Blame


1 08708877 2003-11-25 devnull #include <u.h>
2 08708877 2003-11-25 devnull #include <libc.h>
3 08708877 2003-11-25 devnull #include <bio.h>
4 08708877 2003-11-25 devnull #include "dict.h"
5 08708877 2003-11-25 devnull
6 08708877 2003-11-25 devnull Dict dicts[] = {
7 08708877 2003-11-25 devnull {"oed", "Oxford English Dictionary, 2nd Ed.",
8 32f69c36 2003-12-11 devnull "#9/dict/oed2", "#9/dict/oed2index",
9 08708877 2003-11-25 devnull oednextoff, oedprintentry, oedprintkey},
10 08708877 2003-11-25 devnull {"ahd", "American Heritage Dictionary, 2nd College Ed.",
11 08708877 2003-11-25 devnull "ahd/DICT.DB", "ahd/index",
12 08708877 2003-11-25 devnull ahdnextoff, ahdprintentry, ahdprintkey},
13 08708877 2003-11-25 devnull {"pgw", "Project Gutenberg Webster Dictionary",
14 32f69c36 2003-12-11 devnull "#9/dict/pgw", "#9/dict/pgwindex",
15 08708877 2003-11-25 devnull pgwnextoff, pgwprintentry, pgwprintkey},
16 08708877 2003-11-25 devnull {"thesaurus", "Collins Thesaurus",
17 32f69c36 2003-12-11 devnull "#9/dict/thesaurus", "#9/dict/thesindex",
18 08708877 2003-11-25 devnull thesnextoff, thesprintentry, thesprintkey},
19 08708877 2003-11-25 devnull
20 08708877 2003-11-25 devnull {"ce", "Gendai Chinese->English",
21 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/sandic24.dat",
22 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/ceindex",
23 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
24 08708877 2003-11-25 devnull {"ceh", "Gendai Chinese->English (Hanzi index)",
25 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/sandic24.dat",
26 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/cehindex",
27 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
28 08708877 2003-11-25 devnull {"ec", "Gendai English->Chinese",
29 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/sandic24.dat",
30 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/ecindex",
31 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
32 08708877 2003-11-25 devnull
33 08708877 2003-11-25 devnull {"dae", "Gyldendal Danish->English",
34 32f69c36 2003-12-11 devnull "#9/dict/world/gylddata/sandic30.dat",
35 32f69c36 2003-12-11 devnull "#9/dict/world/gylddata/daeindex",
36 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
37 08708877 2003-11-25 devnull {"eda", "Gyldendal English->Danish",
38 32f69c36 2003-12-11 devnull "#9/dict/world/gylddata/sandic29.dat",
39 32f69c36 2003-12-11 devnull "#9/dict/world/gylddata/edaindex",
40 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
41 08708877 2003-11-25 devnull
42 08708877 2003-11-25 devnull {"due", "Wolters-Noordhoff Dutch->English",
43 32f69c36 2003-12-11 devnull "#9/dict/world/woltdata/sandic07.dat",
44 32f69c36 2003-12-11 devnull "#9/dict/world/woltdata/deindex",
45 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
46 08708877 2003-11-25 devnull {"edu", "Wolters-Noordhoff English->Dutch",
47 32f69c36 2003-12-11 devnull "#9/dict/world/woltdata/sandic06.dat",
48 32f69c36 2003-12-11 devnull "#9/dict/world/woltdata/edindex",
49 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
50 08708877 2003-11-25 devnull
51 08708877 2003-11-25 devnull {"fie", "WSOY Finnish->English",
52 32f69c36 2003-12-11 devnull "#9/dict/world/werndata/sandic32.dat",
53 32f69c36 2003-12-11 devnull "#9/dict/world/werndata/fieindex",
54 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
55 08708877 2003-11-25 devnull {"efi", "WSOY English->Finnish",
56 32f69c36 2003-12-11 devnull "#9/dict/world/werndata/sandic31.dat",
57 32f69c36 2003-12-11 devnull "#9/dict/world/werndata/efiindex",
58 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
59 08708877 2003-11-25 devnull
60 08708877 2003-11-25 devnull {"fe", "Collins French->English",
61 32f69c36 2003-12-11 devnull "#9/dict/fe", "#9/dict/feindex",
62 08708877 2003-11-25 devnull pcollnextoff, pcollprintentry, pcollprintkey},
63 08708877 2003-11-25 devnull {"ef", "Collins English->French",
