PostgreSQL Source Code git master
Functions | Variables
stem_UTF_8_norwegian.c File Reference
#include "header.h"
Include dependency graph for stem_UTF_8_norwegian.c:

Go to the source code of this file.

Functions

int norwegian_UTF_8_stem (struct SN_env *z)
 
static int r_other_suffix (struct SN_env *z)
 
static int r_consonant_pair (struct SN_env *z)
 
static int r_main_suffix (struct SN_env *z)
 
static int r_mark_regions (struct SN_env *z)
 
struct SN_env * norwegian_UTF_8_create_env (void)
 
void norwegian_UTF_8_close_env (struct SN_env *z)
 

Variables

static const symbol s_0_0 [1] = { 'a' }
 
static const symbol s_0_1 [1] = { 'e' }
 
static const symbol s_0_2 [3] = { 'e', 'd', 'e' }
 
static const symbol s_0_3 [4] = { 'a', 'n', 'd', 'e' }
 
static const symbol s_0_4 [4] = { 'e', 'n', 'd', 'e' }
 
static const symbol s_0_5 [3] = { 'a', 'n', 'e' }
 
static const symbol s_0_6 [3] = { 'e', 'n', 'e' }
 
static const symbol s_0_7 [6] = { 'h', 'e', 't', 'e', 'n', 'e' }
 
static const symbol s_0_8 [4] = { 'e', 'r', 't', 'e' }
 
static const symbol s_0_9 [2] = { 'e', 'n' }
 
static const symbol s_0_10 [5] = { 'h', 'e', 't', 'e', 'n' }
 
static const symbol s_0_11 [2] = { 'a', 'r' }
 
static const symbol s_0_12 [2] = { 'e', 'r' }
 
static const symbol s_0_13 [5] = { 'h', 'e', 't', 'e', 'r' }
 
static const symbol s_0_14 [1] = { 's' }
 
static const symbol s_0_15 [2] = { 'a', 's' }
 
static const symbol s_0_16 [2] = { 'e', 's' }
 
static const symbol s_0_17 [4] = { 'e', 'd', 'e', 's' }
 
static const symbol s_0_18 [5] = { 'e', 'n', 'd', 'e', 's' }
 
static const symbol s_0_19 [4] = { 'e', 'n', 'e', 's' }
 
static const symbol s_0_20 [7] = { 'h', 'e', 't', 'e', 'n', 'e', 's' }
 
static const symbol s_0_21 [3] = { 'e', 'n', 's' }
 
static const symbol s_0_22 [6] = { 'h', 'e', 't', 'e', 'n', 's' }
 
static const symbol s_0_23 [3] = { 'e', 'r', 's' }
 
static const symbol s_0_24 [3] = { 'e', 't', 's' }
 
static const symbol s_0_25 [2] = { 'e', 't' }
 
static const symbol s_0_26 [3] = { 'h', 'e', 't' }
 
static const symbol s_0_27 [3] = { 'e', 'r', 't' }
 
static const symbol s_0_28 [3] = { 'a', 's', 't' }
 
static const struct among a_0 [29]
 
static const symbol s_1_0 [2] = { 'd', 't' }
 
static const symbol s_1_1 [2] = { 'v', 't' }
 
static const struct among a_1 [2]
 
static const symbol s_2_0 [3] = { 'l', 'e', 'g' }
 
static const symbol s_2_1 [4] = { 'e', 'l', 'e', 'g' }
 
static const symbol s_2_2 [2] = { 'i', 'g' }
 
static const symbol s_2_3 [3] = { 'e', 'i', 'g' }
 
static const symbol s_2_4 [3] = { 'l', 'i', 'g' }
 
static const symbol s_2_5 [4] = { 'e', 'l', 'i', 'g' }
 
static const symbol s_2_6 [3] = { 'e', 'l', 's' }
 
static const symbol s_2_7 [3] = { 'l', 'o', 'v' }
 
static const symbol s_2_8 [4] = { 'e', 'l', 'o', 'v' }
 
static const symbol s_2_9 [4] = { 's', 'l', 'o', 'v' }
 
static const symbol s_2_10 [7] = { 'h', 'e', 't', 's', 'l', 'o', 'v' }
 
static const struct among a_2 [11]
 
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 }
 
static const unsigned char g_s_ending [] = { 119, 125, 149, 1 }
 
static const symbol s_0 [] = { 'e', 'r' }
 

Function Documentation

◆ norwegian_UTF_8_close_env()

void norwegian_UTF_8_close_env ( struct SN_env *  z)

Definition at line 275 of file stem_UTF_8_norwegian.c.

