def reverseDic(self, wordsArg=None, opt={}):
opt = addDefaultOptions(opt, {
'matchWord': True,
'showRel': 'None',
'includeDefs': False,
'background': False,
'reportStep': 300,
'autoSaveStep': 1000, ## set this to zero to disable auto saving.
'savePath': '',
})
self.stoped = False
ui = self.ui
try:
c = self.continueFrom
except AttributeError:
c = 0
savePath = opt['savePath']
if c == -1:
log.debug('c=%s'%c)
return
elif c==0:
saveFile = open(savePath, 'wb')
ui.progressStart()
ui.progress(0.0, 'Starting...')
elif c>0:
saveFile = open(savePath, 'ab')
if wordsArg is None:
words = self.takeOutputWords()
elif isinstance(wordsArg, file):
words = wordsArg.read().split('\n')
elif isinstance(wordsArg, (list, tuple)):
words = wordsArg[:]
elif isinstance(wordsArg, basestring):
words = open(wordsArg).read().split('\n')
else:
raise TypeError('Argumant wordsArg to function reverseDic is not valid!')
autoSaveStep = opt['autoSaveStep']
if not opt['savePath']:
opt['savePath'] = self.getInfo('name')+'.txt'
revG = Glossary(
info = self.info[:],
)
revG.setInfo('name', self.getInfo('name')+'_reversed')
revG.setInfo('inputlang', self.getInfo('outputlang'))
revG.setInfo('outputlang', self.getInfo('inputlang'))
wNum = len(words)
#steps = opt['reportStep']
#div = 0
#mod = 0
#total = int(wNum/steps)
'''
if c==0:
log.info('Number of input words:', wNum)
log.info('Reversing glossary...')
else:
log.info('continue reversing from index %d ...'%c)
'''
t0 = time.time()
if not ui:
log.info('passed ratio\ttime:\tpassed\tremain\ttotal\tprocess')
n = len(words)
for i in xrange(c, n):
word = words[i]
rat = float(i+1)/n
ui.progress(rat, '%d / %d words completed'%(i, n))
if ui.reverseStop:
saveFile.close() ## if with KeyboardInterrupt it will be closed ??????????????
self.continueFrom = i
self.stoped = True
#thread.exit_thread()
return
else:
self.i = i
'''
if mod == steps:
mod = 0 ; div += 1
t = time.time()
dt = t-t0
tRem = (total-div)*dt/div ## (n-i)*dt/n
rat = float(i)/n
if ui:
############# FIXME
#ui.progressbar.set_text(
'%d/%d words completed (%%%2f) remaining %d seconds'%(i,n,rat*100,tRem)
)
ui.progressbar.update(rat)
while gtk.events_pending():
gtk.main_iteration_do(False)
else:
log.info('%4d / %4d\t%8s\t%8s\t%8s\t%s'%(
div,
total,
timeHMS(dt),
timeHMS(tRem),
timeHMS(dt + tRem),
sys.argv[0],
))
else:
mod += 1
'''
if autoSaveStep>0 and i%autoSaveStep==0 and i>0:
saveFile.close()
saveFile = open(savePath, 'ab')
result = self.searchWordInDef(word, opt)
if len(result)>0:
try:
if opt['includeDefs']:
defi = '\\n\\n'.join(result)
else:
defi = ', '.join(result) + '.'
except:
open('result', 'wb').write(str(result))
log.exception('')
return False
if autoSaveStep>0:
saveFile.write('%s\t%s\n'%(word, defi))
else:
revG.data.append((word, defi))
if autoSaveStep>0 and i==n-1:
saveFile.close()
if autoSaveStep==0:
revG.writeTabfile(opt['savePath'])
ui.r_finished()
ui.progressEnd()
return True
def searchWordInDef(self, st, opt):
#seachs word 'st' in meanings(definitions) of the glossary 'self'
opt = addDefaultOptions(opt, {
'minRel': 0.0,
'maxNum': 100,
'sep': commaFa,
'matchWord': True,
'showRel': 'Percent',
})
sep = opt['sep']
matchWord = opt['matchWord']
maxNum = opt['maxNum']
minRel = opt['minRel']
defs = opt['includeDefs']
outRel = []
for item in self.data:
(word, defi) = item[:2]
defiParts = defi.split(sep)
if defi.find(st) == -1:
continue
rel = 0 ## relation value of word (as a float number between 0 and 1
for part in defiParts:
for ch in sch:
part = part.replace(ch, ' ')
pRel = 0 # part relation
if matchWord:
pNum = 0
partWords = takeStrWords(part)
pLen = len(partWords)
if pLen==0:
continue
for pw in partWords:
if pw == st:
pNum += 1
pRel = float(pNum)/pLen ## part relation
else:
pLen = len(part.replace(' ', ''))
if pLen==0:
continue
pNum = len(findAll(part, st))*len(st)
pRel = float(pNum)/pLen ## part relation
if pRel > rel:
rel = pRel
if rel <= minRel:
continue
if defs:
outRel.append((word, rel, defi))
else:
outRel.append((word, rel))
#sortby_inplace(outRel, 1, True)##???
