| 66 | |
| 67 | |
| 68 | def rettt(sentence): |
| 69 | # handle abbreviations for all languages |
| 70 | sentence = sentence.replace("&nd", "and") |
| 71 | sentence = sentence.replace("Jan.", "january") |
| 72 | sentence = sentence.replace("Feb.", "febrary") |
| 73 | sentence = sentence.replace("Mar.", "march") |
| 74 | sentence = sentence.replace("Apr.", "april") |
| 75 | sentence = sentence.replace("May.", "may") |
| 76 | sentence = sentence.replace("Jun.", "june") |
| 77 | sentence = sentence.replace("Jul.", "july") |
| 78 | sentence = sentence.replace("Aug.", "august") |
| 79 | sentence = sentence.replace("Sept.", "september") |
| 80 | sentence = sentence.replace("Sep.", "september") |
| 81 | sentence = sentence.replace("Oct.", "october") |
| 82 | sentence = sentence.replace("Nov.", "november") |
| 83 | sentence = sentence.replace("Dec.", "december") |
| 84 | sentence = sentence.replace("Mon.", "monday") |
| 85 | sentence = sentence.replace("Tues.", "tuesday") |
| 86 | sentence = sentence.replace("Wed.", "wednesday") |
| 87 | sentence = sentence.replace("Thur.", "thursday") |
| 88 | sentence = sentence.replace("Fri.", "friday") |
| 89 | sentence = sentence.replace("Sat.", "saturday") |
| 90 | if sentence != "Sun.": |
| 91 | sentence = sentence.replace("Sun.", "sunday") |
| 92 | sentence = re.sub(r" St\. ([A-Z])", r" saint \1", sentence) |
| 93 | sentence = re.sub(r" St\.", " street", sentence) |
| 94 | sentence = re.sub(r" Rd\.", " road", sentence) |
| 95 | sentence = re.sub(r"[Aa]\.[Mm]\.", "A_M", sentence) |
| 96 | sentence = re.sub(r"[Pp]\.[Mm]\.", "P_M", sentence) |
| 97 | sentence = re.sub(r"[Bb]\.[Cc]\.", "B_C", sentence) |
| 98 | sentence = re.sub(r"[Ad]\.[Dd]\.", "A_D", sentence) |
| 99 | sentence = sentence.replace("Mr.", "mister") |
| 100 | sentence = sentence.replace("Ms.", "miss") |
| 101 | sentence = sentence.replace("Mrs.", "misses") |
| 102 | sentence = sentence.replace("Ph.D", "P_H_D") |
| 103 | sentence = sentence.replace("i.e.", "that is") |
| 104 | sentence = sentence.replace("e.g.", "for example") |
| 105 | sentence = sentence.replace("btw.", "by the way") |
| 106 | sentence = sentence.replace("btw", "by the way") |
| 107 | sentence = sentence.replace("b.t.w.", "by the way") |
| 108 | sentence = sentence.replace("@", " at ") |
| 109 | return sentence |
| 110 | |
| 111 | |
| 112 | class TextNormalizer: |