275{ SN_close_env(z, 0); }
SN_close_env
void SN_close_env(struct SN_env *z, int S_size)
Definition: api.c:34

References SN_close_env().

◆ norwegian_UTF_8_create_env()

struct SN_env * norwegian_UTF_8_create_env ( void  )

Definition at line 273 of file stem_UTF_8_norwegian.c.

273{ return SN_create_env(0, 2); }
SN_create_env
struct SN_env * SN_create_env(int S_size, int I_size)
Definition: api.c:3

References SN_create_env().

◆ norwegian_UTF_8_stem()

int norwegian_UTF_8_stem ( struct SN_env *  z)

Definition at line 242 of file stem_UTF_8_norwegian.c.

242 {
243 { int c1 = z->c;
244 { int ret = r_mark_regions(z);
245 if (ret < 0) return ret;
246 }
247 z->c = c1;
248 }
249 z->lb = z->c; z->c = z->l;
250
251 { int m2 = z->l - z->c; (void)m2;
252 { int ret = r_main_suffix(z);
253 if (ret < 0) return ret;
254 }
255 z->c = z->l - m2;
256 }
257 { int m3 = z->l - z->c; (void)m3;
258 { int ret = r_consonant_pair(z);
259 if (ret < 0) return ret;
260 }
261 z->c = z->l - m3;
262 }
263 { int m4 = z->l - z->c; (void)m4;
264 { int ret = r_other_suffix(z);
265 if (ret < 0) return ret;
266 }
267 z->c = z->l - m4;
268 }
269 z->c = z->lb;
270 return 1;
271}
r_other_suffix
static int r_other_suffix(struct SN_env *z)
Definition: stem_UTF_8_norwegian.c:225
r_consonant_pair
static int r_consonant_pair(struct SN_env *z)
Definition: stem_UTF_8_norwegian.c:200
r_main_suffix
static int r_main_suffix(struct SN_env *z)
Definition: stem_UTF_8_norwegian.c:157
r_mark_regions
static int r_mark_regions(struct SN_env *z)
Definition: stem_UTF_8_norwegian.c:133
SN_env::lb
int lb
Definition: api.h:16
SN_env::c
int c
Definition: api.h:16
SN_env::l
int l
Definition: api.h:16

References SN_env::c, SN_env::l, SN_env::lb, r_consonant_pair(), r_main_suffix(), r_mark_regions(), and r_other_suffix().

◆ r_consonant_pair()

static int r_consonant_pair ( struct SN_env *  z)
static

Definition at line 200 of file stem_UTF_8_norwegian.c.

200 {
201 { int m_test1 = z->l - z->c;
202
203 { int mlimit2;
204 if (z->c < z->I[1]) return 0;
205 mlimit2 = z->lb; z->lb = z->I[1];
206 z->ket = z->c;
207 if (z->c - 1 <= z->lb || z->p[z->c - 1] != 116) { z->lb = mlimit2; return 0; }
208 if (!(find_among_b(z, a_1, 2))) { z->lb = mlimit2; return 0; }
209 z->bra = z->c;
210 z->lb = mlimit2;
211 }
212 z->c = z->l - m_test1;
213 }
214 { int ret = skip_b_utf8(z->p, z->c, z->lb, 1);
215 if (ret < 0) return 0;
216 z->c = ret;
217 }
218 z->bra = z->c;
219 { int ret = slice_del(z);
220 if (ret < 0) return ret;
221 }
222 return 1;
223}
a_1
static const struct among a_1[2]
Definition: stem_UTF_8_norwegian.c:94
SN_env::p
symbol * p
Definition: api.h:15
SN_env::I
int * I
Definition: api.h:18
SN_env::ket
int ket
Definition: api.h:16
SN_env::bra
int bra
Definition: api.h:16
find_among_b
int find_among_b(struct SN_env *z, const struct among *v, int v_size)
Definition: utilities.c:298
slice_del
int slice_del(struct SN_env *z)
Definition: utilities.c:431
skip_b_utf8
int skip_b_utf8(const symbol *p, int c, int limit, int n)
Definition: utilities.c:52

