19
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
1 /*
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
2 * Normalization rules for Latin text
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
3 * [this is a JFlex specification]
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
4 *
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
5 * Wolfgang Schmidle
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
6 * version 2011-07-12
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
7 *
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
8 */
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
9
|
23
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
diff
changeset
|
10 package de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang;
|
19
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
11
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
12 %%
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
13
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
14 %public
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
15 %class MpdlNormalizerLexLA
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
16 %type java.lang.String
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
17 %unicode
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
18
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
19 // Latin: la, lat
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
20
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
21 %states DISP, DICT, SEARCH
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
22 %states RENAISSANCE_DISP, RENAISSANCE_DICT, RENAISSANCE_SEARCH
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
23
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
24 %{
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
25 private static final int CONS = 1;
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
26 private static final int VOWEL = 2;
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
27 private int cv = 0; // consonant = 1, vowel = 2, everything else = 0
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
28
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
29 private String original = "";
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
30 private String normalized = "";
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
31 private int problem = 0;
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
32
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
33 private void add (String norm) {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
34 original += yytext();
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
35 normalized += norm;
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
36 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
37
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
38 private static final String LB = "[\u002d\u00ad] ";
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
39 %}
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
40
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
41 Vowel = [AEIOUaeiouÆæęœ] // without àèòù etc.
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
42 Cons = [BCDFGHKLMNPQRSTVWXZbcdfghklmnpqrstvwxzſß]
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
43 // y counts neither as Vowel nor as Cons, see the default rule below: [yY] { cv = 0; add(yytext()); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
44
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
45 LR = [lLrR]
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
46
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
47 hyphen = [\u002d\u00ad] // hyphen and soft hyphen
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
48 LB = {hyphen} \u0020
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
49 lb = ({hyphen} \u0020)?
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
50
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
51 END = \n
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
52
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
53 que = (que)? // optional -que
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
54 enclitic = (que | ve | ne)
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
55 prefixCons = (in{lb}ter | per | ſu{lb}per | ſer) // "ſer" for forms of ſervare
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
56
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
57 %%
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
58
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
59
|
23
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
diff
changeset
|
60 // jump over empty xml elements
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
diff
changeset
|
61 "<"[^><]+"/>" { cv = 0; add(yytext()); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
diff
changeset
|
62 "-<"[^><]+"/>" { cv = 0; add(yytext()); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
diff
changeset
|
63 "<"[^><]+"></"[^><]+">" { cv = 0; add(yytext()); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
diff
changeset
|
64 "-<"[^><]+"></"[^><]+">" { cv = 0; add(yytext()); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
diff
changeset
|
65
|
19
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
66 // TEST, siehe Benedetti Seite 444
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
67 𐆑 { add("X"); } // (U+10191; D800+DD91)
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
68
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
69
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
70 <DISP, DICT, SEARCH,
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
71 RENAISSANCE_DISP, RENAISSANCE_DICT, RENAISSANCE_SEARCH> {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
72
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
73 // 1. simple replacements
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
74
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
75 // 1.1 single characters
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
76 ſ { cv = CONS; add("s"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
77 ß { cv = CONS; add("ss"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
78 [æę] { cv = VOWEL; add("ae"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
79 Æ { cv = VOWEL; add("AE"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
80 œ { cv = VOWEL; add("oe"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
81
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
82 // 1.2 character combinations
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
83 ij { cv = VOWEL; add("ii"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
84
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
85 // 2. superfluous diacritics
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
86
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
87 // 2.1 acute accent
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
88 q́ue / {END} { add("que"); } // G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
89 á / [mrst]? {enclitic} {END} { add("a"); } // G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
90 é / [mrst]? {enclitic} {END} { add("e"); } // G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
91 í / [mrst]? {enclitic} {END} { add("i"); } // G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
92 ó / [mrst]? {enclitic} {END} { add("o"); } // G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
93 ú / [mrst]? {enclitic} {END} { add("u"); } // G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
94
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
95 úe / {END} { add("ve"); } // W ??
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
96
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
97 // 2.2 grave accent
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
98 à / {que} {END} { add("a"); } // W G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
99 àm / {que} {END} { add("am"); } // W (G)
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
100 às / {que} {END} { add("as"); } // W (G) (-àsque will likely never occur)
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
101 è / {que} {END} { add("e"); } // W G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
102 ò / {que} {END} { add("o"); } // W G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
103 òd / {que} {END} { add("od"); } // W (G)
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
104 ùm / {que} {END} { add("um"); } // W (G)
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
105 ùs / {que} {END} { add("us"); } // W G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
106
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
107 ès / {que} {END} { add("es"); } // (G)
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
108 ^ quì / {END} { add("qui"); } // W ??
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
109 ^ Quì / {END} { add("Qui"); } // W ??
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
110 àc / {END} { add("ac"); } // W ??
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
111 èr / {END} { add("er"); } // W ??
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
112 èt / {END} { add("et"); } // W ??
