chipper 0.4.2
Sign up to get free protection for your applications and to get access to all the features.
- data/README.rdoc +51 -0
- data/ext/extconf.rb +58 -0
- data/ext/libstemmer_c/Makefile +10 -0
- data/ext/libstemmer_c/examples/stemwords.c +209 -0
- data/ext/libstemmer_c/include/libstemmer.h +79 -0
- data/ext/libstemmer_c/libstemmer/libstemmer.c +95 -0
- data/ext/libstemmer_c/libstemmer/libstemmer_utf8.c +95 -0
- data/ext/libstemmer_c/libstemmer/modules.h +190 -0
- data/ext/libstemmer_c/libstemmer/modules_utf8.h +121 -0
- data/ext/libstemmer_c/mkinc.mak +82 -0
- data/ext/libstemmer_c/mkinc_utf8.mak +52 -0
- data/ext/libstemmer_c/runtime/api.c +66 -0
- data/ext/libstemmer_c/runtime/api.h +26 -0
- data/ext/libstemmer_c/runtime/header.h +58 -0
- data/ext/libstemmer_c/runtime/utilities.c +478 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_danish.c +337 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_danish.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_dutch.c +624 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_dutch.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_english.c +1117 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_english.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_finnish.c +762 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_finnish.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_french.c +1246 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_french.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_german.c +521 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_german.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_hungarian.c +1230 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_hungarian.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_italian.c +1065 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_italian.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_norwegian.c +297 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_norwegian.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_porter.c +749 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_porter.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_portuguese.c +1017 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_portuguese.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_spanish.c +1093 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_spanish.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_swedish.c +307 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_1_swedish.h +16 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_2_romanian.c +998 -0
- data/ext/libstemmer_c/src_c/stem_ISO_8859_2_romanian.h +16 -0
- data/ext/libstemmer_c/src_c/stem_KOI8_R_russian.c +700 -0
- data/ext/libstemmer_c/src_c/stem_KOI8_R_russian.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_danish.c +339 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_danish.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_dutch.c +634 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_dutch.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_english.c +1125 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_english.