ruby-stemmer 0.9.3-x86-mingw32
Sign up to get free protection for your applications and to get access to all the features.
- data/MIT-LICENSE +21 -0
- data/README.rdoc +113 -0
- data/Rakefile +70 -0
- data/TODO +0 -0
- data/VERSION +1 -0
- data/ext/lingua/extconf.rb +40 -0
- data/ext/lingua/stemmer.c +115 -0
- data/lib/lingua/1.8/stemmer_native.so +0 -0
- data/lib/lingua/1.9/stemmer_native.so +0 -0
- data/lib/lingua/stemmer.rb +60 -0
- data/libstemmer_c/MANIFEST +72 -0
- data/libstemmer_c/Makefile +9 -0
- data/libstemmer_c/Makefile.windows +15 -0
- data/libstemmer_c/README +125 -0
- data/libstemmer_c/examples/stemwords.c +209 -0
- data/libstemmer_c/include/libstemmer.h +79 -0
- data/libstemmer_c/libstemmer/libstemmer.c +93 -0
- data/libstemmer_c/libstemmer/libstemmer_utf8.c +93 -0
- data/libstemmer_c/libstemmer/modules.h +195 -0
- data/libstemmer_c/libstemmer/modules.txt +51 -0
- data/libstemmer_c/libstemmer/modules_utf8.h +123 -0
- data/libstemmer_c/libstemmer/modules_utf8.txt +50 -0
- data/libstemmer_c/mkinc.mak +86 -0
- data/libstemmer_c/mkinc_utf8.mak +54 -0
- data/libstemmer_c/runtime/api.c +66 -0
- data/libstemmer_c/runtime/api.h +26 -0
- data/libstemmer_c/runtime/header.h +58 -0
- data/libstemmer_c/runtime/utilities.c +478 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_danish.c +337 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_danish.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_dutch.c +624 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_dutch.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_english.c +1117 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_english.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_finnish.c +762 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_finnish.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_french.c +1246 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_french.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_german.c +503 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_german.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_hungarian.c +1230 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_hungarian.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_italian.c +1065 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_italian.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_latin.c +443 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_latin.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_norwegian.c +297 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_norwegian.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_porter.c +749 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_porter.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_portuguese.c +1017 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_portuguese.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_spanish.c +1093 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_spanish.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_swedish.c +307 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_swedish.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_2_romanian.c +998 -0
- data/libstemmer_c/src_c/stem_ISO_8859_2_romanian.h +16 -0
- data/libstemmer_c/src_c/stem_KOI8_R_russian.c +700 -0
- data/libstemmer_c/src_c/stem_KOI8_R_russian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_danish.c +339 -0
- data/libstemmer_c/src_c/stem_UTF_8_danish.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_dutch.c +634 -0
- data/libstemmer_c/src_c/stem_UTF_8_dutch.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_english.c +1125 -0
- data/libstemmer_c/src_c/stem_UTF_8_english.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_finnish.c +768 -0
- data/libstemmer_c/src_c/stem_UTF_8_finnish.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_french.c +1256 -0
- data/libstemmer_c/src_c/stem_UTF_8_french.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_german.c +509 -0
- data/libstemmer_c/src_c/stem_UTF_8_german.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_hungarian.c +1234 -0
- data/libstemmer_c/src_c/stem_UTF_8_hungarian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_italian.c +1073 -0
- data/libstemmer_c/src_c/stem_UTF_8_italian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_latin.c +443 -0
- data/libstemmer_c/src_c/stem_UTF_8_latin.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_norwegian.c +299 -0
- data/libstemmer_c/src_c/stem_UTF_8_norwegian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_porter.c +755 -0
- data/libstemmer_c/src_c/stem_UTF_8_porter.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_portuguese.c +1023 -0
- data/libstemmer_c/src_c/stem_UTF_8_portuguese.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_romanian.c +1004 -0
- data/libstemmer_c/src_c/stem_UTF_8_romanian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_russian.c +694 -0
- data/libstemmer_c/src_c/stem_UTF_8_russian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_spanish.c +1097 -0
- data/libstemmer_c/src_c/stem_UTF_8_spanish.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_swedish.c +309 -0