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
113 ù / {END} { add("u"); } // W ??
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
114 ùl / {END} { add("ul"); } // W ??
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
115
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
116 // 2.3 circumflex accent
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
117 ^ hîc / {END} { add("hic"); } // W G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
118 ^ Hîc / {END} { add("Hic"); } // W G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
119 ^ ô / {END} { add("o"); } // G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
120 â / {que} {END} { add("a"); } // W G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
121 ûs / {END} { add("us"); } // W G
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
122 âr { add("ar"); } // W (G) --> this is only a rough approximation!
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
123
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
124 // 2.4 trema
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
125 // 2.4.1 common cases
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
126 aë { cv = VOWEL; add("ae"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
127 oë { cv = VOWEL; add("oe"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
128 // 2.4.2 rare cases
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
129 oï { cv = VOWEL; add("oi"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
130 uï { cv = VOWEL; add("ui"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
131 // 2.4.3 extremely rare cases
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
132 uü { cv = VOWEL; add("uu"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
133
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
134
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
135 // 3. rules for u and v
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
136
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
137 // 3.1 rules for u --> v
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
138
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
139 // peruenias --> pervenias, interuallum --> intervallum
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
140 ^ {prefixCons} / {lb} { cv = VOWEL; add(yytext().replace("ſ", "s")); } // not cv = CONS !
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
141
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
142 // uellet --> vellet
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
143 ^ [uU] / {Vowel} { cv = VOWEL; add(yytext().replaceAll("u", "v").replaceAll("U", "V")); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
144
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
145 // diuidatur --> dividatur
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
146 // ut, volui: unchanged
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
147 // no rule for veruina because we cannot distinguish it from volui
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
148 [uU] / {Vowel} {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
149 switch(cv) {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
150 case VOWEL: add(yytext().replace("u", "v").replace("U", "V")); break;
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
151 default: cv = VOWEL; add(yytext()); break;
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
152 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
153 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
154
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
155 // 3.2 rules for v --> u
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
156
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
157 // qvam --> quam
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
158 qv { cv = CONS; add("qu"); } // the replaced v still counts as consonant
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
159 Qv { cv = CONS; add("Qu"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
160 QV { cv = CONS; add("QU"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
161
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
162 // febrvarius --> februarius
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
163 // curva: unchanged
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
164 {LR} [vV] {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
165 switch(cv) {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
166 case CONS: add(yytext().replace("v", "u").replace("V", "U")); break;
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
167 default: cv = CONS; add(yytext()); break;
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
168 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
169 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
170
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
171 // februarivs --> februarius
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
172 v / {lb} {Cons} { cv = CONS; add("u"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
173 V / {lb} {Cons} { cv = CONS; add("U"); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
174
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
175 // 3.3 override default rule for .
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
176
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
177 {Vowel} { cv = VOWEL; add(yytext()); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
178 {Cons} { cv = CONS; add(yytext()); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
179 [yY] { cv = 0; add(yytext()); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
180
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
181 @ { problem = 1; cv = 0; add(yytext()); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
182 {LB} { add(yytext()); }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
183 . { problem = 1; cv = 0; add(yytext()); } // in particular from Arboreal: "〈" (2329), "〉" (232A), Ç, ç
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
184
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
185 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
186
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
187
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
188 <DISP, RENAISSANCE_DISP> {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
189
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
190 {END} {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
191 switch (problem) {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
192 case 1: return original;
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
193 default: return normalized;
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
194 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
195 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
196 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
197
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
198 <DICT, RENAISSANCE_DICT> {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
199
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
200 {END} {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
201 switch (problem) {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
202 case 1: return "";
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
203 default: return normalized.replaceAll(LB, "");
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
204 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
205 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
206 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
207
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
208 <SEARCH, RENAISSANCE_SEARCH> {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
209
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
210 {END} {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
211 switch (problem) {
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
212 case 1: return original;
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
213 default: return normalized.replaceAll(LB, "").toLowerCase();
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
214 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
215 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
216 }
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
217
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
218
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
219 /*
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
220
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
221 Annahmen:
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
222 - die Routine wird wortweise aufgerufen, mit einem \n am Ende des Strings
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
223 - Zeilenumbrüche innerhalb des Wortes sind als hyphen/soft hyphen plus space markiert
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
224
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
225
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
226 TO DO:
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
227
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
228 LA: Nochmal überlegen, ob man Ææęàèòùœ in der Vokal-Klasse weglassen kann. Sie schaden aber auch nicht. (Oder doch !?) Unterscheide Vokal-Klassen vor und nach dem u ?
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
229 LA: Diakritika nochmal mit Paul durchgehen
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
230 LA: Die Disambiguierungen durch die Diakritika fehlen noch.
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
231 LA: ist J wirklich ein Problemfall?
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
232 LA: gibt es Wörter wie super-rv... oder super-lv... in Klein- oder Großbuchstaben?
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
233
|
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff
changeset
|
234 */
|