def test_sentence(self):
self.assertEqual(
avro.parse("ami banglay gan gai"), utf("আমি বাংলায় গান গাই"))
self.assertEqual(
avro.parse(
"amader valObasa hoye gel ghas, kheye gel goru ar diye gelo ba^sh"),
utf("আমাদের ভালোবাসা হয়ে গেল ঘাস, খেয়ে গেল গরু আর দিয়ে গেল বাঁশ"))
def test_sentence(self):
self.assertEqual(avro.parse("ami banglay gan gai"),
utf("আমি বাংলায় গান গাই"))
self.assertEqual(
avro.parse(
"amader valObasa hoye gel ghas, kheye gel goru ar diye gelo ba^sh"
),
utf("আমাদের ভালোবাসা হয়ে গেল ঘাস, খেয়ে গেল গরু আর দিয়ে গেল বাঁশ"))
def test_non_ascii(self):
"""Test parser response for non ascii characters
Parser should return any non-ascii characters passed to it
"""
self.assertEquals(utf('ব'), avro.parse('ব'))
self.assertEquals(utf('অভ্র'), avro.parse('অভ্র'))
# mixed string
self.assertEquals(utf('বআবা গো'), avro.parse('বaba gO'))
self.assertEquals(utf('আমি বাংলায় গান গাই'),
avro.parse('aমি বাংলায় gaন গাi'))
def test_non_ascii(self):
"""Test parser response for non ascii characters
Parser should return any non-ascii characters passed to it
"""
self.assertEqual(utf('ব'), avro.parse('ব'))
self.assertEqual(utf('অভ্র'), avro.parse('অভ্র'))
# mixed string
self.assertEqual(utf('বআবা গো'), avro.parse('বaba gO'))
self.assertEqual(utf('আমি বাংলায় গান গাই'),
avro.parse('aমি বাংলায় gaন গাi'))
def test_patterns_without_rules_not_from_config(self):
"""Tests all patterns not from config that don't have rules
This test is done in addition to
test_patterns_without_rules_from_config() to ensure that text
passed manually to avro.parse are properly parsed when they
don't exact match a pattern that has no rules specified.
"""
# Test some conjunctions
self.assertEquals(utf("ভ্ল"), avro.parse("bhl"))
self.assertEquals(utf("ব্ধ"), avro.parse("bdh"))
self.assertEquals(utf("ব্ধ"), avro.parse("bdh"))
self.assertEquals(utf("ড্ড"), avro.parse("DD"))
# stunned stork!
self.assertEquals(utf("স্তব্ধ বক"), avro.parse("stbdh bk"))
def test_patterns_without_rules_not_from_config(self):
"""Tests all patterns not from config that don't have rules
This test is done in addition to
test_patterns_without_rules_from_config() to ensure that text
passed manually to avro.parse are properly parsed when they
don't exact match a pattern that has no rules specified.
"""
# Test some conjunctions
self.assertEquals(utf("ভ্ল"), avro.parse("bhl"))
self.assertEquals(utf("ব্ধ"), avro.parse("bdh"))
self.assertEquals(utf("ব্ধ"), avro.parse("bdh"))
self.assertEquals(utf("ড্ড"), avro.parse("DD"))
# stunned stork!
self.assertEquals(utf("স্তব্ধ বক"),
avro.parse("stbdh bk"))
def onPressHotkey(self, event) -> None:
self.bengali = parse(self.current) #parsed value will be stored in
# fire backspace to delete previous word
# then place the bengali word followed by triggered key
replace: str = '\b' * (len(self.current) +
1) + self.bengali + CODE_DICT[event.name]
self.previous = self.current # store the current value to previous for `smart backspace`
self.current = ""
write(replace) # write the string
def get_book_info(query,
method,
price_from=0,
price_to=50000,
rating_from=0,
rating_to=5,
sort="TOTAL_SOLD DESC",
book_name="",
author_name=""):
if method == "UPDATE":
quercmd = "SELECT B.BOOK_ID,B.BOOK_NAME,A.AUTHOR_NAME,B.PRICE,B.RATINGS,C.PUBLISHER_NAME, B.DISCOUNT, B.BOOK_EDITION, B.PAGES, B.BOOK_GENRE, B.SUMMARY FROM BOOK B JOIN AUTHOR A USING(AUTHOR_ID) JOIN PUBLISHER C USING(PUBLISHER_ID) WHERE B.BOOK_ID = " + str(
query)
elif is_banglish(query):
# print(avro.parse(query))
query = avro.parse(query)
# print(sort)
if method == "GET":
quercmd = "SELECT B.BOOK_ID,B.BOOK_NAME,A.AUTHOR_NAME,B.PRICE, B.BOOK_GENRE, B.BOOK_EDITION, B.RATINGS, B.NO_OF_RATINGS, C.PUBLISHER_NAME, B.DISCOUNT, B.COUNTRY, B.LANGUAGE, B.SUMMARY, B.ISBN, B.PAGES, B.TOTAL_SOLD FROM BOOK B JOIN AUTHOR A USING(AUTHOR_ID) JOIN PUBLISHER C USING(PUBLISHER_ID) ORDER BY B.BOOK_ID"
elif method == "POST":
quercmd = "SELECT B.BOOK_ID,B.BOOK_NAME,A.AUTHOR_NAME,B.PRICE, B.BOOK_GENRE, B.BOOK_EDITION, B.RATINGS, B.NO_OF_RATINGS, C.PUBLISHER_NAME, B.DISCOUNT, B.COUNTRY, B.LANGUAGE, B.SUMMARY, B.ISBN, B.PAGES, B.TOTAL_SOLD FROM BOOK B JOIN AUTHOR A USING(AUTHOR_ID) JOIN PUBLISHER C USING(PUBLISHER_ID) WHERE ((B.PRICE >= " + str(
price_from
) + " AND B.PRICE <= " + str(price_to) + " AND B.RATINGS >= " + str(
rating_from
) + " AND B.RATINGS <= " + str(
rating_to
) + ") AND (B.BOOK_NAME LIKE '%" + book_name + "%') AND (A.AUTHOR_NAME LIKE '%" + author_name + "%') AND (C.PUBLISHER_NAME LIKE '%" + query + "%') AND (B.BOOK_GENRE LIKE '%" + query + "%')) ORDER BY B." + sort
db_cursor = conn.cursor()
db_cursor.execute(quercmd)
search_results = db_cursor.fetchall()
li = []
for row in search_results:
l2 = []
for item in row:
if item == None:
item = ""
l2.append(item)
l2[1] = slice_name(l2[1])
if check_if_image_exists(l2[0]):
l2.append(l2[0])
# print(l2)
else:
l2.append("book_image")
# print(l2)
li.append(l2)
return li
def getSenti(self, input, models):
sentence, phase = input
if phase is PHASE.BANGLISH:
print("Banglish is :", sentence)
sentence = avro.parse(sentence)
phase = PHASE.BANGLA
print("Bangla is :", sentence)
if phase is PHASE.BANGLA:
# 1 : TextBlob
# 2 : googletrans
# 3 : translate
use_api = self.language.currentIndex()
if use_api == 2:
print("Using Textblob translator")
sentence = str(TextBlob(sentence).translate(to='en'))
elif use_api == 1:
print("Using googletrans translator")
from googletrans import Translator
translator = Translator()
sentence = str(translator.translate(sentence))
elif use_api == 0:
print("Using translate translator")
from translate import Translator
translator = Translator(to_lang="en", from_lang="bn")
sentence = translator.translate(sentence)
phase = PHASE.ENGLISH
print("English is :", sentence)
results = []
for uniModel in models:
model = uniModel(sentence)
result = {
'name': model.name,
'result': model.result(),
'conclusion': model.conclusion()
}
results.append(result)
return results
def get_book_info(query,
method,
price_from=0,
price_to=50000,
rating_from=0,
rating_to=5,
sort="TOTAL_SOLD DESC"):
if is_banglish(query):
# print(avro.parse(query))
query = avro.parse(query)
# print(sort)
if method == "GET":
quercmd = "SELECT B.BOOK_ID,B.BOOK_NAME,A.AUTHOR_NAME,B.PRICE,B.RATINGS,C.PUBLISHER_NAME, B.DISCOUNT FROM BOOK B JOIN AUTHOR A USING(AUTHOR_ID) JOIN PUBLISHER C USING(PUBLISHER_ID) WHERE ((B.BOOK_NAME LIKE '%" + query + "%') OR (A.AUTHOR_NAME LIKE '%" + query + "%') OR (C.PUBLISHER_NAME LIKE '%" + query + "%') OR (B.BOOK_GENRE LIKE '%" + query + "%')) ORDER BY B.TOTAL_SOLD DESC"
elif method == "POST":
quercmd = "SELECT B.BOOK_ID,B.BOOK_NAME,A.AUTHOR_NAME,B.PRICE,B.RATINGS,C.PUBLISHER_NAME, B.DISCOUNT FROM BOOK B JOIN AUTHOR A USING(AUTHOR_ID) JOIN PUBLISHER C USING(PUBLISHER_ID) WHERE ((B.PRICE >= " + str(
price_from
) + " AND B.PRICE <= " + str(price_to) + " AND B.RATINGS >= " + str(
rating_from
) + " AND B.RATINGS <= " + str(
rating_to
) + ") AND ((B.BOOK_NAME LIKE '%" + query + "%') OR (A.AUTHOR_NAME LIKE '%" + query + "%') OR (C.PUBLISHER_NAME LIKE '%" + query + "%')OR (B.BOOK_GENRE LIKE '%" + query + "%'))) ORDER BY B." + sort
# print(quercmd)
# print(query)
db_cursor = conn.cursor()
db_cursor.execute(quercmd)
search_results = db_cursor.fetchall()
li = []
for row in search_results:
l2 = []
for item in row:
l2.append(item)
l2[1] = slice_name(l2[1])
if check_if_image_exists(l2[0]):
l2.append(l2[0])
# print(l2)
else:
l2.append("book_image")
# print(l2)
li.append(l2)
return li
def test_words_with_punctuations(self):
"""Test parsing of words with punctuations"""
self.assertEqual(utf('আয়রে,'), avro.parse('ayre,'))
self.assertEqual(utf('ভোলা'), avro.parse('bhOla'))
self.assertEqual(utf('খেয়াল'), avro.parse('kheyal'))
self.assertEqual(utf('খোলা'), avro.parse('khOla'))
def test_sentences(self):
"""Test parsing of sentences"""
self.assertEquals(utf('আমি বাংলায় গান গাই'),
avro.parse('ami banglay gan gai'))
def test_patterns_without_rules_from_config(self):
"""Tests all patterns from config that don't have rules"""
for pattern in AVRO_DICT['data']['patterns']:
if 'rules' not in pattern:
self.assertEquals(pattern['replace'],
avro.parse(pattern['find']))
def test_patterns_punctuations(self):
"""Tests patterns - punctuations"""
# Test some punctuations
self.assertEqual(utf("।"), avro.parse("."))
self.assertEqual(utf("।।"), avro.parse(".."))
self.assertEqual(utf("..."), avro.parse("..."))