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_finnish.c +768 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_finnish.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_french.c +1256 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_french.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_german.c +527 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_german.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_hungarian.c +1234 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_hungarian.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_italian.c +1073 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_italian.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_norwegian.c +299 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_norwegian.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_porter.c +755 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_porter.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_portuguese.c +1023 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_portuguese.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_romanian.c +1004 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_romanian.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_russian.c +694 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_russian.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_spanish.c +1097 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_spanish.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_swedish.c +309 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_swedish.h +16 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_turkish.c +2205 -0
- data/ext/libstemmer_c/src_c/stem_UTF_8_turkish.h +16 -0
- data/ext/re2/bitstate.cc +378 -0
- data/ext/re2/compile.cc +1138 -0
- data/ext/re2/dfa.cc +2086 -0
- data/ext/re2/filtered_re2.cc +100 -0
- data/ext/re2/filtered_re2.h +99 -0
- data/ext/re2/hash.cc +231 -0
- data/ext/re2/mimics_pcre.cc +185 -0
- data/ext/re2/nfa.cc +709 -0
- data/ext/re2/onepass.cc +614 -0
- data/ext/re2/parse.cc +2202 -0
- data/ext/re2/perl_groups.cc +119 -0
- data/ext/re2/prefilter.cc +671 -0
- data/ext/re2/prefilter.h +105 -0
- data/ext/re2/prefilter_tree.cc +398 -0
- data/ext/re2/prefilter_tree.h +130 -0
- data/ext/re2/prog.cc +341 -0
- data/ext/re2/prog.h +376 -0
- data/ext/re2/re2.cc +1180 -0
- data/ext/re2/re2.h +837 -0
- data/ext/re2/regexp.cc +920 -0
- data/ext/re2/regexp.h +632 -0
- data/ext/re2/rune.cc +258 -0
- data/ext/re2/set.cc +113 -0
- data/ext/re2/set.h +55 -0
- data/ext/re2/simplify.cc +393 -0
- data/ext/re2/stringpiece.cc +87 -0
- data/ext/re2/stringpiece.h +182 -0
- data/ext/re2/tostring.cc +341 -0
- data/ext/re2/unicode_casefold.cc +469 -0
- data/ext/re2/unicode_casefold.h +75 -0
- data/ext/re2/unicode_groups.cc +4851 -0
- data/ext/re2/unicode_groups.h +64 -0
- data/ext/re2/valgrind.cc +24 -0
- data/ext/re2/variadic_function.h +346 -0
- data/ext/re2/walker-inl.h +244 -0
- data/ext/src/chipper.cc +626 -0
- data/ext/src/version.h +1 -0
- data/ext/stemmer.rb +40 -0
- data/ext/util/arena.h +103 -0
- data/ext/util/atomicops.h +79 -0
- data/ext/util/benchmark.h +41 -0
- data/ext/util/flags.h +27 -0
- data/ext/util/logging.h +78 -0
- data/ext/util/mutex.h +190 -0
- data/ext/util/pcre.h +679 -0
