|
3 | 3 |
|
4 | 4 | trannum = str.maketrans("0123456789", "1111111111") |
5 | 5 |
|
6 | | - |
7 | | -# Will we cover IBAN?? |
8 | | -# IBAN - see https://github.com/microsoft/presidio/blob/main/presidio-analyzer/presidio_analyzer/predefined_recognizers/iban_patterns.py which is under MIT |
9 | | - |
10 | | -# IBAN parts format |
11 | | -CC = "[A-Z]{2}" # country code |
12 | | -CK = "[0-9]{2}[ ]?" # checksum |
13 | | -BOS = "^" |
14 | | -EOS = "$" # end of string |
15 | | - |
16 | | -A = "[A-Z][ ]?" |
17 | | -A2 = "([A-Z][ ]?){2}" |
18 | | -A3 = "([A-Z][ ]?){3}" |
19 | | -A4 = "([A-Z][ ]?){4}" |
20 | | - |
21 | | -C = "[a-zA-Z0-9][ ]?" |
22 | | -C2 = "([a-zA-Z0-9][ ]?){2}" |
23 | | -C3 = "([a-zA-Z0-9][ ]?){3}" |
24 | | -C4 = "([a-zA-Z0-9][ ]?){4}" |
25 | | - |
26 | | -N = "[0-9][ ]?" |
27 | | -N2 = "([0-9][ ]?){2}" |
28 | | -N3 = "([0-9][ ]?){3}" |
29 | | -N4 = "([0-9][ ]?){4}" |
30 | | - |
31 | | -# WIP - fix the country codes and group by languages |
32 | | -# move this to financial record |
33 | | -iban_regex = { |
34 | | - # Albania (8n, 16c) ALkk bbbs sssx cccc cccc cccc cccc |
35 | | - "al_AL": "(AL)" + CK + N4 + N4 + C4 + C4 + C4 + C4, |
36 | | - # Andorra (8n, 12c) ADkk bbbb ssss cccc cccc cccc |
37 | | - "ad_AD": "(AD)" + CK + N4 + N4 + C4 + C4 + C4, |
38 | | - # Austria (16n) ATkk bbbb bccc cccc cccc |
39 | | - "en_AT": "(AT)" + CK + N4 + N4 + N4 + N4, |
40 | | - # Azerbaijan (4c,20n) AZkk bbbb cccc cccc cccc cccc cccc |
41 | | - "az_AZ": "(AZ)" + CK + C4 + N4 + N4 + N4 + N4 + N4, |
42 | | - # Bahrain (4a,14c) BHkk bbbb cccc cccc cccc cc |
43 | | - "ar_BH": "(BH)" + CK + A4 + C4 + C4 + C4 + C2, |
44 | | - # Belarus (4c, 4n, 16c) BYkk bbbb aaaa cccc cccc cccc cccc |
45 | | - "bel_BY": "(BY)" + CK + C4 + N4 + C4 + C4 + C4 + C4, |
46 | | - # Belgium (12n) BEkk bbbc cccc ccxx |
47 | | - "fr_BE": "(BE)" + CK + N4 + N4 + N4, |
48 | | - # Bosnia and Herzegovina (16n) BAkk bbbs sscc cccc ccxx |
49 | | - "bos_BA": "(BA)" + CK + N4 + N4 + N4 + N4, |
50 | | - # Brazil (23n,1a,1c) BRkk bbbb bbbb ssss sccc cccc ccct n |
51 | | - "pt_BR": "(BR)" + CK + N4 + N4 + N4 + N4 + N4 + N3 + A + C, |
52 | | - # Bulgaria (4a,6n,8c) BGkk bbbb ssss ttcc cccc cc |
53 | | - "bg_BG": "(BG)" + CK + A4 + N4 + N + N + C2 + C4 + C2, |
54 | | - # Costa Rica (18n) CRkk 0bbb cccc cccc cccc cc (0 = always zero) |
55 | | - "es_CR": "(CR)" + CK + "[0]" + N3 + N4 + N4 + N4 + N2, |
56 | | - # Croatia (17n) HRkk bbbb bbbc cccc cccc c |
57 | | - "hr_HR": "(HR)" + CK + N4 + N4 + N4 + N4 + N, |
58 | | - # Cyprus (8n,16c) CYkk bbbs ssss cccc cccc cccc cccc |