64 32f69c36 2003-12-11 devnull "#9/dict/ef", "#9/dict/efindex",
65 08708877 2003-11-25 devnull pcollnextoff, pcollprintentry, pcollprintkey},
66 08708877 2003-11-25 devnull
67 08708877 2003-11-25 devnull {"ge", "Collins German->English",
68 32f69c36 2003-12-11 devnull "#9/dict/ge", "#9/dict/geindex",
69 08708877 2003-11-25 devnull pcollgnextoff, pcollgprintentry, pcollgprintkey},
70 08708877 2003-11-25 devnull {"eg", "Collins English->German",
71 32f69c36 2003-12-11 devnull "#9/dict/eg", "#9/dict/egindex",
72 08708877 2003-11-25 devnull pcollgnextoff, pcollgprintentry, pcollgprintkey},
73 08708877 2003-11-25 devnull
74 08708877 2003-11-25 devnull {"ie", "Collins Italian->English",
75 32f69c36 2003-12-11 devnull "#9/dict/ie", "#9/dict/ieindex",
76 08708877 2003-11-25 devnull pcollnextoff, pcollprintentry, pcollprintkey},
77 08708877 2003-11-25 devnull {"ei", "Collins English->Italian",
78 32f69c36 2003-12-11 devnull "#9/dict/ei", "#9/dict/eiindex",
79 08708877 2003-11-25 devnull pcollnextoff, pcollprintentry, pcollprintkey},
80 08708877 2003-11-25 devnull
81 08708877 2003-11-25 devnull {"je", "Sanshusha Japanese->English",
82 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/sandic18.dat",
83 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/jeindex",
84 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
85 08708877 2003-11-25 devnull {"jek", "Sanshusha Japanese->English (Kanji index)",
86 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/sandic18.dat",
87 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/jekindex",
88 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
89 08708877 2003-11-25 devnull {"ej", "Sanshusha English->Japanese",
90 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/sandic18.dat",
91 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/ejindex",
92 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
93 08708877 2003-11-25 devnull
94 08708877 2003-11-25 devnull {"tjeg", "Sanshusha technical Japanese->English,German",
95 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/sandic16.dat",
96 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/tjegindex",
97 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
98 08708877 2003-11-25 devnull {"tjegk", "Sanshusha technical Japanese->English,German (Kanji index)",
99 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/sandic16.dat",
100 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/tjegkindex",
101 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
102 08708877 2003-11-25 devnull {"tegj", "Sanshusha technical English->German,Japanese",
103 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/sandic16.dat",
104 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/tegjindex",
105 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
106 08708877 2003-11-25 devnull {"tgje", "Sanshusha technical German->Japanese,English",
107 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/sandic16.dat",
108 32f69c36 2003-12-11 devnull "#9/dict/world/sansdata/tgjeindex",
109 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
110 08708877 2003-11-25 devnull
111 08708877 2003-11-25 devnull {"ne", "Kunnskapforlaget Norwegian->English",
112 32f69c36 2003-12-11 devnull "#9/dict/world/kunndata/sandic28.dat",
113 32f69c36 2003-12-11 devnull "#9/dict/world/kunndata/neindex",
114 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
115 08708877 2003-11-25 devnull {"en", "Kunnskapforlaget English->Norwegian",
116 32f69c36 2003-12-11 devnull "#9/dict/world/kunndata/sandic27.dat",
117 32f69c36 2003-12-11 devnull "#9/dict/world/kunndata/enindex",
118 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
119 08708877 2003-11-25 devnull
120 08708877 2003-11-25 devnull {"re", "Leon Ungier Russian->English",
121 32f69c36 2003-12-11 devnull "#9/dict/re", "#9/dict/reindex",
122 08708877 2003-11-25 devnull simplenextoff, simpleprintentry, simpleprintkey},
123 08708877 2003-11-25 devnull {"er", "Leon Ungier English->Russian",