def test_words_with_punctuations(self):
"""Test parsing of words with punctuations"""
self.assertEquals(utf('আয়রে,'), avro.parse('ayre,'))
self.assertEquals(utf('ভোলা'), avro.parse('bhOla'))
self.assertEquals(utf('খেয়াল'), avro.parse('kheyal'))
self.assertEquals(utf('খোলা'), avro.parse('khOla'))
def test_patterns_with_rules_svaravarna(self):
"""Test patterns - with rules - svaravarna"""
# Test some numbers
self.assertEquals(utf("অ"), avro.parse("o"))
self.assertEquals(utf("আ"), avro.parse("a"))
self.assertEquals(utf("ই"), avro.parse("i"))
self.assertEquals(utf("ঈ"), avro.parse("I"))
self.assertEquals(utf("উ"), avro.parse("u"))
self.assertEquals(utf("উ"), avro.parse("oo"))
self.assertEquals(utf("ঊ"), avro.parse("U"))
self.assertEquals(utf("এ"), avro.parse("e"))
self.assertEquals(utf("ঐ"), avro.parse("OI"))
self.assertEquals(utf("ও"), avro.parse("O"))
self.assertEquals(utf("ঔ"), avro.parse("OU"))
import wget
wget.download('http://0xhosts.com/words.txt')
wget.download('http://0xhosts.com/translate.txt')
my_input=str(raw_input("Enter Your Sentence:"))
from pyavrophonetic import avro
import codecs
spit_it=my_input.split()
check=spit_it
print check
print "Total Words",len(check)
if len(check)==4:
if check[0] and check[1] and check[2] and check[3] in open("words.txt").read().split():
print ("False")
else:
s1=avro.parse(check[0])
s2=avro.parse(check[1])
s3=avro.parse(check[2])
s4=avro.parse(check[3])
p=codecs.open('translate.txt','r+','utf-8')
p.write("Translation : %s %s %s %s" %(s1,s2,s3,s4))
p.close()
print("Translation Completed.Saved to translate.txt")
if s1 and s2 and s3 and s4 in codecs.open('words.txt','r','utf-8').read().split():
print "Banglish Style"
else:
print "Murad Takla Style"
else:
print ("Sorry max/min 4 words")
def test_basic(self):
self.assertEqual(avro.parse("bhl"), utf("ভ্ল"))
self.assertEqual(avro.parse("bj"), utf("ব্জ"))
self.assertEqual(avro.parse("bd"), utf("ব্দ"))
self.assertEqual(avro.parse("bb"), utf("ব্ব"))
self.assertEqual(avro.parse("bl"), utf("ব্ল"))
self.assertEqual(avro.parse("bh"), utf("ভ"))
self.assertEqual(avro.parse("vl"), utf("ভ্ল"))
self.assertEqual(avro.parse("b"), utf("ব"))
self.assertEqual(avro.parse("v"), utf("ভ"))
self.assertEqual(avro.parse("cNG"), utf("চ্ঞ"))
self.assertEqual(avro.parse("cch"), utf("চ্ছ"))
self.assertEqual(avro.parse("cc"), utf("চ্চ"))
self.assertEqual(avro.parse("ch"), utf("ছ"))
self.assertEqual(avro.parse("c"), utf("চ"))
self.assertEqual(avro.parse("dhn"), utf("ধ্ন"))
self.assertEqual(avro.parse("dhm"), utf("ধ্ম"))
self.assertEqual(avro.parse("dgh"), utf("দ্ঘ"))
self.assertEqual(avro.parse("ddh"), utf("দ্ধ"))
self.assertEqual(avro.parse("dbh"), utf("দ্ভ"))
self.assertEqual(avro.parse("dv"), utf("দ্ভ"))
self.assertEqual(avro.parse("dm"), utf("দ্ম"))
self.assertEqual(avro.parse("DD"), utf("ড্ড"))
self.assertEqual(avro.parse("Dh"), utf("ঢ"))
self.assertEqual(avro.parse("dh"), utf("ধ"))
self.assertEqual(avro.parse("dg"), utf("দ্গ"))
self.assertEqual(avro.parse("dd"), utf("দ্দ"))
self.assertEqual(avro.parse("D"), utf("ড"))
self.assertEqual(avro.parse("d"), utf("দ"))
self.assertEqual(avro.parse("..."), utf("..."))
self.assertEqual(avro.parse(".`"), utf("."))
self.assertEqual(avro.parse(".."), utf("।।"))
self.assertEqual(avro.parse("."), utf("।"))
self.assertEqual(avro.parse("ghn"), utf("ঘ্ন"))
self.assertEqual(avro.parse("Ghn"), utf("ঘ্ন"))
self.assertEqual(avro.parse("gdh"), utf("গ্ধ"))
self.assertEqual(avro.parse("gN"), utf("গ্ণ"))
self.assertEqual(avro.parse("GN"), utf("গ্ণ"))
self.assertEqual(avro.parse("gn"), utf("গ্ন"))
self.assertEqual(avro.parse("gm"), utf("গ্ম"))
self.assertEqual(avro.parse("Gm"), utf("গ্ম"))
self.assertEqual(avro.parse("gl"), utf("গ্ল"))
self.assertEqual(avro.parse("Gl"), utf("গ্ল"))
self.assertEqual(avro.parse("gg"), utf("জ্ঞ"))
self.assertEqual(avro.parse("GG"), utf("জ্ঞ"))
self.assertEqual(avro.parse("Gg"), utf("জ্ঞ"))
self.assertEqual(avro.parse("gG"), utf("জ্ঞ"))
self.assertEqual(avro.parse("gh"), utf("ঘ"))
self.assertEqual(avro.parse("Gh"), utf("ঘ"))
self.assertEqual(avro.parse("g"), utf("গ"))
self.assertEqual(avro.parse("hN"), utf("হ্ণ"))
self.assertEqual(avro.parse("hn"), utf("হ্ন"))
self.assertEqual(avro.parse("hm"), utf("হ্ম"))
self.assertEqual(avro.parse("hl"), utf("হ্ল"))
self.assertEqual(avro.parse("h"), utf("হ"))
self.assertEqual(avro.parse("jjh"), utf("জ্ঝ"))
self.assertEqual(avro.parse("jNG"), utf("জ্ঞ"))
self.assertEqual(avro.parse("jh"), utf("ঝ"))
self.assertEqual(avro.parse("jj"), utf("জ্জ"))
self.assertEqual(avro.parse("j"), utf("জ"))
self.assertEqual(avro.parse("J"), utf("জ"))
self.assertEqual(avro.parse("kkhN"), utf("ক্ষ্ণ"))
self.assertEqual(avro.parse("kShN"), utf("ক্ষ্ণ"))
self.assertEqual(avro.parse("kkhm"), utf("ক্ষ্ম"))
self.assertEqual(avro.parse("kShm"), utf("ক্ষ্ম"))
self.assertEqual(avro.parse("kxN"), utf("ক্ষ্ণ"))
self.assertEqual(avro.parse("kxm"), utf("ক্ষ্ম"))
self.assertEqual(avro.parse("kkh"), utf("ক্ষ"))
self.assertEqual(avro.parse("kSh"), utf("ক্ষ"))
self.assertEqual(avro.parse("ksh"), utf("কশ"))
self.assertEqual(avro.parse("kx"), utf("ক্ষ"))
self.assertEqual(avro.parse("kk"), utf("ক্ক"))
self.assertEqual(avro.parse("kT"), utf("ক্ট"))
self.assertEqual(avro.parse("kt"), utf("ক্ত"))
self.assertEqual(avro.parse("kl"), utf("ক্ল"))
self.assertEqual(avro.parse("ks"), utf("ক্স"))
self.assertEqual(avro.parse("kh"), utf("খ"))
self.assertEqual(avro.parse("k"), utf("ক"))
self.assertEqual(avro.parse("lbh"), utf("ল্ভ"))
self.assertEqual(avro.parse("ldh"), utf("ল্ধ"))
self.assertEqual(avro.parse("lkh"), utf("লখ"))
self.assertEqual(avro.parse("lgh"), utf("লঘ"))
self.assertEqual(avro.parse("lph"), utf("লফ"))
self.assertEqual(avro.parse("lk"), utf("ল্ক"))
self.assertEqual(avro.parse("lg"), utf("ল্গ"))
self.assertEqual(avro.parse("lT"), utf("ল্ট"))
self.assertEqual(avro.parse("lD"), utf("ল্ড"))
self.assertEqual(avro.parse("lp"), utf("ল্প"))
self.assertEqual(avro.parse("lv"), utf("ল্ভ"))
self.assertEqual(avro.parse("lm"), utf("ল্ম"))
self.assertEqual(avro.parse("ll"), utf("ল্ল"))
self.assertEqual(avro.parse("lb"), utf("ল্ব"))
self.assertEqual(avro.parse("l"), utf("ল"))
self.assertEqual(avro.parse("mth"), utf("ম্থ"))
self.assertEqual(avro.parse("mph"), utf("ম্ফ"))
self.assertEqual(avro.parse("mbh"), utf("ম্ভ"))
self.assertEqual(avro.parse("mpl"), utf("মপ্ল"))
self.assertEqual(avro.parse("mn"), utf("ম্ন"))
self.assertEqual(avro.parse("mp"), utf("ম্প"))
self.assertEqual(avro.parse("mv"), utf("ম্ভ"))
self.assertEqual(avro.parse("mm"), utf("ম্ম"))
self.assertEqual(avro.parse("ml"), utf("ম্ল"))
self.assertEqual(avro.parse("mb"), utf("ম্ব"))
self.assertEqual(avro.parse("mf"), utf("ম্ফ"))
self.assertEqual(avro.parse("m"), utf("ম"))
self.assertEqual(avro.parse("0"), utf("০"))
self.assertEqual(avro.parse("1"), utf("১"))
self.assertEqual(avro.parse("2"), utf("২"))
self.assertEqual(avro.parse("3"), utf("৩"))
self.assertEqual(avro.parse("4"), utf("৪"))
self.assertEqual(avro.parse("5"), utf("৫"))
self.assertEqual(avro.parse("6"), utf("৬"))
self.assertEqual(avro.parse("7"), utf("৭"))
self.assertEqual(avro.parse("8"), utf("৮"))
self.assertEqual(avro.parse("9"), utf("৯"))
self.assertEqual(avro.parse("NgkSh"), utf("ঙ্ক্ষ"))
self.assertEqual(avro.parse("Ngkkh"), utf("ঙ্ক্ষ"))
self.assertEqual(avro.parse("NGch"), utf("ঞ্ছ"))
self.assertEqual(avro.parse("Nggh"), utf("ঙ্ঘ"))
self.assertEqual(avro.parse("Ngkh"), utf("ঙ্খ"))
self.assertEqual(avro.parse("NGjh"), utf("ঞ্ঝ"))
self.assertEqual(avro.parse("ngOU"), utf("ঙ্গৌ"))
self.assertEqual(avro.parse("ngOI"), utf("ঙ্গৈ"))
self.assertEqual(avro.parse("Ngkx"), utf("ঙ্ক্ষ"))
self.assertEqual(avro.parse("NGc"), utf("ঞ্চ"))
self.assertEqual(avro.parse("nch"), utf("ঞ্ছ"))
self.assertEqual(avro.parse("njh"), utf("ঞ্ঝ"))
self.assertEqual(avro.parse("ngh"), utf("ঙ্ঘ"))
