def main(filein, fileout):
f = codecs.open(filein, "r", "utf-8")
fout = codecs.open(fileout, "w", "utf-8")
nin = 0
nout = 0
for line in f:
line = line.rstrip('\r\n')
nin = nin + 1
if line == '':
continue
if nin > 10:
#print "DEBUG. stop after",nin
#break
pass
rec = Input(line)
subanta = rec.hstem
genderstr = rec.hgenderstr
words1 = rec.mwkey2 # gaRi-mat
dbg = False
#print "dictstr=",dictstr,"\ndictwords=",dictwords
for g in genderstr:
try:
(key1, key2) = deduce_gender_stem(subanta, words1, g,
genderstr, rec.mwtype)
outputs = s_file_init_alt_helper(key1, g, key2, dbg=dbg)
except (NameError, MyException) as err:
print "\ncase=", nin, "line=", line
print err
outputs = None
if isinstance(outputs, list):
#print "outputs is a list of length",len(outputs)
pass
elif outputs == None:
outputs = []
elif ';' in outputs: ##outputs.startswith(':adj'):
outputs = outputs.split(';')
else:
#print "outputs is not a list"
outputs = [outputs]
for output in outputs:
# change form so comparable with huet_noun_tables.txt
(p0, p1, p2, p3, decl) = output.split(':')
output1 = "%s %s:%s" % (subanta, g, decl)
fout.write("%s\n" % output1)
nout = nout + 1
if len(outputs) == 0:
decl0 = ['nil' for i in xrange(0, 24)]
decl = '[' + (' '.join(decl0)) + ']'
output1 = "%s %s:%s" % (subanta, g, decl)
fout.write("%s\n" % output1)
nout = nout + 1
#key1=subanta
#key2 = words1
#print "Warning",key1,key2,":",g
f.close()
fout.close()
print nout, 'lines written to', fileout
def main(filein,fileout):
f = codecs.open(filein,"r","utf-8")
fout = codecs.open(fileout,"w","utf-8")
nin = 0
nout=0
for line in f:
line = line.rstrip('\r\n')
nin = nin + 1
if line == '':
continue
if nin > 10:
#print "DEBUG. stop after",nin
#break
pass
rec = Input(line)
subanta = rec.hstem
genderstr = rec.hgenderstr
words1 = rec.mwkey2 # gaRi-mat
dbg=False
#print "dictstr=",dictstr,"\ndictwords=",dictwords
for g in genderstr:
try:
(key1,key2) = deduce_gender_stem(subanta,words1,g,genderstr,rec.mwtype)
outputs = s_file_init_alt_helper(key1,g,key2,dbg=dbg)
except (NameError,MyException) as err:
print "\ncase=",nin,"line=",line
print err
outputs = None
if isinstance(outputs,list):
#print "outputs is a list of length",len(outputs)
pass
elif outputs == None:
outputs = []
elif ';' in outputs: ##outputs.startswith(':adj'):
outputs = outputs.split(';')
else:
#print "outputs is not a list"
outputs = [outputs]
for output in outputs:
# change form so comparable with huet_noun_tables.txt
(p0,p1,p2,p3,decl) = output.split(':')
output1 = "%s %s:%s" %(subanta,g,decl)
fout.write("%s\n" % output1)
nout = nout + 1
if len(outputs) == 0:
decl0 = ['nil' for i in xrange(0,24)]
decl = '[' + (' '.join(decl0)) + ']'
output1 = "%s %s:%s" %(subanta,g,decl)
fout.write("%s\n" % output1)
nout = nout + 1
#key1=subanta
#key2 = words1
#print "Warning",key1,key2,":",g
f.close()
fout.close()
print nout,'lines written to',fileout
def s_file_init_alt1(intab,indir,outtab,outdir,n1,n2):
nin = 0
nout = 0
filein = "%s/%s" %(indir,intab)
fileout = "%s/%s" %(outdir,outtab)
f = codecs.open(filein,"r","utf-8")
fout = codecs.open(fileout,"w","utf-8")
for line in f:
line = line.rstrip('\r\n')
nin = nin + 1
if line == '':
continue
if not ((n1 <= nin) and (nin <= n2)):
continue
# Example. line = gaRimat : S m : <MW=gaRi-mat,83017,1>
#print line
words = line.split(':')
subanta = words[0].strip() # gaRimat
try:
fg = re.findall(r"[^ ]+",words[1]) # ["S","m"]
except:
print "Case",nin,"Problem with fg.Line=",line
exit(1)
g = fg[1]
# gender is what current version of s_file_init_alt_helper expects
#print "fg=",fg
dictstr = words[2].strip() # <MW=gaRi-mat,83017,1>
dictwords = re.findall(r"[^ <=,>]+",dictstr)
words1 = dictwords[1] # gaRi-mat
dbg=False
#print "dictstr=",dictstr,"\ndictwords=",dictwords
try:
outputs = s_file_init_alt_helper(subanta,g,words1,dbg=dbg)
except (NameError,MyException) as err:
print "\ncase=",nin,"line=",line
print err
outputs = None
if isinstance(outputs,list):
#print "outputs is a list of length",len(outputs)
pass
elif outputs == None:
outputs = []
elif ';' in outputs: ##outputs.startswith(':adj'):
outputs = outputs.split(';')
else:
#print "outputs is not a list"
outputs = [outputs]
for output in outputs:
fout.write("%s\n" % output)
nout = nout + 1
if len(outputs) == 0:
key1=subanta
key2 = words1
print "Warning",key1,fg,key2,":",g
f.close()
fout.close()
def s_file_init_alt1(intab, indir, outtab, outdir, n1, n2):
nin = 0
nout = 0
filein = "%s/%s" % (indir, intab)
fileout = "%s/%s" % (outdir, outtab)
f = codecs.open(filein, "r", "utf-8")
fout = codecs.open(fileout, "w", "utf-8")
for line in f:
line = line.rstrip('\r\n')
nin = nin + 1
if line == '':
continue
if not ((n1 <= nin) and (nin <= n2)):
continue
# Example. line = gaRimat : S m : <MW=gaRi-mat,83017,1>
#print line
words = line.split(':')
subanta = words[0].strip() # gaRimat
try:
fg = re.findall(r"[^ ]+", words[1]) # ["S","m"]
except:
print "Case", nin, "Problem with fg.Line=", line
exit(1)
g = fg[1]
# gender is what current version of s_file_init_alt_helper expects
#print "fg=",fg
dictstr = words[2].strip() # <MW=gaRi-mat,83017,1>
dictwords = re.findall(r"[^ <=,>]+", dictstr)
words1 = dictwords[1] # gaRi-mat
dbg = False
#print "dictstr=",dictstr,"\ndictwords=",dictwords
try:
outputs = s_file_init_alt_helper(subanta, g, words1, dbg=dbg)
except (NameError, MyException) as err:
print "\ncase=", nin, "line=", line
print err
outputs = None
if isinstance(outputs, list):
#print "outputs is a list of length",len(outputs)
pass
elif outputs == None:
outputs = []
elif ';' in outputs: ##outputs.startswith(':adj'):
outputs = outputs.split(';')
else:
#print "outputs is not a list"
outputs = [outputs]
for output in outputs:
fout.write("%s\n" % output)
nout = nout + 1
if len(outputs) == 0:
key1 = subanta
key2 = words1
print "Warning", key1, fg, key2, ":", g
f.close()
fout.close()