124 32f69c36 2003-12-11 devnull "#9/dict/re", "#9/dict/erindex",
125 08708877 2003-11-25 devnull simplenextoff, simpleprintentry, simpleprintkey},
126 08708877 2003-11-25 devnull
127 08708877 2003-11-25 devnull {"se", "Collins Spanish->English",
128 32f69c36 2003-12-11 devnull "#9/dict/se", "#9/dict/seindex",
129 08708877 2003-11-25 devnull pcollnextoff, pcollprintentry, pcollprintkey},
130 08708877 2003-11-25 devnull {"es", "Collins English->Spanish",
131 32f69c36 2003-12-11 devnull "#9/dict/es", "#9/dict/esindex",
132 08708877 2003-11-25 devnull pcollnextoff, pcollprintentry, pcollprintkey},
133 08708877 2003-11-25 devnull
134 08708877 2003-11-25 devnull {"swe", "Esselte Studium Swedish->English",
135 32f69c36 2003-12-11 devnull "#9/dict/world/essedata/sandic34.dat",
136 32f69c36 2003-12-11 devnull "#9/dict/world/essedata/sweindex",
137 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
138 08708877 2003-11-25 devnull {"esw", "Esselte Studium English->Swedish",
139 32f69c36 2003-12-11 devnull "#9/dict/world/essedata/sandic33.dat",
140 32f69c36 2003-12-11 devnull "#9/dict/world/essedata/eswindex",
141 08708877 2003-11-25 devnull worldnextoff, worldprintentry, worldprintkey},
142 08708877 2003-11-25 devnull
143 08708877 2003-11-25 devnull {"movie", "Movies -- by title",
144 32f69c36 2003-12-11 devnull "movie/data", "#9/dict/movtindex",
145 08708877 2003-11-25 devnull movienextoff, movieprintentry, movieprintkey},
146 08708877 2003-11-25 devnull {"moviea", "Movies -- by actor",
147 32f69c36 2003-12-11 devnull "movie/data", "#9/dict/movaindex",
148 08708877 2003-11-25 devnull movienextoff, movieprintentry, movieprintkey},
149 08708877 2003-11-25 devnull {"movied", "Movies -- by director",
150 32f69c36 2003-12-11 devnull "movie/data", "#9/dict/movdindex",
151 08708877 2003-11-25 devnull movienextoff, movieprintentry, movieprintkey},
152 08708877 2003-11-25 devnull
153 08708877 2003-11-25 devnull {"slang", "English Slang",
154 32f69c36 2003-12-11 devnull "#9/dict/slang", "#9/dict/slangindex",
155 08708877 2003-11-25 devnull slangnextoff, slangprintentry, slangprintkey},
156 08708877 2003-11-25 devnull
157 08708877 2003-11-25 devnull {"robert", "Robert Électronique",
158 32f69c36 2003-12-11 devnull "#9/dict/robert/_pointers", "#9/dict/robert/_index",
159 08708877 2003-11-25 devnull robertnextoff, robertindexentry, robertprintkey},
160 08708877 2003-11-25 devnull {"robertv", "Robert Électronique - formes des verbes",
161 32f69c36 2003-12-11 devnull "#9/dict/robert/flex.rob", "#9/dict/robert/_flexindex",
162 08708877 2003-11-25 devnull robertnextflex, robertflexentry, robertprintkey},
163 08708877 2003-11-25 devnull
164 08708877 2003-11-25 devnull {0, 0, 0, 0, 0}
165 08708877 2003-11-25 devnull };
166 08708877 2003-11-25 devnull
167 08708877 2003-11-25 devnull typedef struct Lig Lig;
168 08708877 2003-11-25 devnull struct Lig {
169 08708877 2003-11-25 devnull Rune start; /* accent rune */
170 08708877 2003-11-25 devnull Rune pairs[100]; /* <char,accented version> pairs */
171 08708877 2003-11-25 devnull };
172 08708877 2003-11-25 devnull
173 d126ebb5 2004-03-26 devnull /* keep in sync with dict.h */
174 08708877 2003-11-25 devnull static Lig ligtab[Nligs] = {
175 d126ebb5 2004-03-26 devnull {0xb4, {0x41, 0xc1, 0x61, 0xe1, 0x43, 0x106, 0x63, 0x107, 0x45, 0xc9, 0x65, 0xe9, 0x67, 0x123, 0x49, 0xcd, 0x69, 0xed, 0x131, 0xed, 0x4c, 0x139, 0x6c, 0x13a, 0x4e, 0x143, 0x6e, 0x144, 0x4f, 0xd3, 0x6f, 0xf3, 0x52, 0x154, 0x72, 0x155, 0x53, 0x15a, 0x73, 0x15b, 0x55, 0xda, 0x75, 0xfa, 0x59, 0xdd, 0x79, 0xfd, 0x5a, 0x179, 0x7a, 0x17a, 0}},
176 d126ebb5 2004-03-26 devnull {0x2cb, {0x41, 0xc0, 0x61, 0xe0, 0x45, 0xc8, 0x65, 0xe8, 0x49, 0xcc, 0x69, 0xec, 0x131, 0xec, 0x4f, 0xd2, 0x6f, 0xf2, 0x55, 0xd9, 0x75, 0xf9, 0}},
177 d126ebb5 2004-03-26 devnull {0xa8, {0x41, 0xc4, 0x61, 0xe4, 0x45, 0xcb, 0x65, 0xeb, 0x49, 0xcf, 0x69, 0xef, 0x4f, 0xd6, 0x6f, 0xf6, 0x55, 0xdc, 0x75, 0xfc, 0x59, 0x178, 0x79, 0xff, 0}},
