| 1 | #!/usr/bin/env python |
|---|
| 2 | # -*- coding: iso-8859-1 -*- |
|---|
| 3 | |
|---|
| 4 | """ |
|---|
| 5 | |
|---|
| 6 | =================== |
|---|
| 7 | build_surname_id.py |
|---|
| 8 | =================== |
|---|
| 9 | |
|---|
| 10 | DESCRIPTION |
|---|
| 11 | =========== |
|---|
| 12 | |
|---|
| 13 | generation de la partie "nom" de l'id |
|---|
| 14 | |
|---|
| 15 | |
|---|
| 16 | SEE ALSO |
|---|
| 17 | ======== |
|---|
| 18 | |
|---|
| 19 | :ref:`surname_id.xsl` |
|---|
| 20 | |
|---|
| 21 | EXAMPLES |
|---|
| 22 | ======== |
|---|
| 23 | |
|---|
| 24 | cf. doctest |
|---|
| 25 | |
|---|
| 26 | :: |
|---|
| 27 | |
|---|
| 28 | python -v build_surname_id |
|---|
| 29 | |
|---|
| 30 | TODO |
|---|
| 31 | ==== |
|---|
| 32 | |
|---|
| 33 | EVOLUTIONS |
|---|
| 34 | ========== |
|---|
| 35 | |
|---|
| 36 | $Id$ |
|---|
| 37 | |
|---|
| 38 | $URL$ |
|---|
| 39 | |
|---|
| 40 | - fplod 20120409 |
|---|
| 41 | |
|---|
| 42 | * work with fake data |
|---|
| 43 | |
|---|
| 44 | - fplod 20120407 |
|---|
| 45 | |
|---|
| 46 | * creation |
|---|
| 47 | |
|---|
| 48 | """ |
|---|
| 49 | |
|---|
| 50 | import string |
|---|
| 51 | import sys |
|---|
| 52 | |
|---|
| 53 | try: |
|---|
| 54 | from unidecode import unidecode |
|---|
| 55 | except ImportError: |
|---|
| 56 | print("Failed to import unidecode from any known place") |
|---|
| 57 | sys.exit(1) |
|---|
| 58 | |
|---|
| 59 | def build_surname_id(surname): |
|---|
| 60 | |
|---|
| 61 | |
|---|
| 62 | """ |
|---|
| 63 | >>> surname = [] |
|---|
| 64 | [] |
|---|
| 65 | >>> surname.append(' Tartempillion ') |
|---|
| 66 | [' Tartempillion '] |
|---|
| 67 | >>> surname.append('Tar tempillion ') |
|---|
| 68 | [' Tartempillion ', 'Tar tempillion '] |
|---|
| 69 | >>> surname.append("T\'ar tempillion ") |
|---|
| 70 | [' Tartempillion ', 'Tar tempillion ', "T\'ar tempillion "] |
|---|
| 71 | >>> surname_id = build_surname_id(surname) |
|---|
| 72 | ['TARTEMPILLION','TARTEMPILLION','TARTEMPILLION'] |
|---|
| 73 | """ |
|---|
| 74 | |
|---|
| 75 | if len(surname) == 0: |
|---|
| 76 | print('eee : surname empty') |
|---|
| 77 | sys.exit(-1) |
|---|
| 78 | |
|---|
| 79 | # convert to str |
|---|
| 80 | surname_str = [str(item) for item in surname] |
|---|
| 81 | #for item in surname: |
|---|
| 82 | # print ('iii : item type %s : %s ' % (item, type(item))) |
|---|
| 83 | |
|---|
| 84 | # remove white space before and after |
|---|
| 85 | surname_id = map(str.strip,surname_str) |
|---|
| 86 | #print ('1 sans blan debut fin %s' % surname_id) |
|---|
| 87 | # |
|---|
| 88 | # upper |
|---|
| 89 | surname_id = map(str.upper,surname_id) |
|---|
| 90 | #print ('2 upper %s' % surname_id) |
|---|
| 91 | # |
|---|
| 92 | # remove white space inside |
|---|
| 93 | surname_id_no_spaces = [x.replace(' ', '') for x in surname_id] |
|---|
| 94 | #print ('3 sans blanc milieu %s' % surname_id_no_spaces) |
|---|
| 95 | surname_id = surname_id_no_spaces |
|---|
| 96 | # |
|---|
| 97 | # remove punctuation |
|---|
| 98 | for c in string.punctuation: |
|---|
| 99 | surname_id_no_punctuation = [x.replace(c, '') for x in surname_id] |
|---|
| 100 | #print ('4 sans %s avec replace string.punctuation %s' % (c, surname_id_no_punctuation)) |
|---|
| 101 | surname_id = surname_id_no_punctuation |
|---|
| 102 | |
|---|
| 103 | return surname_id |
|---|
| 104 | |
|---|
| 105 | if __name__ == "__main__": |
|---|
| 106 | import doctest |
|---|
| 107 | doctest.testmod() |
|---|
