1 |
torben |
2807 |
package dk.daoas.daoadresseservice;
|
2 |
|
|
|
3 |
torben |
2808 |
import java.util.ArrayList;
|
4 |
torben |
2807 |
import java.util.HashSet;
|
5 |
|
|
import java.util.Set;
|
6 |
|
|
|
7 |
|
|
public class AliasGenerator {
|
8 |
torben |
2808 |
|
9 |
|
|
static class Replacement {
|
10 |
|
|
public String search;
|
11 |
|
|
public String replace;
|
12 |
torben |
2807 |
|
13 |
torben |
2808 |
public Replacement(String search, String replace) {
|
14 |
|
|
this.search = search;
|
15 |
|
|
this.replace = replace;
|
16 |
|
|
}
|
17 |
|
|
}
|
18 |
|
|
|
19 |
|
|
ArrayList<Replacement> replacements = new ArrayList<Replacement>();
|
20 |
|
|
|
21 |
|
|
public AliasGenerator() {
|
22 |
|
|
replacements.add( new Replacement("alle", "allé") );
|
23 |
|
|
replacements.add( new Replacement("allé", "alle") );
|
24 |
torben |
2807 |
|
25 |
|
|
// forkortelser
|
26 |
torben |
2808 |
replacements.add( new Replacement("dronningens", "dr") );
|
27 |
|
|
replacements.add( new Replacement("dr.", "dronningens") );
|
28 |
|
|
replacements.add( new Replacement("dr ", "dronningens") );
|
29 |
torben |
2807 |
|
30 |
torben |
2808 |
replacements.add( new Replacement("dronning", "dr") );
|
31 |
|
|
replacements.add( new Replacement("dr.", "dronning") );
|
32 |
|
|
replacements.add( new Replacement("dr ", "dronning") );
|
33 |
torben |
2807 |
|
34 |
torben |
2808 |
replacements.add( new Replacement("kng", "kongen") );
|
35 |
|
|
replacements.add( new Replacement("kongen", "kng") );
|
36 |
torben |
2807 |
|
37 |
torben |
2808 |
replacements.add( new Replacement("kvt", "kvarter") );
|
38 |
|
|
replacements.add( new Replacement("kvarter", "kvt") );
|
39 |
torben |
2807 |
|
40 |
torben |
2808 |
replacements.add( new Replacement("gl", "gammel") );
|
41 |
|
|
replacements.add( new Replacement("gammel", "gl") );
|
42 |
torben |
2807 |
|
43 |
torben |
2808 |
replacements.add( new Replacement("gl", "gamle") );
|
44 |
|
|
replacements.add( new Replacement("gamle", "gl") );
|
45 |
torben |
2807 |
|
46 |
torben |
2808 |
replacements.add( new Replacement("lille", "ll") );
|
47 |
|
|
replacements.add( new Replacement("ll ", "lille") );
|
48 |
|
|
replacements.add( new Replacement("ll.", "lille") );
|
49 |
torben |
2807 |
|
50 |
torben |
2808 |
replacements.add( new Replacement("store", "st") );
|
51 |
|
|
replacements.add( new Replacement("st ", "store") );
|
52 |
|
|
replacements.add( new Replacement("st.", "store") );
|
53 |
torben |
2807 |
|
54 |
torben |
2808 |
replacements.add( new Replacement("søndre", "sdr") );
|
55 |
|
|
replacements.add( new Replacement("sdr", "søndre") );
|
56 |
|
|
replacements.add( new Replacement("sønder", "sdr") );
|
57 |
|
|
replacements.add( new Replacement("sdr", "sønder") );
|
58 |
torben |
2807 |
|
59 |
|
|
|
60 |
torben |
2808 |
replacements.add( new Replacement("nørre", "nr") );
|
61 |
|
|
replacements.add( new Replacement("nr", "nørre") );
|
62 |
torben |
2807 |
|
63 |
torben |
2808 |
replacements.add( new Replacement("nordre", "ndr") );
|
64 |
|
|
replacements.add( new Replacement("ndr", "nordre") );
|
65 |
torben |
2807 |
|
66 |
|
|
|
67 |
torben |
2808 |
replacements.add( new Replacement("vestre", "vester") );
|
68 |
|
|
replacements.add( new Replacement("vester", "vestre") );
|
69 |
torben |
2807 |
|
70 |
torben |
2808 |
replacements.add( new Replacement("søndre", "sønder") );
|
71 |
|
|
replacements.add( new Replacement("sønder", "søndre") );
|
72 |
torben |
2807 |
|
73 |
torben |
2808 |
replacements.add( new Replacement("østre", "øster") );
|
74 |
|
|
replacements.add( new Replacement("øster", "østre") );
|
75 |
torben |
2807 |
|
76 |
torben |
2808 |
replacements.add( new Replacement("borgmester", "borgm") );
|
77 |
|
|