178 d126ebb5 2004-03-26 devnull {0xb8, {0x43, 0xc7, 0x63, 0xe7, 0x47, 0x122, 0x4b, 0x136, 0x6b, 0x137, 0x4c, 0x13b, 0x6c, 0x13c, 0x4e, 0x145, 0x6e, 0x146, 0x52, 0x156, 0x72, 0x157, 0x53, 0x15e, 0x73, 0x15f, 0x54, 0x162, 0x74, 0x163, 0}},
179 d126ebb5 2004-03-26 devnull {0x2dc, {0x41, 0xc3, 0x61, 0xe3, 0x49, 0x128, 0x69, 0x129, 0x131, 0x129, 0x4e, 0xd1, 0x6e, 0xf1, 0x4f, 0xd5, 0x6f, 0xf5, 0x55, 0x168, 0x75, 0x169, 0}},
180 d126ebb5 2004-03-26 devnull {0x2d8, {0x41, 0x102, 0x61, 0x103, 0x45, 0x114, 0x65, 0x115, 0x47, 0x11e, 0x67, 0x11f, 0x49, 0x12c, 0x69, 0x12d, 0x131, 0x12d, 0x4f, 0x14e, 0x6f, 0x14f, 0x55, 0x16c, 0x75, 0x16d, 0}},
181 d126ebb5 2004-03-26 devnull {0x2da, {0x41, 0xc5, 0x61, 0xe5, 0x55, 0x16e, 0x75, 0x16f, 0}},
182 d126ebb5 2004-03-26 devnull {0x2d9, {0x43, 0x10a, 0x63, 0x10b, 0x45, 0x116, 0x65, 0x117, 0x47, 0x120, 0x67, 0x121, 0x49, 0x130, 0x4c, 0x13f, 0x6c, 0x140, 0x5a, 0x17b, 0x7a, 0x17c, 0}},
183 d126ebb5 2004-03-26 devnull {0x2e, {0}},
184 d126ebb5 2004-03-26 devnull {0x2322, {0x41, 0xc2, 0x61, 0xe2, 0x43, 0x108, 0x63, 0x109, 0x45, 0xca, 0x65, 0xea, 0x47, 0x11c, 0x67, 0x11d, 0x48, 0x124, 0x68, 0x125, 0x49, 0xce, 0x69, 0xee, 0x131, 0xee, 0x4a, 0x134, 0x6a, 0x135, 0x4f, 0xd4, 0x6f, 0xf4, 0x53, 0x15c, 0x73, 0x15d, 0x55, 0xdb, 0x75, 0xfb, 0x57, 0x174, 0x77, 0x175, 0x59, 0x176, 0x79, 0x177, 0}},
185 d126ebb5 2004-03-26 devnull {0x32f, {0}},
186 d126ebb5 2004-03-26 devnull {0x2db, {0x41, 0x104, 0x61, 0x105, 0x45, 0x118, 0x65, 0x119, 0x49, 0x12e, 0x69, 0x12f, 0x131, 0x12f, 0x55, 0x172, 0x75, 0x173, 0}},
187 d126ebb5 2004-03-26 devnull {0xaf, {0x41, 0x100, 0x61, 0x101, 0x45, 0x112, 0x65, 0x113, 0x49, 0x12a, 0x69, 0x12b, 0x131, 0x12b, 0x4f, 0x14c, 0x6f, 0x14d, 0x55, 0x16a, 0x75, 0x16b, 0}},
188 d126ebb5 2004-03-26 devnull {0x2c7, {0x43, 0x10c, 0x63, 0x10d, 0x44, 0x10e, 0x64, 0x10f, 0x45, 0x11a, 0x65, 0x11b, 0x4c, 0x13d, 0x6c, 0x13e, 0x4e, 0x147, 0x6e, 0x148, 0x52, 0x158, 0x72, 0x159, 0x53, 0x160, 0x73, 0x161, 0x54, 0x164, 0x74, 0x165, 0x5a, 0x17d, 0x7a, 0x17e, 0}},
189 d126ebb5 2004-03-26 devnull {0x2bd, {0}},
190 d126ebb5 2004-03-26 devnull {0x2bc, {0}},
191 d126ebb5 2004-03-26 devnull {0x32e, {0}}
192 08708877 2003-11-25 devnull };
193 08708877 2003-11-25 devnull
194 08708877 2003-11-25 devnull Rune multitab[Nmulti][5] = {
195 d126ebb5 2004-03-26 devnull {0x2bd, 0x3b1, 0},
196 d126ebb5 2004-03-26 devnull {0x2bc, 0x3b1, 0},
197 d126ebb5 2004-03-26 devnull {0x61, 0x6e, 0x64, 0},
198 d126ebb5 2004-03-26 devnull {0x61, 0x2f, 0x71, 0},
199 d126ebb5 2004-03-26 devnull {0x3c, 0x7c, 0},
200 d126ebb5 2004-03-26 devnull {0x2e, 0x2e, 0},
201 d126ebb5 2004-03-26 devnull {0x2e, 0x2e, 0x2e, 0},
202 d126ebb5 2004-03-26 devnull {0x2bd, 0x3b5, 0},
203 d126ebb5 2004-03-26 devnull {0x2bc, 0x3b5, 0},
204 d126ebb5 2004-03-26 devnull {0x2014, 0x2014, 0},
205 d126ebb5 2004-03-26 devnull {0x2bd, 0x3b7, 0},
206 d126ebb5 2004-03-26 devnull {0x2bc, 0x3b7, 0},
207 d126ebb5 2004-03-26 devnull {0x2bd, 0x3b9, 0},
208 d126ebb5 2004-03-26 devnull {0x2bc, 0x3b9, 0},
209 d126ebb5 2004-03-26 devnull {0x63, 0x74, 0},
210 d126ebb5 2004-03-26 devnull {0x66, 0x66, 0},
211 d126ebb5 2004-03-26 devnull {0x66, 0x66, 0x69, 0},
212 d126ebb5 2004-03-26 devnull {0x66, 0x66, 0x6c, 0},
213 d126ebb5 2004-03-26 devnull {0x66, 0x6c, 0},
214 d126ebb5 2004-03-26 devnull {0x66, 0x69, 0},
215 d126ebb5 2004-03-26 devnull {0x26b, 0x26b, 0},
216 d126ebb5 2004-03-26 devnull {0x73, 0x74, 0},
217 d126ebb5 2004-03-26 devnull {0x2bd, 0x3bf, 0},
218 d126ebb5 2004-03-26 devnull {0x2bc, 0x3bf, 0},
219 d126ebb5 2004-03-26 devnull {0x6f, 0x72, 0},
220 d126ebb5 2004-03-26 devnull {0x2bd, 0x3c1, 0},
221 d126ebb5 2004-03-26 devnull {0x2bc, 0x3c1, 0},
222 d126ebb5 2004-03-26 devnull {0x7e, 0x7e, 0},
223 d126ebb5 2004-03-26 devnull {0x2bd, 0x3c5, 0},
224 d126ebb5 2004-03-26 devnull {0x2bc, 0x3c5, 0},
225 d126ebb5 2004-03-26 devnull {0x2bd, 0x3c9, 0},
226 d126ebb5 2004-03-26 devnull {0x2bc, 0x3c9, 0},
227 d126ebb5 2004-03-26 devnull {0x6f, 0x65, 0},
