Merge pull request #32 from project-kxkg/Macrodove/FixAttempt

Attempt spellcheck fix with levenshtein threshold

Former-commit-id: 9aa2eb93eea153a5416cb94eaf854d5a86ca9e1d

SRT.py

@@ -371,8 +371,18 @@ class SRT_script():
                 [real_word, pos] = self.get_real_word(word)
                 if not dict.check(word[:pos]):
                     suggest = term_spellDict.suggest(real_word)
-                    if suggest:  # relax spell check
+                    if suggest and enchant.utils.levenshtein(word, suggest[0]) < (len(word)+len(suggest[0]))/4:  # relax spell check
+                        
+                        #with open("dislog.log","a") as log:
+                        #    if not os.path.exists("dislog.log"):
+                        #        log.write("word \t suggest \t levenshtein \n")
+                        #    log.write(word + "\t" + suggest[0] + "\t" + str(enchant.utils.levenshtein(word, suggest[0]))+'\n')
+                        print(word + ":" + suggest[0] + ":---:levenshtein:" + str(enchant.utils.levenshtein(word, suggest[0])))
+                        
+                        
                         new_word = word.replace(word[:pos],suggest[0])
+                    else:
+                        new_word = word
                 else: 
                     new_word = word
                 ready_words[i] = new_word

pipeline.py

@@ -255,7 +255,7 @@ def main():
 
     # SRT class preprocess
     srt.form_whole_sentence()
-    # srt.spell_check_term()
+    srt.spell_check_term()
     srt.correct_with_force_term()
     srt.write_srt_file_src(srt_file_en)
     script_input = srt.get_source_only()