self.assertEqual(avro.parse("Ngk"), utf("ঙ্ক"))
self.assertEqual(avro.parse("Ngx"), utf("ঙ্ষ"))
self.assertEqual(avro.parse("Ngg"), utf("ঙ্গ"))
self.assertEqual(avro.parse("Ngm"), utf("ঙ্ম"))
self.assertEqual(avro.parse("NGj"), utf("ঞ্জ"))
self.assertEqual(avro.parse("ndh"), utf("ন্ধ"))
self.assertEqual(avro.parse("nTh"), utf("ন্ঠ"))
self.assertEqual(avro.parse("NTh"), utf("ণ্ঠ"))
self.assertEqual(avro.parse("nth"), utf("ন্থ"))
self.assertEqual(avro.parse("nkh"), utf("ঙ্খ"))
self.assertEqual(avro.parse("ngo"), utf("ঙ্গ"))
self.assertEqual(avro.parse("nga"), utf("ঙ্গা"))
self.assertEqual(avro.parse("ngi"), utf("ঙ্গি"))
self.assertEqual(avro.parse("ngI"), utf("ঙ্গী"))
self.assertEqual(avro.parse("ngu"), utf("ঙ্গু"))
self.assertEqual(avro.parse("ngU"), utf("ঙ্গূ"))
self.assertEqual(avro.parse("nge"), utf("ঙ্গে"))
self.assertEqual(avro.parse("ngO"), utf("ঙ্গো"))
self.assertEqual(avro.parse("NDh"), utf("ণ্ঢ"))
self.assertEqual(avro.parse("nsh"), utf("নশ"))
self.assertEqual(avro.parse("Ngr"), utf("ঙর"))
self.assertEqual(avro.parse("NGr"), utf("ঞর"))
self.assertEqual(avro.parse("ngr"), utf("ংর"))
self.assertEqual(avro.parse("nj"), utf("ঞ্জ"))
self.assertEqual(avro.parse("Ng"), utf("ঙ"))
self.assertEqual(avro.parse("NG"), utf("ঞ"))
self.assertEqual(avro.parse("nk"), utf("ঙ্ক"))
self.assertEqual(avro.parse("ng"), utf("ং"))
self.assertEqual(avro.parse("nn"), utf("ন্ন"))
self.assertEqual(avro.parse("NN"), utf("ণ্ণ"))
self.assertEqual(avro.parse("Nn"), utf("ণ্ন"))
self.assertEqual(avro.parse("nm"), utf("ন্ম"))
self.assertEqual(avro.parse("Nm"), utf("ণ্ম"))
self.assertEqual(avro.parse("nd"), utf("ন্দ"))
self.assertEqual(avro.parse("nT"), utf("ন্ট"))
self.assertEqual(avro.parse("NT"), utf("ণ্ট"))
self.assertEqual(avro.parse("nD"), utf("ন্ড"))
self.assertEqual(avro.parse("ND"), utf("ণ্ড"))
self.assertEqual(avro.parse("nt"), utf("ন্ত"))
self.assertEqual(avro.parse("ns"), utf("ন্স"))
self.assertEqual(avro.parse("nc"), utf("ঞ্চ"))
self.assertEqual(avro.parse("n"), utf("ন"))
self.assertEqual(avro.parse("N"), utf("ণ"))
self.assertEqual(avro.parse("OI`"), utf("ৈ"))
self.assertEqual(avro.parse("OU`"), utf("ৌ"))
self.assertEqual(avro.parse("O`"), utf("ো"))
self.assertEqual(avro.parse("OI"), utf("ঐ"))
self.assertEqual(avro.parse("kOI"), utf("কৈ"))
self.assertEqual(avro.parse(" OI"), utf(" ঐ"))
self.assertEqual(avro.parse("(OI"), utf("(ঐ"))
self.assertEqual(avro.parse(".OI"), utf("।ঐ"))
self.assertEqual(avro.parse("OU"), utf("ঔ"))
self.assertEqual(avro.parse("kOU"), utf("কৌ"))
self.assertEqual(avro.parse(" OU"), utf(" ঔ"))
self.assertEqual(avro.parse("-OU"), utf("-ঔ"))
self.assertEqual(avro.parse(",,OU"), utf("্ঔ"))
self.assertEqual(avro.parse("O"), utf("ও"))
self.assertEqual(avro.parse("pO"), utf("পো"))
self.assertEqual(avro.parse(" O"), utf(" ও"))
self.assertEqual(avro.parse("iO"), utf("ইও"))
self.assertEqual(avro.parse("`O"), utf("ও"))
self.assertEqual(avro.parse("phl"), utf("ফ্ল"))
self.assertEqual(avro.parse("pT"), utf("প্ট"))
self.assertEqual(avro.parse("pt"), utf("প্ত"))
self.assertEqual(avro.parse("pn"), utf("প্ন"))
self.assertEqual(avro.parse("pp"), utf("প্প"))
self.assertEqual(avro.parse("pl"), utf("প্ল"))
self.assertEqual(avro.parse("ps"), utf("প্স"))
self.assertEqual(avro.parse("ph"), utf("ফ"))
self.assertEqual(avro.parse("fl"), utf("ফ্ল"))
self.assertEqual(avro.parse("f"), utf("ফ"))
self.assertEqual(avro.parse("p"), utf("প"))
self.assertEqual(avro.parse("rri`"), utf("ৃ"))
self.assertEqual(avro.parse("rri"), utf("ঋ"))
self.assertEqual(avro.parse("krri"), utf("কৃ"))
self.assertEqual(avro.parse("Irri"), utf("ঈঋ"))
self.assertEqual(avro.parse("^rri"), utf("ঁঋ"))
self.assertEqual(avro.parse(":rri"), utf("ঃঋ"))
self.assertEqual(avro.parse("rZ"), utf("র্য"))
self.assertEqual(avro.parse("krZ"), utf("ক্র্য"))
self.assertEqual(avro.parse("rrZ"), utf("রর্য"))
self.assertEqual(avro.parse("yrZ"), utf("ইয়র্য"))
self.assertEqual(avro.parse("wrZ"), utf("ওর্য"))
self.assertEqual(avro.parse("xrZ"), utf("এক্সর্য"))
self.assertEqual(avro.parse("irZ"), utf("ইর্য"))
self.assertEqual(avro.parse("-rZ"), utf("-র্য"))
self.assertEqual(avro.parse("rrrZ"), utf("ররর্য"))
self.assertEqual(avro.parse("ry"), utf("র্য"))
self.assertEqual(avro.parse("qry"), utf("ক্র্য"))
self.assertEqual(avro.parse("rry"), utf("রর্য"))
self.assertEqual(avro.parse("yry"), utf("ইয়র্য"))
self.assertEqual(avro.parse("wry"), utf("ওর্য"))
self.assertEqual(avro.parse("xry"), utf("এক্সর্য"))
self.assertEqual(avro.parse("0ry"), utf("০র্য"))
self.assertEqual(avro.parse("rrrry"), utf("রররর্য"))
self.assertEqual(avro.parse("Rry"), utf("ড়্র্য"))
self.assertEqual(avro.parse("rr"), utf("রর"))
self.assertEqual(avro.parse("arr"), utf("আরর"))
self.assertEqual(avro.parse("arrk"), utf("আর্ক"))
self.assertEqual(avro.parse("arra"), utf("আররা"))
self.assertEqual(avro.parse("arr"), utf("আরর"))
self.assertEqual(avro.parse("arr!"), utf("আরর!"))
self.assertEqual(avro.parse("krr"), utf("ক্রর"))
self.assertEqual(avro.parse("krra"), utf("ক্ররা"))
self.assertEqual(avro.parse("Rg"), utf("ড়্গ"))
self.assertEqual(avro.parse("Rh"), utf("ঢ়"))
self.assertEqual(avro.parse("R"), utf("ড়"))
self.assertEqual(avro.parse("r"), utf("র"))
self.assertEqual(avro.parse("or"), utf("অর"))
self.assertEqual(avro.parse("mr"), utf("ম্র"))
self.assertEqual(avro.parse("1r"), utf("১র"))
self.assertEqual(avro.parse("+r"), utf("+র"))
self.assertEqual(avro.parse("rr"), utf("রর"))
self.assertEqual(avro.parse("yr"), utf("ইয়র"))
self.assertEqual(avro.parse("wr"), utf("ওর"))
self.assertEqual(avro.parse("xr"), utf("এক্সর"))
self.assertEqual(avro.parse("zr"), utf("য্র"))
self.assertEqual(avro.parse("mri"), utf("ম্রি"))
self.assertEqual(avro.parse("shch"), utf("শ্ছ"))
self.assertEqual(avro.parse("ShTh"), utf("ষ্ঠ"))
self.assertEqual(avro.parse("Shph"), utf("ষ্ফ"))
self.assertEqual(avro.parse("Sch"), utf("শ্ছ"))
self.assertEqual(avro.parse("skl"), utf("স্ক্ল"))
self.assertEqual(avro.parse("skh"), utf("স্খ"))
self.assertEqual(avro.parse("sth"), utf("স্থ"))
self.assertEqual(avro.parse("sph"), utf("স্ফ"))
self.assertEqual(avro.parse("shc"), utf("শ্চ"))
self.assertEqual(avro.parse("sht"), utf("শ্ত"))
self.assertEqual(avro.parse("shn"), utf("শ্ন"))
self.assertEqual(avro.parse("shm"), utf("শ্ম"))
self.assertEqual(avro.parse("shl"), utf("শ্ল"))
self.assertEqual(avro.parse("Shk"), utf("ষ্ক"))
self.assertEqual(avro.parse("ShT"), utf("ষ্ট"))
self.assertEqual(avro.parse("ShN"), utf("ষ্ণ"))
self.assertEqual(avro.parse("Shp"), utf("ষ্প"))
self.assertEqual(avro.parse("Shf"), utf("ষ্ফ"))
self.assertEqual(avro.parse("Shm"), utf("ষ্ম"))
self.assertEqual(avro.parse("spl"), utf("স্প্ল"))
self.assertEqual(avro.parse("sk"), utf("স্ক"))
self.assertEqual(avro.parse("Sc"), utf("শ্চ"))
self.assertEqual(avro.parse("sT"), utf("স্ট"))
self.assertEqual(avro.parse("st"), utf("স্ত"))
self.assertEqual(avro.parse("sn"), utf("স্ন"))
self.assertEqual(avro.parse("sp"), utf("স্প"))
self.assertEqual(avro.parse("sf"), utf("স্ফ"))
self.assertEqual(avro.parse("sm"), utf("স্ম"))
self.assertEqual(avro.parse("sl"), utf("স্ল"))
self.assertEqual(avro.parse("sh"), utf("শ"))
self.assertEqual(avro.parse("Sc"), utf("শ্চ"))
self.assertEqual(avro.parse("St"), utf("শ্ত"))
self.assertEqual(avro.parse("Sn"), utf("শ্ন"))
self.assertEqual(avro.parse("Sm"), utf("শ্ম"))
self.assertEqual(avro.parse("Sl"), utf("শ্ল"))
self.assertEqual(avro.parse("Sh"), utf("ষ"))
self.assertEqual(avro.parse("s"), utf("স"))
self.assertEqual(avro.parse("S"), utf("শ"))
self.assertEqual(avro.parse("oo"), utf("উ"))
self.assertEqual(avro.parse("OO"), utf("ওও"))
self.assertEqual(avro.parse("oo`"), utf("ু"))
self.assertEqual(avro.parse("koo"), utf("কু"))
self.assertEqual(avro.parse("ooo"), utf("উঅ"))
self.assertEqual(avro.parse("!oo"), utf("!উ"))