- data/ext/util/random.h +29 -0
- data/ext/util/sparse_array.h +451 -0
- data/ext/util/sparse_set.h +177 -0
- data/ext/util/test.h +57 -0
- data/ext/util/thread.h +26 -0
- data/ext/util/utf.h +43 -0
- data/ext/util/util.h +127 -0
- data/ext/util/valgrind.h +4517 -0
- data/test/helper.rb +5 -0
- data/test/test_entities.rb +57 -0
- data/test/test_tokens.rb +118 -0
- metadata +199 -0
@@ -0,0 +1,16 @@
|
|
1
|
+
|
2
|
+
/* This file was generated automatically by the Snowball to ANSI C compiler */
|
3
|
+
|
4
|
+
#ifdef __cplusplus
|
5
|
+
extern "C" {
|
6
|
+
#endif
|
7
|
+
|
8
|
+
extern struct SN_env * french_ISO_8859_1_create_env(void);
|
9
|
+
extern void french_ISO_8859_1_close_env(struct SN_env * z);
|
10
|
+
|
11
|
+
extern int french_ISO_8859_1_stem(struct SN_env * z);
|
12
|
+
|
13
|
+
#ifdef __cplusplus
|
14
|
+
}
|
15
|
+
#endif
|
16
|
+
|
@@ -0,0 +1,521 @@
|
|
1
|
+
|
2
|
+
/* This file was generated automatically by the Snowball to ANSI C compiler */
|
3
|
+
|
4
|
+
#include "../runtime/header.h"
|
5
|
+
|
6
|
+
#ifdef __cplusplus
|
7
|
+
extern "C" {
|
8
|
+
#endif
|
9
|
+
extern int german_ISO_8859_1_stem(struct SN_env * z);
|
10
|
+
#ifdef __cplusplus
|
11
|
+
}
|
12
|
+
#endif
|
13
|
+
static int r_standard_suffix(struct SN_env * z);
|
14
|
+
static int r_R2(struct SN_env * z);
|
15
|
+
static int r_R1(struct SN_env * z);
|
16
|
+
static int r_mark_regions(struct SN_env * z);
|
17
|
+
static int r_postlude(struct SN_env * z);
|
18
|
+
static int r_prelude(struct SN_env * z);
|
19
|
+
#ifdef __cplusplus
|
20
|
+
extern "C" {
|
21
|
+
#endif
|
22
|
+
|
23
|
+
|
24
|
+
extern struct SN_env * german_ISO_8859_1_create_env(void);
|
25
|
+
extern void german_ISO_8859_1_close_env(struct SN_env * z);
|
26
|
+
|
27
|
+
|
28
|
+
#ifdef __cplusplus
|
29
|
+
}
|
30
|
+
#endif
|
31
|
+
static const symbol s_0_1[1] = { 'U' };
|
32
|
+
static const symbol s_0_2[1] = { 'Y' };
|
33
|
+
static const symbol s_0_3[1] = { 0xE4 };
|
34
|
+
static const symbol s_0_4[1] = { 0xF6 };
|
35
|
+
static const symbol s_0_5[1] = { 0xFC };
|
36
|
+
|
37
|
+
static const struct among a_0[6] =
|
38
|
+
{
|
39
|
+
/* 0 */ { 0, 0, -1, 6, 0},
|
40
|
+
/* 1 */ { 1, s_0_1, 0, 2, 0},
|
41
|
+
/* 2 */ { 1, s_0_2, 0, 1, 0},
|
42
|
+
/* 3 */ { 1, s_0_3, 0, 3, 0},
|
43
|
+
/* 4 */ { 1, s_0_4, 0, 4, 0},
|
44
|
+
/* 5 */ { 1, s_0_5, 0, 5, 0}
|
45
|
+
};
|
46
|
+
|
47
|
+
static const symbol s_1_0[1] = { 'e' };
|
48
|
+
static const symbol s_1_1[2] = { 'e', 'm' };
|
49
|
+
static const symbol s_1_2[2] = { 'e', 'n' };
|
50
|
+
static const symbol s_1_3[3] = { 'e', 'r', 'n' };
|
51
|
+
static const symbol s_1_4[2] = { 'e', 'r' };
|
52
|
+
static const symbol s_1_5[1] = { 's' };
|
53
|
+
static const symbol s_1_6[2] = { 'e', 's' };
|
54
|
+
|
55
|
+
static const struct among a_1[7] =
|
56
|
+
{
|
57
|
+
/* 0 */ { 1, s_1_0, -1, 2, 0},
|
58
|
+
/* 1 */ { 2, s_1_1, -1, 1, 0},
|
59
|
+
/* 2 */ { 2, s_1_2, -1, 2, 0},
|
60
|
+
/* 3 */ { 3, s_1_3, -1, 1, 0},
|
61
|
+
/* 4 */ { 2, s_1_4, -1, 1, 0},
|
62
|
+
/* 5 */ { 1, s_1_5, -1, 3, 0},
|
63
|
+
/* 6 */ { 2, s_1_6, 5, 2, 0}
|
64
|
+
};
|
65
|
+
|
66
|
+
static const symbol s_2_0[2] = { 'e', 'n' };
|
67
|
+
static const symbol s_2_1[2] = { 'e', 'r' };
|
68
|
+