59 | | - "el_CY": "(CY)" + CK + N4 + N4 + C4 + C4 + C4 + C4, |
60 | | - # Czech Republic (20n) CZkk bbbb ssss sscc cccc cccc |
61 | | - "cz_CZ": "(CZ)" + CK + N4 + N4 + N4 + N4 + N4, |
62 | | - # Denmark (14n) DKkk bbbb cccc cccc cc |
63 | | - "dan_DK": "(DK)" + CK + N4 + N4 + N4 + N2, |
64 | | - # Dominican Republic (4a,20n) DOkk bbbb cccc cccc cccc cccc cccc |
65 | | - "es_DO": "(DO)" + CK + A4 + N4 + N4 + N4 + N4 + N4, |
66 | | - # EAt Timor (19n) TLkk bbbc cccc cccc cccc cxx |
67 | | - "tl_TL": "(TL)" + CK + N4 + N4 + N4 + N4 + N3, |
68 | | - # Estonia (16n) EEkk bbss cccc cccc cccx |
69 | | - "ee_EE": "(EE)" + CK + N4 + N4 + N4 + N4, |
70 | | - # Faroe Islands (14n) FOkk bbbb cccc cccc cx |
71 | | - "FO": "(FO)" + CK + N4 + N4 + N4 + N2, |
72 | | - # Finland (14n) FIkk bbbb bbcc cccc cx |
73 | | - "fi_FI": "(FI)" + CK + N4 + N4 + N4 + N2, |
74 | | - # France (10n,11c,2n) FRkk bbbb bsss sscc cccc cccc cxx |
75 | | - "fr_FR": "(FR)" + CK + N4 + N4 + N2 + C2 + C4 + C4 + C + N2, |
76 | | - # Georgia (2c,16n) GEkk bbcc cccc cccc cccc cc |
77 | | - "ge_GE": "(GE)" + CK + C2 + N2 + N4 + N4 + N4 + N2, |
78 | | - # Germany (18n) DEkk bbbb bbbb cccc cccc cc |
79 | | - "de_DE": "(DE)" + CK + N4 + N4 + N4 + N4 + N2, |
80 | | - # Gibraltar (4a,15c) GIkk bbbb cccc cccc cccc ccc |
81 | | - "GI": "(GI)" + CK + A4 + C4 + C4 + C4 + C3, |
82 | | - # Greece (7n,16c) GRkk bbbs sssc cccc cccc cccc ccc |
83 | | - "el_GR": "(GR)" + CK + N4 + N3 + C + C4 + C4 + C4 + C3, |
84 | | - # Greenland (14n) GLkk bbbb cccc cccc cc |
85 | | - "kl_GL": "(GL)" + CK + N4 + N4 + N4 + N2, |
86 | | - # Guatemala (4c,20c) GTkk bbbb mmtt cccc cccc cccc cccc |
87 | | - "es_GT": "(GT)" + CK + C4 + C4 + C4 + C4 + C4 + C4, |
88 | | - # Hungary (24n) HUkk bbbs sssx cccc cccc cccc cccx |
89 | | - "hu_HU": "(HU)" + CK + N4 + N4 + N4 + N4 + N4 + N4, |
90 | | - # Iceland (22n) ISkk bbbb sscc cccc iiii iiii ii |
91 | | - "is_IS": "(IS)" + CK + N4 + N4 + N4 + N4 + N4 + N2, |
92 | | - # Ireland (4c,14n) IEkk aaaa bbbb bbcc cccc cc |
93 | | - "en_IE": "(IE)" + CK + C4 + N4 + N4 + N4 + N2, |
94 | | - # Israel (19n) ILkk bbbn nncc cccc cccc ccc |
95 | | - "hb_IL": "(IL)" + CK + N4 + N4 + N4 + N4 + N3, |
96 | | - # Italy (1a,10n,12c) ITkk xbbb bbss sssc cccc cccc ccc |
97 | | - "it_IT": "(IT)" + CK + A + N3 + N4 + N3 + C + C3 + C + C4 + C3, |
98 | | - # Jordan (4a,22n) JOkk bbbb ssss cccc cccc cccc cccc cc |
99 | | - "ar_JO": "(JO)" + CK + A4 + N4 + N4 + N4 + N4 + N4 + N2, |
100 | | - # Kazakhstan (3n,13c) KZkk bbbc cccc cccc cccc |
101 | | - "kz_KZ": "(KZ)" + CK + N3 + C + C4 + C4 + C4, |