self.assertEqual(avro.parse("!ooo"), utf("!উঅ"))
self.assertEqual(avro.parse("aoo"), utf("আউ"))
self.assertEqual(avro.parse("oop"), utf("উপ"))
self.assertEqual(avro.parse("ooo`"), utf("উ"))
self.assertEqual("", avro.parse("o`"))
self.assertEqual(avro.parse("oZ"), utf("অ্য"))
self.assertEqual(avro.parse("oY"), utf("অয়"))
self.assertEqual(avro.parse("o"), utf("অ"))
self.assertEqual(avro.parse("!o"), utf("!অ"))
self.assertEqual(avro.parse("^o"), utf("ঁঅ"))
self.assertEqual(avro.parse("*o"), utf("*অ"))
self.assertEqual(avro.parse("io"), utf("ইও"))
self.assertEqual(avro.parse("yo"), utf("ইয়"))
self.assertEqual(avro.parse("no"), utf("ন"))
self.assertEqual(avro.parse("tth"), utf("ত্থ"))
self.assertEqual(avro.parse("t``"), utf("ৎ"))
self.assertEqual(avro.parse("`t``"), utf("ৎ"))
self.assertEqual(avro.parse("t``t``"), utf("ৎৎ"))
self.assertEqual(avro.parse("t```"), utf("ৎ"))
self.assertEqual(avro.parse("TT"), utf("ট্ট"))
self.assertEqual(avro.parse("Tm"), utf("ট্ম"))
self.assertEqual(avro.parse("Th"), utf("ঠ"))
self.assertEqual(avro.parse("tn"), utf("ত্ন"))
self.assertEqual(avro.parse("tm"), utf("ত্ম"))
self.assertEqual(avro.parse("th"), utf("থ"))
self.assertEqual(avro.parse("tt"), utf("ত্ত"))
self.assertEqual(avro.parse("T"), utf("ট"))
self.assertEqual(avro.parse("t"), utf("ত"))
self.assertEqual(avro.parse("aZ"), utf("অ্যা"))
self.assertEqual(avro.parse("aaZ"), utf("আঅ্যা"))
self.assertEqual(avro.parse("AZ"), utf("অ্যা"))
self.assertEqual(avro.parse("a`"), utf("া"))
self.assertEqual(avro.parse("a``"), utf("া"))
self.assertEqual(avro.parse("ka`"), utf("কা"))
self.assertEqual(avro.parse("A`"), utf("া"))
self.assertEqual(avro.parse("a"), utf("আ"))
self.assertEqual(avro.parse("`a"), utf("আ"))
self.assertEqual(avro.parse("k`a"), utf("কআ"))
self.assertEqual(avro.parse("ia"), utf("ইয়া"))
self.assertEqual(avro.parse("aaaa`"), utf("আআআা"))
self.assertEqual(avro.parse("i`"), utf("ি"))
self.assertEqual(avro.parse("i"), utf("ই"))
self.assertEqual(avro.parse("`i"), utf("ই"))
self.assertEqual(avro.parse("hi"), utf("হি"))
self.assertEqual(avro.parse("ih"), utf("ইহ"))
self.assertEqual(avro.parse("i`h"), utf("িহ"))
self.assertEqual(avro.parse("I`"), utf("ী"))
self.assertEqual(avro.parse("I"), utf("ঈ"))
self.assertEqual(avro.parse("cI"), utf("চী"))
self.assertEqual(avro.parse("Ix"), utf("ঈক্স"))
self.assertEqual(avro.parse("II"), utf("ঈঈ"))
self.assertEqual(avro.parse("0I"), utf("০ঈ"))
self.assertEqual(avro.parse("oI"), utf("অঈ"))
self.assertEqual(avro.parse("u`"), utf("ু"))
self.assertEqual(avro.parse("u"), utf("উ"))
self.assertEqual(avro.parse("ku"), utf("কু"))
self.assertEqual(avro.parse("uk"), utf("উক"))
self.assertEqual(avro.parse("uu"), utf("উউ"))
self.assertEqual(avro.parse("iu"), utf("ইউ"))
self.assertEqual(avro.parse("&u"), utf("&উ"))
self.assertEqual(avro.parse("u&"), utf("উ&"))
self.assertEqual(avro.parse("U`"), utf("ূ"))
self.assertEqual(avro.parse("U"), utf("ঊ"))
self.assertEqual(avro.parse("yU"), utf("ইয়ূ"))
self.assertEqual(avro.parse("Uy"), utf("ঊয়"))
self.assertEqual(avro.parse("^U"), utf("ঁঊ"))
self.assertEqual(avro.parse("U^"), utf("ঊঁ"))
self.assertEqual(avro.parse("EE"), utf("ঈ"))
self.assertEqual(avro.parse("ee"), utf("ঈ"))
self.assertEqual(avro.parse("Ee"), utf("ঈ"))
self.assertEqual(avro.parse("eE"), utf("ঈ"))
self.assertEqual(avro.parse("ee`"), utf("ী"))
self.assertEqual(avro.parse("kee"), utf("কী"))
self.assertEqual(avro.parse("eek"), utf("ঈক"))
self.assertEqual(avro.parse("0ee"), utf("০ঈ"))
self.assertEqual(avro.parse("ee8"), utf("ঈ৮"))
self.assertEqual(avro.parse("(ee)"), utf("(ঈ)"))
self.assertEqual(avro.parse("e`"), utf("ে"))
self.assertEqual(avro.parse("e"), utf("এ"))
self.assertEqual(avro.parse("ke"), utf("কে"))
self.assertEqual(avro.parse("we"), utf("ওয়ে"))
self.assertEqual(avro.parse("#e#"), utf("#এ#"))
self.assertEqual(avro.parse("`e`"), utf("ে"))
self.assertEqual(avro.parse("z"), utf("য"))
self.assertEqual(avro.parse("Z"), utf("্য"))
self.assertEqual(avro.parse("rZ"), utf("র্য"))
self.assertEqual(avro.parse("kZS"), utf("ক্যশ"))
self.assertEqual(avro.parse("y"), utf("ইয়"))
self.assertEqual(avro.parse("oy"), utf("অয়"))
self.assertEqual(avro.parse("ky"), utf("ক্য"))
self.assertEqual(avro.parse("ya"), utf("ইয়া"))
self.assertEqual(avro.parse("yaa"), utf("ইয়াআ"))
self.assertEqual(avro.parse("Y"), utf("য়"))
self.assertEqual(avro.parse("YY"), utf("য়য়"))
self.assertEqual(avro.parse("iY"), utf("ইয়"))
self.assertEqual(avro.parse("kY"), utf("কয়"))
self.assertEqual(avro.parse("q"), utf("ক"))
self.assertEqual(avro.parse("Q"), utf("ক"))
self.assertEqual(avro.parse("w"), utf("ও"))
self.assertEqual(avro.parse("wa"), utf("ওয়া"))
self.assertEqual(avro.parse("-wa-"), utf("-ওয়া-"))
self.assertEqual(avro.parse("woo"), utf("ওয়ু"))
self.assertEqual(avro.parse("wre"), utf("ওরে"))
self.assertEqual(avro.parse("kw"), utf("ক্ব"))
self.assertEqual(avro.parse("x"), utf("এক্স"))
self.assertEqual(avro.parse("ex"), utf("এক্স"))
self.assertEqual(avro.parse("bx"), utf("বক্স"))
self.assertEqual(avro.parse(":`"), utf(":"))
self.assertEqual(avro.parse(":"), utf("ঃ"))
self.assertEqual(avro.parse("^`"), utf("^"))
self.assertEqual(avro.parse("^"), utf("ঁ"))
self.assertEqual(avro.parse("k^"), utf("কঁ"))
self.assertEqual(avro.parse("k^i"), utf("কঁই"))
self.assertEqual(avro.parse("ki^"), utf("কিঁ"))
self.assertEqual(avro.parse(",,"), utf("্"))
self.assertEqual(avro.parse(",,,"), utf("্,"))
self.assertEqual(avro.parse(",,`,"), utf("্,"))
self.assertEqual(avro.parse("`,,"), utf("্"))
self.assertEqual(avro.parse(",`,"), utf(",,"))
self.assertEqual(avro.parse("$"), utf("৳"))
self.assertEqual("", avro.parse("`"))
self.assertEqual(avro.parse("bdh"), utf("ব্ধ"))
def test_other(self):
self.assertEqual(avro.parse("!"), utf("!"))
def test_other(self):
self.assertEqual(avro.parse("!"), utf("!"))
def test_patterns_numbers(self):
"""Test patterns - numbers"""
# Test some numbers
self.assertEqual(utf("০"), avro.parse("0"))
self.assertEqual(utf("১"), avro.parse("1"))
self.assertEqual(utf("২"), avro.parse("2"))
self.assertEqual(utf("৩"), avro.parse("3"))
self.assertEqual(utf("৪"), avro.parse("4"))
self.assertEqual(utf("৫"), avro.parse("5"))
self.assertEqual(utf("৬"), avro.parse("6"))
self.assertEqual(utf("৭"), avro.parse("7"))
self.assertEqual(utf("৮"), avro.parse("8"))
self.assertEqual(utf("৯"), avro.parse("9"))
self.assertEqual(utf("১১২"), avro.parse("112"))
def test_basic(self):
self.assertEqual(avro.parse("bhl"), utf("ভ্ল"))
self.assertEqual(avro.parse("bj"), utf("ব্জ"))
self.assertEqual(avro.parse("bd"), utf("ব্দ"))
self.assertEqual(avro.parse("bb"), utf("ব্ব"))
self.assertEqual(avro.parse("bl"), utf("ব্ল"))
self.assertEqual(avro.parse("bh"), utf("ভ"))
self.assertEqual(avro.parse("vl"), utf("ভ্ল"))
self.assertEqual(avro.parse("b"), utf("ব"))
self.assertEqual(avro.parse("v"), utf("ভ"))
self.assertEqual(avro.parse("cNG"), utf("চ্ঞ"))
self.assertEqual(avro.parse("cch"), utf("চ্ছ"))
self.assertEqual(avro.parse("cc"), utf("চ্চ"))
self.assertEqual(avro.parse("ch"), utf("ছ"))
self.assertEqual(avro.parse("c"), utf("চ"))
self.assertEqual(avro.parse("dhn"), utf("ধ্ন"))
self.assertEqual(avro.parse("dhm"), utf("ধ্ম"))
self.assertEqual(avro.parse("dgh"), utf("দ্ঘ"))
self.assertEqual(avro.parse("ddh"), utf("দ্ধ"))
self.assertEqual(avro.parse("dbh"), utf("দ্ভ"))
self.assertEqual(avro.parse("dv"), utf("দ্ভ"))
self.assertEqual(avro.parse("dm"), utf("দ্ম"))
self.assertEqual(avro.parse("DD"), utf("ড্ড"))
self.assertEqual(avro.parse("Dh"), utf("ঢ"))
self.assertEqual(avro.parse("dh"), utf("ধ"))
self.assertEqual(avro.parse("dg"), utf("দ্গ"))
self.assertEqual(avro.parse("dd"), utf("দ্দ"))
self.assertEqual(avro.parse("D"), utf("ড"))
self.assertEqual(avro.parse("d"), utf("দ"))
self.assertEqual(avro.parse("..."), utf("..."))