static const symbol s_2_2[2] = { 's', 't' };
|
69
|
+
static const symbol s_2_3[3] = { 'e', 's', 't' };
|
70
|
+
|
71
|
+
static const struct among a_2[4] =
|
72
|
+
{
|
73
|
+
/* 0 */ { 2, s_2_0, -1, 1, 0},
|
74
|
+
/* 1 */ { 2, s_2_1, -1, 1, 0},
|
75
|
+
/* 2 */ { 2, s_2_2, -1, 2, 0},
|
76
|
+
/* 3 */ { 3, s_2_3, 2, 1, 0}
|
77
|
+
};
|
78
|
+
|
79
|
+
static const symbol s_3_0[2] = { 'i', 'g' };
|
80
|
+
static const symbol s_3_1[4] = { 'l', 'i', 'c', 'h' };
|
81
|
+
|
82
|
+
static const struct among a_3[2] =
|
83
|
+
{
|
84
|
+
/* 0 */ { 2, s_3_0, -1, 1, 0},
|
85
|
+
/* 1 */ { 4, s_3_1, -1, 1, 0}
|
86
|
+
};
|
87
|
+
|
88
|
+
static const symbol s_4_0[3] = { 'e', 'n', 'd' };
|
89
|
+
static const symbol s_4_1[2] = { 'i', 'g' };
|
90
|
+
static const symbol s_4_2[3] = { 'u', 'n', 'g' };
|
91
|
+
static const symbol s_4_3[4] = { 'l', 'i', 'c', 'h' };
|
92
|
+
static const symbol s_4_4[4] = { 'i', 's', 'c', 'h' };
|
93
|
+
static const symbol s_4_5[2] = { 'i', 'k' };
|
94
|
+
static const symbol s_4_6[4] = { 'h', 'e', 'i', 't' };
|
95
|
+
static const symbol s_4_7[4] = { 'k', 'e', 'i', 't' };
|
96
|
+
|
97
|
+
static const struct among a_4[8] =
|
98
|
+
{
|
99
|
+
/* 0 */ { 3, s_4_0, -1, 1, 0},
|
100
|
+
/* 1 */ { 2, s_4_1, -1, 2, 0},
|
101
|
+
/* 2 */ { 3, s_4_2, -1, 1, 0},
|
102
|
+
/* 3 */ { 4, s_4_3, -1, 3, 0},
|
103
|
+
/* 4 */ { 4, s_4_4, -1, 2, 0},
|
104
|
+
/* 5 */ { 2, s_4_5, -1, 2, 0},
|
105
|
+
/* 6 */ { 4, s_4_6, -1, 3, 0},
|
106
|
+
/* 7 */ { 4, s_4_7, -1, 4, 0}
|
107
|
+
};
|
108
|
+
|
109
|
+
static const unsigned char g_v[] = { 17, 65, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 0, 32, 8 };
|
110
|
+
|
111
|
+
static const unsigned char g_s_ending[] = { 117, 30, 5 };
|
112
|
+
|
113
|
+
static const unsigned char g_st_ending[] = { 117, 30, 4 };
|
114
|
+
|
115
|
+
static const symbol s_0[] = { 0xDF };
|
116
|
+
static const symbol s_1[] = { 's', 's' };
|
117
|
+
static const symbol s_2[] = { 'u' };
|
118
|
+
static const symbol s_3[] = { 'U' };
|
119
|
+
static const symbol s_4[] = { 'y' };
|
120
|
+
static const symbol s_5[] = { 'Y' };
|
121
|
+
static const symbol s_6[] = { 'y' };
|
122
|
+
static const symbol s_7[] = { 'u' };
|
123
|
+
static const symbol s_8[] = { 'a' };
|
124
|
+
static const symbol s_9[] = { 'o' };
|
125
|
+
static const symbol s_10[] = { 'u' };
|
126
|
+
static const symbol s_11[] = { 's' };
|
127
|
+
static const symbol s_12[] = { 'n', 'i', 's' };
|
128
|
+
static const symbol s_13[] = { 'i', 'g' };
|
129
|
+
static const symbol s_14[] = { 'e' };
|
130
|
+
static const symbol s_15[] = { 'e' };
|
131
|
+
static const symbol s_16[] = { 'e', 'r' };
|
132
|
+
static const symbol s_17[] = { 'e', 'n' };
|
133
|
+
|
134
|
+
static int r_prelude(struct SN_env * z) {
|
135
|
+
{ int c_test = z->c; /* test, line 35 */
|
136
|
+
while(1) { /* repeat, line 35 */
|
137
|
+
int c1 = z->c;
|
138
|
+
{ int c2 = z->c; /* or, line 38 */
|
139
|
+
z->bra = z->c; /* [, line 37 */
|
140
|
+
if (!(eq_s(z, 1, s_0))) goto lab2;
|
141
|
+
z->ket = z->c; /* ], line 37 */
|
142
|
+
{ int ret = slice_from_s(z, 2, s_1); /* <-, line 37 */
|
143
|
+
if (ret < 0) return ret;
|
144
|
+
}
|
145
|
+
goto lab1;
|
146
|
+
lab2:
|
147
|
+
z->c = c2;
|
148
|
+