replacements.add( new Replacement("borgm", "borgmester") );
|
78 |
torben |
2807 |
|
79 |
|
|
|
80 |
torben |
2808 |
replacements.add( new Replacement("sankt", "skt") );
|
81 |
|
|
replacements.add( new Replacement("sankt", "sct") );
|
82 |
|
|
replacements.add( new Replacement("skt", "sankt") );
|
83 |
torben |
2807 |
|
84 |
torben |
2808 |
replacements.add( new Replacement("skt", "sct") );
|
85 |
|
|
replacements.add( new Replacement("sct", "skt") );
|
86 |
torben |
2807 |
|
87 |
|
|
|
88 |
|
|
|
89 |
|
|
// fornavn forkortelser
|
90 |
torben |
2808 |
replacements.add( new Replacement("john", "j") );
|
91 |
|
|
replacements.add( new Replacement("thomas", "th") );
|
92 |
|
|
replacements.add( new Replacement("thorvald", "th") );
|
93 |
|
|
replacements.add( new Replacement("frederik", "fr") );
|
94 |
|
|
replacements.add( new Replacement("frederiks", "fr") );
|
95 |
torben |
2807 |
|
96 |
|
|
|
97 |
torben |
2808 |
replacements.add( new Replacement("christian", "chr") );
|
98 |
|
|
replacements.add( new Replacement("christians", "chr") );
|
99 |
|
|
replacements.add( new Replacement("kristian", "kr") );
|
100 |
|
|
replacements.add( new Replacement("kristians", "kr") );
|
101 |
torben |
2807 |
|
102 |
torben |
2808 |
replacements.add( new Replacement("mylius", "myl") );
|
103 |
|
|
replacements.add( new Replacement("myl", "mylius") );
|
104 |
torben |
2807 |
|
105 |
torben |
2808 |
replacements.add( new Replacement("ludvig", "ludv") );
|
106 |
|
|
replacements.add( new Replacement("ludv", "ludvig") );
|
107 |
torben |
2807 |
|
108 |
|
|
|
109 |
|
|
// alternative stavninger
|
110 |
torben |
2808 |
replacements.add( new Replacement("peder", "peter") );
|
111 |
|
|
replacements.add( new Replacement("peter", "peder") );
|
112 |
torben |
2807 |
|
113 |
|
|
|
114 |
|
|
|
115 |
|
|
// Romertal
|
116 |
torben |
2808 |
replacements.add( new Replacement("1", "i") );
|
117 |
|
|
replacements.add( new Replacement("i", "1") );
|
118 |
torben |
2807 |
|
119 |
torben |
2808 |
replacements.add( new Replacement("2", "ii") );
|
120 |
|
|
replacements.add( new Replacement("ii", "2") );
|
121 |
torben |
2807 |
|
122 |
torben |
2808 |
replacements.add( new Replacement("3", "iii") );
|
123 |
|
|
replacements.add( new Replacement("iii", "3") );
|
124 |
torben |
2807 |
|
125 |
torben |
2808 |
replacements.add( new Replacement("4", "iv") );
|
126 |
|
|
replacements.add( new Replacement("iv", "4") );
|
127 |
torben |
2807 |
|
128 |
torben |
2808 |
replacements.add( new Replacement("5", "v") );
|
129 |
|
|
replacements.add( new Replacement("v", "5") );
|
130 |
torben |
2807 |
|
131 |
torben |
2808 |
replacements.add( new Replacement("6", "vi") );
|
132 |
|
|
replacements.add( new Replacement("vi", "6") );
|
133 |
torben |
2807 |
|
134 |
torben |
2808 |
replacements.add( new Replacement("7", "vii") );
|
135 |
|
|
replacements.add( new Replacement("vii", "7") );
|
136 |
torben |
2807 |
|
137 |
torben |
2808 |
replacements.add( new Replacement("8", "viii") );
|
138 |
|
|
replacements.add( new Replacement("viii", "8") );
|
139 |
torben |
2807 |
|
140 |
torben |
2808 |
replacements.add( new Replacement("9", "ix") );
|
141 |
|
|
replacements.add( new Replacement("ix", "9") );
|
142 |
torben |
2807 |
|
143 |
torben |
2808 |
replacements.add( new Replacement("10", "x") );
|
144 |
|
|
replacements.add( new Replacement("x", "10") );
|
145 |
torben |
2807 |
|
146 |
|
|
|
147 |
|
|
//alternative måder at stave vej/gade/alle
|
148 |
torben |
2808 |
replacements.add( new Replacement("vej", "ve") );
|
149 |
|
|
replacements.add( new Replacement("vej", "vj") );
|
150 |
|
|
replacements.add( new Replacement("vej", "v") );
|
151 |
|
|
replacements.add( new Replacement("vej", "vejen") );
|
152 |
|
|
replacements.add( new Replacement("vejen", "vej") );