References a_1, SN_env::bra, SN_env::c, find_among_b(), SN_env::I, SN_env::ket, SN_env::l, SN_env::lb, SN_env::p, skip_b_utf8(), and slice_del().

Referenced by norwegian_UTF_8_stem().

◆ r_main_suffix()

static int r_main_suffix ( struct SN_env *  z)
static

Definition at line 157 of file stem_UTF_8_norwegian.c.

157 {
158 int among_var;
159
160 { int mlimit1;
161 if (z->c < z->I[1]) return 0;
162 mlimit1 = z->lb; z->lb = z->I[1];
163 z->ket = z->c;
164 if (z->c <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1851426 >> (z->p[z->c - 1] & 0x1f)) & 1)) { z->lb = mlimit1; return 0; }
165 among_var = find_among_b(z, a_0, 29);
166 if (!(among_var)) { z->lb = mlimit1; return 0; }
167 z->bra = z->c;
168 z->lb = mlimit1;
169 }
170 switch (among_var) {
171 case 1:
172 { int ret = slice_del(z);
173 if (ret < 0) return ret;
174 }
175 break;
176 case 2:
177 { int m2 = z->l - z->c; (void)m2;
178 if (in_grouping_b_U(z, g_s_ending, 98, 122, 0)) goto lab1;
179 goto lab0;
180 lab1:
181 z->c = z->l - m2;
182 if (z->c <= z->lb || z->p[z->c - 1] != 'k') return 0;
183 z->c--;
184 if (out_grouping_b_U(z, g_v, 97, 248, 0)) return 0;
185 }
186 lab0:
187 { int ret = slice_del(z);
188 if (ret < 0) return ret;
189 }
190 break;
191 case 3:
192 { int ret = slice_from_s(z, 2, s_0);
193 if (ret < 0) return ret;
194 }
195 break;
196 }
197 return 1;
198}
a_0
static const struct among a_0[29]
Definition: stem_UTF_8_norwegian.c:58
g_v
static const unsigned char g_v[]
Definition: stem_UTF_8_norwegian.c:127
g_s_ending
static const unsigned char g_s_ending[]
Definition: stem_UTF_8_norwegian.c:129
s_0
static const symbol s_0[]
Definition: stem_UTF_8_norwegian.c:131
in_grouping_b_U
int in_grouping_b_U(struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
Definition: utilities.c:129
out_grouping_b_U
int out_grouping_b_U(struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
Definition: utilities.c:153
slice_from_s
int slice_from_s(struct SN_env *z, int s_size, const symbol *s)
Definition: utilities.c:422

References a_0, SN_env::bra, SN_env::c, find_among_b(), g_s_ending, g_v, SN_env::I, in_grouping_b_U(), SN_env::ket, SN_env::l, SN_env::lb, out_grouping_b_U(), SN_env::p, s_0, slice_del(), and slice_from_s().

Referenced by norwegian_UTF_8_stem().

◆ r_mark_regions()

static int r_mark_regions ( struct SN_env *  z)
static

Definition at line 133 of file stem_UTF_8_norwegian.c.

