def updateDisplay( self ):
bs = parseWhitelist( self.blacklist.text() )
ws = parseWhitelist( self.whitelist.text() )
for m in self.db.db.keys():
if m.pos in bs:
self.db.db.pop( m )
elif ws and m.pos not in ws:
self.db.db.pop( m )
if self.col4Mode.isChecked():
self.morphDisplay.setText( self.db.showMs() )
else:
self.morphDisplay.setText( u'\n'.join( [ m.base for m in self.db.db ] ) )
self.analysisDisplay.setText( self.db.analyze2str() )
def onGo( self ):
ws = parseWhitelist( self.whitelist.text() )
bs = parseWhitelist( self.blacklist.text() )
mFmt = unicode( self.matureFmt.text() )
kFmt = unicode( self.knownFmt.text() )
uFmt = unicode( self.unknownFmt.text() )
inFile = QFileDialog.getOpenFileName( caption='Dueling subs to process', directory=dbsPath )
if not inFile: return
outFile = QFileDialog.getSaveFileName( caption='Save adaptive subs to', directory=dbsPath )
if not outFile: return
adaptiveSubs.run( inFile, outFile, ws, bs, mFmt, kFmt, uFmt )
infoMsg( 'Completed successfully' )
def onGo( self ):
ws = parseWhitelist( self.whitelist.text() )
bs = parseWhitelist( self.blacklist.text() )
mFmt = unicode( self.matureFmt.text() )
kFmt = unicode( self.knownFmt.text() )
uFmt = unicode( self.unknownFmt.text() )
inFile = QFileDialog.getOpenFileName( caption='Dueling subs to process', directory=util.dbPath )
if not inFile: return
outFile = QFileDialog.getSaveFileName( caption='Save adaptive subs to', directory=util.dbPath )
if not outFile: return
adaptiveSubs.run( inFile, outFile, ws, bs, mFmt, kFmt, uFmt )
infoMsg( 'Completed successfully' )
def mkAll( self ): # IO ()
log( 'Getting initial all.db...' )
if not hasattr( self, '_allDb' ):
try:
self._allDb = M.MorphDb( self.allPath )
debug( ' * Updating existing all.db' )
except IOError:
self._allDb = M.MorphDb()
debug( ' * Creating new all.db from scratch' )
allDb = self._allDb
log( '...done' )
mp = M.mecab()
# pre-cache lookups
fieldNames = self.cfg['morph fields']
whitelist, blacklist = parseWhitelist( self.cfg['whitelist'] ), parseWhitelist( self.cfg['blacklist'] )
fid2cardsDb = self.fid2cardsDb()
fidDb = allDb.fidDb()
locDb = allDb.locDb()
fs = self.getFacts()
i, lfs = 0, len( fs )
start = time.time()
last = time.time()
for f in fs:
mats = [ c.interval for c in fid2cardsDb[ f.id ] ]
for fieldName in fieldNames:
try:
fieldValue = normalizeFieldValue( f[ fieldName ] )
except KeyError: # if fact doesn't have the field just skip it
continue
try: # existing location
loc = fidDb[ (f.id, fieldName) ]
# new loc only; no morpheme change
if loc.fieldValue == fieldValue and loc.maturities != mats:
debug(' .mats for %d[%s]' % ( f.id, fieldName ) )
newLoc = M.AnkiDeck( f.id, fieldName, fieldValue, self.deckPath, self.deckName, mats )
ms = locDb.pop( loc )
locDb[ newLoc ] = ms
# new loc and new morphemes
elif loc.fieldValue != fieldValue:
debug(' .morphs for %d[%s]' % ( f.id, fieldName ) )
newLoc = M.AnkiDeck( f.id, fieldName, fieldValue, self.deckPath, self.deckName, mats )
ms = M.getMorphemes( mp, fieldValue, ws=whitelist, bs=blacklist )
locDb.pop( loc )
locDb[ newLoc ] = ms
except KeyError: # new location
loc = M.AnkiDeck( f.id, fieldName, fieldValue, self.deckPath, self.deckName, mats )
ms = M.getMorphemes( mp, fieldValue, ws=whitelist, bs=blacklist )
if ms:
debug(' .loc for %d[%s]' % ( f.id, fieldName ) )
locDb[ loc ] = ms
i += 1
if i % 100 == 0:
log(' %d / %d = %d%% in %f sec' % ( i, lfs, 100.*i/lfs, time.time()-last ) )
last = time.time()
log( 'Proccessed all facts in %f sec. Now saving...' % ( time.time()-start ) )
allDb.clear()
allDb.addFromLocDb( locDb )
allDb.save( self.allPath )
self.cfg['last db update'][ self.allPath ] = time.time()
self.cfg['last all.db update took'] = time.time() - start
log( '...done' )
sigterm( mp )
return self._allDb