self.assertEqual(avro.parse(".`"), utf("."))
self.assertEqual(avro.parse(".."), utf("।।"))
self.assertEqual(avro.parse("."), utf("।"))
self.assertEqual(avro.parse("ghn"), utf("ঘ্ন"))
self.assertEqual(avro.parse("Ghn"), utf("ঘ্ন"))
self.assertEqual(avro.parse("gdh"), utf("গ্ধ"))
self.assertEqual(avro.parse("gN"), utf("গ্ণ"))
self.assertEqual(avro.parse("GN"), utf("গ্ণ"))
self.assertEqual(avro.parse("gn"), utf("গ্ন"))
self.assertEqual(avro.parse("gm"), utf("গ্ম"))
self.assertEqual(avro.parse("Gm"), utf("গ্ম"))
self.assertEqual(avro.parse("gl"), utf("গ্ল"))
self.assertEqual(avro.parse("Gl"), utf("গ্ল"))
self.assertEqual(avro.parse("gg"), utf("জ্ঞ"))
self.assertEqual(avro.parse("GG"), utf("জ্ঞ"))
self.assertEqual(avro.parse("Gg"), utf("জ্ঞ"))
self.assertEqual(avro.parse("gG"), utf("জ্ঞ"))
self.assertEqual(avro.parse("gh"), utf("ঘ"))
self.assertEqual(avro.parse("Gh"), utf("ঘ"))
self.assertEqual(avro.parse("g"), utf("গ"))
self.assertEqual(avro.parse("hN"), utf("হ্ণ"))
self.assertEqual(avro.parse("hn"), utf("হ্ন"))
self.assertEqual(avro.parse("hm"), utf("হ্ম"))
self.assertEqual(avro.parse("hl"), utf("হ্ল"))
self.assertEqual(avro.parse("h"), utf("হ"))
self.assertEqual(avro.parse("jjh"), utf("জ্ঝ"))
self.assertEqual(avro.parse("jNG"), utf("জ্ঞ"))
self.assertEqual(avro.parse("jh"), utf("ঝ"))
self.assertEqual(avro.parse("jj"), utf("জ্জ"))
self.assertEqual(avro.parse("j"), utf("জ"))
self.assertEqual(avro.parse("J"), utf("জ"))
self.assertEqual(avro.parse("kkhN"), utf("ক্ষ্ণ"))
self.assertEqual(avro.parse("kShN"), utf("ক্ষ্ণ"))
self.assertEqual(avro.parse("kkhm"), utf("ক্ষ্ম"))
self.assertEqual(avro.parse("kShm"), utf("ক্ষ্ম"))
self.assertEqual(avro.parse("kxN"), utf("ক্ষ্ণ"))
self.assertEqual(avro.parse("kxm"), utf("ক্ষ্ম"))
self.assertEqual(avro.parse("kkh"), utf("ক্ষ"))
self.assertEqual(avro.parse("kSh"), utf("ক্ষ"))
self.assertEqual(avro.parse("ksh"), utf("কশ"))
self.assertEqual(avro.parse("kx"), utf("ক্ষ"))
self.assertEqual(avro.parse("kk"), utf("ক্ক"))
self.assertEqual(avro.parse("kT"), utf("ক্ট"))
self.assertEqual(avro.parse("kt"), utf("ক্ত"))
self.assertEqual(avro.parse("kl"), utf("ক্ল"))
self.assertEqual(avro.parse("ks"), utf("ক্স"))
self.assertEqual(avro.parse("kh"), utf("খ"))
self.assertEqual(avro.parse("k"), utf("ক"))
self.assertEqual(avro.parse("lbh"), utf("ল্ভ"))
self.assertEqual(avro.parse("ldh"), utf("ল্ধ"))
self.assertEqual(avro.parse("lkh"), utf("লখ"))
self.assertEqual(avro.parse("lgh"), utf("লঘ"))
self.assertEqual(avro.parse("lph"), utf("লফ"))
self.assertEqual(avro.parse("lk"), utf("ল্ক"))
self.assertEqual(avro.parse("lg"), utf("ল্গ"))
self.assertEqual(avro.parse("lT"), utf("ল্ট"))
self.assertEqual(avro.parse("lD"), utf("ল্ড"))
self.assertEqual(avro.parse("lp"), utf("ল্প"))
self.assertEqual(avro.parse("lv"), utf("ল্ভ"))
self.assertEqual(avro.parse("lm"), utf("ল্ম"))
self.assertEqual(avro.parse("ll"), utf("ল্ল"))
self.assertEqual(avro.parse("lb"), utf("ল্ব"))
self.assertEqual(avro.parse("l"), utf("ল"))
self.assertEqual(avro.parse("mth"), utf("ম্থ"))
self.assertEqual(avro.parse("mph"), utf("ম্ফ"))
self.assertEqual(avro.parse("mbh"), utf("ম্ভ"))
self.assertEqual(avro.parse("mpl"), utf("মপ্ল"))
self.assertEqual(avro.parse("mn"), utf("ম্ন"))
self.assertEqual(avro.parse("mp"), utf("ম্প"))
self.assertEqual(avro.parse("mv"), utf("ম্ভ"))
self.assertEqual(avro.parse("mm"), utf("ম্ম"))
self.assertEqual(avro.parse("ml"), utf("ম্ল"))
self.assertEqual(avro.parse("mb"), utf("ম্ব"))
self.assertEqual(avro.parse("mf"), utf("ম্ফ"))
self.assertEqual(avro.parse("m"), utf("ম"))
self.assertEqual(avro.parse("0"), utf("০"))
self.assertEqual(avro.parse("1"), utf("১"))
self.assertEqual(avro.parse("2"), utf("২"))
self.assertEqual(avro.parse("3"), utf("৩"))
self.assertEqual(avro.parse("4"), utf("৪"))
self.assertEqual(avro.parse("5"), utf("৫"))
self.assertEqual(avro.parse("6"), utf("৬"))
self.assertEqual(avro.parse("7"), utf("৭"))
self.assertEqual(avro.parse("8"), utf("৮"))
self.assertEqual(avro.parse("9"), utf("৯"))
self.assertEqual(avro.parse("NgkSh"), utf("ঙ্ক্ষ"))
self.assertEqual(avro.parse("Ngkkh"), utf("ঙ্ক্ষ"))
self.assertEqual(avro.parse("NGch"), utf("ঞ্ছ"))
self.assertEqual(avro.parse("Nggh"), utf("ঙ্ঘ"))
self.assertEqual(avro.parse("Ngkh"), utf("ঙ্খ"))
self.assertEqual(avro.parse("NGjh"), utf("ঞ্ঝ"))
self.assertEqual(avro.parse("ngOU"), utf("ঙ্গৌ"))
self.assertEqual(avro.parse("ngOI"), utf("ঙ্গৈ"))
self.assertEqual(avro.parse("Ngkx"), utf("ঙ্ক্ষ"))
self.assertEqual(avro.parse("NGc"), utf("ঞ্চ"))
self.assertEqual(avro.parse("nch"), utf("ঞ্ছ"))
self.assertEqual(avro.parse("njh"), utf("ঞ্ঝ"))
self.assertEqual(avro.parse("ngh"), utf("ঙ্ঘ"))
self.assertEqual(avro.parse("Ngk"), utf("ঙ্ক"))
self.assertEqual(avro.parse("Ngx"), utf("ঙ্ষ"))
self.assertEqual(avro.parse("Ngg"), utf("ঙ্গ"))
self.assertEqual(avro.parse("Ngm"), utf("ঙ্ম"))
self.assertEqual(avro.parse("NGj"), utf("ঞ্জ"))
self.assertEqual(avro.parse("ndh"), utf("ন্ধ"))
self.assertEqual(avro.parse("nTh"), utf("ন্ঠ"))
self.assertEqual(avro.parse("NTh"), utf("ণ্ঠ"))
self.assertEqual(avro.parse("nth"), utf("ন্থ"))
self.assertEqual(avro.parse("nkh"), utf("ঙ্খ"))
self.assertEqual(avro.parse("ngo"), utf("ঙ্গ"))
self.assertEqual(avro.parse("nga"), utf("ঙ্গা"))
self.assertEqual(avro.parse("ngi"), utf("ঙ্গি"))
self.assertEqual(avro.parse("ngI"), utf("ঙ্গী"))
self.assertEqual(avro.parse("ngu"), utf("ঙ্গু"))
self.assertEqual(avro.parse("ngU"), utf("ঙ্গূ"))
self.assertEqual(avro.parse("nge"), utf("ঙ্গে"))
self.assertEqual(avro.parse("ngO"), utf("ঙ্গো"))
self.assertEqual(avro.parse("NDh"), utf("ণ্ঢ"))
self.assertEqual(avro.parse("nsh"), utf("নশ"))
self.assertEqual(avro.parse("Ngr"), utf("ঙর"))
self.assertEqual(avro.parse("NGr"), utf("ঞর"))
self.assertEqual(avro.parse("ngr"), utf("ংর"))
self.assertEqual(avro.parse("nj"), utf("ঞ্জ"))
self.assertEqual(avro.parse("Ng"), utf("ঙ"))
self.assertEqual(avro.parse("NG"), utf("ঞ"))
self.assertEqual(avro.parse("nk"), utf("ঙ্ক"))
self.assertEqual(avro.parse("ng"), utf("ং"))
self.assertEqual(avro.parse("nn"), utf("ন্ন"))
self.assertEqual(avro.parse("NN"), utf("ণ্ণ"))
self.assertEqual(avro.parse("Nn"), utf("ণ্ন"))
self.assertEqual(avro.parse("nm"), utf("ন্ম"))
self.assertEqual(avro.parse("Nm"), utf("ণ্ম"))
self.assertEqual(avro.parse("nd"), utf("ন্দ"))
self.assertEqual(avro.parse("nT"), utf("ন্ট"))
self.assertEqual(avro.parse("NT"), utf("ণ্ট"))
self.assertEqual(avro.parse("nD"), utf("ন্ড"))
self.assertEqual(avro.parse("ND"), utf("ণ্ড"))
self.assertEqual(avro.parse("nt"), utf("ন্ত"))
self.assertEqual(avro.parse("ns"), utf("ন্স"))
self.assertEqual(avro.parse("nc"), utf("ঞ্চ"))
self.assertEqual(avro.parse("n"), utf("ন"))
self.assertEqual(avro.parse("N"), utf("ণ"))
self.assertEqual(avro.parse("OI`"), utf("ৈ"))
self.assertEqual(avro.parse("OU`"), utf("ৌ"))
self.assertEqual(avro.parse("O`"), utf("ো"))
self.assertEqual(avro.parse("OI"), utf("ঐ"))
self.assertEqual(avro.parse("kOI"), utf("কৈ"))
self.assertEqual(avro.parse(" OI"), utf(" ঐ"))
self.assertEqual(avro.parse("(OI"), utf("(ঐ"))
self.assertEqual(avro.parse(".OI"), utf("।ঐ"))
self.assertEqual(avro.parse("OU"), utf("ঔ"))
self.assertEqual(avro.parse("kOU"), utf("কৌ"))
self.assertEqual(avro.parse(" OU"), utf(" ঔ"))
self.assertEqual(avro.parse("-OU"), utf("-ঔ"))
self.assertEqual(avro.parse(",,OU"), utf("্ঔ"))
self.assertEqual(avro.parse("O"), utf("ও"))
self.assertEqual(avro.parse("pO"), utf("পো"))
self.assertEqual(avro.parse(" O"), utf(" ও"))
self.assertEqual(avro.parse("iO"), utf("ইও"))
self.assertEqual(avro.parse("`O"), utf("ও"))
self.assertEqual(avro.parse("phl"), utf("ফ্ল"))
self.assertEqual(avro.parse("pT"), utf("প্ট"))
self.assertEqual(avro.parse("pt"), utf("প্ত"))