- data/libstemmer_c/src_c/stem_UTF_8_swedish.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_turkish.c +2205 -0
- data/libstemmer_c/src_c/stem_UTF_8_turkish.h +16 -0
- data/test/helper.rb +3 -0
- data/test/lingua/test_stemmer.rb +99 -0
- metadata +162 -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 * russian_KOI8_R_create_env(void);
|
9
|
+
extern void russian_KOI8_R_close_env(struct SN_env * z);
|
10
|
+
|
11
|
+
extern int russian_KOI8_R_stem(struct SN_env * z);
|
12
|
+
|
13
|
+
#ifdef __cplusplus
|
14
|
+
}
|
15
|
+
#endif
|
16
|
+
|
@@ -0,0 +1,339 @@
|
|
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 danish_UTF_8_stem(struct SN_env * z);
|
10
|
+
#ifdef __cplusplus
|
11
|
+
}
|
12
|
+
#endif
|
13
|
+
static int r_undouble(struct SN_env * z);
|
14
|
+
static int r_other_suffix(struct SN_env * z);
|
15
|
+
static int r_consonant_pair(struct SN_env * z);
|
16
|
+
static int r_main_suffix(struct SN_env * z);
|
17
|
+
static int r_mark_regions(struct SN_env * z);
|
18
|
+
#ifdef __cplusplus
|
19
|
+
extern "C" {
|
20
|
+
#endif
|
21
|
+
|
22
|
+
|
23
|
+
extern struct SN_env * danish_UTF_8_create_env(void);
|
24
|
+
extern void danish_UTF_8_close_env(struct SN_env * z);
|
25
|
+
|
26
|
+
|
27
|
+
#ifdef __cplusplus
|
28
|
+
}
|
29
|
+
#endif
|
30
|
+
static const symbol s_0_0[3] = { 'h', 'e', 'd' };
|
31
|
+
static const symbol s_0_1[5] = { 'e', 't', 'h', 'e', 'd' };
|
32
|
+
static const symbol s_0_2[4] = { 'e', 'r', 'e', 'd' };
|
33
|
+
static const symbol s_0_3[1] = { 'e' };
|
34
|
+
static const symbol s_0_4[5] = { 'e', 'r', 'e', 'd', 'e' };
|
35
|
+
static const symbol s_0_5[4] = { 'e', 'n', 'd', 'e' };
|
36
|
+
static const symbol s_0_6[6] = { 'e', 'r', 'e', 'n', 'd', 'e' };
|
37
|
+
static const symbol s_0_7[3] = { 'e', 'n', 'e' };
|
38
|
+
static const symbol s_0_8[4] = { 'e', 'r', 'n', 'e' };
|
39
|
+
static const symbol s_0_9[3] = { 'e', 'r', 'e' };
|
40
|
+
static const symbol s_0_10[2] = { 'e', 'n' };
|
41
|
+
static const symbol s_0_11[5] = { 'h', 'e', 'd', 'e', 'n' };
|
42
|
+
static const symbol s_0_12[4] = { 'e', 'r', 'e', 'n' };
|
43
|
+
static const symbol s_0_13[2] = { 'e', 'r' };
|
44
|
+
static const symbol s_0_14[5] = { 'h', 'e', 'd', 'e', 'r' };
|
45
|
+
static const symbol s_0_15[4] = { 'e', 'r', 'e', 'r' };
|
46
|
+
static const symbol s_0_16[1] = { 's' };
|
47
|
+
static const symbol s_0_17[4] = { 'h', 'e', 'd', 's' };
|
48
|
+
static const symbol s_0_18[2] = { 'e', 's' };
|
49
|
+
static const symbol s_0_19[5] = { 'e', 'n', 'd', 'e', 's' };
|
50
|
+
static const symbol s_0_20[7] = { 'e', 'r', 'e', 'n', 'd', 'e', 's' };
|
51
|
+
static const symbol s_0_21[4] = { 'e', 'n', 'e', 's' };
|
52
|
+
static const symbol s_0_22[5] = { 'e', 'r', 'n', 'e', 's' };
|
53
|
+
static const symbol s_0_23[4] = { 'e', 'r', 'e', 's' };
|
54
|
+
static const symbol s_0_24[3] = { 'e', 'n', 's' };
|
55
|
+
static const symbol s_0_25[6] = { 'h', 'e', 'd', 'e', 'n', 's' };
|
56
|
+
static const symbol s_0_26[5] = { 'e', 'r', 'e', 'n', 's' };
|
57
|
+
static const symbol s_0_27[3] = { 'e', 'r', 's' };
|
58
|
+
static const symbol s_0_28[3] = { 'e', 't', 's' };
|
59
|
+
static const symbol s_0_29[5] = { 'e', 'r', 'e', 't', 's' };
|
60
|
+
static const symbol s_0_30[2] = { 'e', 't' };
|
61
|
+
static const symbol s_0_31[4] = { 'e', 'r', 'e', 't' };
|
62
|
+
|
63
|
+
static const struct among a_0[32] =
|
64
|
+
{
|
65
|
+
/* 0 */ { 3, s_0_0, -1, 1, 0},
|
66
|
+
/* 1 */ { 5, s_0_1, 0, 1, 0},
|
67
|
+
/* 2 */ { 4, s_0_2, -1, 1, 0},
|
68
|
+
/* 3 */ { 1, s_0_3, -1, 1, 0},
|
69
|
+
/* 4 */ { 5, s_0_4, 3, 1, 0},
|
70
|
+
/* 5 */ { 4, s_0_5, 3, 1, 0},
|
71
|
+
/* 6 */ { 6, s_0_6, 5, 1, 0},
|
72
|
+
/* 7 */ { 3, s_0_7, 3, 1, 0},
|
73
|
+
/* 8 */ { 4, s_0_8, 3, 1, 0},
|
74
|
+
/* 9 */ { 3, s_0_9, 3, 1, 0},
|
75
|
+
/* 10 */ { 2, s_0_10, -1, 1, 0},
|
76
|
+
/* 11 */ { 5, s_0_11, 10, 1, 0},
|
77
|
+
/* 12 */ { 4, s_0_12, 10, 1, 0},
|
78
|
+
/* 13 */ { 2, s_0_13, -1, 1, 0},
|
79
|
+
/* 14 */ { 5, s_0_14, 13, 1, 0},
|
80
|
+
/* 15 */ { 4, s_0_15, 13, 1, 0},
|
81
|
+
/* 16 */ { 1, s_0_16, -1, 2, 0},
|
82
|
+
/* 17 */ { 4, s_0_17, 16, 1, 0},
|
83
|
+
/* 18 */ { 2, s_0_18, 16, 1, 0},
|
84
|
+
/* 19 */ { 5, s_0_19, 18, 1, 0},
|
85
|
+
/* 20 */ { 7, s_0_20, 19, 1, 0},
|
86
|
+
/* 21 */ { 4, s_0_21, 18, 1, 0},
|
87
|
+
/* 22 */ { 5, s_0_22, 18, 1, 0},
|
88
|
+
/* 23 */ { 4, s_0_23, 18, 1, 0},
|
89
|
+
/* 24 */ { 3, s_0_24, 16, 1, 0},
|
90
|
+
/* 25 */ { 6, s_0_25, 24, 1, 0},
|
91
|
+
/* 26 */ { 5, s_0_26, 24, 1, 0},
|
92
|
+
/* 27 */ { 3, s_0_27, 16, 1, 0},
|
93
|
+
/* 28 */ { 3, s_0_28, 16, 1, 0},
|
94
|
+
/* 29 */ { 5, s_0_29, 28, 1, 0},
|
95
|
+
/* 30 */ { 2, s_0_30, -1, 1, 0},
|
96
|
+
/* 31 */ { 4, s_0_31, 30, 1, 0}
|
97
|
+
};
|
98
|
+
|
99
|
+
static const symbol s_1_0[2] = { 'g', 'd' };
|
100
|
+
static const symbol s_1_1[2] = { 'd', 't' };
|
101
|
+
static const symbol s_1_2[2] = { 'g', 't' };
|
102
|
+
static const symbol s_1_3[2] = { 'k', 't' };
|
103
|
+
|
104
|
+
static const struct among a_1[4] =
|
105
|
+
{
|
106
|
+
/* 0 */ { 2, s_1_0, -1, -1, 0},
|
107
|
+
/* 1 */ { 2, s_1_1, -1, -1, 0},
|
108
|
+
/* 2 */ { 2, s_1_2, -1, -1, 0},
|
109
|
+