228 d126ebb5 2004-03-26 devnull {0x20, 0x20, 0},
229 08708877 2003-11-25 devnull };
230 08708877 2003-11-25 devnull
231 08708877 2003-11-25 devnull #define risupper(r) (0x41 <= (r) && (r) <= 0x5a)
232 08708877 2003-11-25 devnull #define rislatin1(r) (0xC0 <= (r) && (r) <= 0xFF)
233 08708877 2003-11-25 devnull #define rtolower(r) ((r)-'A'+'a')
234 08708877 2003-11-25 devnull
235 08708877 2003-11-25 devnull static Rune latin_fold_tab[] =
236 08708877 2003-11-25 devnull {
237 08708877 2003-11-25 devnull /* Table to fold latin 1 characters to ASCII equivalents
238 08708877 2003-11-25 devnull based at Rune value 0xc0
239 08708877 2003-11-25 devnull
240 08708877 2003-11-25 devnull À Á Â Ã Ä Å Æ Ç
241 08708877 2003-11-25 devnull È É Ê Ë Ì Í Î Ï
242 08708877 2003-11-25 devnull Ð Ñ Ò Ó Ô Õ Ö ×
243 08708877 2003-11-25 devnull Ø Ù Ú Û Ü Ý Þ ß
244 08708877 2003-11-25 devnull à á â ã ä å æ ç
245 08708877 2003-11-25 devnull è é ê ë ì í î ï
246 08708877 2003-11-25 devnull ð ñ ò ó ô õ ö ÷
247 08708877 2003-11-25 devnull ø ù ú û ü ý þ ÿ
248 08708877 2003-11-25 devnull */
249 08708877 2003-11-25 devnull 'a', 'a', 'a', 'a', 'a', 'a', 'a', 'c',
250 08708877 2003-11-25 devnull 'e', 'e', 'e', 'e', 'i', 'i', 'i', 'i',
251 08708877 2003-11-25 devnull 'd', 'n', 'o', 'o', 'o', 'o', 'o', 0 ,
252 08708877 2003-11-25 devnull 'o', 'u', 'u', 'u', 'u', 'y', 0 , 0 ,
253 08708877 2003-11-25 devnull 'a', 'a', 'a', 'a', 'a', 'a', 'a', 'c',
254 08708877 2003-11-25 devnull 'e', 'e', 'e', 'e', 'i', 'i', 'i', 'i',
255 08708877 2003-11-25 devnull 'd', 'n', 'o', 'o', 'o', 'o', 'o', 0 ,
256 08708877 2003-11-25 devnull 'o', 'u', 'u', 'u', 'u', 'y', 0 , 'y',
257 08708877 2003-11-25 devnull };
258 08708877 2003-11-25 devnull
259 08708877 2003-11-25 devnull static Rune *ttabstack[20];
260 08708877 2003-11-25 devnull static int ntt;
261 08708877 2003-11-25 devnull
262 08708877 2003-11-25 devnull /*
263 08708877 2003-11-25 devnull * tab is an array of n Assoc's, sorted by key.
264 08708877 2003-11-25 devnull * Look for key in tab, and return corresponding val
265 08708877 2003-11-25 devnull * or -1 if not there
266 08708877 2003-11-25 devnull */
267 08708877 2003-11-25 devnull long
268 08708877 2003-11-25 devnull lookassoc(Assoc *tab, int n, char *key)
269 08708877 2003-11-25 devnull {
270 08708877 2003-11-25 devnull Assoc *q;
271 08708877 2003-11-25 devnull long i, low, high;
272 08708877 2003-11-25 devnull int r;
273 08708877 2003-11-25 devnull
274 08708877 2003-11-25 devnull for(low = -1, high = n; high > low+1; ){
275 08708877 2003-11-25 devnull i = (high+low)/2;
276 08708877 2003-11-25 devnull q = &tab[i];
277 08708877 2003-11-25 devnull if((r=strcmp(key, q->key))<0)
278 08708877 2003-11-25 devnull high = i;
279 08708877 2003-11-25 devnull else if(r == 0)
280 08708877 2003-11-25 devnull return q->val;
281 08708877 2003-11-25 devnull else
282 08708877 2003-11-25 devnull low=i;
283 08708877 2003-11-25 devnull }
284 08708877 2003-11-25 devnull return -1;
285 08708877 2003-11-25 devnull }
286 08708877 2003-11-25 devnull
287 08708877 2003-11-25 devnull long
288 08708877 2003-11-25 devnull looknassoc(Nassoc *tab, int n, long key)
289 08708877 2003-11-25 devnull {
290 08708877 2003-11-25 devnull Nassoc *q;
291 08708877 2003-11-25 devnull long i, low, high;
292 08708877 2003-11-25 devnull
293 08708877 2003-11-25 devnull for(low = -1, high = n; high > low+1; ){
294 08708877 2003-11-25 devnull i = (high+low)/2;
295 08708877 2003-11-25 devnull q = &tab[i];
296 08708877 2003-11-25 devnull if(key < q->key)
297 08708877 2003-11-25 devnull high = i;
298 08708877 2003-11-25 devnull else if(key == q->key)
299 08708877 2003-11-25 devnull return q->val;
300 08708877 2003-11-25 devnull else
301 08708877 2003-11-25 devnull low=i;
302 08708877 2003-11-25 devnull }
303 08708877 2003-11-25 devnull return -1;
304 08708877 2003-11-25 devnull }
305 08708877 2003-11-25 devnull
306 08708877 2003-11-25 devnull void
307 08708877 2003-11-25 devnull err(char *fmt, ...)