self.assertEqual(avro.parse("pn"), utf("প্ন"))
self.assertEqual(avro.parse("pp"), utf("প্প"))
self.assertEqual(avro.parse("pl"), utf("প্ল"))
self.assertEqual(avro.parse("ps"), utf("প্স"))
self.assertEqual(avro.parse("ph"), utf("ফ"))
self.assertEqual(avro.parse("fl"), utf("ফ্ল"))
self.assertEqual(avro.parse("f"), utf("ফ"))
self.assertEqual(avro.parse("p"), utf("প"))
self.assertEqual(avro.parse("rri`"), utf("ৃ"))
self.assertEqual(avro.parse("rri"), utf("ঋ"))
self.assertEqual(avro.parse("krri"), utf("কৃ"))
self.assertEqual(avro.parse("Irri"), utf("ঈঋ"))
self.assertEqual(avro.parse("^rri"), utf("ঁঋ"))
self.assertEqual(avro.parse(":rri"), utf("ঃঋ"))
self.assertEqual(avro.parse("rZ"), utf("র্য"))
self.assertEqual(avro.parse("krZ"), utf("ক্র্য"))
self.assertEqual(avro.parse("rrZ"), utf("রর্য"))
self.assertEqual(avro.parse("yrZ"), utf("ইয়র্য"))
self.assertEqual(avro.parse("wrZ"), utf("ওর্য"))
self.assertEqual(avro.parse("xrZ"), utf("এক্সর্য"))
self.assertEqual(avro.parse("irZ"), utf("ইর্য"))
self.assertEqual(avro.parse("-rZ"), utf("-র্য"))
self.assertEqual(avro.parse("rrrZ"), utf("ররর্য"))
self.assertEqual(avro.parse("ry"), utf("র্য"))
self.assertEqual(avro.parse("qry"), utf("ক্র্য"))
self.assertEqual(avro.parse("rry"), utf("রর্য"))
self.assertEqual(avro.parse("yry"), utf("ইয়র্য"))
self.assertEqual(avro.parse("wry"), utf("ওর্য"))
self.assertEqual(avro.parse("xry"), utf("এক্সর্য"))
self.assertEqual(avro.parse("0ry"), utf("০র্য"))
self.assertEqual(avro.parse("rrrry"), utf("রররর্য"))
self.assertEqual(avro.parse("Rry"), utf("ড়্র্য"))
self.assertEqual(avro.parse("rr"), utf("রর"))
self.assertEqual(avro.parse("arr"), utf("আরর"))
self.assertEqual(avro.parse("arrk"), utf("আর্ক"))
self.assertEqual(avro.parse("arra"), utf("আররা"))
self.assertEqual(avro.parse("arr"), utf("আরর"))
self.assertEqual(avro.parse("arr!"), utf("আরর!"))
self.assertEqual(avro.parse("krr"), utf("ক্রর"))
self.assertEqual(avro.parse("krra"), utf("ক্ররা"))
self.assertEqual(avro.parse("Rg"), utf("ড়্গ"))
self.assertEqual(avro.parse("Rh"), utf("ঢ়"))
self.assertEqual(avro.parse("R"), utf("ড়"))
self.assertEqual(avro.parse("r"), utf("র"))
self.assertEqual(avro.parse("or"), utf("অর"))
self.assertEqual(avro.parse("mr"), utf("ম্র"))
self.assertEqual(avro.parse("1r"), utf("১র"))
self.assertEqual(avro.parse("+r"), utf("+র"))
self.assertEqual(avro.parse("rr"), utf("রর"))
self.assertEqual(avro.parse("yr"), utf("ইয়র"))
self.assertEqual(avro.parse("wr"), utf("ওর"))
self.assertEqual(avro.parse("xr"), utf("এক্সর"))
self.assertEqual(avro.parse("zr"), utf("য্র"))
self.assertEqual(avro.parse("mri"), utf("ম্রি"))
self.assertEqual(avro.parse("shch"), utf("শ্ছ"))
self.assertEqual(avro.parse("ShTh"), utf("ষ্ঠ"))
self.assertEqual(avro.parse("Shph"), utf("ষ্ফ"))
self.assertEqual(avro.parse("Sch"), utf("শ্ছ"))
self.assertEqual(avro.parse("skl"), utf("স্ক্ল"))
self.assertEqual(avro.parse("skh"), utf("স্খ"))
self.assertEqual(avro.parse("sth"), utf("স্থ"))
self.assertEqual(avro.parse("sph"), utf("স্ফ"))
self.assertEqual(avro.parse("shc"), utf("শ্চ"))
self.assertEqual(avro.parse("sht"), utf("শ্ত"))
self.assertEqual(avro.parse("shn"), utf("শ্ন"))
self.assertEqual(avro.parse("shm"), utf("শ্ম"))
self.assertEqual(avro.parse("shl"), utf("শ্ল"))
self.assertEqual(avro.parse("Shk"), utf("ষ্ক"))
self.assertEqual(avro.parse("ShT"), utf("ষ্ট"))
self.assertEqual(avro.parse("ShN"), utf("ষ্ণ"))
self.assertEqual(avro.parse("Shp"), utf("ষ্প"))
self.assertEqual(avro.parse("Shf"), utf("ষ্ফ"))
self.assertEqual(avro.parse("Shm"), utf("ষ্ম"))
self.assertEqual(avro.parse("spl"), utf("স্প্ল"))
self.assertEqual(avro.parse("sk"), utf("স্ক"))
self.assertEqual(avro.parse("Sc"), utf("শ্চ"))
self.assertEqual(avro.parse("sT"), utf("স্ট"))
self.assertEqual(avro.parse("st"), utf("স্ত"))
self.assertEqual(avro.parse("sn"), utf("স্ন"))
self.assertEqual(avro.parse("sp"), utf("স্প"))
self.assertEqual(avro.parse("sf"), utf("স্ফ"))
self.assertEqual(avro.parse("sm"), utf("স্ম"))
self.assertEqual(avro.parse("sl"), utf("স্ল"))
self.assertEqual(avro.parse("sh"), utf("শ"))
self.assertEqual(avro.parse("Sc"), utf("শ্চ"))
self.assertEqual(avro.parse("St"), utf("শ্ত"))
self.assertEqual(avro.parse("Sn"), utf("শ্ন"))
self.assertEqual(avro.parse("Sm"), utf("শ্ম"))
self.assertEqual(avro.parse("Sl"), utf("শ্ল"))
self.assertEqual(avro.parse("Sh"), utf("ষ"))
self.assertEqual(avro.parse("s"), utf("স"))
self.assertEqual(avro.parse("S"), utf("শ"))
self.assertEqual(avro.parse("oo"), utf("উ"))
self.assertEqual(avro.parse("OO"), utf("ওও"))
self.assertEqual(avro.parse("oo`"), utf("ু"))
self.assertEqual(avro.parse("koo"), utf("কু"))
self.assertEqual(avro.parse("ooo"), utf("উঅ"))
self.assertEqual(avro.parse("!oo"), utf("!উ"))
self.assertEqual(avro.parse("!ooo"), utf("!উঅ"))
self.assertEqual(avro.parse("aoo"), utf("আউ"))
self.assertEqual(avro.parse("oop"), utf("উপ"))
self.assertEqual(avro.parse("ooo`"), utf("উ"))
self.assertEqual("", avro.parse("o`"))
self.assertEqual(avro.parse("oZ"), utf("অ্য"))
self.assertEqual(avro.parse("oY"), utf("অয়"))
self.assertEqual(avro.parse("o"), utf("অ"))
self.assertEqual(avro.parse("!o"), utf("!অ"))
self.assertEqual(avro.parse("^o"), utf("ঁঅ"))
self.assertEqual(avro.parse("*o"), utf("*অ"))
self.assertEqual(avro.parse("io"), utf("ইও"))
self.assertEqual(avro.parse("yo"), utf("ইয়"))
self.assertEqual(avro.parse("no"), utf("ন"))
self.assertEqual(avro.parse("tth"), utf("ত্থ"))
self.assertEqual(avro.parse("t``"), utf("ৎ"))
self.assertEqual(avro.parse("`t``"), utf("ৎ"))
self.assertEqual(avro.parse("t``t``"), utf("ৎৎ"))
self.assertEqual(avro.parse("t```"), utf("ৎ"))
self.assertEqual(avro.parse("TT"), utf("ট্ট"))
self.assertEqual(avro.parse("Tm"), utf("ট্ম"))
self.assertEqual(avro.parse("Th"), utf("ঠ"))
self.assertEqual(avro.parse("tn"), utf("ত্ন"))
self.assertEqual(avro.parse("tm"), utf("ত্ম"))
self.assertEqual(avro.parse("th"), utf("থ"))
self.assertEqual(avro.parse("tt"), utf("ত্ত"))
self.assertEqual(avro.parse("T"), utf("ট"))
self.assertEqual(avro.parse("t"), utf("ত"))
self.assertEqual(avro.parse("aZ"), utf("অ্যা"))
self.assertEqual(avro.parse("aaZ"), utf("আঅ্যা"))
self.assertEqual(avro.parse("AZ"), utf("অ্যা"))
self.assertEqual(avro.parse("a`"), utf("া"))
self.assertEqual(avro.parse("a``"), utf("া"))
self.assertEqual(avro.parse("ka`"), utf("কা"))
self.assertEqual(avro.parse("A`"), utf("া"))
self.assertEqual(avro.parse("a"), utf("আ"))
self.assertEqual(avro.parse("`a"), utf("আ"))
self.assertEqual(avro.parse("k`a"), utf("কআ"))
self.assertEqual(avro.parse("ia"), utf("ইয়া"))
self.assertEqual(avro.parse("aaaa`"), utf("আআআা"))
self.assertEqual(avro.parse("i`"), utf("ি"))
self.assertEqual(avro.parse("i"), utf("ই"))
self.assertEqual(avro.parse("`i"), utf("ই"))
self.assertEqual(avro.parse("hi"), utf("হি"))
self.assertEqual(avro.parse("ih"), utf("ইহ"))
self.assertEqual(avro.parse("i`h"), utf("িহ"))
self.assertEqual(avro.parse("I`"), utf("ী"))
self.assertEqual(avro.parse("I"), utf("ঈ"))
self.assertEqual(avro.parse("cI"), utf("চী"))
self.assertEqual(avro.parse("Ix"), utf("ঈক্স"))
self.assertEqual(avro.parse("II"), utf("ঈঈ"))
self.assertEqual(avro.parse("0I"), utf("০ঈ"))
self.assertEqual(avro.parse("oI"), utf("অঈ"))
self.assertEqual(avro.parse("u`"), utf("ু"))
self.assertEqual(avro.parse("u"), utf("উ"))
self.assertEqual(avro.parse("ku"), utf("কু"))
self.assertEqual(avro.parse("uk"), utf("উক"))
self.assertEqual(avro.parse("uu"), utf("উউ"))
self.assertEqual(avro.parse("iu"), utf("ইউ"))
self.assertEqual(avro.parse("&u"), utf("&উ"))
self.assertEqual(avro.parse("u&"), utf("উ&"))
self.assertEqual(avro.parse("U`"), utf("ূ"))
self.assertEqual(avro.parse("U"), utf("ঊ"))
self.assertEqual(avro.parse("yU"), utf("ইয়ূ"))
