class Payyans():
def __init__(self):
self.input_filename = ""
self.output_filename = ""
self.mapping_filename = ""
self.rulesDict = None
self.pdf = 0
self.normalizer = Normalizer()
def Unicode2ASCII(self, unicode_text, font):
unicode_text = self.normalizer.normalize(unicode_text)
index = 0
ascii_text = ""
self.direction = "u2a"
self.mapping_filename = os.path.join(os.path.dirname(__file__), 'maps',
font + ".map")
self.rulesDict = self.LoadRules()
while index < len(unicode_text):
'''കൂട്ടക്ഷരങ്ങള്ക്കൊരു കുറുക്കുവഴി'''
for charNo in [3, 2, 1]:
letter = unicode_text[index:index + charNo]
if letter in self.rulesDict:
ascii_letter = self.rulesDict[letter]
letter = letter.encode('utf-8')
'''
കിട്ടിയ അക്ഷരങ്ങളുടെ അപ്പുറത്തും ഇപ്പുറത്തും
സ്വരചിഹ്നങ്ങള് ഫിറ്റ് ചെയ്യാനുള്ള ബദ്ധപ്പാട്
'''
if letter == 'ൈ': # പിറകില് രണ്ടു സാധനം പിടിപ്പിക്കുക
ascii_text = ascii_text[:-1] + ascii_letter + \
ascii_text[-1:]
elif (letter == 'ോ') | (letter == 'ൊ') \
| (letter == 'ൌ'): # മുമ്പിലൊന്നും പിറകിലൊന്നും
ascii_text = ascii_text[:-1] + ascii_letter[0] + \
ascii_text[-1:] + ascii_letter[1]
elif (letter == 'െ') | (letter == 'േ') | \
(letter == '്ര'): # പിറകിലൊന്നുമാത്രം
ascii_text = ascii_text[:-1] + ascii_letter + \
ascii_text[-1:]
else:
ascii_text = ascii_text + ascii_letter
index = index + charNo
break
else:
if (charNo == 1):
index = index + 1
ascii_text = ascii_text + letter
break
'''നോക്കിയിട്ടു കിട്ടുന്നില്ല ബായി'''
ascii_letter = letter
# ascii_text = ascii_text + ascii_letter
# index = index+1
return ascii_text
def ASCII2Unicode(self, ascii_text, font):
ascii_text = self.normalizer.normalize(ascii_text)
index = 0
post_index = 0
prebase_letter = ""
postbase_letter = "" # "്യ", "്വ"
unicode_text = ""
next_ucode_letter = ""
self.direction = "a2u"
self.mapping_filename = os.path.join(os.path.dirname(__file__), 'maps',
font + ".map")
self.rulesDict = self.LoadRules()
while index < len(ascii_text):
for charNo in [2, 1]:
letter = ascii_text[index:index + charNo]
if letter in self.rulesDict:
unicode_letter = self.rulesDict[letter]
if (self.isPrebase(unicode_letter)): # സ്വരചിഹ്നമാണോ?
prebase_letter = unicode_letter
else: # സ്വരചിഹ്നമല്ല
'''
എങ്കില് വ്യഞ്ജനത്തിനു ശേഷം
പോസ്റ്റ്-ബേസ് ഉണ്ടോ എന്നു നോക്കൂ
'''
post_index = index + charNo
if post_index < len(ascii_text):
letter = ascii_text[post_index]
if letter in self.rulesDict:
next_ucode_letter = self.rulesDict[letter]
if self.isPostbase(next_ucode_letter):
postbase_letter = next_ucode_letter
index = index + 1
if ((unicode_letter.encode('utf-8') == "എ") |
(unicode_letter.encode('utf-8') == "ഒ")):
unicode_text = unicode_text + postbase_letter + \
self.getVowelSign(prebase_letter,
unicode_letter)
else:
unicode_text = unicode_text + unicode_letter + \
postbase_letter + prebase_letter
prebase_letter = ""
postbase_letter = ""
index = index + charNo
break
else:
if charNo == 1:
unicode_text = unicode_text + letter
index = index + 1
break
unicode_letter = letter
return unicode_text # മതം മാറ്റി തിരിച്ചു കൊടുക്ക്വാ !