outRel.sort(key=1, reverse=True)
n = len(outRel)
if n > maxNum > 0:
outRel = outRel[:maxNum]
n = maxNum
num = 0
out = []
if defs:
for j in xrange(n):
numP = num
(w, num, m) = outRel[j]
m = m.replace('\n', '\\n').replace('\t', '\\t')
onePer = int(1.0/num)
if onePer == 1.0:
out.append('%s\\n%s'%(w, m))
elif opt['showRel'] == 'Percent':
out.append('%s(%%%d)\\n%s'%(w, 100*num, m))
elif opt['showRel'] == 'Percent At First':
if num == numP:
out.append('%s\\n%s'%(w, m))
else:
out.append('%s(%%%d)\\n%s'%(w, 100*num, m))
else:
out.append('%s\\n%s'%(w, m))
return out
for j in xrange(n):
numP = num
(w, num) = outRel[j]
onePer = int(1.0/num)
if onePer == 1.0:
out.append(w)
elif opt['showRel'] == 'Percent':
out.append('%s(%%%d)'%(w, 100*num))
elif opt['showRel'] == 'Percent At First':
if num == numP:
out.append(w)
else:
out.append('%s(%%%d)'%(w, 100*num))
else:
out.append(w)
return out
def reverseDic(self, wordsArg=None, opt={}):
opt = addDefaultOptions(opt, {
'matchWord': True,
'showRel': 'None',
'includeDefs': False,
'background': False,
'reportStep': 300,
'autoSaveStep': 1000, ## set this to zero to disable auto saving.
'savePath': '',
})
self.stoped = False
ui = self.ui
try:
c = self.continueFrom
except AttributeError:
c = 0
savePath = opt['savePath']
if c == -1:
print('c=%s'%c)
return
elif c==0:
saveFile = open(savePath, 'wb')
ui.progressStart()
ui.progress(0.0, 'Starting...')
elif c>0:
saveFile = open(savePath, 'ab')
if wordsArg is None:
words = self.takeOutputWords()
elif isinstance(wordsArg, file):
words = wordsArg.read().split('\n')
elif isinstance(wordsArg, (list, tuple)):
words = wordsArg[:]
elif isinstance(wordsArg, basestring):
words = open(wordsArg).read().split('\n')
else:
raise TypeError('Argumant wordsArg to function reverseDic is not valid!')
autoSaveStep = opt['autoSaveStep']
if not opt['savePath']:
opt['savePath'] = self.getInfo('name')+'.txt'
revG = Glossary(self.info[:])
revG.setInfo('name', self.getInfo('name')+'_reversed')
revG.setInfo('inputlang' , self.getInfo('outputlang'))
revG.setInfo('outputlang', self.getInfo('inputlang'))
wNum = len(words)
#steps = opt['reportStep']
#div = 0
#mod = 0
#total = int(wNum/steps)
'''
if c==0:
print('Number of input words:', wNum)
print('Reversing glossary...')
else:
print('continue reversing from index %d ...'%c)
'''
t0 = time.time()
if not ui:
print('passed ratio\ttime:\tpassed\tremain\ttotal\tprocess')
n = len(words)
for i in xrange(c, n):
word = words[i]
rat = float(i+1)/n
ui.progress(rat, '%d / %d words completed'%(i,n))
if ui.reverseStop:
saveFile.close() ## if with KeyboardInterrupt it will be closed ??????????????
self.continueFrom = i
self.stoped = True
#thread.exit_thread()
return
else:
self.i = i
'''
if mod == steps:
mod = 0 ; div += 1
t = time.time()
dt = t-t0
tRem = (total-div)*dt/div ## (n-i)*dt/n
rat = float(i)/n
if ui:
############# FIXME
#ui.progressbar.set_text(
'%d/%d words completed (%%%2f) remaining %d seconds'%(i,n,rat*100,tRem)
)
ui.progressbar.update(rat)
while gtk.events_pending():
gtk.main_iteration_do(False)
else:
print('%4d / %4d\t%8s\t%8s\t%8s\t%s'%(
div,
total,
timeHMS(dt),
timeHMS(tRem),
timeHMS(dt + tRem),
sys.argv[0],
))
else:
mod += 1
'''
if autoSaveStep>0 and i%autoSaveStep==0 and i>0:
saveFile.close()
saveFile = open(savePath, 'ab')
result = self.searchWordInDef(word, opt)
if len(result)>0:
try:
if opt['includeDefs']:
defi = '\\n\\n'.join(result)
else:
defi = ', '.join(result) + '.'