133 {
134 z->I[1] = z->l;
135 { int c_test1 = z->c;
136 { int ret = skip_utf8(z->p, z->c, z->l, 3);
137 if (ret < 0) return 0;
138 z->c = ret;
139 }
140 z->I[0] = z->c;
141 z->c = c_test1;
142 }
143 if (out_grouping_U(z, g_v, 97, 248, 1) < 0) return 0;
144 {
145 int ret = in_grouping_U(z, g_v, 97, 248, 1);
146 if (ret < 0) return 0;
147 z->c += ret;
148 }
149 z->I[1] = z->c;
150
151 if (!(z->I[1] < z->I[0])) goto lab0;
152 z->I[1] = z->I[0];
153lab0:
154 return 1;
155}
out_grouping_U
int out_grouping_U(struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
Definition: utilities.c:141
in_grouping_U
int in_grouping_U(struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
Definition: utilities.c:117
skip_utf8
int skip_utf8(const symbol *p, int c, int limit, int n)
Definition: utilities.c:27

References SN_env::c, g_v, SN_env::I, in_grouping_U(), SN_env::l, out_grouping_U(), SN_env::p, and skip_utf8().

Referenced by norwegian_UTF_8_stem().

◆ r_other_suffix()

static int r_other_suffix ( struct SN_env *  z)
static

Definition at line 225 of file stem_UTF_8_norwegian.c.

225 {
226
227 { int mlimit1;
228 if (z->c < z->I[1]) return 0;
229 mlimit1 = z->lb; z->lb = z->I[1];
230 z->ket = z->c;
231 if (z->c - 1 <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((4718720 >> (z->p[z->c - 1] & 0x1f)) & 1)) { z->lb = mlimit1; return 0; }
232 if (!(find_among_b(z, a_2, 11))) { z->lb = mlimit1; return 0; }
233 z->bra = z->c;
234 z->lb = mlimit1;
235 }
236 { int ret = slice_del(z);
237 if (ret < 0) return ret;
238 }
239 return 1;
240}
a_2
static const struct among a_2[11]
Definition: stem_UTF_8_norwegian.c:112

References a_2, SN_env::bra, SN_env::c, find_among_b(), SN_env::I, SN_env::ket, SN_env::lb, SN_env::p, and slice_del().

Referenced by norwegian_UTF_8_stem().

Variable Documentation

◆ a_0

const struct among a_0[29]
static

Definition at line 58 of file stem_UTF_8_norwegian.c.

Referenced by r_main_suffix().

◆ a_1

const struct among a_1[2]
static
Initial value:
=
{
{ 2, s_1_0, -1, -1, 0},
{ 2, s_1_1, -1, -1, 0}
}
s_1_0
static const symbol s_1_0[2]
Definition: stem_UTF_8_norwegian.c:91
s_1_1
static const symbol s_1_1[2]
Definition: stem_UTF_8_norwegian.c:92

Definition at line 94 of file stem_UTF_8_norwegian.c.

Referenced by r_consonant_pair().

◆ a_2

const struct among a_2[11]
static
Initial value:
=
{
{ 3, s_2_0, -1, 1, 0},
{ 4, s_2_1, 0, 1, 0},
{ 2, s_2_2, -1, 1, 0},
{ 3, s_2_3, 2, 1, 0},
{ 3, s_2_4, 2, 1, 0},
{ 4, s_2_5, 4, 1, 0},
{ 3, s_2_6, -1, 1, 0},
{ 3, s_2_7, -1, 1, 0},
{ 4, s_2_8, 7, 1, 0},
{ 4, s_2_9, 7, 1, 0},
{ 7, s_2_10, 9, 1, 0}
}
s_2_2
static const symbol s_2_2[2]
Definition: stem_UTF_8_norwegian.c:102
s_2_3
static const symbol s_2_3[3]
Definition: stem_UTF_8_norwegian.c:103
s_2_10
static const symbol s_2_10[7]
Definition: stem_UTF_8_norwegian.c:110
s_2_7
static const symbol s_2_7[3]
Definition: stem_UTF_8_norwegian.c:107
s_2_1
static const symbol s_2_1[4]
Definition: stem_UTF_8_norwegian.c:101
s_2_0
static const symbol s_2_0[3]
Definition: stem_UTF_8_norwegian.c:100
s_2_4
static const symbol s_2_4[3]
Definition: stem_UTF_8_norwegian.c:104
s_2_9
static const symbol s_2_9[4]
Definition: stem_UTF_8_norwegian.c:109
s_2_6
static const symbol s_2_6[3]
Definition: stem_UTF_8_norwegian.c:106
s_2_8
static const symbol s_2_8[4]
Definition: stem_UTF_8_norwegian.c:108
s_2_5
static const symbol s_2_5[4]
Definition: stem_UTF_8_norwegian.c:105

Definition at line 112 of file stem_UTF_8_norwegian.c.

