not in [ notReadyTag, compTag, vocabTag, freshTag ] ]
# determine card type
if N_m == 0: # sentence comprehension card, m+0
ts = ts + [ compTag ]
setField( mid, fs, jcfg('Field_FocusMorph'), u'' )
elif N_k == 1: # new vocab card, k+1
ts = ts + [ vocabTag ]
setField( mid, fs, jcfg('Field_FocusMorph'), u'%s' % focusMorph.base )
elif N_k > 1: # M+1+ and K+2+
ts = ts + [ notReadyTag ]
setField( mid, fs, jcfg('Field_FocusMorph'), u'')
elif N_m == 1: # we have k+0, and m+1, so this card does not introduce a new vocabulary -> card for newly learned morpheme
ts = ts + [ freshTag ]
setField( mid, fs, jcfg('Field_FocusMorph'), u'%s' % list(unmatures)[0].base)
else: # only case left: we have k+0, but m+2 or higher, so this card does not introduce a new vocabulary -> card for newly learned morpheme
ts = ts + [ freshTag ]
setField( mid, fs, jcfg('Field_FocusMorph'), u'')
# set type agnostic fields
setField( mid, fs, jcfg('Field_UnknownMorphCount'), u'%d' % N_k )
setField( mid, fs, jcfg('Field_UnmatureMorphCount'), u'%d' % N_m )
setField( mid, fs, jcfg('Field_MorphManIndex'), u'%d' % mmi )
setField( mid, fs, jcfg('Field_Unknowns'), u', '.join( u.base for u in unknowns ) )
setField( mid, fs, jcfg('Field_Unmatures'), u', '.join( u.base for u in unmatures ) )
setField( mid, fs, jcfg('Field_UnknownFreq'), u'%d' % F_k_avg )
# remove deprecated tag
if badLengthTag is not None and badLengthTag in ts:
ts.remove( badLengthTag )
# other tags
if priorityTag in ts: ts.remove( priorityTag )
if isPriority: ts.append( priorityTag )
if tooShortTag in ts: ts.remove( tooShortTag )
if lenDiffRaw < 0: ts.append( tooShortTag )
if tooLongTag in ts: ts.remove( tooLongTag )
if lenDiffRaw > 0: ts.append( tooLongTag )
# remove unnecessary tags
if not jcfg('Option_SetNotRequiredTags'):
unnecessary = [priorityTag, tooShortTag, tooLongTag]
ts = [tag for tag in ts if tag not in unnecessary]
# update sql db
tags_ = TAG.join( TAG.canonify( ts ) )
flds_ = joinFields( fs )
if flds != flds_ or tags != tags_: # only update notes that have changed
csum = fieldChecksum( fs[0] )
sfld = stripHTML( fs[ getSortFieldIndex( mid ) ] )
ds.append( { 'now':now, 'tags':tags_, 'flds':flds_, 'sfld':sfld, 'csum':csum, 'usn':mw.col.usn(), 'nid':nid } )
mw.progress.update( value=i, label='Updating anki database...' )
mw.col.db.executemany( 'update notes set tags=:tags, flds=:flds, sfld=:sfld, csum=:csum, mod=:now, usn=:usn where id=:nid', ds )
# Now reorder new cards based on MMI
mw.progress.update( value=i, label='Updating new card ordering...' )
ds = []
# "type = 0": new cards
# "type = 1": learning cards [is supposed to be learning: in my case no learning card had this type]
# "type = 2": review cards
for ( cid, nid, due ) in db.execute( 'select id, nid, due from cards where type = 0' ):
if nid in nid2mmi: # owise it was disabled
due_ = nid2mmi[ nid ]
if due != due_: # only update cards that have changed
ds.append( { 'now':now, 'due':due_, 'usn':mw.col.usn(), 'cid':cid } )
mw.col.db.executemany( 'update cards set due=:due, mod=:now, usn=:usn where id=:cid', ds )
mw.reset()
printf( 'Updated notes in %f sec' % ( time.time() - t_0 ) )
mw.progress.finish()
return knownDb
def importNotes(self, notes):
"Convert each card into a note, apply attributes and add to col."
assert self.mappingOk()
# note whether tags are mapped
self._tagsMapped = False
for f in self.mapping:
if f == "_tags":
self._tagsMapped = True
# gather checks for duplicate comparison
csums = {}
for csum, id in self.col.db.execute(
"select csum, id from notes where mid = ?", self.model['id']):
if csum in csums:
csums[csum].append(id)
else:
csums[csum] = [id]
firsts = {}
fld0idx = self.mapping.index(self.model['flds'][0]['name'])
self._fmap = self.col.models.fieldMap(self.model)
self._nextID = timestampID(self.col.db, "notes")
# loop through the notes
updates = []
updateLog = []
updateLogTxt = _("Update as first field matched: %s")
dupeLogTxt = _("Added duplicate with first field: %s")
new = []
self._ids = []
self._cards = []
self._emptyNotes = False
for n in notes:
if not self.allowHTML:
for c in range(len(n.fields)):
n.fields[c] = cgi.escape(n.fields[c])
fld0 = n.fields[fld0idx]
csum = fieldChecksum(fld0)
# first field must exist
if not fld0:
self.log.append(_("Empty first field: %s") %
" ".join(n.fields))
continue
# earlier in import?
if fld0 in firsts and self.importMode != 2:
# duplicates in source file; log and ignore
self.log.append(_("Appeared twice in file: %s") %
fld0)
continue
firsts[fld0] = True
# already exists?
found = False
if csum in csums:
# csum is not a guarantee; have to check
for id in csums[csum]:
flds = self.col.db.scalar(
"select flds from notes where id = ?", id)
sflds = splitFields(flds)
if fld0 == sflds[0]:
# duplicate
found = True
if self.importMode == 0:
data = self.updateData(n, id, sflds)
if data:
updates.append(data)
updateLog.append(updateLogTxt % fld0)
found = True
break
elif self.importMode == 2:
# allow duplicates in this case
updateLog.append(dupeLogTxt % fld0)
found = False
# newly add
if not found:
data = self.newData(n)
if data:
new.append(data)
# note that we've seen this note once already
firsts[fld0] = True
self.addNew(new)
self.addUpdates(updates)
# make sure to update sflds, etc
self.col.updateFieldCache(self._ids)
# generate cards
if self.col.genCards(self._ids):
self.log.insert(0, _(
"Empty cards found. Please run Tools>Empty Cards."))
# apply scheduling updates
self.updateCards()
self.col.sched.maybeRandomizeDeck()
part1 = ngettext("%d note added", "%d notes added", len(new)) % len(new)
part2 = ngettext("%d note updated", "%d notes updated", self.updateCount) % self.updateCount
self.log.append("%s, %s." % (part1, part2))
self.log.extend(updateLog)
if self._emptyNotes:
self.log.append(_("""\
One or more notes were not imported, because they didn't generate any cards. \
This can happen when you have empty fields or when you have not mapped the \
content in the text file to the correct fields."""))
self.total = len(self._ids)
def scmhash(self, m):
"Return a hash of the schema, to see if models are compatible."
s = ""
for f in m["flds"]:
s += f["name"]
return fieldChecksum(s)
def _getOriginal(note):
csum = fieldChecksum(note.fields[0])
original_nid = mw.col.db.first(
"select id from notes where csum == {} and id != {} and mid == {}".
format(csum, note.id, note.mid))[0]
return mw.col.getNote(original_nid)
def importNotes(self, notes):
"Convert each card into a note, apply attributes and add to col."
assert self.mappingOk()
# note whether tags are mapped
self._tagsMapped = False
for f in self.mapping:
if f == "_tags":
self._tagsMapped = True
# gather checks for duplicate comparison
csums = {}
for csum, id in self.col.db.execute(
"select csum, id from notes where mid = ?", self.model['id']):
if csum in csums:
csums[csum].append(id)
else:
csums[csum] = [id]
firsts = {}
fld0idx = self.mapping.index(self.model['flds'][0]['name'])
self._fmap = self.col.models.fieldMap(self.model)
self._nextID = timestampID(self.col.db, "notes")
# loop through the notes
updates = []
updateLog = []
updateLogTxt = _("First field matched: %s")
dupeLogTxt = _("Added duplicate with first field: %s")
new = []
self._ids = []
self._cards = []
self._emptyNotes = False
dupeCount = 0
dupes = []
for n in notes:
for c in range(len(n.fields)):
if not self.allowHTML:
n.fields[c] = html.escape(n.fields[c], quote=False)
n.fields[c] = n.fields[c].strip()
if not self.allowHTML:
n.fields[c] = n.fields[c].replace("\n", "<br>")
n.fields[c] = unicodedata.normalize("NFC", n.fields[c])
n.tags = [unicodedata.normalize("NFC", t) for t in n.tags]
fld0 = n.fields[fld0idx]
csum = fieldChecksum(fld0)
# first field must exist
if not fld0:
self.log.append(
_("Empty first field: %s") % " ".join(n.fields))
continue
# earlier in import?
if fld0 in firsts and self.importMode != 2:
# duplicates in source file; log and ignore
self.log.append(_("Appeared twice in file: %s") % fld0)
continue
firsts[fld0] = True
# already exists?
found = False
if csum in csums:
# csum is not a guarantee; have to check
for id in csums[csum]:
flds = self.col.db.scalar(
"select flds from notes where id = ?", id)
sflds = splitFields(flds)
if fld0 == sflds[0]:
# duplicate
found = True
if self.importMode == 0:
data = self.updateData(n, id, sflds)
if data:
updates.append(data)
updateLog.append(updateLogTxt % fld0)
dupeCount += 1
found = True
elif self.importMode == 1:
dupeCount += 1
elif self.importMode == 2:
# allow duplicates in this case
if fld0 not in dupes:
# only show message once, no matter how many
# duplicates are in the collection already
updateLog.append(dupeLogTxt % fld0)
dupes.append(fld0)
found = False
# newly add
if not found:
data = self.newData(n)
if data:
new.append(data)
# note that we've seen this note once already
firsts[fld0] = True
self.addNew(new)
self.addUpdates(updates)
# make sure to update sflds, etc
self.col.updateFieldCache(self._ids)
# generate cards
if self.col.genCards(self._ids):
self.log.insert(
0, _("Empty cards found. Please run Tools>Empty Cards."))