102 | | - # Kosovo (4n,10n,2n) XKkk bbbb cccc cccc cccc |
103 | | - "xk_XK": "(XK)" + CK + N4 + N4 + N4 + N4, |
104 | | - # Kuwait (4a,22c) KWkk bbbb cccc cccc cccc cccc cccc cc |
105 | | - "ar_KW": "(KW)" + CK + A4 + C4 + C4 + C4 + C4 + C4 + C2, |
106 | | - # Latvia (4a,13c) LVkk bbbb cccc cccc cccc c |
107 | | - "lv_LV": "(LV)" + CK + A4 + C4 + C4 + C4 + C, |
108 | | - # Lebanon (4n,20c) LBkk bbbb cccc cccc cccc cccc cccc |
109 | | - "lb_LB": "(LB)" + CK + N4 + C4 + C4 + C4 + C4 + C4, |
110 | | - # de_LiechteNtein (5n,12c) LIkk bbbb bccc cccc cccc c |
111 | | - "li_LI": "(LI)" + CK + N4 + N + C3 + C4 + C4 + C, |
112 | | - # Lithuania (16n) LTkk bbbb bccc cccc cccc |
113 | | - "lt_LT": "(LT)" + CK + N4 + N4 + N4 + N4, |
114 | | - # Luxembourg (3n,13c) LUkk bbbc cccc cccc cccc |
115 | | - "lu_LU": "(LU)" + CK + N3 + C + C4 + C4 + C4, |
116 | | - # Malta (4a,5n,18c) MTkk bbbb ssss sccc cccc cccc cccc ccc |
117 | | - "mt_MT": "(MT)" + CK + A4 + N4 + N + C3 + C4 + C4 + C4 + C3, |
118 | | - # Mauritania (23n) MRkk bbbb bsss sscc cccc cccc cxx |
119 | | - "mr_MR": "(MR)" + CK + N4 + N4 + N4 + N4 + N4 + N3, |
120 | | - # Mauritius (4a,19n,3a) MUkk bbbb bbss cccc cccc cccc 000m mm |
121 | | - "mu_MU": "(MU)" + CK + A4 + N4 + N4 + N4 + N4 + N3 + A, |
122 | | - # Moldova (2c,18c) MDkk bbcc cccc cccc cccc cccc |
123 | | - "md_MD": "(MD)" + CK + C4 + C4 + C4 + C4 + C4, |
124 | | - # Monaco (10n,11c,2n) MCkk bbbb bsss sscc cccc cccc cxx |
125 | | - "mc_MC": "(MC)" + CK + N4 + N4 + N2 + C2 + C4 + C4 + C + N2, |
126 | | - # Montenegro (18n) MEkk bbbc cccc cccc cccc xx |
127 | | - "me_ME": "(ME)" + CK + N4 + N4 + N4 + N4 + N2, |
128 | | - # Netherlands (4a,10n) NLkk bbbb cccc cccc cc |
129 | | - "nl_NL": "(NL)" + CK + A4 + N4 + N4 + N2, |
130 | | - # North Macedonia (3n,10c,2n) MKkk bbbc cccc cccc cxx |
131 | | - "mk_MK": "(MK)" + CK + N3 + C + C4 + C4 + C + N2, |
132 | | - # Norway (11n) NOkk bbbb cccc ccx |
133 | | - "no_NO": "(NO)" + CK + N4 + N4 + N3, |
134 | | - # Pakistan (4c,16n) PKkk bbbb cccc cccc cccc cccc |
135 | | - "pk_PK": "(PK)" + CK + C4 + N4 + N4 + N4 + N4, |
136 | | - # Palestinian territories (4c,21n) PSkk bbbb xxxx xxxx xccc cccc cccc c |
137 | | - "ps_PS": "(PS)" + CK + C4 + N4 + N4 + N4 + N4 + N, |
138 | | - # Poland (24n) PLkk bbbs sssx cccc cccc cccc cccc |
139 | | - "pl_PL": "(PL)" + CK + N4 + N4 + N4 + N4 + N4 + N4, |
140 | | - # Portugal (21n) PTkk bbbb ssss cccc cccc cccx x |
141 | | - "pt_PT": "(PT)" + CK + N4 + N4 + N4 + N4 + N, |
142 | | - # Qatar (4a,21c) QAkk bbbb cccc cccc cccc cccc cccc c |