|
153 |
torben |
2807 |
|
154 |
torben |
2808 |
replacements.add( new Replacement("v", "vej") );//Vi kan have vejnavne i DB der bare slutter på v istedet for vej
|
155 |
torben |
2807 |
|
156 |
torben |
2808 |
replacements.add( new Replacement("alle", "all") );
|
157 |
|
|
replacements.add( new Replacement("allé", "all") );
|
158 |
|
|
replacements.add( new Replacement("alle", "allú") ); //Fundet i logs.hentruteinfo
|
159 |
|
|
replacements.add( new Replacement("allé", "allú") );
|
160 |
torben |
2807 |
|
161 |
torben |
2808 |
replacements.add( new Replacement("gade", "gaed") ); //Fundet i logs.hentruteinfo
|
162 |
|
|
replacements.add( new Replacement("gade", "gde") );
|
163 |
torben |
2807 |
|
164 |
torben |
2808 |
replacements.add( new Replacement("plads", "pl") );
|
165 |
|
|
replacements.add( new Replacement("vænget", "vænge") );
|
166 |
torben |
2807 |
|
167 |
torben |
2808 |
replacements.add( new Replacement("boulevard", "blvd") );
|
168 |
|
|
replacements.add( new Replacement("blvd", "boulevard") );
|
169 |
torben |
2807 |
|
170 |
torben |
2808 |
replacements.add( new Replacement("boulevard", "boule") );
|
171 |
|
|
replacements.add( new Replacement("boule", "boulevard") );
|
172 |
torben |
2807 |
|
173 |
torben |
2808 |
replacements.add( new Replacement("kronborg", "krbg") ); //Krbg ladegårdsvej, 3000
|
174 |
|
|
replacements.add( new Replacement("krbg", "kronborg") );
|
175 |
|
|
|
176 |
|
|
}
|
177 |
|
|
|
178 |
|
|
|
179 |
|
|
|
180 |
|
|
public Set<String> findVejAliaser(String vejnavn) {
|
181 |
|
|
vejnavn = vejnavn.toLowerCase();
|
182 |
|
|
HashSet<String> aliasSet = new HashSet<String>();
|
183 |
|
|
aliasSet.add(vejnavn);
|
184 |
torben |
2807 |
|
185 |
torben |
2808 |
for (Replacement repl : replacements) {
|
186 |
|
|
aliasSet.add( vejnavn.replace(repl.search, repl.replace) );
|
187 |
|
|
}
|
188 |
|
|
|
189 |
|
|
|
190 |
|
|
|
191 |
torben |
2807 |
// Opbyg æøå varianter over alle fundne aliaser
|
192 |
|
|
|
193 |
torben |
2808 |
HashSet<String> variants = new HashSet<String>( aliasSet );//brug Copy-Constructor
|
194 |
torben |
2807 |
|
195 |
|
|
for (String vVejnavn : variants) {
|
196 |
|
|
// danske tegn 1
|
197 |
|
|
aliasSet.add( vVejnavn.replace("æ", "ae") );
|
198 |
|
|
aliasSet.add( vVejnavn.replace("ø", "oe") );
|
199 |
|
|
aliasSet.add( vVejnavn.replace("å", "aa") );
|
200 |
|
|
aliasSet.add( vVejnavn.replace("ae", "æ") );
|
201 |
|
|
aliasSet.add( vVejnavn.replace("oe", "ø") );
|
202 |
|
|
aliasSet.add( vVejnavn.replace("aa", "å") );
|
203 |
|
|
|
204 |
|
|
//danske tegn 2
|
205 |
|
|
aliasSet.add( vVejnavn.replace("æ", "ae").replace("ø", "oe") );
|
206 |
|
|
aliasSet.add( vVejnavn.replace("æ", "ae").replace("å", "aa") );
|
207 |
|
|
aliasSet.add( vVejnavn.replace("ø", "oe").replace("å", "aa") );
|
208 |
|
|
aliasSet.add( vVejnavn.replace("ae", "æ").replace("oe","ø") );
|
209 |
|
|
aliasSet.add( vVejnavn.replace("ae", "æ").replace("aa","å") );
|
210 |
|
|
aliasSet.add( vVejnavn.replace("oe", "ø").replace("aa", "å") );
|
211 |
|
|
|
212 |
|
|
//danske tegn 3
|
213 |
|
|
aliasSet.add( vVejnavn.replace("æ", "ae").replace("ø", "oe").replace("å", "aa") );
|
214 |
|
|
aliasSet.add( vVejnavn.replace("ae", "æ").replace("oe", "ø").replace("aa", "å") );
|
215 |
|
|
|
216 |
|
|
|
217 |
|
|
//udenlandsketegn
|
218 |
|
|
aliasSet.add( vVejnavn.replace("u", "ü") );
|
219 |
|
|
aliasSet.add( vVejnavn.replace("ü", "u") );
|
220 |
|
|
|
221 |
|
|
aliasSet.add( vVejnavn.replace("ä", "æ") );
|
222 |
|
|
aliasSet.add( vVejnavn.replace("æ", "ä") );
|
223 |
|
|
|
224 |
|
|
aliasSet.add( vVejnavn.replace("ö", "ø") );
|
225 |
|
|
aliasSet.add( vVejnavn.replace("ø", "ö") );
|
226 |
|
|
}
|
227 |
|
|
|
228 |
|
|
return aliasSet;
|
229 |
|
|
}
|
230 |
|
|
}
|