# apply scheduling updates
self.updateCards()
# we randomize or order here, to ensure that siblings
# have the same due#
did = self.col.decks.selected()
conf = self.col.decks.confForDid(did)
# in order due?
if conf['new']['order'] == NEW_CARDS_RANDOM:
self.col.sched.randomizeCards(did)
part1 = ngettext("%d note added", "%d notes added",
len(new)) % len(new)
part2 = ngettext("%d note updated", "%d notes updated",
self.updateCount) % self.updateCount
if self.importMode == 0:
unchanged = dupeCount - self.updateCount
elif self.importMode == 1:
unchanged = dupeCount
else:
unchanged = 0
part3 = ngettext("%d note unchanged", "%d notes unchanged",
unchanged) % unchanged
self.log.append("%s, %s, %s." % (part1, part2, part3))
self.log.extend(updateLog)
if self._emptyNotes:
self.log.append(
_("""\
One or more notes were not imported, because they didn't generate any cards. \
This can happen when you have empty fields or when you have not mapped the \
content in the text file to the correct fields."""))
self.total = len(self._ids)
def importNotes(self, notes):
"Convert each card into a note, apply attributes and add to col."
assert self.mappingOk()
# note whether tags are mapped
self._tagsMapped = False
for f in self.mapping:
if f == "_tags":
self._tagsMapped = True
# gather checks for duplicate comparison
csums = {}
for csum, id in self.col.db.execute(
"select csum, id from notes where mid = ?", self.model['id']):
if csum in csums:
csums[csum].append(id)
else:
csums[csum] = [id]
firsts = {}
fld0idx = self.mapping.index(self.model['flds'][0]['name'])
self._fmap = self.col.models.fieldMap(self.model)
self._nextID = timestampID(self.col.db, "notes")
# loop through the notes
updates = []
updateLog = []
updateLogTxt = _("First field matched: %s")
dupeLogTxt = _("Added duplicate with first field: %s")
new = []
self._ids = []
self._cards = []
self._emptyNotes = False
dupeCount = 0
dupes = []
for n in notes:
for c in range(len(n.fields)):
if not self.allowHTML:
n.fields[c] = cgi.escape(n.fields[c])
n.fields[c] = n.fields[c].strip().replace("\n", "<br>")
fld0 = n.fields[fld0idx]
csum = fieldChecksum(fld0)
# first field must exist
if not fld0:
self.log.append(_("Empty first field: %s") %
" ".join(n.fields))
continue
# earlier in import?
if fld0 in firsts and self.importMode != 2:
# duplicates in source file; log and ignore
self.log.append(_("Appeared twice in file: %s") %
fld0)
continue
firsts[fld0] = True
# already exists?
found = False
if csum in csums:
# csum is not a guarantee; have to check
for id in csums[csum]:
flds = self.col.db.scalar(
"select flds from notes where id = ?", id)
sflds = splitFields(flds)
if fld0 == sflds[0]:
# duplicate
found = True
if self.importMode == 0:
data = self.updateData(n, id, sflds)
if data:
updates.append(data)
updateLog.append(updateLogTxt % fld0)
dupeCount += 1
found = True
elif self.importMode == 1:
dupeCount += 1
elif self.importMode == 2:
# allow duplicates in this case
if fld0 not in dupes:
# only show message once, no matter how many
# duplicates are in the collection already
updateLog.append(dupeLogTxt % fld0)
dupes.append(fld0)
found = False
# newly add
if not found:
data = self.newData(n)
if data:
new.append(data)
# note that we've seen this note once already
firsts[fld0] = True
self.addNew(new)
self.addUpdates(updates)
# make sure to update sflds, etc
self.col.updateFieldCache(self._ids)
# generate cards
if self.col.genCards(self._ids):
self.log.insert(0, _(
"Empty cards found. Please run Tools>Empty Cards."))
# apply scheduling updates
self.updateCards()
# we randomize or order here, to ensure that siblings
# have the same due#
did = self.col.decks.selected()
conf = self.col.decks.confForDid(did)
# in order due?
if conf['new']['order'] == NEW_CARDS_RANDOM:
self.col.sched.randomizeCards(did)
else:
self.col.sched.orderCards(did)
part1 = ngettext("%d note added", "%d notes added",
len(new)) % len(new)
part2 = ngettext("%d note updated", "%d notes updated",
self.updateCount) % self.updateCount
if self.importMode == 0:
unchanged = dupeCount - self.updateCount
elif self.importMode == 1:
unchanged = dupeCount
else:
unchanged = 0
part3 = ngettext("%d note unchanged", "%d notes unchanged",
unchanged) % unchanged
self.log.append("%s, %s, %s." % (part1, part2, part3))
self.log.extend(updateLog)
if self._emptyNotes:
self.log.append(_("""\
One or more notes were not imported, because they didn't generate any cards. \
This can happen when you have empty fields or when you have not mapped the \
content in the text file to the correct fields."""))
self.total = len(self._ids)
def importNotes(self, notes: list[ForeignNote]) -> None:
"Convert each card into a note, apply attributes and add to col."
assert self.mappingOk()
# note whether tags are mapped
self._tagsMapped = False
for f in self.mapping:
if f == "_tags":
self._tagsMapped = True
# gather checks for duplicate comparison
csums: dict[str, list[NoteId]] = {}
for csum, id in self.col.db.execute(
"select csum, id from notes where mid = ?", self.model["id"]):
if csum in csums:
csums[csum].append(id)
else:
csums[csum] = [id]
firsts: dict[str, bool] = {}
fld0idx = self.mapping.index(self.model["flds"][0]["name"])
self._fmap = self.col.models.field_map(self.model)
self._nextID = NoteId(timestampID(self.col.db, "notes"))
# loop through the notes
updates: list[Updates] = []
updateLog = []
new = []
self._ids: list[NoteId] = []
self._cards: list[tuple] = []
dupeCount = 0
dupes: list[str] = []
for n in notes:
for c, field in enumerate(n.fields):
if not self.allowHTML:
n.fields[c] = html.escape(field, quote=False)
n.fields[c] = field.strip()
if not self.allowHTML:
n.fields[c] = field.replace("\n", "<br>")
fld0 = unicodedata.normalize("NFC", n.fields[fld0idx])
# first field must exist
if not fld0:
self.log.append(
self.col.tr.importing_empty_first_field(
val=" ".join(n.fields)))
continue
csum = fieldChecksum(fld0)
# earlier in import?
if fld0 in firsts and self.importMode != ADD_MODE:
# duplicates in source file; log and ignore
self.log.append(
self.col.tr.importing_appeared_twice_in_file(val=fld0))
continue
firsts[fld0] = True
# already exists?
found = False
if csum in csums:
# csum is not a guarantee; have to check
for id in csums[csum]:
flds = self.col.db.scalar(
"select flds from notes where id = ?", id)
sflds = splitFields(flds)
if fld0 == sflds[0]:
# duplicate
found = True
if self.importMode == UPDATE_MODE:
data = self.updateData(n, id, sflds)
if data:
updates.append(data)
updateLog.append(
self.col.tr.importing_first_field_matched(
val=fld0))
dupeCount += 1
found = True
elif self.importMode == IGNORE_MODE:
dupeCount += 1
elif self.importMode == ADD_MODE:
# allow duplicates in this case
if fld0 not in dupes:
# only show message once, no matter how many
# duplicates are in the collection already
updateLog.append(
self.col.tr.