if (z->c >= z->l) goto lab0;
|
149
|
+
z->c++; /* next, line 38 */
|
150
|
+
}
|
151
|
+
lab1:
|
152
|
+
continue;
|
153
|
+
lab0:
|
154
|
+
z->c = c1;
|
155
|
+
break;
|
156
|
+
}
|
157
|
+
z->c = c_test;
|
158
|
+
}
|
159
|
+
while(1) { /* repeat, line 41 */
|
160
|
+
int c3 = z->c;
|
161
|
+
while(1) { /* goto, line 41 */
|
162
|
+
int c4 = z->c;
|
163
|
+
if (in_grouping(z, g_v, 97, 252, 0)) goto lab4;
|
164
|
+
z->bra = z->c; /* [, line 42 */
|
165
|
+
{ int c5 = z->c; /* or, line 42 */
|
166
|
+
if (!(eq_s(z, 1, s_2))) goto lab6;
|
167
|
+
z->ket = z->c; /* ], line 42 */
|
168
|
+
if (in_grouping(z, g_v, 97, 252, 0)) goto lab6;
|
169
|
+
{ int ret = slice_from_s(z, 1, s_3); /* <-, line 42 */
|
170
|
+
if (ret < 0) return ret;
|
171
|
+
}
|
172
|
+
goto lab5;
|
173
|
+
lab6:
|
174
|
+
z->c = c5;
|
175
|
+
if (!(eq_s(z, 1, s_4))) goto lab4;
|
176
|
+
z->ket = z->c; /* ], line 43 */
|
177
|
+
if (in_grouping(z, g_v, 97, 252, 0)) goto lab4;
|
178
|
+
{ int ret = slice_from_s(z, 1, s_5); /* <-, line 43 */
|
179
|
+
if (ret < 0) return ret;
|
180
|
+
}
|
181
|
+
}
|
182
|
+
lab5:
|
183
|
+
z->c = c4;
|
184
|
+
break;
|
185
|
+
lab4:
|
186
|
+
z->c = c4;
|
187
|
+
if (z->c >= z->l) goto lab3;
|
188
|
+
z->c++; /* goto, line 41 */
|
189
|
+
}
|
190
|
+
continue;
|
191
|
+
lab3:
|
192
|
+
z->c = c3;
|
193
|
+
break;
|
194
|
+
}
|
195
|
+
return 1;
|
196
|
+
}
|
197
|
+
|
198
|
+
static int r_mark_regions(struct SN_env * z) {
|
199
|
+
z->I[0] = z->l;
|
200
|
+
z->I[1] = z->l;
|
201
|
+
{ int c_test = z->c; /* test, line 52 */
|
202
|
+
{ int ret = z->c + 3;
|
203
|
+
if (0 > ret || ret > z->l) return 0;
|
204
|
+
z->c = ret; /* hop, line 52 */
|
205
|
+
}
|
206
|
+
z->I[2] = z->c; /* setmark x, line 52 */
|
207
|
+
z->c = c_test;
|
208
|
+
}
|
209
|
+
{ /* gopast */ /* grouping v, line 54 */
|
210
|
+
int ret = out_grouping(z, g_v, 97, 252, 1);
|
211
|
+
if (ret < 0) return 0;
|
212
|
+
z->c += ret;
|
213
|
+
}
|
214
|
+
{ /* gopast */ /* non v, line 54 */
|
215
|
+
int ret = in_grouping(z, g_v, 97, 252, 1);
|
216
|
+
if (ret < 0) return 0;
|
217
|
+
z->c += ret;
|
218
|
+
}
|
219
|
+
z->I[0] = z->c; /* setmark p1, line 54 */
|
220
|
+
/* try, line 55 */
|
221
|
+
if (!(z->I[0] < z->I[2])) goto lab0;
|
222
|
+
z->I[0] = z->I[2];
|
223
|
+
lab0:
|
224
|
+
{ /* gopast */ /* grouping v, line 56 */
|
225
|
+
int ret = out_grouping(z, g_v, 97, 252, 1);
|
226
|
+
if (ret < 0) return 0;
|
227
|
+
z->c += ret;
|
228
|
+
}
|
229
|
+
{ /* gopast */ /* non v, line 56 */
|
230
|
+
int ret = in_grouping(z, g_v, 97, 252, 1);
|
231
|
+
if (ret < 0) return 0;
|
232
|
+
z->c += ret;
|
233
|
+
}
|
234
|
+
z->I[1] = z->c; /* setmark p2, line 56 */
|
235
|
+
return 1;
|
236
|
+
}
|
237
|
+
|
238
|
+
static int r_postlude(struct SN_env * z) {
|
239
|
+
int among_var;
|
240
|
+
while(1) { /* repeat, line 60 */
|
241
|
+
int c1 = z->c;
|
242
|
+
z->bra = z->c; /* [, line 62 */
|
243
|
+
among_var = find_among(z, a_0, 6); /* substring, line 62 */
|
244
|
+
if (!(among_var)) goto lab0;
|
245
|
+
z->ket = z->c; /* ], line 62 */
|
246
|
+
switch(among_var) {
|
247
|
+
case 0: goto lab0;
|
248
|
+
case 1:
|
249
|
+
{ int ret = slice_from_s(z, 1, s_6); /* <-, line 63 */
|
250
|
+
if (ret < 0) return ret;
|
251
|