def getVowelSign(self, vowel_letter, vowel_sign_letter):
vowel = vowel_letter.encode('utf-8')
vowel_sign = vowel_sign_letter.encode('utf-8')
if vowel == "എ":
if vowel_sign == "െ":
return "ഐ"
if vowel == "ഒ":
if vowel_sign == "ാ":
return "ഓ"
if vowel_sign == "ൗ":
return "ഔ"
return (vowel_letter + vowel_sign_letter)
def isPrebase(self, letter):
'''
ഇതെന്തിനാന്നു ചോദിച്ചാ, ഈ അക്ഷരങ്ങളുടെ ഇടതു വശത്തെഴുതുന്ന
സ്വര ചിഹ്നങ്ങളുണ്ടല്ലോ? അവ ആസ്കി തരികിടയില് എഴുതുന്നതു് ഇടതു വശത്തു
തന്നെയാ. യൂണിക്കോഡില് അക്ഷരത്തിനു ശേഷവും അപ്പൊ ആ വക സംഭവങ്ങളെ
തിരിച്ചറിയാനാണു് ഈ സംഭവം.
"തരികിട തരികിടോ ധീംതരികിട" (തരികിട തരികിടയാല്)
എന്നു പയ്യന്റെ ഗുരു പയ്യഗുരു പയ്യെ മൊഴിഞ്ഞിട്ടുണ്ടു്.
'''
unicode_letter = letter.encode('utf-8')
if ((unicode_letter == "േ") | (unicode_letter == "ൈ") |
(unicode_letter == "ൊ") | (unicode_letter == "ോ") |
(unicode_letter == "ൌ") | (unicode_letter == "്ര") |
(unicode_letter == "െ")):
return True # "ഇതു സത്യം... അ...സത്യം.... അസത്യം...!"
else:
return False
def isPostbase(self, letter):
'''
"ക്യ" എന്നതിലെ "്യ", "ക്വ" എന്നതിലെ "്വ" എന്നിവ പോസ്റ്റ്-ബേസ് ആണ്.
"ത്യേ" എന്നത് ആസ്കിയില് "ഏ+ത+്യ" എന്നാണ് എഴുതുന്നത്.
അപ്പോള് വ്യഞ്ജനം കഴിഞ്ഞ് പോസ്റ്റ്-ബേസ് ഉണ്ടെങ്കില്
വ്യഞ്ജനം+പോസ്റ്റ്-ബേസ് കഴിഞ്ഞേ പ്രീ-ബേസ് ചേര്ക്കാവൂ!
ഹൊ, പയ്യന് പാണിനീശിഷ്യനാണ്!!
'''
unicode_letter = letter.encode('utf-8')
if ((unicode_letter == "്യ") | (unicode_letter == "്വ")):
return True
else:
return False
def LoadRules(self):
'''
ഈ സംഭവമാണു് മാപ്പിങ്ങ് ഫയല് എടുത്തു് വായിച്ചു പഠിക്കുന്നതു്.
