returns:
stemmed Unicode string
"""
if len(strg) < 4: return strg
if strg[-1] == "s" and strg[-2] == "'":
return strg[:-2]
else:
t = ellyToken.EllyToken(strg)
self.apply(t)
return t.toUnicode()
if __name__ == "__main__": # for unit testing
import stemTest
try:
stemmer = InflectionStemmerEN()
except ellyException.TableFailure:
print >> sys.stderr , 'cannot initialize stemmer'
sys.exit(1)
# print stemmer.sLog
# print stemmer.dLog
# print stemmer.gLog
# print stemmer.rLog
# print stemmer.pLog
# print stemmer.uLog
stemTest.stemTest(stemmer)
returns:
stemmed Unicode string
"""
if len(strg) < 4: return strg
if strg[-1] == "s" and ellyChar.isApostrophe(strg[-2]):
return strg[:-2]
else:
t = ellyToken.EllyToken(strg)
self.apply(t)
return t.toUnicode()
if __name__ == "__main__": # for unit testing
import stemTest
try:
stemmer = InflectionStemmerEN()
except ellyException.TableFailure:
print >> sys.stderr, 'cannot initialize stemmer'
sys.exit(1)
# print stemmer.sLog
# print stemmer.dLog
# print stemmer.gLog
# print stemmer.rLog
# print stemmer.pLog
# print stemmer.uLog
stemTest.stemTest(stemmer)
if __name__ == "__main__":
import stemTest
if len(sys.argv) < 2:
print("usage: X stem_logic_file_name")
sys.exit(0)
log = StemLogic()
print('reading', sys.argv[1])
if not log.define(sys.argv[1]):
print("logic definition failed")
sys.exit(1)
tab = log.table
siz = len(tab)
print(" logic size= ", siz)
if siz == 0:
print("NULL table", file=sys.stderr)
sys.exit(1)
M = 16
for i in range(siz):
if i % M == 0: print()
print(tab[i], end=' ')
print()
stemTest.stemTest(log, tab[0][::-1])
if __name__ == "__main__":
import stemTest
if len(sys.argv) < 2:
print "usage: X stem_logic_file_name"
sys.exit(0)
log = StemLogic()
print 'reading' , sys.argv[1]
if not log.define(sys.argv[1]):
print "logic definition failed"
sys.exit(1)
tab = log.table
siz = len(tab)
print " logic size= ", siz
if siz == 0:
print >> sys.stderr , "NULL table"
sys.exit(1)
M = 16
for i in range(siz):
if i%M == 0: print ""
print tab[i],
print ""
stemTest.stemTest(log,tab[0][::-1])