importing_added_duplicate_with_first_field(
val=fld0, ))
dupes.append(fld0)
found = False
# newly add
if not found:
new_data = self.newData(n)
if new_data:
new.append(new_data)
# note that we've seen this note once already
firsts[fld0] = True
self.addNew(new)
self.addUpdates(updates)
# generate cards + update field cache
self.col.after_note_updates(self._ids, mark_modified=False)
# apply scheduling updates
self.updateCards()
# we randomize or order here, to ensure that siblings
# have the same due#
did = self.col.decks.selected()
conf = self.col.decks.config_dict_for_deck_id(did)
# in order due?
if not conf["dyn"] and conf["new"]["order"] == NEW_CARDS_RANDOM:
self.col.sched.randomizeCards(did)
part1 = self.col.tr.importing_note_added(count=len(new))
part2 = self.col.tr.importing_note_updated(count=self.updateCount)
if self.importMode == UPDATE_MODE:
unchanged = dupeCount - self.updateCount
elif self.importMode == IGNORE_MODE:
unchanged = dupeCount
else:
unchanged = 0
part3 = self.col.tr.importing_note_unchanged(count=unchanged)
self.log.append(f"{part1}, {part2}, {part3}.")
self.log.extend(updateLog)
self.total = len(self._ids)
def updateNotes(allDb):
t_0, now, db, TAG = time.time(), intTime(), mw.col.db, mw.col.tags
ds, nid2mmi = [], {}
N_notes = db.scalar('select count() from notes')
mw.progress.start(label='Updating data', max=N_notes, immediate=True)
fidDb = allDb.fidDb()
locDb = allDb.locDb(recalc=False) # fidDb() already forces locDb recalc
# handle secondary databases
mw.progress.update(label='Creating seen/known/mature from all.db')
seenDb = filterDbByMat(allDb, cfg1('threshold_seen'))
knownDb = filterDbByMat(allDb, cfg1('threshold_known'))
matureDb = filterDbByMat(allDb, cfg1('threshold_mature'))
mw.progress.update(label='Loading priority.db')
priorityDb = MorphDb(cfg1('path_priority'), ignoreErrors=True).db
if cfg1('saveDbs'):
mw.progress.update(label='Saving seen/known/mature dbs')
seenDb.save(cfg1('path_seen'))
knownDb.save(cfg1('path_known'))
matureDb.save(cfg1('path_mature'))
mw.progress.update(label='Calculating frequency information')
pops = [len(locs) for locs in allDb.db.values()]
pops = [n for n in pops if n > 1]
mw.progress.update(label='Updating notes')
for i, (nid, mid, flds, guid, tags) in enumerate(
db.execute('select id, mid, flds, guid, tags from notes')):
if i % 500 == 0: mw.progress.update(value=i)
C = partial(cfg, mid, None)
if not C('enabled'): continue
# Get all morphemes for note
ms = set()
for fieldName in C('morph_fields'):
try:
loc = fidDb[(nid, guid, fieldName)]
ms.update(locDb[loc])
except KeyError:
continue
ms = [m for m in ms if m.pos not in C('morph_blacklist')]
# Determine un-seen/known/mature and i+N
unseens, unknowns, unmatures, newKnowns = set(), set(), set(), set()
for m in ms:
if m not in seenDb.db: unseens.add(m)
if m not in knownDb.db: unknowns.add(m)
if m not in matureDb.db: unmatures.add(m)
if m not in matureDb.db and m in knownDb.db:
newKnowns.add(m)
# Determine MMI - Morph Man Index
N, N_s, N_k, N_m = len(ms), len(unseens), len(unknowns), len(unmatures)
# Bail early for lite update
if N_k > 2 and C('only update k+2 and below'): continue
# average frequency of unknowns (ie. how common the word is within your collection)
F_k = 0
for focusMorph in unknowns: # focusMorph used outside loop
F_k += len(allDb.db[focusMorph])
F_k_avg = F_k / N_k if N_k > 0 else F_k
usefulness = F_k_avg
# add bonus for morphs in priority.db
isPriority = False
for focusMorph in unknowns:
if focusMorph in priorityDb:
isPriority = True
usefulness += C('priority.db weight')
# add bonus for studying recent learned knowns (reinforce)
for m in newKnowns:
locs = allDb.db[m]
if locs:
ivl = min(1, max(loc.maturity for loc in locs))
usefulness += C(
'reinforce new vocab weight'
) / ivl #TODO: maybe average this so it doesnt favor long sentences
if any(m.pos == u'動詞'
for m in unknowns): #FIXME: this isn't working???
usefulness += C('verb bonus')
usefulness = 999 - min(999, usefulness)
# difference from optimal length (too little context vs long sentence)
lenDiff = max(0, min(9, abs(C('optimal sentence length') - N) - 2))
tooLong = N > C('optimal sentence length')
# calculate mmi
mmi = 10000 * N_k + 1000 * lenDiff + usefulness
if C('set due based on mmi'):
nid2mmi[nid] = mmi
# Fill in various fields/tags on the note based on cfg
ts, fs = TAG.split(tags), splitFields(flds)
# determine card type
compTag, vocabTag, notReadyTag, alreadyKnownTag, priorityTag, badLengthTag, tooLongTag = tagNames = C(
'tag_comprehension'), C('tag_vocab'), C('tag_notReady'), C(
'tag_alreadyKnown'), C('tag_priority'), C('tag_badLength'), C(
'tag_tooLong')
if N_m == 0: # sentence comprehension card, m+0
ts = [compTag
] + [t for t in ts if t not in [vocabTag, notReadyTag]]
setField(mid, fs, C('focusMorph'), u'')
elif N_k == 1: # new vocab card, k+1
ts = [vocabTag
] + [t for t in ts if t not in [compTag, notReadyTag]]
setField(mid, fs, C('focusMorph'), u'%s' % focusMorph.base)
elif N_k > 1: # M+1+ and K+2+
ts = [notReadyTag
] + [t for t in ts if t not in [compTag, vocabTag]]
# set type agnostic fields
setField(mid, fs, C('k+N'), u'%d' % N_k)
setField(mid, fs, C('m+N'), u'%d' % N_m)
setField(mid, fs, C('morphManIndex'), u'%d' % mmi)
setField(mid, fs, C('unknowns'), u', '.join(u.base for u in unknowns))
setField(mid, fs, C('unmatures'),
u', '.join(u.base for u in unmatures))
setField(mid, fs, C('unknownFreq'), u'%d' % F_k_avg)
# other tags
if priorityTag in ts: ts.remove(priorityTag)
if isPriority: ts.append(priorityTag)
if badLengthTag in ts: ts.remove(badLengthTag)
if lenDiff: ts.append(badLengthTag)
if tooLongTag in ts: ts.remove(tooLongTag)
if tooLong: ts.append(tooLongTag)
# update sql db
tags_ = TAG.join(TAG.canonify(ts))
flds_ = joinFields(fs)
if flds != flds_ or tags != tags_: # only update notes that have changed
csum = fieldChecksum(fs[0])
sfld = stripHTML(fs[getSortFieldIndex(mid)])
ds.append({
'now': now,
'tags': tags_,
'flds': flds_,
'sfld': sfld,
'csum': csum,
'usn': mw.col.usn(),
'nid': nid
})
mw.progress.update(value=i, label='Updating anki database...')
mw.col.db.executemany(
'update notes set tags=:tags, flds=:flds, sfld=:sfld, csum=:csum, mod=:now, usn=:usn where id=:nid',
ds)
TAG.register(tagNames)
# Now reorder new cards based on MMI
mw.progress.update(value=i, label='Updating new card ordering...')