'''
# if(self.rulesDict):
# return self.rulesDict
rules_dict = dict()
line = []
line_number = 0
rules_file = codecs.open(self.mapping_filename,
encoding='utf-8',
errors='ignore')
while True:
'''
ലൈന് നമ്പര് , മാപ്പിങ്ങ് ഫയലില് തെറ്റുണ്ടെങ്കില്
പറയാന് ആവശ്യാണു്
'''
line_number = line_number + 1
original_text = rules_file.readline()
try:
text = unicode(original_text) # noqa: F821
except BaseException:
text = original_text
if text == "":
break
'''കമന്റടിച്ചേ മത്യാവൂന്നു വെച്ചാ ആവാം. ഒട്ടും മുഷിയില്ല്യ'''
if text[0] == '#':
continue
'''
കമന്റടി പതിവുപോലെ മൈന്റ് ചെയ്യണ്ട ഒന്നും കണ്ടില്യാ
കേട്ടില്യാന്നു വെച്ചു നടന്നോളൂ(മനസ്സില് ചിരിച്ചോളൂ)
'''
line = text.strip()
if (line == ""):
continue
'''ലൈനൊന്നും ല്യാ, മോശം.. ങും പോട്ടെ. വേറെ ലൈന് പിടിക്കാം'''
if (len(line.split("=")) != 2):
'''എന്തോ പ്രശ്നണ്ടു്. ന്നാ അതങ്ങടു തുറന്നു പറഞ്ഞേക്കാം'''
print(
"Error: Syntax Error in the Ascii to Unicode Map "
"in line number ", line_number)
print("Line: " + text)
'''പരിപാടി നിര്ത്താം '''
return 2 # Error - Syntax error in Mapping file
'''ഇടതന്'''
lhs = line.split("=")[0]
'''വലതന്'''
rhs = line.split("=")[1]
'''ഇതിനിടക്കിനി മൂന്നാമനു സ്കോപ്പിണ്ടോ? '''
'''മറക്കാതെ ഇരിക്കട്ടെ. ആവശ്യം വരും '''
lhs = lhs.strip()
rhs = rhs.strip()
if self.direction == 'a2u':
rules_dict[lhs] = rhs
else:
rules_dict[rhs] = lhs
return rules_dict
def get_module_name(self):
return "Payyans Unicode-ASCII Converter"
def get_info(self):
return "ASCII data - Unicode Convertor based on font maps"
class Transliterator:
"""
Transliteration class, instantiate this to get access to the transliteration methods
"""
def __init__(self):
self.cmu = CMUDict()
self.normalizer = Normalizer()
def transliterate_en_ml(self, word):
"""
:param word: The word to be transliterated.
:type word: str.
:returns: the translated word.
Transliterate English to Malayalam with the help of
CMU pronuciation dictionary
"""
return self.cmu.pronunciation(word, "ml_IN")
def transliterate_en_kn(self, word):
"""
:param word: The word to be transliterated.
:type word: str.
:returns: the translated word.
Transliterate English to Kannada with the help of
CMU pronuciation dictionary
"""
return self.cmu.pronunciation(word, "kn_IN")
def transliterate_en_hi(self, word):
"""
:param word: The word to be transliterated.
:type word: str.
:returns: the translated word.
Transliterate English to Hindi with the help of
CMU pronuciation dictionary
"""
return self.cmu.pronunciation(word, "hi_IN")
def transliterate_en_xx(self, word, target_lang):
"""
:param word: The word to be transliterated.
:type word: str.
:param target_lang: The language into which word has to be transliterated.
:type target_lang: str.
:returns: the translated word.
Transliterate English to any Indian Language.
"""
if target_lang == "en_IN" or target_lang == "en_US":
return word
if target_lang == "kn_IN":
tx_str = self.transliterate_en_kn(word)
return tx_str
elif target_lang == "hi_IN":
tx_str = self.transliterate_en_hi(word)
return tx_str
else:
tx_str = self.transliterate_en_ml(word)
if target_lang == "ml_IN":
return tx_str
#chain it through indic indic transliteratioin
#first remove malayalam specific zwj
tx_str = tx_str.replace(u'', '') # remove instances of zwnj
if tx_str[-1:] == u'്' and \
(target_lang == "hi_IN"
or target_lang == "gu_IN"
or target_lang == "bn_IN"):
tx_str = tx_str[:-(len(u'്'))]
# remove the last virama'
return self.transliterate_indic_indic(tx_str, "ml_IN", target_lang)
def transliterate_xx_en(self, word, src_lang):
"""
:param word: The word to be transliterated.
:type word: str.
:param src_lang: The language of the word.
:type src_lang: str.
:returns: the translated word.
Transliterate Indian Language to English.