+
}
|
252
|
+
break;
|
253
|
+
case 2:
|
254
|
+
{ int ret = slice_from_s(z, 1, s_7); /* <-, line 64 */
|
255
|
+
if (ret < 0) return ret;
|
256
|
+
}
|
257
|
+
break;
|
258
|
+
case 3:
|
259
|
+
{ int ret = slice_from_s(z, 1, s_8); /* <-, line 65 */
|
260
|
+
if (ret < 0) return ret;
|
261
|
+
}
|
262
|
+
break;
|
263
|
+
case 4:
|
264
|
+
{ int ret = slice_from_s(z, 1, s_9); /* <-, line 66 */
|
265
|
+
if (ret < 0) return ret;
|
266
|
+
}
|
267
|
+
break;
|
268
|
+
case 5:
|
269
|
+
{ int ret = slice_from_s(z, 1, s_10); /* <-, line 67 */
|
270
|
+
if (ret < 0) return ret;
|
271
|
+
}
|
272
|
+
break;
|
273
|
+
case 6:
|
274
|
+
if (z->c >= z->l) goto lab0;
|
275
|
+
z->c++; /* next, line 68 */
|
276
|
+
break;
|
277
|
+
}
|
278
|
+
continue;
|
279
|
+
lab0:
|
280
|
+
z->c = c1;
|
281
|
+
break;
|
282
|
+
}
|
283
|
+
return 1;
|
284
|
+
}
|
285
|
+
|
286
|
+
static int r_R1(struct SN_env * z) {
|
287
|
+
if (!(z->I[0] <= z->c)) return 0;
|
288
|
+
return 1;
|
289
|
+
}
|
290
|
+
|
291
|
+
static int r_R2(struct SN_env * z) {
|
292
|
+
if (!(z->I[1] <= z->c)) return 0;
|
293
|
+
return 1;
|
294
|
+
}
|
295
|
+
|
296
|
+
static int r_standard_suffix(struct SN_env * z) {
|
297
|
+
int among_var;
|
298
|
+
{ int m1 = z->l - z->c; (void)m1; /* do, line 79 */
|
299
|
+
z->ket = z->c; /* [, line 80 */
|
300
|
+
if (z->c <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((811040 >> (z->p[z->c - 1] & 0x1f)) & 1)) goto lab0;
|
301
|
+
among_var = find_among_b(z, a_1, 7); /* substring, line 80 */
|
302
|
+
if (!(among_var)) goto lab0;
|
303
|
+
z->bra = z->c; /* ], line 80 */
|
304
|
+
{ int ret = r_R1(z);
|
305
|
+
if (ret == 0) goto lab0; /* call R1, line 80 */
|
306
|
+
if (ret < 0) return ret;
|
307
|
+
}
|
308
|
+
switch(among_var) {
|
309
|
+
case 0: goto lab0;
|
310
|
+
case 1:
|
311
|
+
{ int ret = slice_del(z); /* delete, line 82 */
|
312
|
+
if (ret < 0) return ret;
|
313
|
+
}
|
314
|
+
break;
|
315
|
+
case 2:
|
316
|
+
{ int ret = slice_del(z); /* delete, line 85 */
|
317
|
+
if (ret < 0) return ret;
|
318
|
+
}
|
319
|
+
{ int m_keep = z->l - z->c;/* (void) m_keep;*/ /* try, line 86 */
|
320
|
+
z->ket = z->c; /* [, line 86 */
|
321
|
+
if (!(eq_s_b(z, 1, s_11))) { z->c = z->l - m_keep; goto lab1; }
|
322
|
+
z->bra = z->c; /* ], line 86 */
|
323
|
+
if (!(eq_s_b(z, 3, s_12))) { z->c = z->l - m_keep; goto lab1; }
|
324
|
+
{ int ret = slice_del(z); /* delete, line 86 */
|
325
|
+
if (ret < 0) return ret;
|
326
|
+
}
|
327
|
+
lab1:
|
328
|
+
;
|
329
|
+
}
|
330
|
+
break;
|
331
|
+
case 3:
|
332
|
+
if (in_grouping_b(z, g_s_ending, 98, 116, 0)) goto lab0;
|
333
|
+
{ int ret = slice_del(z); /* delete, line 89 */
|
334
|
+
if (ret < 0) return ret;
|
335
|
+
}
|
336
|
+
break;
|
337
|
+
}
|
338
|
+
lab0:
|
339
|
+
z->c = z->l - m1;
|
340
|
+
}
|
341
|
+
{ int m2 = z->l - z->c; (void)m2; /* do, line 93 */
|
342
|
+
z->ket = z->c; /* [, line 94 */
|
343
|
+
if (z->c - 1 <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1327104 >> (z->p[z->c - 1] & 0x1f)) & 1)) goto lab2;
|
344
|
+
among_var = find_among_b(z, a_2, 4); /* substring, line 94 */
|
345
|
+
if (!(among_var)) goto lab2;
|
346
|
+
z->bra = z->c; /* ], line 94 */
|
347
|
+
{ int ret = r_R1(z);
|
348
|
+