self.assertEqual(avro.parse("Uy"), utf("ঊয়"))
self.assertEqual(avro.parse("^U"), utf("ঁঊ"))
self.assertEqual(avro.parse("U^"), utf("ঊঁ"))
self.assertEqual(avro.parse("EE"), utf("ঈ"))
self.assertEqual(avro.parse("ee"), utf("ঈ"))
self.assertEqual(avro.parse("Ee"), utf("ঈ"))
self.assertEqual(avro.parse("eE"), utf("ঈ"))
self.assertEqual(avro.parse("ee`"), utf("ী"))
self.assertEqual(avro.parse("kee"), utf("কী"))
self.assertEqual(avro.parse("eek"), utf("ঈক"))
self.assertEqual(avro.parse("0ee"), utf("০ঈ"))
self.assertEqual(avro.parse("ee8"), utf("ঈ৮"))
self.assertEqual(avro.parse("(ee)"), utf("(ঈ)"))
self.assertEqual(avro.parse("e`"), utf("ে"))
self.assertEqual(avro.parse("e"), utf("এ"))
self.assertEqual(avro.parse("ke"), utf("কে"))
self.assertEqual(avro.parse("we"), utf("ওয়ে"))
self.assertEqual(avro.parse("#e#"), utf("#এ#"))
self.assertEqual(avro.parse("`e`"), utf("ে"))
self.assertEqual(avro.parse("z"), utf("য"))
self.assertEqual(avro.parse("Z"), utf("্য"))
self.assertEqual(avro.parse("rZ"), utf("র্য"))
self.assertEqual(avro.parse("kZS"), utf("ক্যশ"))
self.assertEqual(avro.parse("y"), utf("ইয়"))
self.assertEqual(avro.parse("oy"), utf("অয়"))
self.assertEqual(avro.parse("ky"), utf("ক্য"))
self.assertEqual(avro.parse("ya"), utf("ইয়া"))
self.assertEqual(avro.parse("yaa"), utf("ইয়াআ"))
self.assertEqual(avro.parse("Y"), utf("য়"))
self.assertEqual(avro.parse("YY"), utf("য়য়"))
self.assertEqual(avro.parse("iY"), utf("ইয়"))
self.assertEqual(avro.parse("kY"), utf("কয়"))
self.assertEqual(avro.parse("q"), utf("ক"))
self.assertEqual(avro.parse("Q"), utf("ক"))
self.assertEqual(avro.parse("w"), utf("ও"))
self.assertEqual(avro.parse("wa"), utf("ওয়া"))
self.assertEqual(avro.parse("-wa-"), utf("-ওয়া-"))
self.assertEqual(avro.parse("woo"), utf("ওয়ু"))
self.assertEqual(avro.parse("wre"), utf("ওরে"))
self.assertEqual(avro.parse("kw"), utf("ক্ব"))
self.assertEqual(avro.parse("x"), utf("এক্স"))
self.assertEqual(avro.parse("ex"), utf("এক্স"))
self.assertEqual(avro.parse("bx"), utf("বক্স"))
self.assertEqual(avro.parse(":`"), utf(":"))
self.assertEqual(avro.parse(":"), utf("ঃ"))
self.assertEqual(avro.parse("^`"), utf("^"))
self.assertEqual(avro.parse("^"), utf("ঁ"))
self.assertEqual(avro.parse("k^"), utf("কঁ"))
self.assertEqual(avro.parse("k^i"), utf("কঁই"))
self.assertEqual(avro.parse("ki^"), utf("কিঁ"))
self.assertEqual(avro.parse(",,"), utf("্"))
self.assertEqual(avro.parse(",,,"), utf("্,"))
self.assertEqual(avro.parse(",,`,"), utf("্,"))
self.assertEqual(avro.parse("`,,"), utf("্"))
self.assertEqual(avro.parse(",`,"), utf(",,"))
self.assertEqual(avro.parse("$"), utf("৳"))
self.assertEqual("", avro.parse("`"))
self.assertEqual(avro.parse("bdh"), utf("ব্ধ"))
def conv(stre="", delim="!?.", n=0):
'''convert bengali text written in english to bengali while keeping english ony text '''
import string #importing string
from pyavrophonetic import avro
import enchant
d = enchant.DictWithPWL("en_GB", "data/bdict4.txt")
strret = ""
english = 0
bengali = 0
startendsign = ""
global globstr
try:
while stre[n] != "." and stre[n] != "?" and stre[n] != "!":
# Iterating from the back of the sentence
strret += stre[n]
n += 1
else:
endendsign = stre[n]
except:
return "Done"
strlst = strret.split()
lenghty = len(strlst)
for element in range(lenghty):
flag1 = 0
flag2 = 0
if d.check(strlst[element]):
flag1 += 1
elementben = avro.parse(strlst[element])
if d.check(elementben):
flag2 += 1
if flag1 == 1 and flag2 == 0:
english += 1
elif flag1 == 0 and flag2 == 1:
bengali += 1
elif flag1 == 1 and flag2 == 1:
english += 1
bengali += 1
else:
pass
if english > bengali:
pass
strret = " ".join(strlst)
elif bengali > english:
for element in range(lenghty):
flag1 = 0
flag2 = 0
if d.check(strlst[element]):
flag1 += 1
elementben = avro.parse(strlst[element])
if d.check(elementben):
flag2 += 1
if flag1 == 1 and flag2 == 0:
pass
elif flag1 == 0 and flag2 == 1:
strlst[element] = elementben
elif flag1 == 1 and flag2 == 1:
pass
else:
pass
strret = " ".join(strlst)
else:
pass
strret = startendsign + strret + avro.parse(endendsign)
print(strret, )
globstr += strret
conv(stre, n=n + 1)
#!/usr/local/bin/python
from pyavrophonetic import avro
from textblob import TextBlob
from vaderSentiment.vaderSentiment import SentimentIntensityAnalyzer
import sys
b = input("Enter Banglish :")
textBlobTrans = TextBlob(avro.parse(b))
c = str(textBlobTrans.translate(to='en'))
print(c)
analyze = TextBlob(c)
print('By TextBlob Sentiment analysis')
print(analyze.sentiment)
print('By vaderSentiment analyzer')
analyzer = SentimentIntensityAnalyzer()
vs = analyzer.polarity_scores(c)
print(str(vs))
def test_patterns_numbers(self):
"""Test patterns - numbers"""
# Test some numbers
self.assertEquals(utf("০"), avro.parse("0"))
self.assertEquals(utf("১"), avro.parse("1"))
self.assertEquals(utf("২"), avro.parse("2"))
self.assertEquals(utf("৩"), avro.parse("3"))
self.assertEquals(utf("৪"), avro.parse("4"))
self.assertEquals(utf("৫"), avro.parse("5"))
self.assertEquals(utf("৬"), avro.parse("6"))
self.assertEquals(utf("৭"), avro.parse("7"))
self.assertEquals(utf("৮"), avro.parse("8"))
self.assertEquals(utf("৯"), avro.parse("9"))
self.assertEquals(utf("১১২"), avro.parse("112"))
def test_patterns_without_rules_from_config(self):
"""Tests all patterns from config that don't have rules"""
for pattern in AVRO_DICT['data']['patterns']:
if 'rules' not in pattern:
self.assertEqual(pattern['replace'],
avro.parse(pattern['find']))
def test_patterns_punctuations(self):
"""Tests patterns - punctuations"""
# Test some punctuations
self.assertEquals(utf("।"), avro.parse("."))
self.assertEquals(utf("।।"), avro.parse(".."))
self.assertEquals(utf("..."), avro.parse("..."))
def test_sentences(self):
"""Test parsing of sentences"""
self.assertEqual(utf('আমি বাংলায় গান গাই'),
avro.parse('ami banglay gan gai'))
def translateBanglishToBangla(banglishWord):
return avro.parse(banglishWord)
def buttonClick():
speak = wincl.Dispatch("SAPI.SpVoice")
recengine = "google"
mixer.init()
x = 0
y = 0
flag = False
mode = 0
r = sr.Recognizer()
global ok
if ok:
playSound("msg1.mp3")
ok = 0
recognised = u""
with sr.Microphone() as source:
r.adjust_for_ambient_noise(source)
playSound("start.mp3")
f1 = False
print u"আপনার বাক্য বলুন"
with sr.Microphone() as source:
audio = r.listen(source)
try:
recognised = r.recognize_google(audio, language='bn-BD')
f1 = True
except sr.UnknownValueError:
print(u"গুগল স্পিচ রিকগনিশন অডিও বুঝতে পারেনি(১)")
except sr.RequestError as e:
print(u"গুগল স্পিচ রিকগনিশন অডিও বুঝতে পারেনি(২)")
#print("Could not request results from Google Speech Recognition service; {0}".format(e))
if f1:
print u"সফল হয়েছে"
entry1.focus()
entry1.delete(0, END)
entry1.insert(0, recognised)
if findWholeWord('hello')(recognised) or findWholeWord('hi')(
recognised) or findWholeWord('hey')(
recognised) or findWholeWord("what's up")(recognised):
speak.Speak("greetings to you too. I'm ready to do your bidding.")
elif findWholeWord('good morning')(recognised) or findWholeWord(
'good afternoon')(recognised) or findWholeWord('good evening')(
recognised):
speak.Speak("greetings to you too. I'm ready to do your bidding.")
elif findBanglaWord(recognised, u"সুপ্রভাত") or findBanglaWord(
recognised, u"শুভ সকাল"):
textToSpeech(
u"সুপ্রভাত! একটি চমৎকার দিন আশা করছি আপনার জন্য স্যার")
playSound("result.mp3")
elif findBanglaWord(recognised, u"শুভ অপরাহ্ন"):
textToSpeech(u"ধন্যবাদ। শুভ অপরাহ্ন")
playSound("result.mp3")
elif findBanglaWord(recognised, u"শুভ সন্ধ্যা"):
textToSpeech(u"ধন্যবাদ। শুভ সন্ধ্যা")
playSound("result.mp3")
elif findWholeWord('thank you')(recognised) or findWholeWord('thank')(
recognised):
speak.Speak("You are welcome.")