308 08708877 2003-11-25 devnull {
309 08708877 2003-11-25 devnull char buf[1000];
310 08708877 2003-11-25 devnull va_list v;
311 08708877 2003-11-25 devnull
312 08708877 2003-11-25 devnull va_start(v, fmt);
313 08708877 2003-11-25 devnull vsnprint(buf, sizeof(buf), fmt, v);
314 08708877 2003-11-25 devnull va_end(v);
315 08708877 2003-11-25 devnull fprint(2, "%s: %s\n", argv0, buf);
316 08708877 2003-11-25 devnull }
317 08708877 2003-11-25 devnull
318 08708877 2003-11-25 devnull /*
319 08708877 2003-11-25 devnull * Write the rune r to bout, keeping track of line length
320 08708877 2003-11-25 devnull * and breaking the lines (at blanks) when they get too long
321 08708877 2003-11-25 devnull */
322 08708877 2003-11-25 devnull void
323 08708877 2003-11-25 devnull outrune(long r)
324 08708877 2003-11-25 devnull {
325 08708877 2003-11-25 devnull if(outinhibit)
326 08708877 2003-11-25 devnull return;
327 08708877 2003-11-25 devnull if(++linelen > breaklen && r == 0x20) {
328 08708877 2003-11-25 devnull Bputc(bout, '\n');
329 08708877 2003-11-25 devnull linelen = 0;
330 08708877 2003-11-25 devnull } else
331 08708877 2003-11-25 devnull Bputrune(bout, r);
332 08708877 2003-11-25 devnull }
333 08708877 2003-11-25 devnull
334 08708877 2003-11-25 devnull void
335 08708877 2003-11-25 devnull outrunes(Rune *rp)
336 08708877 2003-11-25 devnull {
337 08708877 2003-11-25 devnull Rune r;
338 08708877 2003-11-25 devnull
339 08708877 2003-11-25 devnull while((r = *rp++) != 0)
340 08708877 2003-11-25 devnull outrune(r);
341 08708877 2003-11-25 devnull }
342 08708877 2003-11-25 devnull
343 08708877 2003-11-25 devnull /* like outrune, but when arg is know to be a char */
344 08708877 2003-11-25 devnull void
345 08708877 2003-11-25 devnull outchar(int c)
346 08708877 2003-11-25 devnull {
347 08708877 2003-11-25 devnull if(outinhibit)
348 08708877 2003-11-25 devnull return;
349 08708877 2003-11-25 devnull if(++linelen > breaklen && c == ' ') {
350 08708877 2003-11-25 devnull c ='\n';
351 08708877 2003-11-25 devnull linelen = 0;
352 08708877 2003-11-25 devnull }
353 08708877 2003-11-25 devnull Bputc(bout, c);
354 08708877 2003-11-25 devnull }
355 08708877 2003-11-25 devnull
356 08708877 2003-11-25 devnull void
357 08708877 2003-11-25 devnull outchars(char *s)
358 08708877 2003-11-25 devnull {
359 08708877 2003-11-25 devnull char c;
360 08708877 2003-11-25 devnull
361 08708877 2003-11-25 devnull while((c = *s++) != 0)
362 08708877 2003-11-25 devnull outchar(c);
363 08708877 2003-11-25 devnull }
364 08708877 2003-11-25 devnull
365 08708877 2003-11-25 devnull void
366 08708877 2003-11-25 devnull outprint(char *fmt, ...)
367 08708877 2003-11-25 devnull {
368 08708877 2003-11-25 devnull char buf[1000];
369 08708877 2003-11-25 devnull va_list v;
370 08708877 2003-11-25 devnull
371 08708877 2003-11-25 devnull va_start(v, fmt);
372 08708877 2003-11-25 devnull vsnprint(buf, sizeof(buf), fmt, v);
373 08708877 2003-11-25 devnull va_end(v);
374 08708877 2003-11-25 devnull outchars(buf);
375 08708877 2003-11-25 devnull }
376 08708877 2003-11-25 devnull
377 08708877 2003-11-25 devnull void
378 08708877 2003-11-25 devnull outpiece(char *b, char *e)
379 08708877 2003-11-25 devnull {
380 08708877 2003-11-25 devnull int c, lastc;
381 08708877 2003-11-25 devnull
382 08708877 2003-11-25 devnull lastc = 0;
383 08708877 2003-11-25 devnull while(b < e) {
384 08708877 2003-11-25 devnull c = *b++;
385 08708877 2003-11-25 devnull if(c == '\n')
386 08708877 2003-11-25 devnull c = ' ';
387 08708877 2003-11-25 devnull if(!(c == ' ' && lastc == ' '))
388 08708877 2003-11-25 devnull outchar(c);
389 08708877 2003-11-25 devnull lastc = c;
390 08708877 2003-11-25 devnull }
391 08708877 2003-11-25 devnull }
392 08708877 2003-11-25 devnull
393 08708877 2003-11-25 devnull /*
394 08708877 2003-11-25 devnull * Go to new line if not already there; indent if ind != 0.
395 08708877 2003-11-25 devnull * If ind > 1, leave a blank line too.
396 08708877 2003-11-25 devnull * Slight hack: assume if current line is only one or two
397 08708877 2003-11-25 devnull * characters long, then they were spaces.