"""
if src_lang == "en_IN" or src_lang == "en_US":
return word
# TODO: the function is generic now so no need of testing the lanuguage
# but since the indic_en contains only for kn_IN and ml_IN we need this
# check.
# Add all indic language to indic_en
# remplace this block with single call to indic_en function
if src_lang == "kn_IN":
return self.transliterate_indic_en(word, src_lang)
if not src_lang == "ml_IN":
word = self.transliterate_indic_indic(word, src_lang, "ml_IN")
return self.transliterate_indic_en(word, "ml_IN")
def transliterate_iso15919(self, word, src_language):
tx_str = ""
index = 0
word_length = len(word)
for chr in word:
index += 1
offset = ord(chr) - lang_bases[src_language]
#76 is the virama offset for all indian languages from its base
if offset >= 61 and offset <= 76:
tx_str = tx_str[:-1] # remove the last 'a'
if offset > 0 and offset <= 128:
tx_str = tx_str + charmap["ISO15919"][offset]
#delete the inherent 'a' at the end of the word from hindi
if tx_str[-1:] == 'a' and (src_language == "hi_IN" or src_language
== "gu_IN" or src_language == "bn_IN"):
if word_length == index and word_length > 1: # if last letter
tx_str = tx_str[:-1] # remove the last 'a'
return tx_str.decode("utf-8")
def transliterate_ipa(self, word, src_language):
"""
Transliterate the given word in src_language to
IPA - International Phonetical Alphabet notation.
:param word: The word to be transliterated.
:type word: str.
:returns: the translated word.
:param src_lang: The language of the word.
:type src_lang: str.
"""
tx_str = ""
index = 0
word_length = len(word)
for chr in word:
index += 1
if ord(chr) < 255: # ASCII characters + English
tx_str += chr
continue
offset = ord(chr) - lang_bases[src_language]
#76 is the virama offset for all indian languages from its base
if offset >= 61 and offset <= 76:
tx_str = tx_str[:-(len('ə'))] # remove the last 'ə'
if offset > 0 and offset <= 128:
tx_str = tx_str + charmap_transphon["IPA"][offset]
#delete the inherent 'a' at the end of the word from hindi
if tx_str[-1:] == 'ə' and \
(src_language == "hi_IN"
or src_language == "gu_IN"
or src_language == "bn_IN") and \
(word_length == index
and word_length > 1):
tx_str = tx_str[:-(len('ə'))]
# if last letter
# remove the last 'a'
return tx_str.decode("utf-8")
def _malayalam_fixes(self, text):
try:
text = text.replace(u"മ് ", u"ം ")
text = text.replace(u"മ്,", u"ം,")
text = text.replace(u"മ്.", u"ം.")
text = text.replace(u"മ്)", u"ം)")
text = text.replace(u"ഩ", u"ന")
text = text.replace(u"", u".") # danda by fullstop
except:
pass
return text
def transliterate_indic_indic(self, word, src_lang, target_lang):
"""
Transliterate from an Indian languge word
to another indian language word
:param word: The word to be transliterated.
:type word: str.
:param src_lang: The language of the word.
:type src_lang: str.
:param target_lang: The language into which word has to be transliterated.
:type target_lang: str.
:returns: the translated word.
"""
index = 0
tx_str = ""
word = self.normalizer.normalize(word)
if src_lang == "ml_IN" and target_lang != "ml_IN":
word = word.replace(u"\u200C", u"")
word = word.replace(u"\u200D", u"")
#replace all samvruthokaram by u vowels
word = word.replace(u"ു്", u"")
for character in word:
index += 1
if character in string.punctuation or (ord(character) <= 2304
and ord(character) >= 3071):
tx_str = tx_str + character
continue
offset = ord(character) + self.getOffset(src_lang, target_lang)
if (offset > 0):
tx_str = tx_str + chr(offset)
#schwa deletion
baseoffset = offset - lang_bases[target_lang]
#76 : virama
if (index == len(word) and baseoffset == 76
and (target_lang == "hi_IN" or target_lang == "gu_IN"
or target_lang == "pa_IN" or target_lang == "bn_IN")):
#TODO Add more languages having schwa deletion characteristic
tx_str = tx_str[:-(len(character))] # remove the last 'a'
if target_lang == "ml_IN" and src_lang == "ta_IN":
tx_str = tx_str.replace(u"ഩ", u"ന")
if target_lang == "ta_IN":
tx_str = tx_str.replace(u'\u0B96', u"க")
tx_str = tx_str.replace(u'\u0B97', u"க")
tx_str = tx_str.replace(u'\u0B98', u"க")
tx_str = tx_str.replace(u'\u0B9B', u"ச")
tx_str = tx_str.replace(u'\u0B9D', u"ச")
tx_str = tx_str.replace(u'\u0BA0', u"ட")