/* 3 */ { 2, s_1_3, -1, -1, 0}
|
110
|
+
};
|
111
|
+
|
112
|
+
static const symbol s_2_0[2] = { 'i', 'g' };
|
113
|
+
static const symbol s_2_1[3] = { 'l', 'i', 'g' };
|
114
|
+
static const symbol s_2_2[4] = { 'e', 'l', 'i', 'g' };
|
115
|
+
static const symbol s_2_3[3] = { 'e', 'l', 's' };
|
116
|
+
static const symbol s_2_4[5] = { 'l', 0xC3, 0xB8, 's', 't' };
|
117
|
+
|
118
|
+
static const struct among a_2[5] =
|
119
|
+
{
|
120
|
+
/* 0 */ { 2, s_2_0, -1, 1, 0},
|
121
|
+
/* 1 */ { 3, s_2_1, 0, 1, 0},
|
122
|
+
/* 2 */ { 4, s_2_2, 1, 1, 0},
|
123
|
+
/* 3 */ { 3, s_2_3, -1, 1, 0},
|
124
|
+
/* 4 */ { 5, s_2_4, -1, 2, 0}
|
125
|
+
};
|
126
|
+
|
127
|
+
static const unsigned char g_v[] = { 17, 65, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 48, 0, 128 };
|
128
|
+
|
129
|
+
static const unsigned char g_s_ending[] = { 239, 254, 42, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 16 };
|
130
|
+
|
131
|
+
static const symbol s_0[] = { 's', 't' };
|
132
|
+
static const symbol s_1[] = { 'i', 'g' };
|
133
|
+
static const symbol s_2[] = { 'l', 0xC3, 0xB8, 's' };
|
134
|
+
|
135
|
+
static int r_mark_regions(struct SN_env * z) {
|
136
|
+
z->I[0] = z->l;
|
137
|
+
{ int c_test = z->c; /* test, line 33 */
|
138
|
+
{ int ret = skip_utf8(z->p, z->c, 0, z->l, + 3);
|
139
|
+
if (ret < 0) return 0;
|
140
|
+
z->c = ret; /* hop, line 33 */
|
141
|
+
}
|
142
|
+
z->I[1] = z->c; /* setmark x, line 33 */
|
143
|
+
z->c = c_test;
|
144
|
+
}
|
145
|
+
if (out_grouping_U(z, g_v, 97, 248, 1) < 0) return 0; /* goto */ /* grouping v, line 34 */
|
146
|
+
{ /* gopast */ /* non v, line 34 */
|
147
|
+
int ret = in_grouping_U(z, g_v, 97, 248, 1);
|
148
|
+
if (ret < 0) return 0;
|
149
|
+
z->c += ret;
|
150
|
+
}
|
151
|
+
z->I[0] = z->c; /* setmark p1, line 34 */
|
152
|
+
/* try, line 35 */
|
153
|
+
if (!(z->I[0] < z->I[1])) goto lab0;
|
154
|
+
z->I[0] = z->I[1];
|
155
|
+
lab0:
|
156
|
+
return 1;
|
157
|
+
}
|
158
|
+
|
159
|
+
static int r_main_suffix(struct SN_env * z) {
|
160
|
+
int among_var;
|
161
|
+
{ int mlimit; /* setlimit, line 41 */
|
162
|
+
int m1 = z->l - z->c; (void)m1;
|
163
|
+
if (z->c < z->I[0]) return 0;
|
164
|
+
z->c = z->I[0]; /* tomark, line 41 */
|
165
|
+
mlimit = z->lb; z->lb = z->c;
|
166
|
+
z->c = z->l - m1;
|
167
|
+
z->ket = z->c; /* [, line 41 */
|
168
|
+
if (z->c <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1851440 >> (z->p[z->c - 1] & 0x1f)) & 1)) { z->lb = mlimit; return 0; }
|
169
|
+
among_var = find_among_b(z, a_0, 32); /* substring, line 41 */
|
170
|
+
if (!(among_var)) { z->lb = mlimit; return 0; }
|
171
|
+
z->bra = z->c; /* ], line 41 */
|
172
|
+
z->lb = mlimit;
|
173
|
+
}
|
174
|
+
switch(among_var) {
|
175
|
+
case 0: return 0;
|
176
|
+
case 1:
|
177
|
+
{ int ret = slice_del(z); /* delete, line 48 */
|
178
|
+
if (ret < 0) return ret;
|
179
|
+
}
|
180
|
+
break;
|
181
|
+
case 2:
|
182
|
+
if (in_grouping_b_U(z, g_s_ending, 97, 229, 0)) return 0;
|
183
|
+
{ int ret = slice_del(z); /* delete, line 50 */
|
184
|
+
if (ret < 0) return ret;
|
185
|
+
}
|
186
|
+
break;
|
187
|
+
}
|
188
|
+
return 1;
|
189
|
+
}
|
190
|
+
|
191
|
+
static int r_consonant_pair(struct SN_env * z) {
|
192
|
+
{ int m_test = z->l - z->c; /* test, line 55 */
|
193
|
+
{ int mlimit; /* setlimit, line 56 */
|
194
|
+
int m1 = z->l - z->c; (void)m1;
|
195
|
+
if (z->c < z->I[0]) return 0;
|
196
|
+
z->c = z->I[0]; /* tomark, line 56 */
|
197
|
+
mlimit = z->lb; z->lb = z->c;
|
198
|
+
z->c = z->l - m1;
|
199
|
+
z->ket = z->c; /* [, line 56 */
|
200
|
+
if (z->c - 1 <= z->lb || (z->p[z->c - 1] != 100 && z->p[z->c - 1] != 116)) { z->lb = mlimit; return 0; }
|
201
|
+
if (!(find_among_b(z, a_1, 4))) { z->lb = mlimit; return 0; } /* substring, line 56 */
|
202
|
+
z->bra = z->c; /* ], line 56 */
|
203
|
+
z->lb = mlimit;
|
204
|
+
}
|
205
|
+
z->c = z->l - m_test;
|
206
|
+
}
|
207
|
+
{ int ret = skip_utf8(z->p, z->c, z->lb, 0, -1);
|
208
|
+
if (ret < 0) return 0;
|
209
|
+
z->c = ret; /* next, line 62 */
|
210
|
+
}
|
211
|
+
z->bra = z->c; /* ], line 62 */
|
212
|
+
{ int ret = slice_del(z); /* delete, line 62 */
|
213
|
+
if (ret < 0) return ret;
|
214
|
+
}
|
215
|
+
return 1;
|
216
|
+
}
|
217
|
+
|
218
|
+
static int r_other_suffix(struct SN_env * z) {
|
219
|
+
int among_var;
|
220
|
+
{ int m1 = z->l - z->c; (void)m1; /* do, line 66 */
|
221
|
+
z->ket = z->c; /* [, line 66 */
|
222
|
+
if (!(eq_s_b(z, 2, s_0))) goto lab0;
|
223
|
+
z->bra = z->c; /* ], line 66 */
|
224
|
+
if (!(eq_s_b(z, 2, s_1))) goto lab0;
|
225
|
+
{ int ret = slice_del(z); /* delete, line 66 */
|
226
|
+
if (ret < 0) return ret;
|
227
|
+
}
|
228
|
+
lab0:
|
229
|
+
z->c = z->l - m1;
|
230
|
+
}
|
231
|
+
{ int mlimit; /* setlimit, line 67 */
|
232
|
+
int m2 = z->l - z->c; (void)m2;
|
233
|
+
if (z->c < z->I[0]) return 0;
|
234
|
+
z->c = z->I[0]; /* tomark, line 67 */
|
235
|
+
mlimit = z->lb; z->lb = z->c;
|
236
|
+
z->c = z->l - m2;
|
237
|
+
z->ket = z->c; /* [, line 67 */
|
238
|
+
if (z->c - 1 <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1572992 >> (z->p[z->c - 1] & 0x1f)) & 1)) { z->lb = mlimit; return 0; }
|
239
|
+