ds = []
for (cid, nid,
due) in db.execute('select id, nid, due from cards where type = 0'):
if nid in nid2mmi: # owise it was disabled
due_ = nid2mmi[nid]
if due != due_: # only update cards that have changed
ds.append({
'now': now,
'due': due_,
'usn': mw.col.usn(),
'cid': cid
})
mw.col.db.executemany(
'update cards set due=:due, mod=:now, usn=:usn where id=:cid', ds)
mw.reset()
printf('Updated notes in %f sec' % (time.time() - t_0))
mw.progress.finish()
return knownDb
def updateNotes(allDb):
t_0, now, db, TAG = time.time(), intTime(), mw.col.db, mw.col.tags
ds, nid2mmi = [], {}
N_notes = db.scalar('select count() from notes')
mw.progress.start(label='Updating data', max=N_notes, immediate=True)
fidDb = allDb.fidDb()
locDb = allDb.locDb(recalc=False) # fidDb() already forces locDb recalc
# read tag names
compTag, vocabTag, freshTag, notReadyTag, alreadyKnownTag, priorityTag, tooShortTag, tooLongTag, frequencyTag = tagNames = jcfg(
'Tag_Comprehension'), jcfg('Tag_Vocab'), jcfg('Tag_Fresh'), jcfg(
'Tag_NotReady'), jcfg('Tag_AlreadyKnown'), jcfg(
'Tag_Priority'), jcfg('Tag_TooShort'), jcfg(
'Tag_TooLong'), jcfg('Tag_Frequency')
TAG.register(tagNames)
badLengthTag = jcfg2().get('Tag_BadLength')
# handle secondary databases
mw.progress.update(label='Creating seen/known/mature from all.db')
seenDb = filterDbByMat(allDb, cfg1('threshold_seen'))
knownDb = filterDbByMat(allDb, cfg1('threshold_known'))
matureDb = filterDbByMat(allDb, cfg1('threshold_mature'))
mw.progress.update(label='Loading priority.db')
priorityDb = MorphDb(cfg1('path_priority'), ignoreErrors=True).db
mw.progress.update(label='Loading frequency.txt')
frequencyListPath = cfg1('path_frequency')
try:
with codecs.open(frequencyListPath, 'r', 'utf-8') as f:
frequencyList = [line.strip() for line in f.readlines()]
frequencyListLength = len(frequencyList)
except FileNotFoundError:
pass # User does not have a frequency.txt
if cfg1('saveDbs'):
mw.progress.update(label='Saving seen/known/mature dbs')
seenDb.save(cfg1('path_seen'))
knownDb.save(cfg1('path_known'))
matureDb.save(cfg1('path_mature'))
mw.progress.update(label='Updating notes')
for i, (nid, mid, flds, guid, tags) in enumerate(
db.execute('select id, mid, flds, guid, tags from notes')):
if i % 500 == 0: mw.progress.update(value=i)
C = partial(cfg, mid, None)
note = mw.col.getNote(nid)
notecfg = getFilter(note)
if notecfg is None or not notecfg['Modify']: continue
# Get all morphemes for note
morphemes = set()
for fieldName in notecfg['Fields']:
try:
loc = fidDb[(nid, guid, fieldName)]
morphemes.update(locDb[loc])
except KeyError:
continue
# Determine un-seen/known/mature and i+N
unseens, unknowns, unmatures, newKnowns = set(), set(), set(), set()
for morpheme in morphemes:
if morpheme not in seenDb.db: unseens.add(morpheme)
if morpheme not in knownDb.db: unknowns.add(morpheme)
if morpheme not in matureDb.db: unmatures.add(morpheme)
if morpheme not in matureDb.db and morpheme in knownDb.db:
newKnowns.add(morpheme)
# Determine MMI - Morph Man Index
N, N_s, N_k, N_m = len(morphemes), len(unseens), len(unknowns), len(
unmatures)
# Bail early for lite update
if N_k > 2 and C('only update k+2 and below'): continue
# average frequency of unknowns (ie. how common the word is within your collection)
F_k = 0
for focusMorph in unknowns: # focusMorph used outside loop
F_k += allDb.frequency(focusMorph)
F_k_avg = F_k // N_k if N_k > 0 else F_k
usefulness = F_k_avg
# add bonus for morphs in priority.db and frequency.txt
isPriority = False
isFrequency = False
for focusMorph in unknowns:
if focusMorph in priorityDb:
isPriority = True
usefulness += C('priority.db weight')
focusMorphString = focusMorph.show().split()[0]
try:
focusMorphIndex = frequencyList.index(focusMorphString)
isFrequency = True
frequencyWeight = C('frequency.txt weight scale')
# The bigger this number, the lower mmi becomes
usefulness += (frequencyListLength -
focusMorphIndex) * frequencyWeight
except:
pass
# add bonus for studying recent learned knowns (reinforce)
for morpheme in newKnowns:
locs = allDb.db[morpheme]
if locs:
ivl = min(1, max(loc.maturity for loc in locs))
usefulness += C(
'reinforce new vocab weight'
) // ivl #TODO: maybe average this so it doesnt favor long sentences
if any(morpheme.pos == '動詞'
for morpheme in unknowns): #FIXME: this isn't working???
usefulness += C('verb bonus')
usefulness = 99999 - min(99999, usefulness)
# difference from optimal length range (too little context vs long sentence)
lenDiffRaw = min(N - C('min good sentence length'),
max(0, N - C('max good sentence length')))
lenDiff = min(9, abs(lenDiffRaw))
# calculate mmi
mmi = 100000 * N_k + 1000 * lenDiff + usefulness
if C('set due based on mmi'):
nid2mmi[nid] = mmi
# Fill in various fields/tags on the note based on cfg
ts, fs = TAG.split(tags), splitFields(flds)
# clear any 'special' tags, the appropriate will be set in the next few lines
ts = [
t for t in ts
if t not in [notReadyTag, compTag, vocabTag, freshTag]
]
# determine card type
if N_m == 0: # sentence comprehension card, m+0
ts = ts + [compTag]
elif N_k == 1: # new vocab card, k+1
ts = ts + [vocabTag]
setField(mid, fs, jcfg('Field_FocusMorph'), '%s' % focusMorph.base)
elif N_k > 1: # M+1+ and K+2+
ts = ts + [notReadyTag]
setField(mid, fs, jcfg('Field_FocusMorph'), '')
elif N_m == 1: # we have k+0, and m+1, so this card does not introduce a new vocabulary -> card for newly learned morpheme
ts = ts + [freshTag]
setField(mid, fs, jcfg('Field_FocusMorph'),
'%s' % list(unmatures)[0].base)
else: # only case left: we have k+0, but m+2 or higher, so this card does not introduce a new vocabulary -> card for newly learned morpheme
ts = ts + [freshTag]
setField(mid, fs, jcfg('Field_FocusMorph'), '')
# set type agnostic fields
setField(mid, fs, jcfg('Field_UnknownMorphCount'), '%d' % N_k)
setField(mid, fs, jcfg('Field_UnmatureMorphCount'), '%d' % N_m)
setField(mid, fs, jcfg('Field_MorphManIndex'), '%d' % mmi)
setField(mid, fs, jcfg('Field_Unknowns'),
', '.join(u.base for u in unknowns))
setField(mid, fs, jcfg('Field_Unmatures'),
', '.join(u.base for u in unmatures))
setField(mid, fs, jcfg('Field_UnknownFreq'), '%d' % F_k_avg)
# remove deprecated tag
if badLengthTag is not None and badLengthTag in ts:
ts.remove(badLengthTag)
# other tags
if priorityTag in ts: ts.remove(priorityTag)
if isPriority: ts.append(priorityTag)
if frequencyTag in ts: ts.remove(frequencyTag)
if isFrequency: ts.append(frequencyTag)
if tooShortTag in ts: ts.remove(tooShortTag)
if lenDiffRaw < 0: ts.append(tooShortTag)
if tooLongTag in ts: ts.remove(tooLongTag)
if lenDiffRaw > 0: ts.append(tooLongTag)
# remove unnecessary tags
if not jcfg('Option_SetNotRequiredTags'):
unnecessary = [priorityTag, tooShortTag, tooLongTag]
ts = [tag for tag in ts if tag not in unnecessary]
# update sql db
tags_ = TAG.join(TAG.canonify(ts))
flds_ = joinFields(fs)
if flds != flds_ or tags != tags_: # only update notes that have changed
csum = fieldChecksum(fs[0])
sfld = stripHTML(fs[getSortFieldIndex(mid)])
ds.append({
'now': now,
'tags': tags_,
'flds': flds_,
'sfld': sfld,
'csum': csum,
'usn': mw.col.usn(),
'nid': nid
})
mw.progress.update(value=i, label='Updating anki database...')
mw.col.db.executemany(
'update notes set tags=:tags, flds=:flds, sfld=:sfld, csum=:csum, mod=:now, usn=:usn where id=:nid',
ds)
# Now reorder new cards based on MMI
mw.progress.update(value=i, label='Updating new card ordering...')