398 08708877 2003-11-25 devnull */
399 08708877 2003-11-25 devnull void
400 08708877 2003-11-25 devnull outnl(int ind)
401 08708877 2003-11-25 devnull {
402 08708877 2003-11-25 devnull if(outinhibit)
403 08708877 2003-11-25 devnull return;
404 08708877 2003-11-25 devnull if(ind) {
405 08708877 2003-11-25 devnull if(ind > 1) {
406 08708877 2003-11-25 devnull if(linelen > 2)
407 08708877 2003-11-25 devnull Bputc(bout, '\n');
408 08708877 2003-11-25 devnull Bprint(bout, "\n ");
409 08708877 2003-11-25 devnull } else if(linelen == 0)
410 08708877 2003-11-25 devnull Bprint(bout, " ");
411 08708877 2003-11-25 devnull else if(linelen == 1)
412 08708877 2003-11-25 devnull Bputc(bout, ' ');
413 08708877 2003-11-25 devnull else if(linelen != 2)
414 08708877 2003-11-25 devnull Bprint(bout, "\n ");
415 08708877 2003-11-25 devnull linelen = 2;
416 08708877 2003-11-25 devnull } else {
417 08708877 2003-11-25 devnull if(linelen) {
418 08708877 2003-11-25 devnull Bputc(bout, '\n');
419 08708877 2003-11-25 devnull linelen = 0;
420 08708877 2003-11-25 devnull }
421 08708877 2003-11-25 devnull }
422 08708877 2003-11-25 devnull }
423 08708877 2003-11-25 devnull
424 08708877 2003-11-25 devnull /*
425 08708877 2003-11-25 devnull * Fold the runes in null-terminated rp.
426 08708877 2003-11-25 devnull * Use the sort(1) definition of folding (uppercase to lowercase,
427 08708877 2003-11-25 devnull * latin1-accented characters to corresponding unaccented chars)
428 08708877 2003-11-25 devnull */
429 08708877 2003-11-25 devnull void
430 08708877 2003-11-25 devnull fold(Rune *rp)
431 08708877 2003-11-25 devnull {
432 08708877 2003-11-25 devnull Rune r;
433 08708877 2003-11-25 devnull
434 08708877 2003-11-25 devnull while((r = *rp) != 0) {
435 08708877 2003-11-25 devnull if (rislatin1(r) && latin_fold_tab[r-0xc0])
436 08708877 2003-11-25 devnull r = latin_fold_tab[r-0xc0];
437 08708877 2003-11-25 devnull if(risupper(r))
438 08708877 2003-11-25 devnull r = rtolower(r);
439 08708877 2003-11-25 devnull *rp++ = r;
440 08708877 2003-11-25 devnull }
441 08708877 2003-11-25 devnull }
442 08708877 2003-11-25 devnull
443 08708877 2003-11-25 devnull /*
444 08708877 2003-11-25 devnull * Like fold, but put folded result into new
445 08708877 2003-11-25 devnull * (assumed to have enough space).
446 08708877 2003-11-25 devnull * old is a regular expression, but we know that
447 08708877 2003-11-25 devnull * metacharacters aren't affected
448 08708877 2003-11-25 devnull */
449 08708877 2003-11-25 devnull void
450 08708877 2003-11-25 devnull foldre(char *new, char *old)
451 08708877 2003-11-25 devnull {
452 08708877 2003-11-25 devnull Rune r;
453 08708877 2003-11-25 devnull
454 08708877 2003-11-25 devnull while(*old) {
455 08708877 2003-11-25 devnull old += chartorune(&r, old);
456 08708877 2003-11-25 devnull if (rislatin1(r) && latin_fold_tab[r-0xc0])
457 08708877 2003-11-25 devnull r = latin_fold_tab[r-0xc0];
458 08708877 2003-11-25 devnull if(risupper(r))
459 08708877 2003-11-25 devnull r = rtolower(r);
460 08708877 2003-11-25 devnull new += runetochar(new, &r);
461 08708877 2003-11-25 devnull }
462 08708877 2003-11-25 devnull *new = 0;
463 08708877 2003-11-25 devnull }
464 08708877 2003-11-25 devnull
465 08708877 2003-11-25 devnull /*
466 08708877 2003-11-25 devnull * acomp(s, t) returns:
467 08708877 2003-11-25 devnull * -2 if s strictly precedes t
468 08708877 2003-11-25 devnull * -1 if s is a prefix of t
469 08708877 2003-11-25 devnull * 0 if s is the same as t
470 08708877 2003-11-25 devnull * 1 if t is a prefix of s
471 08708877 2003-11-25 devnull * 2 if t strictly precedes s
472 08708877 2003-11-25 devnull */
473 08708877 2003-11-25 devnull
474 08708877 2003-11-25 devnull int
475 08708877 2003-11-25 devnull acomp(Rune *s, Rune *t)
476 08708877 2003-11-25 devnull {
477 08708877 2003-11-25 devnull int cs, ct;
478 08708877 2003-11-25 devnull
479 08708877 2003-11-25 devnull for(;;) {
480 08708877 2003-11-25 devnull cs = *s;
481 08708877 2003-11-25 devnull ct = *t;
482 08708877 2003-11-25 devnull if(cs != ct)
483 08708877 2003-11-25 devnull break;
484 08708877 2003-11-25 devnull if(cs == 0)
485 08708877 2003-11-25 devnull return 0;
486 08708877 2003-11-25 devnull s++;
487 08708877 2003-11-25 devnull t++;
488 08708877 2003-11-25 devnull }
489 08708877 2003-11-25 devnull if(cs == 0)
490 08708877 2003-11-25 devnull return -1;
491 08708877 2003-11-25 devnull if(ct == 0)
492 08708877 2003-11-25 devnull return 1;
493 08708877 2003-11-25 devnull if(cs < ct)
494 08708877 2003-11-25 devnull return -2;
495 08708877 2003-11-25 devnull return 2;
496 08708877 2003-11-25 devnull }
497 08708877 2003-11-25 devnull
498 08708877 2003-11-25 devnull /*
499 08708877 2003-11-25 devnull * Copy null terminated Runes from 'from' to 'to'.