among_var = find_among_b(z, a_2, 5); /* substring, line 67 */
|
240
|
+
if (!(among_var)) { z->lb = mlimit; return 0; }
|
241
|
+
z->bra = z->c; /* ], line 67 */
|
242
|
+
z->lb = mlimit;
|
243
|
+
}
|
244
|
+
switch(among_var) {
|
245
|
+
case 0: return 0;
|
246
|
+
case 1:
|
247
|
+
{ int ret = slice_del(z); /* delete, line 70 */
|
248
|
+
if (ret < 0) return ret;
|
249
|
+
}
|
250
|
+
{ int m3 = z->l - z->c; (void)m3; /* do, line 70 */
|
251
|
+
{ int ret = r_consonant_pair(z);
|
252
|
+
if (ret == 0) goto lab1; /* call consonant_pair, line 70 */
|
253
|
+
if (ret < 0) return ret;
|
254
|
+
}
|
255
|
+
lab1:
|
256
|
+
z->c = z->l - m3;
|
257
|
+
}
|
258
|
+
break;
|
259
|
+
case 2:
|
260
|
+
{ int ret = slice_from_s(z, 4, s_2); /* <-, line 72 */
|
261
|
+
if (ret < 0) return ret;
|
262
|
+
}
|
263
|
+
break;
|
264
|
+
}
|
265
|
+
return 1;
|
266
|
+
}
|
267
|
+
|
268
|
+
static int r_undouble(struct SN_env * z) {
|
269
|
+
{ int mlimit; /* setlimit, line 76 */
|
270
|
+
int m1 = z->l - z->c; (void)m1;
|
271
|
+
if (z->c < z->I[0]) return 0;
|
272
|
+
z->c = z->I[0]; /* tomark, line 76 */
|
273
|
+
mlimit = z->lb; z->lb = z->c;
|
274
|
+
z->c = z->l - m1;
|
275
|
+
z->ket = z->c; /* [, line 76 */
|
276
|
+
if (out_grouping_b_U(z, g_v, 97, 248, 0)) { z->lb = mlimit; return 0; }
|
277
|
+
z->bra = z->c; /* ], line 76 */
|
278
|
+
z->S[0] = slice_to(z, z->S[0]); /* -> ch, line 76 */
|
279
|
+
if (z->S[0] == 0) return -1; /* -> ch, line 76 */
|
280
|
+
z->lb = mlimit;
|
281
|
+
}
|
282
|
+
if (!(eq_v_b(z, z->S[0]))) return 0; /* name ch, line 77 */
|
283
|
+
{ int ret = slice_del(z); /* delete, line 78 */
|
284
|
+
if (ret < 0) return ret;
|
285
|
+
}
|
286
|
+
return 1;
|
287
|
+
}
|
288
|
+
|
289
|
+
extern int danish_UTF_8_stem(struct SN_env * z) {
|
290
|
+
{ int c1 = z->c; /* do, line 84 */
|
291
|
+
{ int ret = r_mark_regions(z);
|
292
|
+
if (ret == 0) goto lab0; /* call mark_regions, line 84 */
|
293
|
+
if (ret < 0) return ret;
|
294
|
+
}
|
295
|
+
lab0:
|
296
|
+
z->c = c1;
|
297
|
+
}
|
298
|
+
z->lb = z->c; z->c = z->l; /* backwards, line 85 */
|
299
|
+
|
300
|
+
{ int m2 = z->l - z->c; (void)m2; /* do, line 86 */
|
301
|
+
{ int ret = r_main_suffix(z);
|
302
|
+
if (ret == 0) goto lab1; /* call main_suffix, line 86 */
|
303
|
+
if (ret < 0) return ret;
|
304
|
+
}
|
305
|
+
lab1:
|
306
|
+
z->c = z->l - m2;
|
307
|
+
}
|
308
|
+
{ int m3 = z->l - z->c; (void)m3; /* do, line 87 */
|
309
|
+
{ int ret = r_consonant_pair(z);
|
310
|
+
if (ret == 0) goto lab2; /* call consonant_pair, line 87 */
|
311
|
+
if (ret < 0) return ret;
|
312
|
+
}
|
313
|
+
lab2:
|
314
|
+
z->c = z->l - m3;
|
315
|
+
}
|
316
|
+
{ int m4 = z->l - z->c; (void)m4; /* do, line 88 */
|
317
|
+
{ int ret = r_other_suffix(z);
|
318
|
+
if (ret == 0) goto lab3; /* call other_suffix, line 88 */
|
319
|
+
if (ret < 0) return ret;
|
320
|
+
}
|
321
|
+
lab3:
|
322
|
+
z->c = z->l - m4;
|
323
|
+
}
|
324
|
+
{ int m5 = z->l - z->c; (void)m5; /* do, line 89 */
|
325
|
+
{ int ret = r_undouble(z);
|
326
|
+
if (ret == 0) goto lab4; /* call undouble, line 89 */
|
327
|
+
if (ret < 0) return ret;
|
328
|
+
}
|
329
|
+
lab4:
|
330
|
+
z->c = z->l - m5;
|
331
|
+
}
|
332
|
+
z->c = z->lb;
|
333
|
+
return 1;
|
334
|
+
}
|
335
|
+
|
336
|
+
extern struct SN_env * danish_UTF_8_create_env(void) { return SN_create_env(1, 2, 0); }
|
337
|
+
|
338
|
+
extern void danish_UTF_8_close_env(struct SN_env * z) { SN_close_env(z, 1); }
|
339
|
+
|
@@ -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 * danish_UTF_8_create_env(void);
|
9
|
+
extern void danish_UTF_8_close_env(struct SN_env * z);
|
10
|
+
|
11
|
+
extern int danish_UTF_8_stem(struct SN_env * z);
|
12
|
+
|
13
|
+
#ifdef __cplusplus
|
14
|
+
}
|
15
|
+
#endif
|
16
|
+
|
@@ -0,0 +1,634 @@
|
|
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 dutch_UTF_8_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_undouble(struct SN_env * z);
|
15
|
+
static int r_R2(struct SN_env * z);
|
16
|
+
static int r_R1(struct SN_env * z);
|
17
|
+
static int r_mark_regions(struct SN_env * z);
|
18
|
+
static int r_en_ending(struct SN_env * z);
|
19
|
+
static int r_e_ending(struct SN_env * z);
|
20
|
+
static int r_postlude(struct SN_env * z);
|
21
|
+
static int r_prelude(struct SN_env * z);
|
22
|
+
#ifdef __cplusplus
|
23
|
+
extern "C" {
|
24
|
+
#endif
|
25
|
+
|
26
|
+
|
27
|
+
extern struct SN_env * dutch_UTF_8_create_env(void);
|
28
|
+
extern void dutch_UTF_8_close_env(struct SN_env * z);
|
29
|
+
|
30
|
+
|
31
|
+
#ifdef __cplusplus
|
32
|
+
}
|
33
|
+
#endif
|
34
|
+
static const symbol s_0_1[2] = { 0xC3, 0xA1 };
|
35
|
+
static const symbol s_0_2[2] = { 0xC3, 0xA4 };
|
36
|
+
static const symbol s_0_3[2] = { 0xC3, 0xA9 };
|
37
|
+
static const symbol s_0_4[2] = { 0xC3, 0xAB };
|
38
|
+
static const symbol s_0_5[2] = { 0xC3, 0xAD };
|
39
|
+
static const symbol s_0_6[2] = { 0xC3, 0xAF };
|
40
|
+
static const symbol s_0_7[2] = { 0xC3, 0xB3 };
|
41
|
+
static const symbol s_0_8[2] = { 0xC3, 0xB6 };
|
42
|
+
static const symbol s_0_9[2] = { 0xC3, 0xBA };
|
43
|
+
static const symbol s_0_10[2] = { 0xC3, 0xBC };
|
44
|
+
|
45
|
+
static const struct among a_0[11] =