ds = []
# "type = 0": new cards
# "type = 1": learning cards [is supposed to be learning: in my case no learning card had this type]
# "type = 2": review cards
for (cid, nid,
due) in db.execute('select id, nid, due from cards where type = 0'):
if nid in nid2mmi: # owise it was disabled
due_ = nid2mmi[nid]
if due != due_: # only update cards that have changed
ds.append({
'now': now,
'due': due_,
'usn': mw.col.usn(),
'cid': cid
})
mw.col.db.executemany(
'update cards set due=:due, mod=:now, usn=:usn where id=:cid', ds)
mw.reset()
printf('Updated notes in %f sec' % (time.time() - t_0))
mw.progress.finish()
return knownDb
def updateNotes( allDb ):
t_0, now, db, TAG = time.time(), intTime(), mw.col.db, mw.col.tags
ds, nid2mmi = [], {}
N_notes = db.scalar( 'select count() from notes' )
mw.progress.start( label='Updating data', max=N_notes, immediate=True )
fidDb = allDb.fidDb()
locDb = allDb.locDb( recalc=False ) # fidDb() already forces locDb recalc
# handle secondary databases
mw.progress.update( label='Creating seen/known/mature from all.db' )
seenDb = filterDbByMat( allDb, cfg1('threshold_seen') )
knownDb = filterDbByMat( allDb, cfg1('threshold_known') )
matureDb = filterDbByMat( allDb, cfg1('threshold_mature') )
mw.progress.update( label='Loading priority.db' )
priorityDb = MorphDb( cfg1('path_priority'), ignoreErrors=True ).db
if cfg1('saveDbs'):
mw.progress.update( label='Saving seen/known/mature dbs' )
seenDb.save( cfg1('path_seen') )
knownDb.save( cfg1('path_known') )
matureDb.save( cfg1('path_mature') )
mw.progress.update( label='Calculating frequency information' )
pops = [ len( locs ) for locs in allDb.db.values() ]
pops = [ n for n in pops if n > 1 ]
mw.progress.update( label='Updating notes' )
for i,( nid, mid, flds, guid, tags ) in enumerate( db.execute( 'select id, mid, flds, guid, tags from notes' ) ):
if i % 500 == 0: mw.progress.update( value=i )
C = partial( cfg, mid, None )
if not C('enabled'): continue
# Get all morphemes for note
ms = set()
for fieldName in C('morph_fields'):
try:
loc = fidDb[ ( nid, guid, fieldName ) ]
ms.update( locDb[ loc ] )
except KeyError: continue
ms = [ m for m in ms if m.pos not in C('morph_blacklist') ]
# Determine un-seen/known/mature and i+N
unseens, unknowns, unmatures, newKnowns = set(), set(), set(), set()
for m in ms:
if m not in seenDb.db: unseens.add( m )
if m not in knownDb.db: unknowns.add( m )
if m not in matureDb.db: unmatures.add( m )
if m not in matureDb.db and m in knownDb.db:
newKnowns.add( m )
# Determine MMI - Morph Man Index
N, N_s, N_k, N_m = len( ms ), len( unseens ), len( unknowns ), len( unmatures )
# Bail early for lite update
if N_k > 2 and C('only update k+2 and below'): continue
# average frequency of unknowns (ie. how common the word is within your collection)
F_k = 0
for focusMorph in unknowns: # focusMorph used outside loop
F_k += len( allDb.db[ focusMorph ] )
F_k_avg = F_k / N_k if N_k > 0 else F_k
usefulness = F_k_avg
# add bonus for morphs in priority.db
isPriority = False
for focusMorph in unknowns:
if focusMorph in priorityDb:
isPriority = True
usefulness += C('priority.db weight')
# add bonus for studying recent learned knowns (reinforce)
for m in newKnowns:
locs = allDb.db[ m ]
if locs:
ivl = min( 1, max( loc.maturity for loc in locs ) )
usefulness += C('reinforce new vocab weight') / ivl #TODO: maybe average this so it doesnt favor long sentences
if any( m.pos == u'動詞' for m in unknowns ): #FIXME: this isn't working???
usefulness += C('verb bonus')
usefulness = 999 - min( 999, usefulness )
# difference from optimal length (too little context vs long sentence)
lenDiff = max( 0, min( 9, abs( C('optimal sentence length') - N ) -2 ) )
# calculate mmi
mmi = 10000*N_k + 1000*lenDiff + usefulness
if C('set due based on mmi'):
nid2mmi[ nid ] = mmi
# Fill in various fields/tags on the note based on cfg
ts, fs = TAG.split( tags ), splitFields( flds )
# determine card type
compTag, vocabTag, notReadyTag, alreadyKnownTag, priorityTag = tagNames = C('tag_comprehension'), C('tag_vocab'), C('tag_notReady'), C('tag_alreadyKnown'), C('tag_priority')
if N_m == 0: # sentence comprehension card, m+0
ts = [ compTag ] + [ t for t in ts if t not in [ vocabTag, notReadyTag ] ]
setField( mid, fs, C('focusMorph'), u'' )
elif N_k == 1: # new vocab card, k+1
ts = [ vocabTag ] + [ t for t in ts if t not in [ compTag, notReadyTag ] ]
setField( mid, fs, C('focusMorph'), u'%s' % focusMorph.base )
elif N_k > 1: # M+1+ and K+2+
ts = [ notReadyTag ] + [ t for t in ts if t not in [ compTag, vocabTag ] ]
# set type agnostic fields
setField( mid, fs, C('k+N'), u'%d' % N_k )
setField( mid, fs, C('m+N'), u'%d' % N_m )
setField( mid, fs, C('morphManIndex'), u'%d' % mmi )
setField( mid, fs, C('unknowns'), u', '.join( u.base for u in unknowns ) )
setField( mid, fs, C('unmatures'), u', '.join( u.base for u in unmatures ) )
setField( mid, fs, C('unknownFreq'), u'%d' % F_k_avg )
# other tags
if priorityTag in ts: ts.remove( priorityTag )
if isPriority: ts.append( priorityTag )
# update sql db
tags_ = TAG.join( TAG.canonify( ts ) )
flds_ = joinFields( fs )
if flds != flds_ or tags != tags_: # only update notes that have changed
csum = fieldChecksum( fs[0] )
sfld = stripHTML( fs[ getSortFieldIndex( mid ) ] )
ds.append( { 'now':now, 'tags':tags_, 'flds':flds_, 'sfld':sfld, 'csum':csum, 'usn':mw.col.usn(), 'nid':nid } )
mw.progress.update( value=i, label='Updating anki database...' )
mw.col.db.executemany( 'update notes set tags=:tags, flds=:flds, sfld=:sfld, csum=:csum, mod=:now, usn=:usn where id=:nid', ds )
TAG.register( tagNames )
# Now reorder new cards based on MMI
mw.progress.update( value=i, label='Updating new card ordering...' )
ds = []
for ( cid, nid, due ) in db.execute( 'select id, nid, due from cards where type = 0' ):
if nid in nid2mmi: # owise it was disabled
due_ = nid2mmi[ nid ]
if due != due_: # only update cards that have changed
ds.append( { 'now':now, 'due':due_, 'usn':mw.col.usn(), 'cid':cid } )
mw.col.db.executemany( 'update cards set due=:due, mod=:now, usn=:usn where id=:cid', ds )
mw.reset()
printf( 'Updated notes in %f sec' % ( time.time() - t_0 ) )
mw.progress.finish()
def _ignore_dupes(self_note=None, self_expression=None, self_deck=None):
"""We will override Anki's Note.dupeOrEmpty function with this function,
This method is meant to return
1 if self.fields[0] is empty
2 if the note is a duplicate
False elsewise (i.e. "nice" note).
:param self_note: Anki note object.
:param self_expression: String. Will overwrite note.fields[0]
:param self_deck: Deck the note belongs to.
"""
# Nomenclature: We compare the note given as argument to other notes.
# Everything that has to do with that initial note has the prefix 'self',
# everything that has to do with one of the other notes has the prefix
# other.
# Some explanation for abbreviations used in Anki:
# * id: Note id (as variable belonging to a note)
# * nid: Note id (as variable belonging to a card)
# * did: Deck id
# * mid: Model id
# 1. Default values & Co.
if self_note is None and self_expression is None:
# maybe we should raise a ValueError instead, but well...
return False
if self_note:
self_search_value = self_note.fields[0] # might be None!
self_note_id = self_note.id
self_note_mid = self_note.mid
else:
self_search_value = None
self_note_id = None
self_note_mid = None
if self_expression:
# Note: If self_note was given as well, self_search_value will be
# overwritten.
self_search_value = self_expression
# 2. Check if we have a key field/Expression
logger.debug("key field = '%s'" % self_search_value)
if not self_search_value or isinstance(self_search_value, str) and not \
self_search_value.strip():
# note that self_note.fields[0] might be None!
logger.debug("Key field empty.")
return 1
# 3. Get Note Ids of notes that might be duplicates.
csum = fieldChecksum(self_search_value)
if self_note_mid:
# we don't have to check for the note id, because it defaults to 0 in
# the search query (mostly copied from anki's source).
# Select all note ids from notes
# 1. whose key field has the same check sum
# 2. whose note id is different (i.e. we're excluding self_note)
# 3. whose model id is the same
other_note_ids = mw.col.db.list(
"select id from notes where csum = ? "
"and id != ? and mid = ?", csum, self_note_id or 0, self_note_mid)
else:
# don't apply any criteria for note id and mid model id, just seach
# for the checksum.
other_note_ids = mw.col.db.list("select id from notes where csum = ?",
csum)
logger.debug("other_note_ids: {}".format(other_note_ids))
if not other_note_ids:
logger.debug("Did not find any notes with the same key field checksum "
"as self.")
return False
# 4. get the self_deck ids from the decks the self card belonged to
if self_deck:
# use the deck supplied as argument
self_deck_ids = did_from_dname(self_deck)
else:
# try to get the deck from anki
self_deck_ids = mw.col.db.list("select did from cards where nid = ?",
self_note_id)
if not self_deck_ids:
# We tried to get the denk name from anki, but the result was None.
# use the self_deck id of the currently active self_deck
self_deck_ids = [mw.col.conf['curDeck']]
logger.debug("self_deck_ids {}".format(self_deck_ids))
# 5. Loop over the other_note_ids
for other_note_id in other_note_ids:
# 5a. Get the field values of note with other_note_id
other_fields = mw.col.db.list("select flds from notes where id = ?",
other_note_id)
if not other_fields:
# note with no fields
logger.debug("No fields.")