143 | | - "ar_QA": "(QA)" + CK + A4 + C4 + C4 + C4 + C4 + C, |
144 | | - # Romania (4a,16c) ROkk bbbb cccc cccc cccc cccc |
145 | | - "ro_RO": "(RO)" + CK + A4 + C4 + C4 + C4 + C4, |
146 | | - # San Marino (1a,10n,12c) SMkk xbbb bbss sssc cccc cccc ccc |
147 | | - "sm": "(SM)" + CK + A + N3 + N4 + N3 + C + C4 + C4 + C3, |
148 | | - # Saudi Arabia (2n,18c) SAkk bbcc cccc cccc cccc cccc |
149 | | - "ar_SA": "(SA)" + CK + N2 + C2 + C4 + C4 + C4 + C4, |
150 | | - # Serbia (18n) RSkk bbbc cccc cccc cccc xx |
151 | | - "rs_RS": "(RS)" + CK + N4 + N4 + N4 + N4 + N2, |
152 | | - # Slovakia (20n) SKkk bbbb ssss sscc cccc cccc |
153 | | - "sk_SK": "(SK)" + CK + N4 + N4 + N4 + N4 + N4, |
154 | | - # Slovenia (15n) SIkk bbss sccc cccc cxx |
155 | | - "si_SI": "(SI)" + CK + N4 + N4 + N4 + N3, |
156 | | - # Spain (20n) ESkk bbbb ssss xxcc cccc cccc |
157 | | - "es_ES": "(ES)" + CK + N4 + N4 + N4 + N4 + N4, |
158 | | - # Sweden (20n) SEkk bbbc cccc cccc cccc cccc |
159 | | - "se_SE": "(SE)" + CK + N4 + N4 + N4 + N4 + N4, |
160 | | - # Switzerland (5n,12c) CHkk bbbb bccc cccc cccc c |
161 | | - "gsw_CH": "(CH)" + CK + N4 + N + C3 + C4 + C4 + C, |
162 | | - # Tunisia (20n) TNkk bbss sccc cccc cccc cccc |
163 | | - "ar_TN": "(TN)" + CK + N4 + N4 + N4 + N4 + N4, |
164 | | - # Turkey (5n,17c) TRkk bbbb bxcc cccc cccc cccc cc |
165 | | - "tr_TR": "(TR)" + CK + N4 + N + C3 + C4 + C4 + C4 + C2, |
166 | | - # United Arab Emirates (3n,16n) AEkk bbbc cccc cccc cccc ccc |
167 | | - "ar_AE": "(AE)" + CK + N4 + N4 + N4 + N4 + N3, |
168 | | - # United Kingdom (4a,14n) GBkk bbbb ssss sscc cccc cc |
169 | | - "en_GB": "(GB)" + CK + A4 + N4 + N4 + N4 + N2, |
170 | | - # Vatican City (3n,15n) VAkk bbbc cccc cccc cccc cc |
171 | | - "it_VA": "(VA)" + CK + N4 + N4 + N4 + N4 + N2, |
172 | | - # Virgin Islands, British (4c,16n) VGkk bbbb cccc cccc cccc cccc |
173 | | - "en_VG": "(VG)" + CK + C4 + N4 + N4 + N4 + N4, |
174 | | -} |
175 | | - |
176 | | - |
177 | | -# should we move license plate to govt_id? |
178 | | -# ("LICENSE_PLATE", regex.compile('^(?:[京津沪渝冀豫云辽黑湘皖鲁新苏浙赣鄂桂甘晋蒙陕吉闽贵粤青藏川宁琼使领 A-Z]{1}[A-HJ-NP-Z]{1}(?:(?:[0-9]{5}[DF])|(?:[DF](?:[A-HJ-NP-Z0-9])[0-9]{4})))|(?:[京津沪渝冀豫云辽黑湘皖鲁新苏浙赣鄂桂甘晋蒙陕吉闽贵粤青藏川宁琼使领 A-Z]{1}[A-Z]{1}[A-HJ-NP-Z0-9]{4}[A-HJ-NP-Z0-9 挂学警港澳]{1})$'), None, None, None), |
179 | | -# ("LICENSE_PLATE", regex.compile('\b[A-Z]{3}-\d{4}\b'), None, None, None), |
180 | | - |
181 | | -# Code below needs to be updated/completed. |
182 | | - |
183 | | - |
184 | 6 | def apply_regex_anonymization( |
185 | 7 | sentence: str, |
186 | 8 | lang_id: str, |
|
0 commit comments