|
46
|
+
{
|
47
|
+
/* 0 */ { 0, 0, -1, 6, 0},
|
48
|
+
/* 1 */ { 2, s_0_1, 0, 1, 0},
|
49
|
+
/* 2 */ { 2, s_0_2, 0, 1, 0},
|
50
|
+
/* 3 */ { 2, s_0_3, 0, 2, 0},
|
51
|
+
/* 4 */ { 2, s_0_4, 0, 2, 0},
|
52
|
+
/* 5 */ { 2, s_0_5, 0, 3, 0},
|
53
|
+
/* 6 */ { 2, s_0_6, 0, 3, 0},
|
54
|
+
/* 7 */ { 2, s_0_7, 0, 4, 0},
|
55
|
+
/* 8 */ { 2, s_0_8, 0, 4, 0},
|
56
|
+
/* 9 */ { 2, s_0_9, 0, 5, 0},
|
57
|
+
/* 10 */ { 2, s_0_10, 0, 5, 0}
|
58
|
+
};
|
59
|
+
|
60
|
+
static const symbol s_1_1[1] = { 'I' };
|
61
|
+
static const symbol s_1_2[1] = { 'Y' };
|
62
|
+
|
63
|
+
static const struct among a_1[3] =
|
64
|
+
{
|
65
|
+
/* 0 */ { 0, 0, -1, 3, 0},
|
66
|
+
/* 1 */ { 1, s_1_1, 0, 2, 0},
|
67
|
+
/* 2 */ { 1, s_1_2, 0, 1, 0}
|
68
|
+
};
|
69
|
+
|
70
|
+
static const symbol s_2_0[2] = { 'd', 'd' };
|
71
|
+
static const symbol s_2_1[2] = { 'k', 'k' };
|
72
|
+
static const symbol s_2_2[2] = { 't', 't' };
|
73
|
+
|
74
|
+
static const struct among a_2[3] =
|
75
|
+
{
|
76
|
+
/* 0 */ { 2, s_2_0, -1, -1, 0},
|
77
|
+
/* 1 */ { 2, s_2_1, -1, -1, 0},
|
78
|
+
/* 2 */ { 2, s_2_2, -1, -1, 0}
|
79
|
+
};
|
80
|
+
|
81
|
+
static const symbol s_3_0[3] = { 'e', 'n', 'e' };
|
82
|
+
static const symbol s_3_1[2] = { 's', 'e' };
|
83
|
+
static const symbol s_3_2[2] = { 'e', 'n' };
|
84
|
+
static const symbol s_3_3[5] = { 'h', 'e', 'd', 'e', 'n' };
|
85
|
+
static const symbol s_3_4[1] = { 's' };
|
86
|
+
|
87
|
+
static const struct among a_3[5] =
|
88
|
+
{
|
89
|
+
/* 0 */ { 3, s_3_0, -1, 2, 0},
|
90
|
+
/* 1 */ { 2, s_3_1, -1, 3, 0},
|
91
|
+
/* 2 */ { 2, s_3_2, -1, 2, 0},
|
92
|
+
/* 3 */ { 5, s_3_3, 2, 1, 0},
|
93
|
+
/* 4 */ { 1, s_3_4, -1, 3, 0}
|
94
|
+
};
|
95
|
+
|
96
|
+
static const symbol s_4_0[3] = { 'e', 'n', 'd' };
|
97
|
+
static const symbol s_4_1[2] = { 'i', 'g' };
|
98
|
+
static const symbol s_4_2[3] = { 'i', 'n', 'g' };
|
99
|
+
static const symbol s_4_3[4] = { 'l', 'i', 'j', 'k' };
|
100
|
+
static const symbol s_4_4[4] = { 'b', 'a', 'a', 'r' };
|
101
|
+
static const symbol s_4_5[3] = { 'b', 'a', 'r' };
|
102
|
+
|
103
|
+
static const struct among a_4[6] =
|
104
|
+
{
|
105
|
+
/* 0 */ { 3, s_4_0, -1, 1, 0},
|
106
|
+
/* 1 */ { 2, s_4_1, -1, 2, 0},
|
107
|
+
/* 2 */ { 3, s_4_2, -1, 1, 0},
|
108
|
+
/* 3 */ { 4, s_4_3, -1, 3, 0},
|
109
|
+
/* 4 */ { 4, s_4_4, -1, 4, 0},
|
110
|
+
/* 5 */ { 3, s_4_5, -1, 5, 0}
|
111
|
+
};
|
112
|
+
|
113
|
+
static const symbol s_5_0[2] = { 'a', 'a' };
|
114
|
+
static const symbol s_5_1[2] = { 'e', 'e' };
|
115
|
+
static const symbol s_5_2[2] = { 'o', 'o' };
|
116
|
+
static const symbol s_5_3[2] = { 'u', 'u' };
|
117
|
+
|
118
|
+
static const struct among a_5[4] =
|
119
|
+
{
|
120
|
+
/* 0 */ { 2, s_5_0, -1, -1, 0},
|
121
|
+
/* 1 */ { 2, s_5_1, -1, -1, 0},
|
122
|
+
/* 2 */ { 2, s_5_2, -1, -1, 0},
|
123
|
+
/* 3 */ { 2, s_5_3, -1, -1, 0}
|
124
|
+
};
|
125
|
+
|
126
|
+
static const unsigned char g_v[] = { 17, 65, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128 };
|
127
|
+
|
128
|
+
static const unsigned char g_v_I[] = { 1, 0, 0, 17, 65, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128 };
|
129
|
+
|
130
|
+
static const unsigned char g_v_j[] = { 17, 67, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128 };
|
131
|
+
|
132
|
+
static const symbol s_0[] = { 'a' };
|
133
|
+
static const symbol s_1[] = { 'e' };
|
134
|
+
static const symbol s_2[] = { 'i' };
|
135
|
+
static const symbol s_3[] = { 'o' };
|
136
|
+
static const symbol s_4[] = { 'u' };
|
137
|
+
static const symbol s_5[] = { 'y' };
|
138
|
+
static const symbol s_6[] = { 'Y' };
|
139
|
+
static const symbol s_7[] = { 'i' };
|
140
|
+
static const symbol s_8[] = { 'I' };
|
141
|
+
static const symbol s_9[] = { 'y' };
|
142
|
+
static const symbol s_10[] = { 'Y' };
|
143
|
+
static const symbol s_11[] = { 'y' };
|
144
|
+
static const symbol s_12[] = { 'i' };
|
145
|
+
static const symbol s_13[] = { 'e' };
|
146
|
+
static const symbol s_14[] = { 'g', 'e', 'm' };
|
147
|
+
static const symbol s_15[] = { 'h', 'e', 'i', 'd' };
|
148
|
+
static const symbol s_16[] = { 'h', 'e', 'i', 'd' };
|
149
|
+
static const symbol s_17[] = { 'c' };
|
150
|
+
static const symbol s_18[] = { 'e', 'n' };
|
151
|
+
static const symbol s_19[] = { 'i', 'g' };
|
152
|
+
static const symbol s_20[] = { 'e' };
|
153
|
+
static const symbol s_21[] = { 'e' };
|
154
|
+
|
155
|
+
static int r_prelude(struct SN_env * z) {
|
156
|
+
int among_var;
|
157
|
+
{ int c_test = z->c; /* test, line 42 */
|
158
|
+
while(1) { /* repeat, line 42 */
|
159
|
+
int c1 = z->c;
|
160
|
+
z->bra = z->c; /* [, line 43 */
|
161
|
+
if (z->c + 1 >= z->l || z->p[z->c + 1] >> 5 != 5 || !((340306450 >> (z->p[z->c + 1] & 0x1f)) & 1)) among_var = 6; else
|
162
|
+
among_var = find_among(z, a_0, 11); /* substring, line 43 */
|
163
|
+
if (!(among_var)) goto lab0;
|
164
|
+
z->ket = z->c; /* ], line 43 */
|
165
|
+
switch(among_var) {
|
166
|
+
case 0: goto lab0;
|
167
|
+
case 1:
|
168
|
+
{ int ret = slice_from_s(z, 1, s_0); /* <-, line 45 */
|
169
|
+
if (ret < 0) return ret;
|
170
|
+
}
|
171
|
+
break;
|
172
|
+
case 2:
|
173
|
+
{ int ret = slice_from_s(z, 1, s_1); /* <-, line 47 */
