@@ -8,8 +8,8 @@ def compile_patterns(patterns):
88class StringUtilsConstants :
99 arabic_vowel_translate_table = str .maketrans (
1010 dict .fromkeys (
11- "\u064E \u064F \u0650 \u0652 \u0651 \u0653 \u064B \u064C \u0621 \u064D \u0670 " # Normal vowels (Fatha, Damma, Kasra, etc)
12- "\u06D6 \u06D7 \u06D8 \u06D9 \u06DA \u06DB " , # Quranic marks
11+ "\u064e \u064f \u0650 \u0652 \u0651 \u0653 \u064b \u064c \u0621 \u064d \u0670 " # Normal vowels (Fatha, Damma, Kasra, etc)
12+ "\u06d6 \u06d7 \u06d8 \u06d9 \u06da \u06db " , # Quranic marks
1313 "" ,
1414 ),
1515 )
@@ -19,7 +19,7 @@ class StringUtilsConstants:
1919
2020 # replace 'ٳ|ٲ|ٱ|إ|ﺍ|أ|ٵ | ﺎ' with 'ا'
2121 alphabet_alef_translate_table = str .maketrans (
22- dict .fromkeys ("\uFE8E \u0672 \u0671 \u0625 \ufe8d \u0623 \u0675 \u0673 " , "\u0627 " ),
22+ dict .fromkeys ("\ufe8e \u0672 \u0671 \u0625 \ufe8d \u0623 \u0675 \u0673 " , "\u0627 " ),
2323 )
2424
2525 # replace 'ٮ|ݕ|ٻ|ﺐ|ﺏ|ﺑ' with "ب"
@@ -77,7 +77,7 @@ class StringUtilsConstants:
7777 alphabet_shin_translate_table = str .maketrans (dict .fromkeys ("\ufeb5 \ufeb6 \ufeb8 \ufeb7 \u069c \u06fa " , "\u0634 " ))
7878
7979 # replace "ڝ|ﺺ|ﺼ|ﺻ |ﺹ" with "ص"
80- alphabet_sad_translate_table = str .maketrans (dict .fromkeys ("\uFEB9 \ufeba \ufebc \ufebb \u069d " , "\u0635 " ))
80+ alphabet_sad_translate_table = str .maketrans (dict .fromkeys ("\ufeb9 \ufeba \ufebc \ufebb \u069d " , "\u0635 " ))
8181
8282 # replace "ڞ|ۻ|ﺽ|ﺾ|ﺿ|ﻀ" with "ض"
8383 alphabet_zad_translate_table = str .maketrans (dict .fromkeys ("\ufebd \ufebe \ufebf \ufec0 \u06fb \u069e " , "\u0636 " ))
@@ -120,7 +120,7 @@ class StringUtilsConstants:
120120
121121 # replace "ڵ|ڶ|ڸ|ڷ|ݪ|ﻝ|ﻞ|ﻠ|ڵ | ﻟ" with "ل"
122122 alphabet_lam_translate_table = str .maketrans (
123- dict .fromkeys ("\uFEDF \ufedd \ufede \ufee0 \u076a \u06b7 \u06b8 \u06b6 \u06b5 " , "\u0644 " ),
123+ dict .fromkeys ("\ufedf \ufedd \ufede \ufee0 \u076a \u06b7 \u06b8 \u06b6 \u06b5 " , "\u0644 " ),
124124 )
125125
126126 # replace "ݥ|ݦ|ﻡ|ﻤ|ﻢ|ﻣ" with "م"
@@ -141,15 +141,15 @@ class StringUtilsConstants:
141141 # replace "ܝ|ܤ|ܣ|ﺔ|ﻬ|ھ|ﻩ|ﻫ|ﻪ|ۀ|ە|ة|ہ|ﮭ|ﺓ" with "ه"
142142 alphabet_ha_translate_table = str .maketrans (
143143 dict .fromkeys (
144- "\uFE93 \uFBAD \ufe94 \ufeec \u06be \ufee9 \ufeeb \ufeea \u06c0 \u06d5 \u0629 \u06c1 \u0723 \u0724 \u071d " ,
144+ "\ufe93 \ufbad \ufe94 \ufeec \u06be \ufee9 \ufeeb \ufeea \u06c0 \u06d5 \u0629 \u06c1 \u0723 \u0724 \u071d " ,
145145 "\u0647 " ,
146146 ),
147147 )
148148
149149 # replace "ﺋ|ؿ|ؾ|ؽ|ۑ|ٸ|ﭛ|ﻯ|ۍ|ﻰ|ﻱ|ﻲ|ﻳ|ﻴ|ﯼ|ې|ﯽ|ﯾ|ﯿ|ێ|ے|ى|ي|ﺉ|ﺌ |ﯨ" with "ی"
150150 alphabet_ye_translate_table = str .maketrans (
151151 dict .fromkeys (
152- "\uFBE8 \ufb5b \ufeef \u06cd \ufef0 \ufef1 \ufef2 \ufef3 \ufef4 \ufbfc "
152+ "\ufbe8 \ufb5b \ufeef \u06cd \ufef0 \ufef1 \ufef2 \ufef3 \ufef4 \ufbfc "
153153 "\u06d0 \ufbfd \ufbfe \ufbff \u06ce \u06d2 \u0649 \u064a \u0678 "
154154 "\u06d1 \u063d \u063e \u063f \ufe89 \ufe8b \ufe8c " ,
155155 "\u06cc " ,
@@ -160,43 +160,43 @@ class StringUtilsConstants:
160160
161161 # replace '•|·|●|·|・|∙|。|ⴰ' with '.'