tx_str = tx_str.replace(u'\u0BA1', u"ட")
tx_str = tx_str.replace(u'\u0BA2', u"ட")
tx_str = tx_str.replace(u'\u0BA5', u"த")
tx_str = tx_str.replace(u'\u0BA6', u"த")
tx_str = tx_str.replace(u'\u0BA7', u"த")
tx_str = tx_str.replace(u'\u0BAB', u"ப")
tx_str = tx_str.replace(u'\u0BAC', u"ப")
tx_str = tx_str.replace(u'\u0BAD', u"ப")
tx_str = tx_str.replace(u'\u0BC3', u"ிரு")
tx_str = tx_str.replace(u'ஂ', u'ம்')
#If target is malayalam, we need to add the virama
if ((target_lang == "ml_IN")
and (src_lang == "hi_IN" or src_lang == "gu_IN"
or src_lang == "pa_IN" or src_lang == "bn_IN")
and tx_str[-1].isalpha()):
tx_str = tx_str + u"്"
return tx_str
def transliterate_indic_en(self, word, src_lang):
"""
Arguments:
- `self`:
- `word`: Word to be transliterated (sentence)
- `src_lang`: Language from which we need to transilterate
"""
# Get all the language related stuffs
dictionary = indic_en.get_dictionary_for(src_lang)
vowels = indic_en.get_vowels_for(src_lang)
vowel_signs = indic_en.get_vowel_signs_for(src_lang)
virama = indic_en.get_virama_for(src_lang)
anuswara = indic_en.get_anuswara_for(src_lang)
word_length = len(word)
index = 0
tx_string = ""
while index < word_length:
# If current charachter is a punctuation symbol
# skip it.
# Added to avoid getting extra 'a' to the begining
# of word next to punctuation symbol
#
if word[index] in string.punctuation:
tx_string += word[index]
index += 1
continue
# Virama = conjucter
if word[index] == virama:
index += 1
continue
# Get english equivalaent of the charachter.
try:
tx_string += dictionary[word[index]]
except KeyError:
# If charachter isn't present in the dict
# just append the charachter to string
# This case is now handled by punctuation checking
tx_string += word[index]
if index + 1 < word_length and not word[index + 1] in vowel_signs \
and word[index + 1] in dictionary \
and not word[index] in vowels \
and not word[index] in vowel_signs:
tx_string += 'a'
if index + 1 == word_length and not word[index] in vowel_signs \
and word[index] in dictionary:
tx_string += 'a'
#handle am sign
if index + 1 < word_length and word[index + 1] == anuswara \
and not word[index] in vowel_signs:
tx_string += 'a'
index += 1
return tx_string
def transliterate(self, text, target_lang_code):
"""
:param text: The text to be transliterated.
:type text: str.
:param target_lang_code: The language into which word has to be transliterated.
:type target_lang_code: str.
:returns: the transliterated text.
The transliteration functioon which can transliterate text to the
supported target languages.
"""
tx_str = ""
lines = text.split("\n")
for line in lines:
words = line.split(" ")
for word in words:
if (word.strip() > ""):
try:
src_lang_code = detect_lang(word)[word]
except:
tx_str = tx_str + " " + word
continue # FIXME
if target_lang_code == "ISO15919":
tx_str = (tx_str + self.transliterate_iso15919(
word, src_lang_code) + " ")
continue
if target_lang_code == "IPA":
tx_str = (tx_str +
self.transliterate_ipa(word, src_lang_code) +
" ")
continue
if src_lang_code == "en_US":
tx_str = (tx_str + self.transliterate_en_xx(
word, target_lang_code) + " ")
continue
if target_lang_code == "en_US" or \
target_lang_code == "en_IN":
tx_str = (tx_str + self.transliterate_xx_en(
word, src_lang_code) + " ")
continue
tx_str += self.transliterate_indic_indic(
word, src_lang_code, target_lang_code)
if len(line) > 1:
tx_str += " "
else:
tx_str = tx_str + word
if len(lines) > 1:
tx_str += "\n"
# Language specific fixes
if target_lang_code == "ml_IN":
tx_str = self._malayalam_fixes(tx_str)
return tx_str
def getOffset(self, src, target):
src_id = 0
target_id = 0
try:
src_id = lang_bases[src]
target_id = lang_bases[target]
return (target_id - src_id)
except:
return 0
def get_module_name(self):
"""
returns module name
"""
return "Transliterator"
def get_info(self):
"""
Returns module info
"""
return "Transliterate the text between any Indian Language"
class Transliterator:
"""
Transliteration class, instantiate this to get access to the transliteration methods
"""
def __init__(self):
self.cmu = CMUDict()
self.normalizer = Normalizer()
def transliterate_en_ml(self, word):
"""
:param word: The word to be transliterated.