except:
open('result', 'wb').write(str(result))
myRaise(__file__)
return False
if autoSaveStep>0:
saveFile.write('%s\t%s\n'%(word, defi))
else:
revG.data.append((word, defi))
if autoSaveStep>0 and i==n-1:
saveFile.close()
if autoSaveStep==0:
revG.writeTabfile(opt['savePath'])
ui.r_finished()
ui.progressEnd()
return True
def searchWordInDef(self, st, opt):
#seachs word 'st' in meanings(definitions) of the glossary 'self'
opt = addDefaultOptions(opt, {
'minRel': 0.0,
'maxNum': 100,
'sep': commaFa,
'matchWord': True,
'showRel': 'Percent',
})
sep = opt['sep']
matchWord = opt['matchWord']
maxNum = opt['maxNum']
minRel = opt['minRel']
defs = opt['includeDefs']
outRel = []
for item in self.data:
(word, defi) = item[:2]
defiParts = defi.split(sep)
if defi.find(st) == -1:
continue
rel = 0 ## relation value of word (as a float number between 0 and 1
for part in defiParts:
for ch in sch:
part = part.replace(ch, ' ')
pRel = 0 # part relation
if matchWord:
pNum = 0
partWords = takeStrWords(part)
pLen = len(partWords)
if pLen==0:
continue
for pw in partWords:
if pw == st:
pNum += 1
pRel = float(pNum)/pLen ## part relation
else:
pLen = len(part.replace(' ', ''))
if pLen==0:
continue
pNum = len(findAll(part, st))*len(st)
pRel = float(pNum)/pLen ## part relation
if pRel > rel:
rel = pRel
if rel <= minRel:
continue
if defs:
outRel.append((word, rel, defi))
else:
outRel.append((word, rel))
#sortby_inplace(outRel, 1, True)##???
outRel.sort(key=1, reverse=True)
n = len(outRel)
if n > maxNum > 0:
outRel = outRel[:maxNum]
n = maxNum
num = 0
out = []
if defs:
for j in xrange(n):
numP = num
(w, num, m) = outRel[j]
m = m.replace('\n', '\\n').replace('\t', '\\t')
onePer = int(1.0/num)
if onePer == 1.0:
out.append('%s\\n%s'%(w, m))
elif opt['showRel'] == 'Percent':
out.append('%s(%%%d)\\n%s'%(w, 100*num, m))
elif opt['showRel'] == 'Percent At First':
if num == numP:
out.append('%s\\n%s'%(w, m))
else:
out.append('%s(%%%d)\\n%s'%(w, 100*num , m))
else:
out.append('%s\\n%s'%(w,m))
return out
for j in xrange(n):
numP = num
(w, num) = outRel[j]
onePer = int(1.0/num)
if onePer == 1.0:
out.append(w)
elif opt['showRel'] == 'Percent':
out.append('%s(%%%d)'%(w, 100*num))
elif opt['showRel'] == 'Percent At First':
if num == numP:
out.append(w)
else:
out.append('%s(%%%d)'%(w, 100*num))
else:
out.append(w)
return out
def reverseDic_ext(self, wordsArg=None, opt={}):
from _reverse_dic import search
tabStr=self.writeTabfile(filename=None)
defOpt={
'matchWord':True,
'showRel':'None',
'background':False,
'reportStep':300,
'autoSaveStep':1000, ## set this to zero to disable auto saving.
'savePath':'',
'sep':commaFa}
opt = addDefaultOptions(opt, defOpt)
self.stoped = False
ui=self.ui
try:
c = self.continueFrom
except AttributeError:
c = 0
if c==-1:
print('c=%s'%c)
return
elif c==0:
ui.progress(0, 'Starting....')
if wordsArg==None:
words = self.takeOutputWords()
elif isinstance(wordsArg, file):
words = [ w[:-1] for w in wordsArg.readlines() ]
elif isinstance(wordsArg, (list, tuple)):
words = wordsArg[:]
elif isinstance(wordsArg, basestring):
fp = open(wordsArg)
words = [ w[:-1] for w in fp.readlines() ]
fp.close()
else:
raise TypeError, 'Argumant wordsArg to function reverseDic is not valid!'
