Mercurial > hg > mpdl-group

comparison software/mpdl-services/mpiwg-mpdl-lt/bin/de/mpg/mpiwg/berlin/mpdl/lt/text/norm/lang/MpdlNormalizerLexEL.lex @ 23:e845310098ba

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
diverse Korrekturen
author Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
date Tue, 27 Nov 2012 12:35:19 +0100
parents
children
comparison
equal deleted inserted replaced
22:6a45a982c333 23:e845310098ba
1 /*
2 * Normalization rules for Greek text
3 * [this is a JFlex specification]
4 *
5 * Wolfgang Schmidle
6 * version 2011-08-03
7 *
8 */
9
10 package de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang;
11
12 %%
13
14 %public
15 %class MpdlNormalizerLexEL
16 %type java.lang.String
17 %unicode
18
19 // Greek: el, grc
20
21 %states DISP, DICT, SEARCH
22 %state SIGMA
23
24 %{
25 private String original = "";
26 private String normalized = "";
27 private int problem = 0;
28
29 private void add (String norm) {
30 original += yytext();
31 normalized += norm;
32 }
33
34 private static final String LB = "[\u002d\u00ad] ";
35 %}
36
37 hyphen = [-\u{00ad}] // hyphen and soft hyphen
38 LB = {hyphen} \u0020
39 // lb = ({hyphen} \u0020)?
40
41 END = \n
42
43 wordend = [νρς]? {END}
44
45 Latin = [abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ]
46
47
48 %%
49
50 // jump over empty xml elements
51 "<"[^><]+"/>" { add(yytext()); }
52 "-<"[^><]+"/>" { add(yytext()); }
53 "<"[^><]+"></"[^><]+">" { add(yytext()); }
54 "-<"[^><]+"></"[^><]+">" { add(yytext()); }
55
56
57 // always replace tonos by oxia
58 // (although this should really be corrected in the text rather than normalized)
59 ά { add("ά"); }
60 έ { add("έ"); }
61 ή { add("ή"); }
62 ί { add("ί"); }
63 ό { add("ό"); }
64 ύ { add("ύ"); }
65 ώ { add("ώ"); }
66
67
68 <DICT, SEARCH, SIGMA> {
69
70 ὰ / {wordend} { add("ά"); }
71 ᾲ / {wordend} { add("ᾴ"); }
72 ὲ / {wordend} { add("έ"); }
73 ὴ / {wordend} { add("ή"); }
74 ῂ / {wordend} { add("ῄ"); }
75 ὶ / {wordend} { add("ί"); }
76 ὸ / {wordend} { add("ό"); }
77 ὺ / {wordend} { add("ύ"); }
78 ὼ / {wordend} { add("ώ"); }
79 ῲ / {wordend} { add("ῴ"); }
80
81 // other candidates: Ὰ Ὲ Ὴ Ὶ Ὺ Ὸ Ὼ
82
83 }
84
85 <SIGMA> {
86
87 ς { add("σ"); }
88
89 }
90
91 // default
92
93 @ { problem = 1; add(yytext()); }
94 {Latin} { problem = 1; add(yytext()); }
95
96 {LB} { add(yytext()); }
97 . { add(yytext()); }
98
99
100 <DISP> {
101
102 {END} {
103 switch (problem) {
104 case 1: return original;
105 default: return normalized;
106 }
107 }
108 }
109
110 <DICT, SIGMA> {
111
112 {END} {
113 switch (problem) {
114 case 1: return "";
115 default: return normalized.replaceAll(LB, "");
116 }
117 }
118 }
119
120 <SEARCH> {
121
122 {END} {
123 switch (problem) {
124 case 1: return original;
125 default: return normalized.replaceAll(LB, "").toLowerCase();
126 }
127 }
128 }
129
130
131 /*
132
133 Annahmen:
134 - die Routine wird wortweise aufgerufen, mit einem \n am Ende des Strings
135 - Zeilenumbrüche innerhalb des Wortes sind als hyphen/soft hyphen plus space markiert
136
137 TO DO:
138
139 EL: tonos --> oxia wieder rausnehmen, weil es im Text geändert werden muss?
140 EL: gibt es noch weitere Fälle, wo legitimerweise ein Gravis vorkommen kann?
141 EL: kommen Großbuchstaben mit Gravis bei uns jemals vor, und sollen sie normalisiert werden?
142 EL: neuer State BETACODE ?
143 EL: nicht falsche Zeichen definieren, sondern erlaubte Zeichen
144
145 */