:type word: str.
:returns: the translated word.
Transliterate English to Malayalam with the help of
CMU pronuciation dictionary
"""
return self.cmu.pronunciation(word, "ml_IN")
def transliterate_en_kn(self, word):
"""
:param word: The word to be transliterated.
:type word: str.
:returns: the translated word.
Transliterate English to Kannada with the help of
CMU pronuciation dictionary
"""
return self.cmu.pronunciation(word, "kn_IN")
def transliterate_en_hi(self, word):
"""
:param word: The word to be transliterated.
:type word: str.
:returns: the translated word.
Transliterate English to Hindi with the help of
CMU pronuciation dictionary
"""
return self.cmu.pronunciation(word, "hi_IN")
def transliterate_en_xx(self, word, target_lang):
"""
:param word: The word to be transliterated.
:type word: str.
:param target_lang: The language into which word has to be transliterated.
:type target_lang: str.
:returns: the translated word.
Transliterate English to any Indian Language.
"""
if target_lang == "en_IN" or target_lang == "en_US":
return word
if target_lang == "kn_IN":
tx_str = self.transliterate_en_kn(word)
return tx_str
elif target_lang == "hi_IN":
tx_str = self.transliterate_en_hi(word)
return tx_str
else:
tx_str = self.transliterate_en_ml(word)
if target_lang == "ml_IN":
return tx_str
#chain it through indic indic transliteratioin
#first remove malayalam specific zwj
tx_str = tx_str.replace(u'', '') # remove instances of zwnj
if tx_str[-1:] == u'്' and \
(target_lang == "hi_IN"
or target_lang == "gu_IN"
or target_lang == "bn_IN"): tx_str = tx_str[:-(len(u'്'))]
# remove the last virama'
return self.transliterate_indic_indic(tx_str, "ml_IN", target_lang)
def transliterate_xx_en(self, word, src_lang):
"""
:param word: The word to be transliterated.
:type word: str.
:param src_lang: The language of the word.
:type src_lang: str.
:returns: the translated word.
Transliterate Indian Language to English.
"""
if src_lang == "en_IN" or src_lang == "en_US":
return word
# TODO: the function is generic now so no need of testing the lanuguage
# but since the indic_en contains only for kn_IN and ml_IN we need this
# check.
# Add all indic language to indic_en
# remplace this block with single call to indic_en function
if src_lang == "kn_IN":
return self.transliterate_indic_en(word, src_lang)
if not src_lang == "ml_IN":
word = self.transliterate_indic_indic(word, src_lang, "ml_IN")
return self.transliterate_indic_en(word, "ml_IN")
def transliterate_iso15919(self, word, src_language):
tx_str = ""
index = 0
word_length = len(word)
for chr in word:
index += 1
offset = ord(chr) - lang_bases[src_language]
#76 is the virama offset for all indian languages from its base
if offset >= 61 and offset <= 76:
tx_str = tx_str[:-1] # remove the last 'a'
if offset > 0 and offset <= 128:
tx_str = tx_str + charmap["ISO15919"][offset]
#delete the inherent 'a' at the end of the word from hindi
if tx_str[-1:] == 'a' and (src_language == "hi_IN"
or src_language == "gu_IN"
or src_language == "bn_IN"):
if word_length == index and word_length > 1: # if last letter
tx_str = tx_str[:-1] # remove the last 'a'
return tx_str .decode("utf-8")
def transliterate_ipa(self, word, src_language):
"""
Transliterate the given word in src_language to
IPA - International Phonetical Alphabet notation.