autoSaveStep = opt['autoSaveStep']
if opt['savePath']=='':
opt['savePath']=self.getInfo('name')+'.txt'
savePath = opt['savePath']
if c > 0:
saveFile = open(savePath, 'ab')
else:
saveFile = open(savePath, 'wb')
ui.progressStart()
revG = Glossary(self.info[:])
revG.setInfo('name', self.getInfo('name')+'_reversed')
revG.setInfo('inputlang' , self.getInfo('outputlang'))
revG.setInfo('outputlang', self.getInfo('inputlang'))
wNum=len(words)
#steps = opt['reportStep']
#div = 0
#mod = 0
#total = int(wNum/steps)
if c==0:
print('Number of input words:', wNum)
print('Reversing glossary...')
else:
print('continue reversing from index %d ...'%c)
t0=time.time()
if ui==None:
print('passed ratio time: passed remain total process')
n = len(words)
for i in xrange(c, n):
word = words[i]
rat = float(i+1)/n
ui.progress(rat, '%d / %d words completed'%(i,n))
if ui.reverseStop:
saveFile.close()
self.continueFrom = i
self.stoped = True
#thread.exit_thread()
return
if autoSaveStep>0 and i%autoSaveStep==0 and i>0:
saveFile.close()
saveFile = open(savePath, 'ab')
result = search(tabStr,word,opt['minRel'],opt['maxNum'],opt['sep'],opt['matchWord'],opt['showRel'])
if len(result)>0:
new = ( word , result )
if autoSaveStep>0:
saveFile.write('%s\t%s\n'%new)
else:
revG.data.append(new)
if autoSaveStep>0 and i==n-1:
saveFile.close()
if autoSaveStep==0:
revG.writeTabfile(opt['savePath'])
ui.r_finished()
ui.progressEnd()
return True
def reverseDic_ext(self, wordsArg=None, opt={}):
from _reverse_dic import search
tabStr = self.writeTabfile(filename=None)
defOpt = {
'matchWord': True,
'showRel': 'None',
'background': False,
'reportStep': 300,
'autoSaveStep': 1000, ## set this to zero to disable auto saving.
'savePath': '',
'sep': commaFa
}
opt = addDefaultOptions(opt, defOpt)
self.stoped = False
ui = self.ui
try:
c = self.continueFrom
except AttributeError:
c = 0
if c == -1:
print('c=%s' % c)
return
elif c == 0:
ui.progress(0, 'Starting....')
if wordsArg == None:
words = self.takeOutputWords()
elif isinstance(wordsArg, file):
words = [w[:-1] for w in wordsArg.readlines()]
elif isinstance(wordsArg, (list, tuple)):
words = wordsArg[:]
elif isinstance(wordsArg, basestring):
fp = open(wordsArg)
words = [w[:-1] for w in fp.readlines()]
fp.close()
else:
raise TypeError, 'Argumant wordsArg to function reverseDic is not valid!'
autoSaveStep = opt['autoSaveStep']
if opt['savePath'] == '':
opt['savePath'] = self.getInfo('name') + '.txt'
savePath = opt['savePath']
if c > 0:
saveFile = open(savePath, 'ab')
else:
saveFile = open(savePath, 'wb')
ui.progressStart()
revG = Glossary(self.info[:])
revG.setInfo('name', self.getInfo('name') + '_reversed')
revG.setInfo('inputlang', self.getInfo('outputlang'))
revG.setInfo('outputlang', self.getInfo('inputlang'))
wNum = len(words)
#steps = opt['reportStep']
#div = 0
#mod = 0
#total = int(wNum/steps)
if c == 0:
print('Number of input words:', wNum)
print('Reversing glossary...')
else:
print('continue reversing from index %d ...' % c)
t0 = time.time()
if ui == None:
print(
'passed ratio time: passed remain total process'
)
n = len(words)
for i in xrange(c, n):
word = words[i]
rat = float(i + 1) / n
ui.progress(rat, '%d / %d words completed' % (i, n))
if ui.reverseStop:
saveFile.close()
self.continueFrom = i
self.stoped = True
#thread.exit_thread()
return
if autoSaveStep > 0 and i % autoSaveStep == 0 and i > 0:
saveFile.close()
saveFile = open(savePath, 'ab')
result = search(tabStr, word, opt['minRel'], opt['maxNum'],
opt['sep'], opt['matchWord'], opt['showRel'])
if len(result) > 0:
new = (word, result)
if autoSaveStep > 0:
saveFile.write('%s\t%s\n' % new)
else:
revG.data.append(new)
if autoSaveStep > 0 and i == n - 1:
saveFile.close()
if autoSaveStep == 0:
revG.writeTabfile(opt['savePath'])
ui.r_finished()
ui.progressEnd()
return True