|
174
|
+
if (ret < 0) return ret;
|
175
|
+
}
|
176
|
+
break;
|
177
|
+
case 3:
|
178
|
+
{ int ret = slice_from_s(z, 1, s_2); /* <-, line 49 */
|
179
|
+
if (ret < 0) return ret;
|
180
|
+
}
|
181
|
+
break;
|
182
|
+
case 4:
|
183
|
+
{ int ret = slice_from_s(z, 1, s_3); /* <-, line 51 */
|
184
|
+
if (ret < 0) return ret;
|
185
|
+
}
|
186
|
+
break;
|
187
|
+
case 5:
|
188
|
+
{ int ret = slice_from_s(z, 1, s_4); /* <-, line 53 */
|
189
|
+
if (ret < 0) return ret;
|
190
|
+
}
|
191
|
+
break;
|
192
|
+
case 6:
|
193
|
+
{ int ret = skip_utf8(z->p, z->c, 0, z->l, 1);
|
194
|
+
if (ret < 0) goto lab0;
|
195
|
+
z->c = ret; /* next, line 54 */
|
196
|
+
}
|
197
|
+
break;
|
198
|
+
}
|
199
|
+
continue;
|
200
|
+
lab0:
|
201
|
+
z->c = c1;
|
202
|
+
break;
|
203
|
+
}
|
204
|
+
z->c = c_test;
|
205
|
+
}
|
206
|
+
{ int c_keep = z->c; /* try, line 57 */
|
207
|
+
z->bra = z->c; /* [, line 57 */
|
208
|
+
if (!(eq_s(z, 1, s_5))) { z->c = c_keep; goto lab1; }
|
209
|
+
z->ket = z->c; /* ], line 57 */
|
210
|
+
{ int ret = slice_from_s(z, 1, s_6); /* <-, line 57 */
|
211
|
+
if (ret < 0) return ret;
|
212
|
+
}
|
213
|
+
lab1:
|
214
|
+
;
|
215
|
+
}
|
216
|
+
while(1) { /* repeat, line 58 */
|
217
|
+
int c2 = z->c;
|
218
|
+
while(1) { /* goto, line 58 */
|
219
|
+
int c3 = z->c;
|
220
|
+
if (in_grouping_U(z, g_v, 97, 232, 0)) goto lab3;
|
221
|
+
z->bra = z->c; /* [, line 59 */
|
222
|
+
{ int c4 = z->c; /* or, line 59 */
|
223
|
+
if (!(eq_s(z, 1, s_7))) goto lab5;
|
224
|
+
z->ket = z->c; /* ], line 59 */
|
225
|
+
if (in_grouping_U(z, g_v, 97, 232, 0)) goto lab5;
|
226
|
+
{ int ret = slice_from_s(z, 1, s_8); /* <-, line 59 */
|
227
|
+
if (ret < 0) return ret;
|
228
|
+
}
|
229
|
+
goto lab4;
|
230
|
+
lab5:
|
231
|
+
z->c = c4;
|
232
|
+
if (!(eq_s(z, 1, s_9))) goto lab3;
|
233
|
+
z->ket = z->c; /* ], line 60 */
|
234
|
+
{ int ret = slice_from_s(z, 1, s_10); /* <-, line 60 */
|
235
|
+
if (ret < 0) return ret;
|
236
|
+
}
|
237
|
+
}
|
238
|
+
lab4:
|
239
|
+
z->c = c3;
|
240
|
+
break;
|
241
|
+
lab3:
|
242
|
+
z->c = c3;
|
243
|
+
{ int ret = skip_utf8(z->p, z->c, 0, z->l, 1);
|
244
|
+
if (ret < 0) goto lab2;
|
245
|
+
z->c = ret; /* goto, line 58 */
|
246
|
+
}
|
247
|
+
}
|
248
|
+
continue;
|
249
|
+
lab2:
|
250
|
+
z->c = c2;
|
251
|
+
break;
|
252
|
+
}
|
253
|
+
return 1;
|
254
|
+
}
|
255
|
+
|
256
|
+
static int r_mark_regions(struct SN_env * z) {
|
257
|
+
z->I[0] = z->l;
|
258
|
+
z->I[1] = z->l;
|
259
|
+
{ /* gopast */ /* grouping v, line 69 */
|
260
|
+
int ret = out_grouping_U(z, g_v, 97, 232, 1);
|
261
|
+
if (ret < 0) return 0;
|
262
|
+
z->c += ret;
|
263
|
+
}
|
264
|
+
{ /* gopast */ /* non v, line 69 */
|
265
|
+
int ret = in_grouping_U(z, g_v, 97, 232, 1);
|
266
|
+
if (ret < 0) return 0;
|
267
|
+
z->c += ret;
|
268
|
+
}
|
269
|
+
z->I[0] = z->c; /* setmark p1, line 69 */
|
270
|
+
/* try, line 70 */
|
271
|
+
if (!(z->I[0] < 3)) goto lab0;
|
272
|
+
z->I[0] = 3;
|
273
|
+
lab0:
|
274
|
+
{ /* gopast */ /* grouping v, line 71 */
|
275
|
+
int ret = out_grouping_U(z, g_v, 97, 232, 1);
|
276
|
+
if (ret < 0) return 0;
|
277
|
+
z->c += ret;
|
278
|
+
}
|
279
|
+
{ /* gopast */ /* non v, line 71 */
|
280
|
+
int ret = in_grouping_U(z, g_v, 97, 232, 1);
|
281
|
+
if (ret < 0) return 0;
|
282
|
+
z->c += ret;
|
283
|
+
}
|
284
|
+
z->I[1] = z->c; /* setmark p2, line 71 */
|
285
|
+
return 1;
|
286
|
+
}
|
287
|
+
|
288
|
+
static int r_postlude(struct SN_env * z) {
|
289
|
+
int among_var;
|
290
|
+
while(1) { /* repeat, line 75 */
|
291
|
+
int c1 = z->c;
|
292
|
+
z->bra = z->c; /* [, line 77 */
|
293
|
+
if (z->c >= z->l || (z->p[z->c + 0] != 73 && z->p[z->c + 0] != 89)) among_var = 3; else
|
294
|
+
among_var = find_among(z, a_1, 3); /* substring, line 77 */
|
295
|
+
if (!(among_var)) goto lab0;
|
296
|
+
z->ket = z->c; /* ], line 77 */
|
297
|
+
switch(among_var) {
|
298
|
+
case 0: goto lab0;
|
299
|
+
case 1:
|
300
|
+
{ int ret = slice_from_s(z, 1, s_11); /* <-, line 78 */
|
301
|
+
if (ret < 0) return ret;
|
302
|
+
}
|
303
|
+
break;
|
304
|
+
case 2:
|
305
|
+
{ int ret = slice_from_s(z, 1, s_12); /* <-, line 79 */
|
306
|
+
if (ret < 0) return ret;
|
307
|
+
}
|
308
|
+
break;
|
309
|
+
case 3:
|
310
|
+
{ int ret = skip_utf8(z->p, z->c, 0, z->l, 1);
|
311
|
+
if (ret < 0) goto lab0;
|
312
|
+
z->c = ret; /* next, line 80 */
|
313
|
+
}
|
314
|
+
break;
|
315
|
+
}
|
316
|
+
continue;
|
317
|
+
lab0:
|
318
|
+
z->c = c1;
|
319
|
+
break;
|
320
|
+
}
|
321
|
+
return 1;
|
322
|
+
}
|
323
|
+
|
324
|
+
static int r_R1(struct SN_env * z) {
|
325
|
+
if (!(z->I[0] <= z->c)) return 0;
|
326
|
+
return 1;
|
327
|
+
}
|
328
|
+
|
329
|
+
static int r_R2(struct SN_env * z) {
|
330
|
+
if (!(z->I[1] <= z->c)) return 0;
|
331
|
+
return 1;
|
332
|
+
}
|
333
|
+
|
334
|
+
static int r_undouble(struct SN_env * z) {
|
335
|
+
{ int m_test = z->l - z->c; /* test, line 91 */
|
336
|
+
if (z->c - 1 <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1050640 >> (z->p[z->c - 1] & 0x1f)) & 1)) return 0;
|
337
|
+
if (!(find_among_b(z, a_2, 3))) return 0; /* among, line 91 */
|
338
|
+
z->c = z->l - m_test;
|
339