:param word: The word to be transliterated.
:type word: str.
:returns: the translated word.
:param src_lang: The language of the word.
:type src_lang: str.
"""
tx_str = ""
index = 0
word_length = len(word)
for chr in word:
index += 1
if ord(chr) < 255: # ASCII characters + English
tx_str += chr
continue
offset = ord(chr) - lang_bases[src_language]
#76 is the virama offset for all indian languages from its base
if offset >= 61 and offset <= 76:
tx_str = tx_str[:-(len('ə'))] # remove the last 'ə'
if offset > 0 and offset <= 128:
tx_str = tx_str + charmap_transphon["IPA"][offset]
#delete the inherent 'a' at the end of the word from hindi
if tx_str[-1:] == 'ə' and \
(src_language == "hi_IN"
or src_language == "gu_IN"
or src_language == "bn_IN") and \
(word_length == index
and word_length > 1): tx_str = tx_str[:-(len('ə'))]
# if last letter
# remove the last 'a'
return tx_str.decode("utf-8")
def _malayalam_fixes(self, text):
try:
text = text.replace(u"മ് ", u"ം ")
text = text.replace(u"മ്,", u"ം,")
text = text.replace(u"മ്.", u"ം.")
text = text.replace(u"മ്)", u"ം)")
text = text.replace(u"ഩ", u"ന")
text = text.replace(u"", u".") # danda by fullstop
except:
pass
return text
def transliterate_indic_indic(self, word, src_lang, target_lang):
"""
Transliterate from an Indian languge word
to another indian language word
:param word: The word to be transliterated.
:type word: str.
:param src_lang: The language of the word.
:type src_lang: str.
:param target_lang: The language into which word has to be transliterated.
:type target_lang: str.
:returns: the translated word.
"""
index = 0
tx_str = ""
word = self.normalizer.normalize(word)
if src_lang == "ml_IN" and target_lang != "ml_IN":
word = word.replace(u"\u200C", u"")
word = word.replace(u"\u200D", u"")
#replace all samvruthokaram by u vowels
word = word.replace(u"ു്", u"")
for character in word:
index += 1
if character in string.punctuation or (ord(character) <= 2304
and ord(character) >= 3071):
tx_str = tx_str + character
continue
offset = ord(character) + self.getOffset(src_lang, target_lang)
if(offset > 0):
tx_str = tx_str + chr(offset)
#schwa deletion
baseoffset = offset - lang_bases[target_lang]
#76 : virama
if (index == len(word) and baseoffset == 76
and (target_lang == "hi_IN"
or target_lang == "gu_IN"
or target_lang == "pa_IN"
or target_lang == "bn_IN")):
#TODO Add more languages having schwa deletion characteristic
tx_str = tx_str[:-(len(character))] # remove the last 'a'
if target_lang == "ml_IN" and src_lang == "ta_IN":
tx_str = tx_str.replace(u"ഩ", u"ന")
if target_lang == "ta_IN":
tx_str = tx_str.replace(u'\u0B96', u"க")
tx_str = tx_str.replace(u'\u0B97', u"க")
tx_str = tx_str.replace(u'\u0B98', u"க")
tx_str = tx_str.replace(u'\u0B9B', u"ச")
tx_str = tx_str.replace(u'\u0B9D', u"ச")
tx_str = tx_str.replace(u'\u0BA0', u"ட")
tx_str = tx_str.replace(u'\u0BA1', u"ட")
tx_str = tx_str.replace(u'\u0BA2', u"ட")
tx_str = tx_str.replace(u'\u0BA5', u"த")
tx_str = tx_str.replace(u'\u0BA6', u"த")
tx_str = tx_str.replace(u'\u0BA7', u"த")
tx_str = tx_str.replace(u'\u0BAB', u"ப")
tx_str = tx_str.replace(u'\u0BAC', u"ப")
tx_str = tx_str.replace(u'\u0BAD', u"ப")
tx_str = tx_str.replace(u'\u0BC3', u"ிரு")
tx_str = tx_str.replace(u'ஂ', u'ம்')
#If target is malayalam, we need to add the virama
if ((target_lang == "ml_IN") and
(src_lang == "hi_IN"
or src_lang == "gu_IN"