Referenced by r_other_suffix().

◆ g_s_ending

const unsigned char g_s_ending[] = { 119, 125, 149, 1 }
static

Definition at line 129 of file stem_UTF_8_norwegian.c.

Referenced by r_main_suffix().

◆ g_v

const unsigned char g_v[] = { 17, 65, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 48, 0, 128 }
static

Definition at line 127 of file stem_UTF_8_norwegian.c.

Referenced by r_main_suffix(), and r_mark_regions().

◆ s_0

const symbol s_0[] = { 'e', 'r' }
static

Definition at line 131 of file stem_UTF_8_norwegian.c.

Referenced by r_main_suffix().

◆ s_0_0

const symbol s_0_0[1] = { 'a' }
static

Definition at line 28 of file stem_UTF_8_norwegian.c.

◆ s_0_1

const symbol s_0_1[1] = { 'e' }
static

Definition at line 29 of file stem_UTF_8_norwegian.c.

◆ s_0_10

const symbol s_0_10[5] = { 'h', 'e', 't', 'e', 'n' }
static

Definition at line 38 of file stem_UTF_8_norwegian.c.

◆ s_0_11

const symbol s_0_11[2] = { 'a', 'r' }
static

Definition at line 39 of file stem_UTF_8_norwegian.c.

◆ s_0_12

const symbol s_0_12[2] = { 'e', 'r' }
static

Definition at line 40 of file stem_UTF_8_norwegian.c.

◆ s_0_13

const symbol s_0_13[5] = { 'h', 'e', 't', 'e', 'r' }
static

Definition at line 41 of file stem_UTF_8_norwegian.c.

◆ s_0_14

const symbol s_0_14[1] = { 's' }
static

Definition at line 42 of file stem_UTF_8_norwegian.c.

◆ s_0_15

const symbol s_0_15[2] = { 'a', 's' }
static

Definition at line 43 of file stem_UTF_8_norwegian.c.

◆ s_0_16

const symbol s_0_16[2] = { 'e', 's' }
static

Definition at line 44 of file stem_UTF_8_norwegian.c.

◆ s_0_17

const symbol s_0_17[4] = { 'e', 'd', 'e', 's' }
static

Definition at line 45 of file stem_UTF_8_norwegian.c.

◆ s_0_18

const symbol s_0_18[5] = { 'e', 'n', 'd', 'e', 's' }
static

Definition at line 46 of file stem_UTF_8_norwegian.c.

◆ s_0_19

const symbol s_0_19[4] = { 'e', 'n', 'e', 's' }
static

Definition at line 47 of file stem_UTF_8_norwegian.c.

◆ s_0_2

const symbol s_0_2[3] = { 'e', 'd', 'e' }
static

Definition at line 30 of file stem_UTF_8_norwegian.c.

◆ s_0_20

const symbol s_0_20[7] = { 'h', 'e', 't', 'e', 'n', 'e', 's' }
static

Definition at line 48 of file stem_UTF_8_norwegian.c.

◆ s_0_21

const symbol s_0_21[3] = { 'e', 'n', 's' }
static

Definition at line 49 of file stem_UTF_8_norwegian.c.

◆ s_0_22

const symbol s_0_22[6] = { 'h', 'e', 't', 'e', 'n', 's' }
static

Definition at line 50 of file stem_UTF_8_norwegian.c.

◆ s_0_23

const symbol s_0_23[3] = { 'e', 'r', 's' }
static

Definition at line 51 of file stem_UTF_8_norwegian.c.

◆ s_0_24

const symbol s_0_24[3] = { 'e', 't', 's' }
static

Definition at line 52 of file stem_UTF_8_norwegian.c.