|
+
}
|
340
|
+
z->ket = z->c; /* [, line 91 */
|
341
|
+
{ int ret = skip_utf8(z->p, z->c, z->lb, 0, -1);
|
342
|
+
if (ret < 0) return 0;
|
343
|
+
z->c = ret; /* next, line 91 */
|
344
|
+
}
|
345
|
+
z->bra = z->c; /* ], line 91 */
|
346
|
+
{ int ret = slice_del(z); /* delete, line 91 */
|
347
|
+
if (ret < 0) return ret;
|
348
|
+
}
|
349
|
+
return 1;
|
350
|
+
}
|
351
|
+
|
352
|
+
static int r_e_ending(struct SN_env * z) {
|
353
|
+
z->B[0] = 0; /* unset e_found, line 95 */
|
354
|
+
z->ket = z->c; /* [, line 96 */
|
355
|
+
if (!(eq_s_b(z, 1, s_13))) return 0;
|
356
|
+
z->bra = z->c; /* ], line 96 */
|
357
|
+
{ int ret = r_R1(z);
|
358
|
+
if (ret == 0) return 0; /* call R1, line 96 */
|
359
|
+
if (ret < 0) return ret;
|
360
|
+
}
|
361
|
+
{ int m_test = z->l - z->c; /* test, line 96 */
|
362
|
+
if (out_grouping_b_U(z, g_v, 97, 232, 0)) return 0;
|
363
|
+
z->c = z->l - m_test;
|
364
|
+
}
|
365
|
+
{ int ret = slice_del(z); /* delete, line 96 */
|
366
|
+
if (ret < 0) return ret;
|
367
|
+
}
|
368
|
+
z->B[0] = 1; /* set e_found, line 97 */
|
369
|
+
{ int ret = r_undouble(z);
|
370
|
+
if (ret == 0) return 0; /* call undouble, line 98 */
|
371
|
+
if (ret < 0) return ret;
|
372
|
+
}
|
373
|
+
return 1;
|
374
|
+
}
|
375
|
+
|
376
|
+
static int r_en_ending(struct SN_env * z) {
|
377
|
+
{ int ret = r_R1(z);
|
378
|
+
if (ret == 0) return 0; /* call R1, line 102 */
|
379
|
+
if (ret < 0) return ret;
|
380
|
+
}
|
381
|
+
{ int m1 = z->l - z->c; (void)m1; /* and, line 102 */
|
382
|
+
if (out_grouping_b_U(z, g_v, 97, 232, 0)) return 0;
|
383
|
+
z->c = z->l - m1;
|
384
|
+
{ int m2 = z->l - z->c; (void)m2; /* not, line 102 */
|
385
|
+
if (!(eq_s_b(z, 3, s_14))) goto lab0;
|
386
|
+
return 0;
|
387
|
+
lab0:
|
388
|
+
z->c = z->l - m2;
|
389
|
+
}
|
390
|
+
}
|
391
|
+
{ int ret = slice_del(z); /* delete, line 102 */
|
392
|
+
if (ret < 0) return ret;
|
393
|
+
}
|
394
|
+
{ int ret = r_undouble(z);
|
395
|
+
if (ret == 0) return 0; /* call undouble, line 103 */
|
396
|
+
if (ret < 0) return ret;
|
397
|
+
}
|
398
|
+
return 1;
|
399
|
+
}
|
400
|
+
|
401
|
+
static int r_standard_suffix(struct SN_env * z) {
|
402
|
+
int among_var;
|
403
|
+
{ int m1 = z->l - z->c; (void)m1; /* do, line 107 */
|
404
|
+
z->ket = z->c; /* [, line 108 */
|
405
|
+
if (z->c <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((540704 >> (z->p[z->c - 1] & 0x1f)) & 1)) goto lab0;
|
406
|
+
among_var = find_among_b(z, a_3, 5); /* substring, line 108 */
|
407
|
+
if (!(among_var)) goto lab0;
|
408
|
+
z->bra = z->c; /* ], line 108 */
|
409
|
+
switch(among_var) {
|
410
|
+
case 0: goto lab0;
|
411
|
+
case 1:
|
412
|
+
{ int ret = r_R1(z);
|
413
|
+
if (ret == 0) goto lab0; /* call R1, line 110 */
|
414
|
+
if (ret < 0) return ret;
|
415
|
+
}
|
416
|
+
{ int ret = slice_from_s(z, 4, s_15); /* <-, line 110 */
|
417
|
+
if (ret < 0) return ret;
|
418
|
+
}
|
419
|
+
break;
|
420
|
+
case 2:
|
421
|
+
{ int ret = r_en_ending(z);
|
422
|
+
if (ret == 0) goto lab0; /* call en_ending, line 113 */
|
423
|
+
if (ret < 0) return ret;
|
424
|
+
}
|
425
|
+
break;
|
426
|
+
case 3:
|
427
|
+
{ int ret = r_R1(z);
|
428
|
+
if (ret == 0) goto lab0; /* call R1, line 116 */
|
429
|
+
if (ret < 0) return ret;
|
430
|
+
}
|
431
|
+
if (out_grouping_b_U(z, g_v_j, 97, 232, 0)) goto lab0;
|
432
|
+
{ int ret = slice_del(z); /* delete, line 116 */
|
433
|
+
if (ret < 0) return ret;
|
434
|
+
}
|
435
|
+
break;
|
436
|
+
}
|
437
|
+
lab0:
|
438
|
+
z->c = z->l - m1;
|
439
|
+
}
|
440
|
+
{ int m2 = z->l - z->c; (void)m2; /* do, line 120 */
|
441
|
+
{ int ret = r_e_ending(z);
|
442
|
+
if (ret == 0) goto lab1; /* call e_ending, line 120 */
|
443
|
+
if (ret < 0) return ret;
|
444
|
+
}
|
445
|
+
lab1:
|
446
|
+
z->c = z->l - m2;
|
447
|
+
}
|
448
|
+
{ int m3 = z->l - z->c; (void)m3; /* do, line 122 */
|
449
|
+
z->ket = z->c; /* [, line 122 */
|
450
|
+
if (!(eq_s_b(z, 4, s_16))) goto lab2;
|
451
|
+
z->bra = z->c; /* ], line 122 */
|
452
|
+
{ int ret = r_R2(z);
|
453
|
+
if (ret == 0) goto lab2; /* call R2, line 122 */
|
454
|
+
if (ret < 0) return ret;
|
455
|
+
}
|
456
|
+
{ int m4 = z->l - z->c; (void)m4; /* not, line 122 */
|
457
|
+
if (!(eq_s_b(z, 1, s_17))) goto lab3;
|
458
|
+
goto lab2;
|
459
|
+
lab3:
|
460
|
+
z->c = z->l - m4;
|
461
|
+
}
|
462
|
+
{ int ret = slice_del(z); /* delete, line 122 */
|
463
|
+
if (ret < 0) return ret;
|
464
|
+
}
|
465
|
+
z->ket = z->c; /* [, line 123 */
|
466
|
+
if (!(eq_s_b(z, 2, s_18))) goto lab2;
|
467
|
+
z->bra = z->c; /* ], line 123 */
|
468
|
+
{ int ret = r_en_ending(z);
|
469
|
+
if (ret == 0) goto lab2; /* call en_ending, line 123 */
|
470
|
+
if (ret < 0) return ret;
|
471
|
+
}
|
472
|
+
lab2:
|
473
|
+
z->c = z->l - m3;
|
474
|
+
}
|
475
|
+
{ int m5 = z->l - z->c; (void)m5; /* do, line 126 */
|
476
|
+
z->ket = z->c; /* [, line 127 */
|
477
|
+
if (z->c - 1 <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((264336 >> (z->p[z->c - 1] & 0x1f)) & 1)) goto lab4;
|
478
|
+
among_var = find_among_b(z, a_4, 6); /* substring, line 127 */
|
479
|
+
if (!(among_var)) goto lab4;
|
480
|
+
z->bra = z->c; /* ], line 127 */
|
481