or src_lang == "pa_IN"
or src_lang == "bn_IN") and tx_str[-1].isalpha()):
tx_str = tx_str + u"്"
return tx_str
def transliterate_indic_en(self, word, src_lang):
"""
Arguments:
- `self`:
- `word`: Word to be transliterated (sentence)
- `src_lang`: Language from which we need to transilterate
"""
# Get all the language related stuffs
dictionary = indic_en.get_dictionary_for(src_lang)
vowels = indic_en.get_vowels_for(src_lang)
vowel_signs = indic_en.get_vowel_signs_for(src_lang)
virama = indic_en.get_virama_for(src_lang)
anuswara = indic_en.get_anuswara_for(src_lang)
word_length = len(word)
index = 0
tx_string = ""
while index < word_length:
# If current charachter is a punctuation symbol
# skip it.
# Added to avoid getting extra 'a' to the begining
# of word next to punctuation symbol
#
if word[index] in string.punctuation:
tx_string += word[index]
index += 1
continue
# Virama = conjucter
if word[index] == virama:
index += 1
continue
# Get english equivalaent of the charachter.
try:
tx_string += dictionary[word[index]]
except KeyError:
# If charachter isn't present in the dict
# just append the charachter to string
# This case is now handled by punctuation checking
tx_string += word[index]
if index + 1 < word_length and not word[index + 1] in vowel_signs \
and word[index + 1] in dictionary \
and not word[index] in vowels \
and not word[index] in vowel_signs:
tx_string += 'a'
if index + 1 == word_length and not word[index] in vowel_signs \
and word[index] in dictionary:
tx_string += 'a'
#handle am sign
if index + 1 < word_length and word[index + 1] == anuswara \
and not word[index] in vowel_signs:
tx_string += 'a'
index += 1
return tx_string
def transliterate(self, text, target_lang_code):
"""
:param text: The text to be transliterated.
:type text: str.
:param target_lang_code: The language into which word has to be transliterated.
:type target_lang_code: str.
:returns: the transliterated text.
The transliteration functioon which can transliterate text to the
supported target languages.
"""
tx_str = ""
lines = text.split("\n")
for line in lines:
words = line.split(" ")
for word in words:
if(word.strip() > ""):
try:
src_lang_code = detect_lang(word)[word]
except:
tx_str = tx_str + " " + word
continue # FIXME
if target_lang_code == "ISO15919":
tx_str = (tx_str
+ self.transliterate_iso15919(word,
src_lang_code)
+ " ")
continue
if target_lang_code == "IPA":
tx_str = (tx_str
+ self.transliterate_ipa(word,
src_lang_code)
+ " ")
continue
if src_lang_code == "en_US":
tx_str = (tx_str
+ self.transliterate_en_xx(word,
target_lang_code)
+ " ")
continue
if target_lang_code == "en_US" or \
target_lang_code == "en_IN":
tx_str = (tx_str
+ self.transliterate_xx_en(word,
src_lang_code)
+ " ")
continue
tx_str += self.transliterate_indic_indic(word,
src_lang_code,
target_lang_code)
if len(line) > 1:
tx_str += " "
else:
tx_str = tx_str + word
if len(lines) > 1:
tx_str += "\n"
# Language specific fixes
if target_lang_code == "ml_IN":
tx_str = self._malayalam_fixes(tx_str)
return tx_str
def getOffset(self, src, target):
src_id = 0
target_id = 0
try:
src_id = lang_bases[src]
target_id = lang_bases[target]
return (target_id - src_id)
except:
return 0
def get_module_name(self):
"""
returns module name
"""
return "Transliterator"
def get_info(self):
"""
Returns module info
"""
return "Transliterate the text between any Indian Language"