return False
# 5b. Get the self_deck ids of all the cards of the note with
# other_note_id (one note can have multiple cards in different decks)
other_deck_ids = mw.col.db.list("select did from cards where nid = ?",
other_note_id)
logger.debug("other_deck_ids {}".format(other_deck_ids))
if not other_deck_ids:
logger.debug("No cards with matching checksum.")
return False
# 5c. Check that the key fields match.
if not stripHTMLMedia(splitFields(other_fields[0])[0]) == \
stripHTMLMedia(self_search_value):
logger.debug("Expressions didn't match after all.")
return False
# 6c. Check if we want to ignore that case.
# Normally a card would be flagged as a duplicate here.
for self_deck_id in self_deck_ids:
for other_deck_id in other_deck_ids:
self_name = dname_from_did(self_deck_id)
other_name = dname_from_did(other_deck_id)
if ignore_duplicates(self_name, other_name):
# don't do anything!
logger.debug("Duplicate! deck1 = '%s', deck2 = '%s' ==> "
"Ignored." % (self_name, other_name))
else:
logger.debug("Duplicate! deck1 = '%s', deck2 = '%s' ==> "
"Flagged." % (self_name, other_name))
return 2
return False
def importNotes(self, notes):
"Convert each card into a note, apply attributes and add to col."
assert self.mappingOk()
# gather checks for duplicate comparison
csums = {}
for csum, id in self.col.db.execute(
"select csum, id from notes where mid = ?", self.model['id']):
if csum in csums:
csums[csum].append(id)
else:
csums[csum] = [id]
firsts = {}
fld0idx = self.mapping.index(self.model['flds'][0]['name'])
self._fmap = self.col.models.fieldMap(self.model)
self._nextID = timestampID(self.col.db, "notes")
# loop through the notes
updates = []
new = []
self._ids = []
self._cards = []
for n in notes:
fld0 = n.fields[fld0idx]
csum = fieldChecksum(fld0)
# first field must exist
if not fld0:
self.log.append(_("Empty first field: %s") %
" ".join(n.fields))
continue
# earlier in import?
if fld0 in firsts:
# duplicates in source file; log and ignore
self.log.append(_("Appeared twice in file: %s") %
fld0)
continue
firsts[fld0] = True
# already exists?
found = False
if csum in csums:
# csum is not a guarantee; have to check
for id in csums[csum]:
flds = self.col.db.scalar(
"select flds from notes where id = ?", id)
sflds = splitFields(flds)
if fld0 == sflds[0]:
# duplicate
found = True
if self.update:
data = self.updateData(n, id, sflds)
if data:
updates.append(data)
found = True
break
# newly add
if not found:
data = self.newData(n)
if data:
new.append(data)
# note that we've seen this note once already
firsts[fld0] = True
self.addNew(new)
self.addUpdates(updates)
self.col.updateFieldCache(self._ids)
# generate cards
if self.col.genCards(self._ids):
self.log.insert(0, _(
"Empty cards found. Please run Tools>Maintenance>Empty Cards."))
# apply scheduling updates
self.updateCards()
# make sure to update sflds, etc
self.log.append(_("%(a)d notes added, %(b)d notes updated.") %
dict(a=len(new), b=self.updateCount))
self.total = len(self._ids)
def maybeChecksum(self, data, unique):
if not unique:
return ""
return fieldChecksum(data)
def importNotes(self, notes):
"Convert each card into a note, apply attributes and add to col."
print "importNotes()"
print notes
assert self.mappingOk()
# note whether tags are mapped
self._tagsMapped = False
for f in self.mapping:
if f == "_tags":
self._tagsMapped = True
# gather checks for duplicate comparison
csums = {}
for csum, id in self.col.db.execute(
"select csum, id from notes where mid = ?", self.model['id']):
if csum in csums:
csums[csum].append(id)
else:
csums[csum] = [id]
firsts = {}
fld0idx = self.mapping.index(self.model['flds'][0]['name'])
self._fmap = self.col.models.fieldMap(self.model)
self._nextID = timestampID(self.col.db, "notes")
# loop through the notes
updates = []
updateLog = []
updateLogTxt = _("First field matched: %s")
dupeLogTxt = _("Added duplicate with first field: %s")
new = []
self._ids = []
self._cards = []
self._emptyNotes = False
for n in notes:
if not self.allowHTML:
for c in range(len(n.fields)):
n.fields[c] = cgi.escape(n.fields[c])
fld0 = n.fields[fld0idx]
csum = fieldChecksum(fld0)
# first field must exist
if not fld0:
self.log.append(_("Empty first field: %s") %
" ".join(n.fields))
continue
# earlier in import?
if fld0 in firsts and self.importMode != 2:
# duplicates in source file; log and ignore
self.log.append(_("Appeared twice in file: %s") %
fld0)
continue
firsts[fld0] = True
# already exists?
found = False
if csum in csums:
# csum is not a guarantee; have to check
for id in csums[csum]:
flds = self.col.db.scalar(
"select flds from notes where id = ?", id)
sflds = splitFields(flds)
if fld0 == sflds[0]:
# duplicate
found = True
if self.importMode == 0:
data = self.updateData(n, id, sflds)
if data:
updates.append(data)
updateLog.append(updateLogTxt % fld0)
found = True
break
elif self.importMode == 2:
# allow duplicates in this case
updateLog.append(dupeLogTxt % fld0)
found = False
# newly add
if not found:
data = self.newData(n)
if data:
new.append(data)
# note that we've seen this note once already
firsts[fld0] = True
did = self.col.decks.id(self.deck)
self.col.decks.select(did)
#SUPER IMPORTANT (setting the associated deck to the model)
self.model['did'] = did
print "Selected: ", self.col.decks.get(self.col.decks.selected());
self.addNew(new)
self.addUpdates(updates)
# make sure to update sflds, etc
self.col.updateFieldCache(self._ids)
# generate cards
if self.col.genCards(self._ids):
self.log.insert(0, _(
"Empty cards found. Please run Tools>Empty Cards."))
# apply scheduling updates
self.updateCards()
# we randomize or order here, to ensure that siblings
# have the same due#
# m = self.col.models.byName("Basic")
# deck = self.col.decks.get(did)
# deck['mid'] = m['id']
# self.col.decks.save(deck)
# print "Deck:", self.col.decks.byName(self.deck)
# print "DID:", did
# save tags to model
# m = self.note.model()
# m['tags'] = self.note.tags
# self.mw.col.models.save(m)
conf = self.col.decks.confForDid(did)
# print "Conf: ",conf
# in order due?
if conf['new']['order'] == NEW_CARDS_RANDOM:
self.col.sched.randomizeCards(did)
else:
self.col.sched.orderCards(did)
part1 = ngettext("%d note added", "%d notes added", len(new)) % len(new)
part2 = ngettext("%d note updated", "%d notes updated", self.updateCount) % self.updateCount
self.log.append("%s, %s." % (part1, part2))
print part1, part2, "on deck: [", self.deck, "]"
self.log.extend(updateLog)
if self._emptyNotes:
print "there were empty notes"
self.log.append(_("""\
One or more notes were not imported, because they didn't generate any cards. \
This can happen when you have empty fields or when you have not mapped the \
content in the text file to the correct fields."""))
self.total = len(self._ids)
try:
if re.search("(?i)^" + regex + "$", strg):
nids.append(id)
except sre_constants.error:
return
if not nids:
return "0"
return "n.id in %s" % ids2str(nids)
def _findDupes(self, (val, args)):
# caller must call stripHTMLMedia on passed val
try:
mid, val = val.split(",", 1)
except OSError:
return
csum = fieldChecksum(val)
nids = []
for nid, flds in self.col.db.execute("select id, flds from notes where mid=? and csum=?", mid, csum):
if stripHTMLMedia(splitFields(flds)[0]) == val:
nids.append(nid)
return "n.id in %s" % ids2str(nids)
# Find and replace
##########################################################################
def findReplace(col, nids, src, dst, regex=False, field=None, fold=True):
"Find and replace fields in a note."
mmap = {}
if field:
def importNotes(self, notes):
"Convert each card into a note, apply attributes and add to col."
assert self.mappingOk()
# note whether tags are mapped
self._tagsMapped = False
for f in self.mapping:
if f == "_tags":
self._tagsMapped = True
# gather checks for duplicate comparison
csums = {}
for csum, id in self.col.db.execute(
"select csum, id from notes where mid = ?", self.model['id']):
if csum in csums:
csums[csum].append(id)
else:
csums[csum] = [id]
firsts = {}
fld0idx = self.mapping.index(self.model['flds'][0]['name'])
self._fmap = self.col.models.fieldMap(self.model)
self._nextID = timestampID(self.col.db, "notes")
# loop through the notes
updates = []
new = []
self._ids = []
self._cards = []
self._emptyNotes = False
for n in notes:
if not self.allowHTML:
for c in range(len(n.fields)):
n.fields[c] = cgi.escape(n.fields[c])
fld0 = n.fields[fld0idx]
csum = fieldChecksum(fld0)
# first field must exist
if not fld0:
self.log.append(_("Empty first field: %s") %
" ".join(n.fields))
continue
# earlier in import?
if fld0 in firsts and self.importMode != 2:
# duplicates in source file; log and ignore
self.log.append(_("Appeared twice in file: %s") %
fld0)
continue
firsts[fld0] = True
# already exists?
found = False
if csum in csums:
# csum is not a guarantee; have to check
for id in csums[csum]:
flds = self.col.db.scalar(
"select flds from notes where id = ?", id)
sflds = splitFields(flds)
if fld0 == sflds[0]:
# duplicate
found = True
if self.importMode == 0:
data = self.updateData(n, id, sflds)
if data:
updates.append(data)
found = True
break
elif self.importMode == 2:
# allow duplicates in this case
found = False
# newly add
if not found:
data = self.newData(n)
if data:
new.append(data)
# note that we've seen this note once already
firsts[fld0] = True
self.addNew(new)
self.addUpdates(updates)
self.col.updateFieldCache(self._ids)
# generate cards
if self.col.genCards(self._ids):
self.log.insert(0, _(
"Empty cards found. Please run Tools>Empty Cards."))