◆ s_0_25

const symbol s_0_25[2] = { 'e', 't' }
static

Definition at line 53 of file stem_UTF_8_norwegian.c.

◆ s_0_26

const symbol s_0_26[3] = { 'h', 'e', 't' }
static

Definition at line 54 of file stem_UTF_8_norwegian.c.

◆ s_0_27

const symbol s_0_27[3] = { 'e', 'r', 't' }
static

Definition at line 55 of file stem_UTF_8_norwegian.c.

◆ s_0_28

const symbol s_0_28[3] = { 'a', 's', 't' }
static

Definition at line 56 of file stem_UTF_8_norwegian.c.

◆ s_0_3

const symbol s_0_3[4] = { 'a', 'n', 'd', 'e' }
static

Definition at line 31 of file stem_UTF_8_norwegian.c.

◆ s_0_4

const symbol s_0_4[4] = { 'e', 'n', 'd', 'e' }
static

Definition at line 32 of file stem_UTF_8_norwegian.c.

◆ s_0_5

const symbol s_0_5[3] = { 'a', 'n', 'e' }
static

Definition at line 33 of file stem_UTF_8_norwegian.c.

◆ s_0_6

const symbol s_0_6[3] = { 'e', 'n', 'e' }
static

Definition at line 34 of file stem_UTF_8_norwegian.c.

◆ s_0_7

const symbol s_0_7[6] = { 'h', 'e', 't', 'e', 'n', 'e' }
static

Definition at line 35 of file stem_UTF_8_norwegian.c.

◆ s_0_8

const symbol s_0_8[4] = { 'e', 'r', 't', 'e' }
static

Definition at line 36 of file stem_UTF_8_norwegian.c.

◆ s_0_9

const symbol s_0_9[2] = { 'e', 'n' }
static

Definition at line 37 of file stem_UTF_8_norwegian.c.

◆ s_1_0

const symbol s_1_0[2] = { 'd', 't' }
static

Definition at line 91 of file stem_UTF_8_norwegian.c.

◆ s_1_1

const symbol s_1_1[2] = { 'v', 't' }
static

Definition at line 92 of file stem_UTF_8_norwegian.c.

◆ s_2_0

const symbol s_2_0[3] = { 'l', 'e', 'g' }
static

Definition at line 100 of file stem_UTF_8_norwegian.c.

◆ s_2_1

const symbol s_2_1[4] = { 'e', 'l', 'e', 'g' }
static

Definition at line 101 of file stem_UTF_8_norwegian.c.

◆ s_2_10

const symbol s_2_10[7] = { 'h', 'e', 't', 's', 'l', 'o', 'v' }
static

Definition at line 110 of file stem_UTF_8_norwegian.c.

◆ s_2_2

const symbol s_2_2[2] = { 'i', 'g' }
static

Definition at line 102 of file stem_UTF_8_norwegian.c.

◆ s_2_3

const symbol s_2_3[3] = { 'e', 'i', 'g' }
static

Definition at line 103 of file stem_UTF_8_norwegian.c.

◆ s_2_4

const symbol s_2_4[3] = { 'l', 'i', 'g' }
static

Definition at line 104 of file stem_UTF_8_norwegian.c.

◆ s_2_5

const symbol s_2_5[4] = { 'e', 'l', 'i', 'g' }
static

Definition at line 105 of file stem_UTF_8_norwegian.c.

◆ s_2_6

const symbol s_2_6[3] = { 'e', 'l', 's' }
static

Definition at line 106 of file stem_UTF_8_norwegian.c.

◆ s_2_7

const symbol s_2_7[3] = { 'l', 'o', 'v' }
static

Definition at line 107 of file stem_UTF_8_norwegian.c.

◆ s_2_8

const symbol s_2_8[4] = { 'e', 'l', 'o', 'v' }
static

Definition at line 108 of file stem_UTF_8_norwegian.c.

◆ s_2_9

const symbol s_2_9[4] = { 's', 'l', 'o', 'v' }
static

Definition at line 109 of file stem_UTF_8_norwegian.c.