elif findBanglaWord(recognised, u"ধন্যবাদ"):
textToSpeech(u"আপনাকে স্বাগতম")
playSound("result.mp3")
elif findWholeWord('how are you')(recognised):
speak.Speak("I am fine. Thank you")
elif findBanglaWord(recognised, u"কেমন আছেন") or findBanglaWord(
recognised, u"আছেন কেমন") or findBanglaWord(
recognised, u"কেমন আছো"):
textToSpeech(u"আমি ভালো। আপনি কেমন আছেন?")
playSound("result.mp3")
elif findWholeWord('your name')(recognised):
speak.Speak("I am Smarty. I can make your work a lot easier.")
elif findBanglaWord(recognised, u"নাম কি") or findBanglaWord(
recognised, u"কি নাম"):
textToSpeech(u"আমি Smarty")
playSound("result.mp3")
elif findWholeWord('open browser')(recognised) or findWholeWord(
'open chrome')(recognised) or findWholeWord('open firefox')(
recognised) or findWholeWord('open internet explorer')(
recognised):
if findWholeWord('open browser')(recognised) or findWholeWord(
'browser')(recognised):
speak.Speak("All right, I will do it for you.")
webbrowser.open('http://www.google.com')
elif findWholeWord('open chrome')(recognised):
speak.Speak("All right, I will open Chrome.")
url = 'chrome://newtab'
chrome_path = 'C:/Program Files (x86)/Google/Chrome/Application/chrome.exe %s'
webbrowser.get(chrome_path).open(url)
elif findWholeWord('open internet explorer')(
recognised) or findWholeWord('internet explorer')(
recognised):
ie = webbrowser.get(webbrowser.iexplore)
ie.open('google.com')
elif findWholeWord('open mozilla firefox')(
recognised) or findWholeWord('open firefox')(
recognised) or findWholeWord('firefox')(recognised):
speak.Speak("All right, I will open Firefox for you.")
webbrowser.get(
'C:/Program Files/Mozilla Firefox/firefox.exe %s').open(
'http://www.google.com')
elif findWholeWord('what time is it')(recognised) or findWholeWord(
'what is the time now')(recognised) or findWholeWord(
'tell me the time')(
recognised) or findWholeWord('what time')(
recognised) and findWholeWord('now')(recognised):
timenow = time.localtime()
timeh = timenow.tm_hour
timem = timenow.tm_min
if timeh < 13: ampm = "AM"
else: ampm = "PM"
if timem > 9: extra = " "
else: extra = "oh"
speak.Speak("the time is " + str(timeh) + extra + str(timem) +
ampm)
elif findBanglaWord(recognised, u"কয়টা বাজে") or findBanglaWord(
recognised, u"বাজে কয়টা") or findBanglaWord(
recognised, u"সময় কত"):
timenow = time.localtime()
timeH = timenow.tm_hour
timeM = timenow.tm_min
ampm = "AM"
if timeH == 0:
timeH = 12
if timeH == 12:
ampm = "PM"
if timeH > 12 and timeH <= 23:
timeH = timeH - 12
ampm = "PM"
if ampm == "AM":
somoy = u"এখন সময় সকাল " + avro.parse(
str(timeH)) + u" টা বেজে " + avro.parse(
str(timeM)) + u" মিনিট"
else:
somoy = u"এখন সময় বিকেল " + avro.parse(
str(timeH)) + u" টা বেজে " + avro.parse(
str(timeM)) + u" মিনিট"
print somoy
textToSpeech(somoy)
playSound("result.mp3")
elif findWholeWord('what is the temperature')(
recognised) or findWholeWord('what is the weather')(
recognised) or findWholeWord('weather in')(recognised):
loc = "Sylhet, Bangladesh"
if "dhaka" in recognised:
loc = "Dhaka, Bangladesh"
elif "chittagong" in recognised:
loc = "Chittagong, Bangladesh"
elif "new york" in recognised:
loc = "New York"
lookup = pywapi.get_location_ids(loc)
#print lookup
for i in lookup:
location_id = i
print "loocation_id = " + location_id
weather = pywapi.get_weather_from_weather_com(location_id)
speak.Speak("the weather in " + loc + " is " +
weather['current_conditions']['text'] +
". Current temp " +
weather['current_conditions']['temperature'] +
" degrees Celsius")
elif findBanglaWord(recognised, u"ওয়েদার কেমন") or findBanglaWord(
recognised, u"আবহাওয়া কেমন") or findBanglaWord(
recognised, u"আবহাওয়ার খবর") or findBanglaWord(
recognised, u"তাপমাত্রা কেমন") or findBanglaWord(
recognised, u"আবহাওয়ার কি অবস্থা"):
loc = "Sylhet, Bangladesh"
lookup = pywapi.get_location_ids(loc)
for i in lookup:
location_id = i
weather = pywapi.get_weather_from_weather_com(location_id)
text = weather['current_conditions']['text']
temp = weather['current_conditions']['temperature']
abohowya = u"বর্তমান আবহাওয়া " + text + u" এবং তাপমাত্রা " + avro.parse(
str(temp)) + u" ডিগ্রি সেলসিয়াস"
textToSpeech(abohowya)
playSound("result.mp3")
elif findWholeWord('go to')(recognised) and ".com" in recognised:
domain = recognised.split("go to ", 1)[1]
speak.Speak("All right, I will take you to" + domain)
webbrowser.open('http://www.' + domain)
elif findWholeWord('close')(recognised):
f = recognised.split("close ", 1)[1]
speak.Speak(f + " is closed Sir.")
p = 'TASKKILL /F /IM ' + f + '.exe'
try:
os.system(p)
except:
speak.Speak(f + " is already closed Sir")
elif findWholeWord('lock')(recognised) and findWholeWord('pc')(
recognised
) or findWholeWord('lock')(recognised) and findWholeWord('computer')(
recognised) or findWholeWord('lock')(
recognised) and findWholeWord('system')(recognised):
ctypes.windll.user32.LockWorkStation()
elif findBanglaWord(recognised, "কম্পিউটার তালা"):
ctypes.windll.user32.LockWorkStation()
elif findWholeWord('shutdown')(recognised) and findWholeWord('pc')(
recognised
) or findWholeWord('shutdown')(recognised) and findWholeWord(
'computer')(recognised) or findWholeWord('shutdown')(
recognised) and findWholeWord('system')(recognised):
os.system('shutdown -s')
elif findBanglaWord(
recognised, "কম্পিউটার বন্ধ করুন") or findBanglaWord(
recognised, "কম্পিউটার বন্ধ করো") or findBanglaWord(
recognised, "কম্পিউটার বন্ধ কর"):
os.system('shutdown -s')
else:
#internal search
qDict = {}
cntDict = {}
for word in recognised.split():
qDict[word] = 1
path = u"""C:/Users/royta/Desktop/Final/"""
files = os.listdir(path)
for txtfiles in files:
if ".txt" in txtfiles:
p = path + txtfiles
cnt = 0
with codecs.open(p, 'r', encoding='utf-8',
errors='ignore') as f:
for line in f:
for word in line.split():
if word in qDict:
cnt += 1
cntDict[txtfiles] = cnt
Max = 0
fileName = ""
for key, value in cntDict.iteritems():
#print key, "-->", value
if value > Max:
Max = value
fileName = key
if Max == 0:
print "Nothing matched. Looking online for solution."
else:
print "Maximum matched ", fileName, " --> ", Max
#end of the internal search
print recognised
if detect(recognised) == "en":
#wikipedia English stuff
querySummary = ""
wikipedia.set_lang("en")
try:
idx = 0
searchResult = wikipedia.search(recognised)
length = len(searchResult)
i = 0
while i < length:
if searchResult[i] == recognised:
idx = i
break
i += 1
querySummary = wikipedia.summary(searchResult[idx],
sentences=3)
result = remove_text_between_parens(querySummary)
print result
speak.Speak(result)
except IndexError:
playSound("error.mp3")
except wikipedia.exceptions.DisambiguationError:
playSound("error.mp3")
except wikipedia.exceptions.PageError:
playSound("error.mp3")
else:
#wikipedia Bangla stuff
querySummary = u""
wikipedia.set_lang("bn")
try:
idx = 0
searchResult = wikipedia.search(recognised)
length = len(searchResult)
i = 0
while i < length:
if searchResult[i] == recognised:
idx = i
break
i += 1
playSound("waitmsg.mp3")
querySummary = wikipedia.summary(searchResult[idx])
#3 line summary
result = sentence(3, querySummary)
result = remove_text_between_parens(result)
#print result
textToSpeech(result)
playSound("result.mp3")
except IndexError:
playSound("error.mp3")
except wikipedia.exceptions.DisambiguationError:
playSound("error.mp3")
except wikipedia.exceptions.PageError:
playSound("error.mp3")
else:
playSound("error.mp3")
#!/usr/local/bin/python
#coding:utf8
from pyavrophonetic import avro
from textblob import TextBlob
from vaderSentiment.vaderSentiment import SentimentIntensityAnalyzer
banglish = "ami to shudhu Ilish je Chandpur er eita confirm korechilam... Shorshe r sathe mile je sorshe ilish namok shushadu khabar toiri hobe ami ki jani."
#banglish = "আগামী ২০ থেকে ২৭ জুলাই পর্যন্ত (শুক্রবার ব্যতীত) নগরীর হ্যানে রেলওয়ে মাধ্যমিক স্কুলে সদরের ২১নং ওয়ার্ডের ভোটারদের স্মার্ট কার্ড বিতরণ করা হবে। এছাড়া ৩০ জুলাই"
print banglish
tr_bng = avro.parse(banglish)
print tr_bng
blob = TextBlob(tr_bng)
tr_eng = blob.translate(to="en")
print tr_eng.sentences
analyser = SentimentIntensityAnalyzer()
def print_sentiment_scores(sentence):
snt = analyser.polarity_scores(sentence)
print("{:-<40} {}".format(sentence, str(snt)))
for sentence in tr_eng.sentences:
#print type(sentence)
stn = unicode(sentence)
print_sentiment_scores(stn)
def test_patterns_with_rules_svaravarna(self):
"""Test patterns - with rules - svaravarna"""
# Test some numbers
self.assertEqual(utf("অ"), avro.parse("o"))
self.assertEqual(utf("আ"), avro.parse("a"))
self.assertEqual(utf("ই"), avro.parse("i"))
self.assertEqual(utf("ঈ"), avro.parse("I"))
self.assertEqual(utf("উ"), avro.parse("u"))
self.assertEqual(utf("উ"), avro.parse("oo"))
self.assertEqual(utf("ঊ"), avro.parse("U"))
self.assertEqual(utf("এ"), avro.parse("e"))
self.assertEqual(utf("ঐ"), avro.parse("OI"))
self.assertEqual(utf("ও"), avro.parse("O"))
self.assertEqual(utf("ঔ"), avro.parse("OU"))
