def main(args):
# exmpale for encode
import mysam.tagmaker as tagmaker
tag_maker = tagmaker.tagMaker()
taglist = [u'اسم', u'مضير متصل', u'مجرور']
tag_maker.encode(taglist)
tagstr = str(tag_maker)
print(tagstr.encode('utf8'))
# exmaple for decode
print("***exmaple for decode***")
tag_maker = tagmaker.tagMaker()
tagcode = 'N--;--I-;----;---'
print(tag_maker.repr(tag_maker.decode(tagcode)).encode('utf8'))
print("***exmaple for inflect***")
tag_maker = tagmaker.tagMaker()
tagcode = 'N--;M3IY;---H;---'
print(tag_maker.inflect(tagcode).encode('utf8'))
print("***exmaple for add tag***")
tag_maker = tagmaker.tagMaker()
tagcode = 'N--;--I-;----;---'
tag_new = u"تعريف"
tag_maker.add(tag_new)
tag_new = u"اسم"
tag_maker.add(tag_new)
print(str(tag_maker).encode('utf8'))
print("***exmaple for has tag***")
tag_maker = tagmaker.tagMaker()
tagcode = 'N--;--I-;----;---'
tag_search = u"مجرور"
print(tag_maker.has_tag(tag_search, tagcode))
def __init__(self, wordtype, version="N/A"):
"""
initiate the dict
"""
csvdict.CsvDict.__init__(self, wordtype, version)
self.affixes_list = []
nb1 = 0
nb2 = 0
file_conf = os.path.join(os.path.dirname(__file__),
"config/tag.config")
self.tagmaker = tagmaker.tagMaker(file_conf)
for procletic in snconst.COMP_PREFIX_LIST_MODEL.keys():
for encletic in snconst.COMP_SUFFIX_LIST_MODEL:
#~ for procletic in snconst.COMP_PREFIX_LIST:
#~ for encletic in snconst.COMP_SUFFIX_LIST:
for suffix in snconst.CONJ_SUFFIX_LIST:
pro_nm = araby.strip_tashkeel(procletic)
enc_nm = araby.strip_tashkeel(encletic)
if u"-".join([pro_nm,
enc_nm]) in snconst.COMP_NOUN_AFFIXES:
nb1 += 1
if nspell.verify_proaffix_affix(
procletic, encletic, suffix):
nb2 += 1
self.affixes_list.append(
(procletic, encletic, suffix))
print nb1, nb2
def __init__(
self,
version="N/A",
):
"""
initiate the dict
"""
csvdict.CsvDict.__init__(self, version)
file_conf = os.path.join(os.path.dirname(__file__),
"config/tag.config")
self.tagmaker = tagmaker.tagMaker(file_conf)
self.affixer = verb_affixer.verb_affixer()
def main(args):
taglists = [[u'اسم', u'هاء', u'مجرور',],
u'تعريف::مرفوع:متحرك:ينون:::'.split(":"),
u'المضارع المعلوم:هو:::n:'.split(":"),
u':مضاف:مجرور:متحرك:ينون:::'.split(':'),
]
for taglist in taglists:
tag_maker = tagmaker.tagMaker()
tag_maker.encode(taglist)
print(u"+".join(taglist).encode('utf8'))
tagstr = str(tag_maker)
print(tagstr.encode('utf8'))
# decode a unifed tag string
print(tag_maker.repr(tag_maker.decode()).encode('utf8'))
def main(args):
import pandas as pd
# test all existing tags
tag_maker = tagmaker.tagMaker()
tagstr = str(tag_maker)
print("----")
for tag in tag_const.TAGSDICT:
tagstr = str(tag_maker)
tag_maker.add(tag)
tagstr_new = str(tag_maker)
if tagstr == tagstr_new:
print(u" ".join(["error:old\t", tagstr, tag, "\n new:\t", tagstr_new]).encode('utf8'))
else:
print(u" ".join([tag, tagstr_new]).encode('utf8'))
decode_tags = tag_maker.decode()
df = pd.DataFrame(decode_tags)
print(df)
tag_maker.add(u"اسم")
print("******Inflect", tag_maker.inflect().encode('utf8'))
tag_maker.add(u"فعل")
print("***Verb***Inflect", tag_maker.inflect().encode('utf8'))