# apply scheduling updates
self.updateCards()
# make sure to update sflds, etc
part1 = ngettext("%d note added", "%d notes added", len(new)) % len(new)
part2 = ngettext("%d note updated", "%d notes updated", self.updateCount) % self.updateCount
self.log.append("%s, %s." % (part1, part2))
if self._emptyNotes:
self.log.append(_("""\
One or more notes were not imported, because they didn't generate any cards. \
This can happen when you have empty fields or when you have not mapped the \
content in the text file to the correct fields."""))
self.total = len(self._ids)
def importNotes(self, notes):
"Convert each card into a note, apply attributes and add to col."
assert self.mappingOk()
# gather checks for duplicate comparison
csums = {}
for csum, id in self.col.db.execute(
"select csum, id from notes where mid = ?", self.model['id']):
if csum in csums:
csums[csum].append(id)
else:
csums[csum] = [id]
firsts = {}
fld0idx = self.mapping.index(self.model['flds'][0]['name'])
self._fmap = self.col.models.fieldMap(self.model)
self._nextID = timestampID(self.col.db, "notes")
# loop through the notes
updates = []
new = []
self._ids = []
self._cards = []
for n in notes:
fld0 = n.fields[fld0idx]
csum = fieldChecksum(fld0)
# first field must exist
if not fld0:
self.log.append(
_("Empty first field: %s") % " ".join(n.fields))
continue
# earlier in import?
if fld0 in firsts:
# duplicates in source file; log and ignore
self.log.append(_("Appeared twice in file: %s") % fld0)
continue
firsts[fld0] = True
# already exists?
found = False
if csum in csums:
# csum is not a guarantee; have to check
for id in csums[csum]:
flds = self.col.db.scalar(
"select flds from notes where id = ?", id)
sflds = splitFields(flds)
if fld0 == sflds[0]:
# duplicate
found = True
if self.update:
data = self.updateData(n, id, sflds)
if data:
updates.append(data)
found = True
break
# newly add
if not found:
data = self.newData(n)
if data:
new.append(data)
# note that we've seen this note once already
firsts[fld0] = True
self.addNew(new)
self.addUpdates(updates)
self.col.updateFieldCache(self._ids)
# generate cards
if self.col.genCards(self._ids):
self.log.insert(
0,
_("Empty cards found. Please run Tools>Maintenance>Empty Cards."
))
# apply scheduling updates
self.updateCards()
# make sure to update sflds, etc
self.log.append(
_("%(a)d notes added, %(b)d notes updated.") %
dict(a=len(new), b=self.updateCount))
self.total = len(self._ids)
def updateNotes(allDb):
t_0, now, db = time.time(), intTime(), mw.col.db
TAG = mw.col.tags # type: TagManager
ds, nid2mmi = [], {}
N_notes = db.scalar('select count() from notes')
mw.progress.start(label='Updating data', max=N_notes, immediate=True)
fidDb = allDb.fidDb(recalc=True)
loc_db = allDb.locDb(recalc=False) # type: Dict[Location, Set[Morpheme]]
# read tag names
compTag, vocabTag, freshTag, notReadyTag, alreadyKnownTag, priorityTag, tooShortTag, tooLongTag, frequencyTag = tagNames = cfg(
'Tag_Comprehension'), cfg('Tag_Vocab'), cfg('Tag_Fresh'), cfg(
'Tag_NotReady'), cfg('Tag_AlreadyKnown'), cfg('Tag_Priority'), cfg(
'Tag_TooShort'), cfg('Tag_TooLong'), cfg('Tag_Frequency')
TAG.register(tagNames)
badLengthTag = cfg('Tag_BadLength')
# handle secondary databases
mw.progress.update(label='Creating seen/known/mature from all.db')
seenDb = filterDbByMat(allDb, cfg('threshold_seen'))
knownDb = filterDbByMat(allDb, cfg('threshold_known'))
matureDb = filterDbByMat(allDb, cfg('threshold_mature'))
mw.progress.update(label='Loading priority.db')
priorityDb = MorphDb(cfg('path_priority'), ignoreErrors=True).db
mw.progress.update(label='Loading frequency.txt')
frequencyListPath = cfg('path_frequency')
try:
with codecs.open(frequencyListPath, encoding='utf-8') as f:
# create a dictionary. key is word, value is its position in the file
frequency_list = dict(
zip([line.strip().split('\t')[0] for line in f.readlines()],
itertools.count(0)))
except FileNotFoundError:
frequency_list = dict()
frequencyListLength = len(frequency_list)
if cfg('saveDbs'):
mw.progress.update(label='Saving seen/known/mature dbs')
seenDb.save(cfg('path_seen'))
knownDb.save(cfg('path_known'))
matureDb.save(cfg('path_mature'))
mw.progress.update(label='Updating notes')
# prefetch cfg for fields
field_focus_morph = cfg('Field_FocusMorph')
field_unknown_count = cfg('Field_UnknownMorphCount')
field_unmature_count = cfg('Field_UnmatureMorphCount')
field_morph_man_index = cfg('Field_MorphManIndex')
field_unknowns = cfg('Field_Unknowns')
field_unmatures = cfg('Field_Unmatures')
field_unknown_freq = cfg('Field_UnknownFreq')
field_focus_morph_pos = cfg("Field_FocusMorphPos")
for i, (nid, mid, flds, guid, tags) in enumerate(
db.execute('select id, mid, flds, guid, tags from notes')):
ts = TAG.split(tags)
if i % 500 == 0:
mw.progress.update(value=i)
C = partial(cfg, model_id=mid)
notecfg = getFilterByMidAndTags(mid, ts)
if notecfg is None or not notecfg['Modify']:
continue
# Get all morphemes for note
morphemes = set()
for fieldName in notecfg['Fields']:
try:
loc = fidDb[(nid, guid, fieldName)]
morphemes.update(loc_db[loc])
except KeyError:
continue
proper_nouns_known = cfg('Option_ProperNounsAlreadyKnown')
# Determine un-seen/known/mature and i+N
unseens, unknowns, unmatures, new_knowns = set(), set(), set(), set()
for morpheme in morphemes:
if proper_nouns_known and morpheme.isProperNoun():
continue
if not seenDb.matches(morpheme):
unseens.add(morpheme)
if not knownDb.matches(morpheme):
unknowns.add(morpheme)
if not matureDb.matches(morpheme):
unmatures.add(morpheme)
if knownDb.matches(morpheme):
new_knowns.add(morpheme)
# Determine MMI - Morph Man Index
N, N_s, N_k, N_m = len(morphemes), len(unseens), len(unknowns), len(
unmatures)
# Bail early for lite update
if N_k > 2 and C('only update k+2 and below'):
continue
# add bonus for morphs in priority.db and frequency.txt
frequencyBonus = C('frequency.txt bonus')
isPriority = False
isFrequency = False
focusMorph = None
F_k = 0
usefulness = 0
for focusMorph in unknowns:
F_k += allDb.frequency(focusMorph)
if focusMorph in priorityDb:
isPriority = True
usefulness += C('priority.db weight')
focusMorphString = focusMorph.base
try:
focusMorphIndex = frequency_list[focusMorphString]
isFrequency = True
# The bigger this number, the lower mmi becomes
usefulness += int(
round(frequencyBonus *
(1 - focusMorphIndex / frequencyListLength)))
except KeyError:
pass
# average frequency of unknowns (ie. how common the word is within your collection)
F_k_avg = F_k // N_k if N_k > 0 else F_k
usefulness += F_k_avg
# add bonus for studying recent learned knowns (reinforce)
for morpheme in new_knowns:
locs = knownDb.getMatchingLocs(morpheme)
if locs:
ivl = min(1, max(loc.maturity for loc in locs))
# TODO: maybe average this so it doesnt favor long sentences
usefulness += C('reinforce new vocab weight') // ivl
if any(morpheme.pos == '動詞'
for morpheme in unknowns): # FIXME: this isn't working???