|
+
switch(among_var) {
|
482
|
+
case 0: goto lab4;
|
483
|
+
case 1:
|
484
|
+
{ int ret = r_R2(z);
|
485
|
+
if (ret == 0) goto lab4; /* call R2, line 129 */
|
486
|
+
if (ret < 0) return ret;
|
487
|
+
}
|
488
|
+
{ int ret = slice_del(z); /* delete, line 129 */
|
489
|
+
if (ret < 0) return ret;
|
490
|
+
}
|
491
|
+
{ int m6 = z->l - z->c; (void)m6; /* or, line 130 */
|
492
|
+
z->ket = z->c; /* [, line 130 */
|
493
|
+
if (!(eq_s_b(z, 2, s_19))) goto lab6;
|
494
|
+
z->bra = z->c; /* ], line 130 */
|
495
|
+
{ int ret = r_R2(z);
|
496
|
+
if (ret == 0) goto lab6; /* call R2, line 130 */
|
497
|
+
if (ret < 0) return ret;
|
498
|
+
}
|
499
|
+
{ int m7 = z->l - z->c; (void)m7; /* not, line 130 */
|
500
|
+
if (!(eq_s_b(z, 1, s_20))) goto lab7;
|
501
|
+
goto lab6;
|
502
|
+
lab7:
|
503
|
+
z->c = z->l - m7;
|
504
|
+
}
|
505
|
+
{ int ret = slice_del(z); /* delete, line 130 */
|
506
|
+
if (ret < 0) return ret;
|
507
|
+
}
|
508
|
+
goto lab5;
|
509
|
+
lab6:
|
510
|
+
z->c = z->l - m6;
|
511
|
+
{ int ret = r_undouble(z);
|
512
|
+
if (ret == 0) goto lab4; /* call undouble, line 130 */
|
513
|
+
if (ret < 0) return ret;
|
514
|
+
}
|
515
|
+
}
|
516
|
+
lab5:
|
517
|
+
break;
|
518
|
+
case 2:
|
519
|
+
{ int ret = r_R2(z);
|
520
|
+
if (ret == 0) goto lab4; /* call R2, line 133 */
|
521
|
+
if (ret < 0) return ret;
|
522
|
+
}
|
523
|
+
{ int m8 = z->l - z->c; (void)m8; /* not, line 133 */
|
524
|
+
if (!(eq_s_b(z, 1, s_21))) goto lab8;
|
525
|
+
goto lab4;
|
526
|
+
lab8:
|
527
|
+
z->c = z->l - m8;
|
528
|
+
}
|
529
|
+
{ int ret = slice_del(z); /* delete, line 133 */
|
530
|
+
if (ret < 0) return ret;
|
531
|
+
}
|
532
|
+
break;
|
533
|
+
case 3:
|
534
|
+
{ int ret = r_R2(z);
|
535
|
+
if (ret == 0) goto lab4; /* call R2, line 136 */
|
536
|
+
if (ret < 0) return ret;
|
537
|
+
}
|
538
|
+
{ int ret = slice_del(z); /* delete, line 136 */
|
539
|
+
if (ret < 0) return ret;
|
540
|
+
}
|
541
|
+
{ int ret = r_e_ending(z);
|
542
|
+
if (ret == 0) goto lab4; /* call e_ending, line 136 */
|
543
|
+
if (ret < 0) return ret;
|
544
|
+
}
|
545
|
+
break;
|
546
|
+
case 4:
|
547
|
+
{ int ret = r_R2(z);
|
548
|
+
if (ret == 0) goto lab4; /* call R2, line 139 */
|
549
|
+
if (ret < 0) return ret;
|
550
|
+
}
|
551
|
+
{ int ret = slice_del(z); /* delete, line 139 */
|
552
|
+
if (ret < 0) return ret;
|
553
|
+
}
|
554
|
+
break;
|
555
|
+
case 5:
|
556
|
+
{ int ret = r_R2(z);
|
557
|
+
if (ret == 0) goto lab4; /* call R2, line 142 */
|
558
|
+
if (ret < 0) return ret;
|
559
|
+
}
|
560
|
+
if (!(z->B[0])) goto lab4; /* Boolean test e_found, line 142 */
|
561
|
+
{ int ret = slice_del(z); /* delete, line 142 */
|
562
|
+
if (ret < 0) return ret;
|
563
|
+
}
|
564
|
+
break;
|
565
|
+
}
|
566
|
+
lab4:
|
567
|
+
z->c = z->l - m5;
|
568
|
+
}
|
569
|
+
{ int m9 = z->l - z->c; (void)m9; /* do, line 146 */
|
570
|
+
if (out_grouping_b_U(z, g_v_I, 73, 232, 0)) goto lab9;
|
571
|
+
{ int m_test = z->l - z->c; /* test, line 148 */
|
572
|
+
if (z->c - 1 <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((2129954 >> (z->p[z->c - 1] & 0x1f)) & 1)) goto lab9;
|
573
|
+
if (!(find_among_b(z, a_5, 4))) goto lab9; /* among, line 149 */
|
574
|
+
if (out_grouping_b_U(z, g_v, 97, 232, 0)) goto lab9;
|
575
|
+
z->c = z->l - m_test;
|
576
|
+
}
|
577
|
+
z->ket = z->c; /* [, line 152 */
|
578
|
+
{ int ret = skip_utf8(z->p, z->c, z->lb, 0, -1);
|
579
|
+
if (ret < 0) goto lab9;
|
580
|
+
z->c = ret; /* next, line 152 */
|
581
|
+
}
|
582
|
+
z->bra = z->c; /* ], line 152 */
|
583
|
+
{ int ret = slice_del(z); /* delete, line 152 */
|
584
|
+
if (ret < 0) return ret;
|
585
|
+
}
|
586
|
+
lab9:
|
587
|
+
z->c = z->l - m9;
|
588
|
+
}
|
589
|
+
return 1;
|
590
|
+
}
|
591
|
+
|
592
|
+
extern int dutch_UTF_8_stem(struct SN_env * z) {
|
593
|
+
{ int c1 = z->c; /* do, line 159 */
|
594
|
+
{ int ret = r_prelude(z);
|
595
|
+
if (ret == 0) goto lab0; /* call prelude, line 159 */
|
596
|
+
if (ret < 0) return ret;
|
597
|
+
}
|
598
|
+
lab0:
|
599
|
+
z->c = c1;
|
600
|
+
}
|
601
|
+
{ int c2 = z->c; /* do, line 160 */
|
602
|
+
{ int ret = r_mark_regions(z);
|
603
|
+
if (ret == 0) goto lab1; /* call mark_regions, line 160 */
|
604
|
+
if (ret < 0) return ret;
|
605
|
+
}
|
606
|
+
lab1:
|
607
|
+
z->c = c2;
|
608
|
+
}
|
609
|
+
z->lb = z->c; z->c = z->l; /* backwards, line 161 */
|
610
|
+
|
611
|
+
{ int m3 = z->l - z->c; (void)m3; /* do, line 162 */
|
612
|
+
{ int ret = r_standard_suffix(z);
|
613
|
+
if (ret == 0) goto lab2; /* call standard_suffix, line 162 */
|
614
|
+
if (ret < 0) return ret;
|
615
|
+
}
|
616
|
+
lab2:
|
617
|
+
z->c = z->l - m3;
|
618
|
+
}
|
619
|
+
z->c = z->lb;
|
620
|
+
{ int c4 = z->c; /* do, line 163 */
|
621
|
+
{ int ret = r_postlude(z);
|
622
|
+
if (ret == 0) goto lab3; /* call postlude, line 163 */
|
623
|
+
if (ret < 0) return ret;
|
624
|
+
}
|
625
|
+
lab3:
|
626
|
+
z->c = c4;
|
627
|
+
}
|
628
|
+
return 1;
|
629
|
+
}
|
630
|
+
|
631
|
+
extern struct SN_env * dutch_UTF_8_create_env(void) { return SN_create_env(0, 2, 1); }
|
632
|
+
|
633
|
+
extern void dutch_UTF_8_close_env(struct SN_env * z) { SN_close_env(z, 0); }
|
634
|
+
|