500 08708877 2003-11-25 devnull */
501 08708877 2003-11-25 devnull void
502 08708877 2003-11-25 devnull runescpy(Rune *to, Rune *from)
503 08708877 2003-11-25 devnull {
504 08708877 2003-11-25 devnull while((*to++ = *from++) != 0)
505 08708877 2003-11-25 devnull continue;
506 08708877 2003-11-25 devnull }
507 08708877 2003-11-25 devnull
508 08708877 2003-11-25 devnull /*
509 08708877 2003-11-25 devnull * Conversion of unsigned number to long, no overflow detection
510 08708877 2003-11-25 devnull */
511 08708877 2003-11-25 devnull long
512 08708877 2003-11-25 devnull runetol(Rune *r)
513 08708877 2003-11-25 devnull {
514 08708877 2003-11-25 devnull int c;
515 08708877 2003-11-25 devnull long n;
516 08708877 2003-11-25 devnull
517 08708877 2003-11-25 devnull n = 0;
518 08708877 2003-11-25 devnull for(;; r++){
519 08708877 2003-11-25 devnull c = *r;
520 08708877 2003-11-25 devnull if(0x30<=c && c<=0x39)
521 08708877 2003-11-25 devnull c -= '0';
522 08708877 2003-11-25 devnull else
523 08708877 2003-11-25 devnull break;
524 08708877 2003-11-25 devnull n = n*10 + c;
525 08708877 2003-11-25 devnull }
526 08708877 2003-11-25 devnull return n;
527 08708877 2003-11-25 devnull }
528 08708877 2003-11-25 devnull
529 08708877 2003-11-25 devnull /*
530 08708877 2003-11-25 devnull * See if there is a rune corresponding to the accented
531 08708877 2003-11-25 devnull * version of r with accent acc (acc in [LIGS..LIGE-1]),
532 08708877 2003-11-25 devnull * and return it if so, else return NONE.
533 08708877 2003-11-25 devnull */
534 08708877 2003-11-25 devnull Rune
535 08708877 2003-11-25 devnull liglookup(Rune acc, Rune r)
536 08708877 2003-11-25 devnull {
537 08708877 2003-11-25 devnull Rune *p;
538 08708877 2003-11-25 devnull
539 08708877 2003-11-25 devnull if(acc < LIGS || acc >= LIGE)
540 08708877 2003-11-25 devnull return NONE;
541 08708877 2003-11-25 devnull for(p = ligtab[acc-LIGS].pairs; *p; p += 2)
542 08708877 2003-11-25 devnull if(*p == r)
543 08708877 2003-11-25 devnull return *(p+1);
544 08708877 2003-11-25 devnull return NONE;
545 08708877 2003-11-25 devnull }
546 08708877 2003-11-25 devnull
547 08708877 2003-11-25 devnull /*
548 08708877 2003-11-25 devnull * Maintain a translation table stack (a translation table
549 08708877 2003-11-25 devnull * is an array of Runes indexed by bytes or 7-bit bytes).
550 08708877 2003-11-25 devnull * If starting is true, push the curtab onto the stack
551 08708877 2003-11-25 devnull * and return newtab; else pop the top of the stack and
552 08708877 2003-11-25 devnull * return it.
553 08708877 2003-11-25 devnull * If curtab is 0, initialize the stack and return.
554 08708877 2003-11-25 devnull */
555 08708877 2003-11-25 devnull Rune *
556 08708877 2003-11-25 devnull changett(Rune *curtab, Rune *newtab, int starting)
557 08708877 2003-11-25 devnull {
558 08708877 2003-11-25 devnull if(curtab == 0) {
559 08708877 2003-11-25 devnull ntt = 0;
560 08708877 2003-11-25 devnull return 0;
561 08708877 2003-11-25 devnull }
562 08708877 2003-11-25 devnull if(starting) {
563 08708877 2003-11-25 devnull if(ntt >= asize(ttabstack)) {
564 08708877 2003-11-25 devnull if(debug)
565 08708877 2003-11-25 devnull err("translation stack overflow");
566 08708877 2003-11-25 devnull return curtab;
567 08708877 2003-11-25 devnull }
568 08708877 2003-11-25 devnull ttabstack[ntt++] = curtab;
569 08708877 2003-11-25 devnull return newtab;
570 08708877 2003-11-25 devnull } else {
571 08708877 2003-11-25 devnull if(ntt == 0) {
572 08708877 2003-11-25 devnull if(debug)
573 08708877 2003-11-25 devnull err("translation stack underflow");
574 08708877 2003-11-25 devnull return curtab;
575 08708877 2003-11-25 devnull }
576 08708877 2003-11-25 devnull return ttabstack[--ntt];
577 08708877 2003-11-25 devnull }
578 08708877 2003-11-25 devnull }