if (ret == 0) goto lab2; /* call R1, line 94 */
|
349
|
+
if (ret < 0) return ret;
|
350
|
+
}
|
351
|
+
switch(among_var) {
|
352
|
+
case 0: goto lab2;
|
353
|
+
case 1:
|
354
|
+
{ int ret = slice_del(z); /* delete, line 96 */
|
355
|
+
if (ret < 0) return ret;
|
356
|
+
}
|
357
|
+
break;
|
358
|
+
case 2:
|
359
|
+
if (in_grouping_b(z, g_st_ending, 98, 116, 0)) goto lab2;
|
360
|
+
{ int ret = z->c - 3;
|
361
|
+
if (z->lb > ret || ret > z->l) goto lab2;
|
362
|
+
z->c = ret; /* hop, line 99 */
|
363
|
+
}
|
364
|
+
{ int ret = slice_del(z); /* delete, line 99 */
|
365
|
+
if (ret < 0) return ret;
|
366
|
+
}
|
367
|
+
break;
|
368
|
+
}
|
369
|
+
lab2:
|
370
|
+
z->c = z->l - m2;
|
371
|
+
}
|
372
|
+
{ int m3 = z->l - z->c; (void)m3; /* do, line 103 */
|
373
|
+
z->ket = z->c; /* [, line 104 */
|
374
|
+
if (z->c - 1 <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1051024 >> (z->p[z->c - 1] & 0x1f)) & 1)) goto lab3;
|
375
|
+
among_var = find_among_b(z, a_4, 8); /* substring, line 104 */
|
376
|
+
if (!(among_var)) goto lab3;
|
377
|
+
z->bra = z->c; /* ], line 104 */
|
378
|
+
{ int ret = r_R2(z);
|
379
|
+
if (ret == 0) goto lab3; /* call R2, line 104 */
|
380
|
+
if (ret < 0) return ret;
|
381
|
+
}
|
382
|
+
switch(among_var) {
|
383
|
+
case 0: goto lab3;
|
384
|
+
case 1:
|
385
|
+
{ int ret = slice_del(z); /* delete, line 106 */
|
386
|
+
if (ret < 0) return ret;
|
387
|
+
}
|
388
|
+
{ int m_keep = z->l - z->c;/* (void) m_keep;*/ /* try, line 107 */
|
389
|
+
z->ket = z->c; /* [, line 107 */
|
390
|
+
if (!(eq_s_b(z, 2, s_13))) { z->c = z->l - m_keep; goto lab4; }
|
391
|
+
z->bra = z->c; /* ], line 107 */
|
392
|
+
{ int m4 = z->l - z->c; (void)m4; /* not, line 107 */
|
393
|
+
if (!(eq_s_b(z, 1, s_14))) goto lab5;
|
394
|
+
{ z->c = z->l - m_keep; goto lab4; }
|
395
|
+
lab5:
|
396
|
+
z->c = z->l - m4;
|
397
|
+
}
|
398
|
+
{ int ret = r_R2(z);
|
399
|
+
if (ret == 0) { z->c = z->l - m_keep; goto lab4; } /* call R2, line 107 */
|
400
|
+
if (ret < 0) return ret;
|
401
|
+
}
|
402
|
+
{ int ret = slice_del(z); /* delete, line 107 */
|
403
|
+
if (ret < 0) return ret;
|
404
|
+
}
|
405
|
+
lab4:
|
406
|
+
;
|
407
|
+
}
|
408
|
+
break;
|
409
|
+
case 2:
|
410
|
+
{ int m5 = z->l - z->c; (void)m5; /* not, line 110 */
|
411
|
+
if (!(eq_s_b(z, 1, s_15))) goto lab6;
|
412
|
+
goto lab3;
|
413
|
+
lab6:
|
414
|
+
z->c = z->l - m5;
|
415
|
+
}
|
416
|
+
{ int ret = slice_del(z); /* delete, line 110 */
|
417
|
+
if (ret < 0) return ret;
|
418
|
+
}
|
419
|
+
break;
|
420
|
+
case 3:
|
421
|
+
{ int ret = slice_del(z); /* delete, line 113 */
|
422
|
+
if (ret < 0) return ret;
|
423
|
+
}
|
424
|
+
{ int m_keep = z->l - z->c;/* (void) m_keep;*/ /* try, line 114 */
|
425
|
+
z->ket = z->c; /* [, line 115 */
|
426
|
+
{ int m6 = z->l - z->c; (void)m6; /* or, line 115 */
|
427
|
+
if (!(eq_s_b(z, 2, s_16))) goto lab9;
|
428
|
+
goto lab8;
|
429
|
+
lab9:
|
430
|
+
z->c = z->l - m6;
|
431
|
+
if (!(eq_s_b(z, 2, s_17))) { z->c = z->l - m_keep; goto lab7; }
|
432
|
+
}
|
433
|
+
lab8:
|
434
|
+
z->bra = z->c; /* ], line 115 */
|
435
|
+
{ int ret = r_R1(z);
|
436
|
+
if (ret == 0) { z->c = z->l - m_keep; goto lab7; } /* call R1, line 115 */
|
437
|
+
