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

diverse Korrekturen
author Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
date Tue, 27 Nov 2012 12:35:19 +0100
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
23
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1 /*
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
2 * Normalization rules for Dutch text
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
3 * [this is a JFlex specification]
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
4 *
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
5 * Wolfgang Schmidle
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
6 * version 2011-07-12
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
7 *
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
8 */
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
9
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
10 package de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang;
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
11
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
12 %%
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
13
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
14 %public
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
15 %class MpdlNormalizerLexNL
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
16 %type java.lang.String
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
17 %unicode
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
18
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
19 // Dutch: nl
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
20
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
21 %states DISP, DICT, SEARCH
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
22
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
23 %{
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
24 private String original = "";
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
25 private String normalized = "";
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
26 private int problem = 0;
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
27
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
28 private void add (String norm) {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
29 original += yytext();
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
30 normalized += norm;
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
31 }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
32
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
33 private static final String LB = "[\u002d\u00ad] ";
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
34 %}
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
35
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
36 hyphen = [-\u{00ad}] // hyphen and soft hyphen
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
37 LB = {hyphen} \u0020
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
38 // lb = ({hyphen} \u0020)?
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
39
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
40 END = \n
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
41
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
42 %%
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
43
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
44 // jump over empty xml elements
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
45 "<"[^><]+"/>" { add(yytext()); }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
46 "-<"[^><]+"/>" { add(yytext()); }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
47 "<"[^><]+"></"[^><]+">" { add(yytext()); }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
48 "-<"[^><]+"></"[^><]+">" { add(yytext()); }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
49
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
50 <DISP, DICT, SEARCH> {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
51
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
52 ſ { add("s"); }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
53
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
54 }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
55
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
56
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
57 // default
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
58
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
59 @ { problem = 1; add(yytext()); }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
60 {LB} { add(yytext()); }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
61 . { add(yytext()); }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
62
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
63
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
64 <DISP> {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
65
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
66 {END} {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
67 switch (problem) {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
68 case 1: return original;
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
69 default: return normalized;
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
70 }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
71 }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
72 }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
73
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
74 <DICT> {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
75
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
76 {END} {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
77 switch (problem) {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
78 case 1: return "";
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
79 default: return normalized.replaceAll(LB, "");
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
80 }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
81 }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
82 }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
83
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
84 <SEARCH> {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
85
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
86 {END} {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
87 switch (problem) {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
88 case 1: return original;
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
89 default: return normalized.replaceAll(LB, "").toLowerCase();
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
90 }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
91 }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
92 }
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
93
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
94
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
95 /*
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
96
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
97 Annahmen:
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
98 - die Routine wird wortweise aufgerufen, mit einem \n am Ende des Strings
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
99 - Zeilenumbrüche innerhalb des Wortes sind als hyphen/soft hyphen plus space markiert
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
100
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
101 TO DO:
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
102
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
103 NL: vollständig?
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
104
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
105 */