usefulness += C('verb bonus')
usefulness = 99999 - min(99999, usefulness)
# difference from optimal length range (too little context vs long sentence)
lenDiffRaw = min(N - C('min good sentence length'),
max(0, N - C('max good sentence length')))
lenDiff = min(9, abs(lenDiffRaw))
# calculate mmi
mmi = 100000 * N_k + 1000 * lenDiff + int(round(usefulness))
if C('set due based on mmi'):
nid2mmi[nid] = mmi
# Fill in various fields/tags on the note based on cfg
fs = splitFields(flds)
# clear any 'special' tags, the appropriate will be set in the next few lines
ts = [
t for t in ts
if t not in (notReadyTag, compTag, vocabTag, freshTag)
]
# determine card type
if N_m == 0: # sentence comprehension card, m+0
ts.append(compTag)
elif N_k == 1: # new vocab card, k+1
ts.append(vocabTag)
setField(mid, fs, field_focus_morph, focusMorph.base)
setField(mid, fs, field_focus_morph_pos, focusMorph.pos)
elif N_k > 1: # M+1+ and K+2+
ts.append(notReadyTag)
elif N_m == 1: # we have k+0, and m+1, so this card does not introduce a new vocabulary -> card for newly learned morpheme
ts.append(freshTag)
focusMorph = next(iter(unmatures))
setField(mid, fs, field_focus_morph, focusMorph.base)
setField(mid, fs, field_focus_morph_pos, focusMorph.pos)
else: # only case left: we have k+0, but m+2 or higher, so this card does not introduce a new vocabulary -> card for newly learned morpheme
ts.append(freshTag)
# set type agnostic fields
setField(mid, fs, field_unknown_count, '%d' % N_k)
setField(mid, fs, field_unmature_count, '%d' % N_m)
setField(mid, fs, field_morph_man_index, '%d' % mmi)
setField(mid, fs, field_unknowns, ', '.join(u.base for u in unknowns))
setField(mid, fs, field_unmatures,
', '.join(u.base for u in unmatures))
setField(mid, fs, field_unknown_freq, '%d' % F_k_avg)
# remove deprecated tag
if badLengthTag is not None and badLengthTag in ts:
ts.remove(badLengthTag)
# other tags
if priorityTag in ts:
ts.remove(priorityTag)
if isPriority:
ts.append(priorityTag)
if frequencyTag in ts:
ts.remove(frequencyTag)
if isFrequency:
ts.append(frequencyTag)
if tooShortTag in ts:
ts.remove(tooShortTag)
if lenDiffRaw < 0:
ts.append(tooShortTag)
if tooLongTag in ts:
ts.remove(tooLongTag)
if lenDiffRaw > 0:
ts.append(tooLongTag)
# remove unnecessary tags
if not cfg('Option_SetNotRequiredTags'):
unnecessary = [priorityTag, tooShortTag, tooLongTag]
ts = [tag for tag in ts if tag not in unnecessary]
# update sql db
tags_ = TAG.join(TAG.canonify(ts))
flds_ = joinFields(fs)
if flds != flds_ or tags != tags_: # only update notes that have changed
csum = fieldChecksum(fs[0])
sfld = stripHTML(fs[getSortFieldIndex(mid)])
ds.append((tags_, flds_, sfld, csum, now, mw.col.usn(), nid))
mw.progress.update(label='Updating anki database...')
mw.col.db.executemany(
'update notes set tags=?, flds=?, sfld=?, csum=?, mod=?, usn=? where id=?',
ds)
# Now reorder new cards based on MMI
mw.progress.update(label='Updating new card ordering...')
ds = []
# "type = 0": new cards
# "type = 1": learning cards [is supposed to be learning: in my case no learning card had this type]
# "type = 2": review cards
for (cid, nid,
due) in db.execute('select id, nid, due from cards where type = 0'):
if nid in nid2mmi: # owise it was disabled
due_ = nid2mmi[nid]
if due != due_: # only update cards that have changed
ds.append((due_, now, mw.col.usn(), cid))
mw.col.db.executemany('update cards set due=?, mod=?, usn=? where id=?',
ds)
mw.reset()
printf('Updated notes in %f sec' % (time.time() - t_0))
mw.progress.finish()
return knownDb
try:
if re.search("(?i)^" + regex + "$", strg):
nids.append(id)
except sre_constants.error:
return
if not nids:
return "0"
return "n.id in %s" % ids2str(nids)
def _findDupes(self, (val, args)):
# caller must call stripHTMLMedia on passed val
try:
mid, val = val.split(",", 1)
except OSError:
return
csum = fieldChecksum(val)
nids = []
for nid, flds in self.col.db.execute(
"select id, flds from notes where mid=? and csum=?", mid,
csum):
if stripHTMLMedia(splitFields(flds)[0]) == val:
nids.append(nid)
return "n.id in %s" % ids2str(nids)
# Find and replace
##########################################################################
def findReplace(col, nids, src, dst, regex=False, field=None, fold=True):
"Find and replace fields in a note."
def importNotes(self, notes: List[ForeignNote]) -> None:
"Convert each card into a note, apply attributes and add to col."
assert self.mappingOk()
# note whether tags are mapped
self._tagsMapped = False
for f in self.mapping:
if f == "_tags":
self._tagsMapped = True
# gather checks for duplicate comparison
csums: Dict[str, List[int]] = {}
for csum, id in self.col.db.execute(
"select csum, id from notes where mid = ?", self.model["id"]):
if csum in csums:
csums[csum].append(id)
else:
csums[csum] = [id]
firsts: Dict[str, bool] = {}
fld0idx = self.mapping.index(self.model["flds"][0]["name"])
self._fmap = self.col.models.fieldMap(self.model)
self._nextID = timestampID(self.col.db, "notes")
# loop through the notes
updates = []
updateLog = []
new = []
self._ids: List[int] = []
self._cards: List[Tuple] = []
dupeCount = 0
dupes: List[str] = []
for n in notes:
for c in range(len(n.fields)):
if not self.allowHTML:
n.fields[c] = html.escape(n.fields[c], quote=False)
n.fields[c] = n.fields[c].strip()
if not self.allowHTML:
n.fields[c] = n.fields[c].replace("\n", "<br>")
fld0 = n.fields[fld0idx]
csum = fieldChecksum(fld0)
# first field must exist
if not fld0:
self.log.append(
self.col.tr(TR.IMPORTING_EMPTY_FIRST_FIELD,
val=" ".join(n.fields)))
continue
# earlier in import?
if fld0 in firsts and self.importMode != ADD_MODE:
# duplicates in source file; log and ignore
self.log.append(
self.col.tr(TR.IMPORTING_APPEARED_TWICE_IN_FILE, val=fld0))
continue
firsts[fld0] = True
# already exists?
found = False
if csum in csums:
# csum is not a guarantee; have to check
for id in csums[csum]:
flds = self.col.db.scalar(
"select flds from notes where id = ?", id)
sflds = splitFields(flds)
if fld0 == sflds[0]:
# duplicate
found = True
if self.importMode == UPDATE_MODE:
data = self.updateData(n, id, sflds)
if data:
updates.append(data)
updateLog.append(
self.col.tr(
TR.IMPORTING_FIRST_FIELD_MATCHED,
val=fld0))
dupeCount += 1
found = True
elif self.importMode == IGNORE_MODE:
dupeCount += 1
elif self.importMode == ADD_MODE:
# allow duplicates in this case
if fld0 not in dupes:
# only show message once, no matter how many
# duplicates are in the collection already
updateLog.append(
self.col.tr(
TR.
IMPORTING_ADDED_DUPLICATE_WITH_FIRST_FIELD,
val=fld0,
))
dupes.append(fld0)
found = False
# newly add
if not found:
data = self.newData(n)
if data:
new.append(data)
# note that we've seen this note once already
firsts[fld0] = True
self.addNew(new)
self.addUpdates(updates)
# generate cards + update field cache
self.col.after_note_updates(self._ids, mark_modified=False)
# apply scheduling updates
self.updateCards()
# we randomize or order here, to ensure that siblings
# have the same due#
did = self.col.decks.selected()
conf = self.col.decks.confForDid(did)
# in order due?
if conf["new"]["order"] == NEW_CARDS_RANDOM:
self.col.sched.randomizeCards(did)
part1 = self.col.tr(TR.IMPORTING_NOTE_ADDED, count=len(new))
part2 = self.col.tr(TR.IMPORTING_NOTE_UPDATED, count=self.updateCount)
if self.importMode == UPDATE_MODE:
unchanged = dupeCount - self.updateCount
elif self.importMode == IGNORE_MODE:
unchanged = dupeCount
else:
unchanged = 0
part3 = self.col.tr(TR.IMPORTING_NOTE_UNCHANGED, count=unchanged)
self.log.append(f"{part1}, {part2}, {part3}.")
self.log.extend(updateLog)
self.total = len(self._ids)
def maybeChecksum(self, data, unique):
if not unique:
return ""
return fieldChecksum(data)