if (ret < 0) return ret;
|
438
|
+
}
|
439
|
+
{ int ret = slice_del(z); /* delete, line 115 */
|
440
|
+
if (ret < 0) return ret;
|
441
|
+
}
|
442
|
+
lab7:
|
443
|
+
;
|
444
|
+
}
|
445
|
+
break;
|
446
|
+
case 4:
|
447
|
+
{ int ret = slice_del(z); /* delete, line 119 */
|
448
|
+
if (ret < 0) return ret;
|
449
|
+
}
|
450
|
+
{ int m_keep = z->l - z->c;/* (void) m_keep;*/ /* try, line 120 */
|
451
|
+
z->ket = z->c; /* [, line 121 */
|
452
|
+
if (z->c - 1 <= z->lb || (z->p[z->c - 1] != 103 && z->p[z->c - 1] != 104)) { z->c = z->l - m_keep; goto lab10; }
|
453
|
+
among_var = find_among_b(z, a_3, 2); /* substring, line 121 */
|
454
|
+
if (!(among_var)) { z->c = z->l - m_keep; goto lab10; }
|
455
|
+
z->bra = z->c; /* ], line 121 */
|
456
|
+
{ int ret = r_R2(z);
|
457
|
+
if (ret == 0) { z->c = z->l - m_keep; goto lab10; } /* call R2, line 121 */
|
458
|
+
if (ret < 0) return ret;
|
459
|
+
}
|
460
|
+
switch(among_var) {
|
461
|
+
case 0: { z->c = z->l - m_keep; goto lab10; }
|
462
|
+
case 1:
|
463
|
+
{ int ret = slice_del(z); /* delete, line 123 */
|
464
|
+
if (ret < 0) return ret;
|
465
|
+
}
|
466
|
+
break;
|
467
|
+
}
|
468
|
+
lab10:
|
469
|
+
;
|
470
|
+
}
|
471
|
+
break;
|
472
|
+
}
|
473
|
+
lab3:
|
474
|
+
z->c = z->l - m3;
|
475
|
+
}
|
476
|
+
return 1;
|
477
|
+
}
|
478
|
+
|
479
|
+
extern int german_ISO_8859_1_stem(struct SN_env * z) {
|
480
|
+
{ int c1 = z->c; /* do, line 134 */
|
481
|
+
{ int ret = r_prelude(z);
|
482
|
+
if (ret == 0) goto lab0; /* call prelude, line 134 */
|
483
|
+
if (ret < 0) return ret;
|
484
|
+
}
|
485
|
+
lab0:
|
486
|
+
z->c = c1;
|
487
|
+
}
|
488
|
+
{ int c2 = z->c; /* do, line 135 */
|
489
|
+
{ int ret = r_mark_regions(z);
|
490
|
+
if (ret == 0) goto lab1; /* call mark_regions, line 135 */
|
491
|
+
if (ret < 0) return ret;
|
492
|
+
}
|
493
|
+
lab1:
|
494
|
+
z->c = c2;
|
495
|
+
}
|
496
|
+
z->lb = z->c; z->c = z->l; /* backwards, line 136 */
|
497
|
+
|
498
|
+
{ int m3 = z->l - z->c; (void)m3; /* do, line 137 */
|
499
|
+
{ int ret = r_standard_suffix(z);
|
500
|
+
if (ret == 0) goto lab2; /* call standard_suffix, line 137 */
|
501
|
+
if (ret < 0) return ret;
|
502
|
+
}
|
503
|
+
lab2:
|
504
|
+
z->c = z->l - m3;
|
505
|
+
}
|
506
|
+
z->c = z->lb;
|
507
|
+
{ int c4 = z->c; /* do, line 138 */
|
508
|
+
{ int ret = r_postlude(z);
|
509
|
+
if (ret == 0) goto lab3; /* call postlude, line 138 */
|
510
|
+
if (ret < 0) return ret;
|
511
|
+
}
|
512
|
+
lab3:
|
513
|
+
z->c = c4;
|
514
|
+
}
|
515
|
+
return 1;
|
516
|
+
}
|
517
|
+
|
518
|
+
extern struct SN_env * german_ISO_8859_1_create_env(void) { return SN_create_env(0, 3, 0); }
|
519
|
+
|
520
|
+
extern void german_ISO_8859_1_close_env(struct SN_env * z) { SN_close_env(z, 0); }
|
521
|
+
|
@@ -0,0 +1,16 @@
|
|
1
|
+
|
2
|
+
/* This file was generated automatically by the Snowball to ANSI C compiler */
|
3
|
+
|
4
|
+
#ifdef __cplusplus
|
5
|
+
extern "C" {
|
6
|
+
#endif
|
7
|
+
|
8
|
+
extern struct SN_env * german_ISO_8859_1_create_env(void);
|
9
|
+
extern void german_ISO_8859_1_close_env(struct SN_env * z);
|
10
|
+
|
11
|
+
extern int german_ISO_8859_1_stem(struct SN_env * z);
|
12
|
+
|
13
|
+
#ifdef __cplusplus
|
14
|
+
}
|
15
|
+
#endif
|
16
|
+
|