162162 punctuation_translate_table2 = str .maketrans (
163- dict .fromkeys ("\u2022 \u00b7 \u25cf \u0387 \u30fb \u2219 \uff61 \u2d30 " , "\u002E " ),
163+ dict .fromkeys ("\u2022 \u00b7 \u25cf \u0387 \u30fb \u2219 \uff61 \u2d30 " , "\u002e " ),
164164 )
165165
166166 # replace ',|٬|٫|‚|,' with '،'
167167 punctuation_translate_table3 = str .maketrans (dict .fromkeys ("\u002c \u066c \u066b \u201a \uff0c " , "\u060c " ))
168168
169169 # replace 'ʕ | ? | ⁉ | � ' with '؟'
170- punctuation_translate_table4 = str .maketrans (dict .fromkeys ("\u0295 \u003f \u2049 \uFFFD " , "\u061f " ))
170+ punctuation_translate_table4 = str .maketrans (dict .fromkeys ("\u0295 \u003f \u2049 \ufffd " , "\u061f " ))
171171
172172 # replace '‼ | ❕ ' with '!'
173- punctuation_translate_table5 = str .maketrans (dict .fromkeys ("\u203C \u2755 " , "\u0021 " ))
173+ punctuation_translate_table5 = str .maketrans (dict .fromkeys ("\u203c \u2755 " , "\u0021 " ))
174174
175175 # replace '_ ' with 'ـ'
176- punctuation_translate_table6 = str .maketrans (dict .fromkeys ("\u005F " , "\u0640 " ))
176+ punctuation_translate_table6 = str .maketrans (dict .fromkeys ("\u005f " , "\u0640 " ))
177177
178178 # replace ' - | ━ | − | ‐ | ‑ | – | — | ─ | − | ー | ⁃ (hyphen bullet : not supported by pycharm) | ' with '-'
179179 punctuation_translate_table7 = str .maketrans (
180- dict .fromkeys ("\uFF0D \u2501 \u2212 \u2010 \u2011 \u2013 \u2014 \u2500 \u2212 \u30FC \u2043 " , "\u002D " ),
180+ dict .fromkeys ("\uff0d \u2501 \u2212 \u2010 \u2011 \u2013 \u2014 \u2500 \u2212 \u30fc \u2043 " , "\u002d " ),
181181 )
182182
183183 # replace '‹ |《 | ﴾ ' with '«'
184- punctuation_translate_table8 = str .maketrans (dict .fromkeys ("\u2039 \u300A \uFD3E " , "\u00AB " ))
184+ punctuation_translate_table8 = str .maketrans (dict .fromkeys ("\u2039 \u300a \ufd3e " , "\u00ab " ))
185185
186186 # replace '› | 》| ﴿ ' with '»'
187- punctuation_translate_table9 = str .maketrans (dict .fromkeys ("\u203A \u300B \uFD3F " , "\u00BB " ))
187+ punctuation_translate_table9 = str .maketrans (dict .fromkeys ("\u203a \u300b \ufd3f " , "\u00bb " ))
188188
189189 # replace ';' with '؛'
190- punctuation_translate_table10 = str .maketrans (dict .fromkeys ("\u003B " , "\u061B " ))
190+ punctuation_translate_table10 = str .maketrans (dict .fromkeys ("\u003b " , "\u061b " ))
191191
192192 # replace '%' with '٪'
193- punctuation_translate_table11 = str .maketrans (dict .fromkeys ("\u0025 " , "\u066A " ))
193+ punctuation_translate_table11 = str .maketrans (dict .fromkeys ("\u0025 " , "\u066a " ))
194194
195195 # replace " ˈ | ‘ | ’ | “ | ” " with " ' "
196- punctuation_translate_table12 = str .maketrans (dict .fromkeys ("\u02C8 \u2018 \u2019 \u201C \u201D " , "\u0027 " ))
196+ punctuation_translate_table12 = str .maketrans (dict .fromkeys ("\u02c8 \u2018 \u2019 \u201c \u201d " , "\u0027 " ))
197197
198198 # replace ':' with ': '
199- punctuation_translate_table13 = str .maketrans (dict .fromkeys ("\uFF1A " , "\u003A " ))
199+ punctuation_translate_table13 = str .maketrans (dict .fromkeys ("\uff1a " , "\u003a " ))
200200
201201 character_refinement_patterns : list = compile_patterns (
202202 [
@@ -258,8 +258,8 @@ class StringUtilsConstants:
258258 # replace ' «|» | . | : | ، | ؛ | ؟ | [|] | (|) | {|} | - | ـ | ٪ | ! | ' | " | # | + | / |' with ' '
259259 punctuation_persian_marks_to_space_translate_table = str .maketrans (
260260 dict .fromkeys (
261- "\u002E \u003A \u0021 \u060C \u061B \u061F \u00BB \u005D "
262- "\u0029 \u007D \u00AB \u005B \u0028 \u007B \u002D \u0640 \u066A \u0021 \u0027 \u0022 \u0023 "
261+ "\u002e \u003a \u0021 \u060c \u061b \u061f \u00bb \u005d "
262+ "\u0029 \u007d \u00ab \u005b \u0028 \u007b \u002d \u0640 \u066a \u0021 \u0027 \u0022 \u0023 "
263263 "\u002b \u002f " ,
264264 "\u0020 " ,
265265 ),
0 commit comments