PostgreSQL Source Code  git master
stem_UTF_8_swedish.c File Reference
#include "header.h"
Include dependency graph for stem_UTF_8_swedish.c:

Go to the source code of this file.

Functions

int swedish_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_envswedish_UTF_8_create_env (void)
 
void swedish_UTF_8_close_env (struct SN_env *z)
 

Variables

static const symbol s_0_0 [1] = { 'a' }
 
static const symbol s_0_1 [4] = { 'a', 'r', 'n', 'a' }
 
static const symbol s_0_2 [4] = { 'e', 'r', 'n', 'a' }
 
static const symbol s_0_3 [7] = { 'h', 'e', 't', 'e', 'r', 'n', 'a' }
 
static const symbol s_0_4 [4] = { 'o', 'r', 'n', 'a' }
 
static const symbol s_0_5 [2] = { 'a', 'd' }
 
static const symbol s_0_6 [1] = { 'e' }
 
static const symbol s_0_7 [3] = { 'a', 'd', 'e' }
 
static const symbol s_0_8 [4] = { 'a', 'n', 'd', 'e' }
 
static const symbol s_0_9 [4] = { 'a', 'r', 'n', 'e' }
 
static const symbol s_0_10 [3] = { 'a', 'r', 'e' }
 
static const symbol s_0_11 [4] = { 'a', 's', 't', 'e' }
 
static const symbol s_0_12 [2] = { 'e', 'n' }
 
static const symbol s_0_13 [5] = { 'a', 'n', 'd', 'e', 'n' }
 
static const symbol s_0_14 [4] = { 'a', 'r', 'e', 'n' }
 
static const symbol s_0_15 [5] = { 'h', 'e', 't', 'e', 'n' }
 
static const symbol s_0_16 [3] = { 'e', 'r', 'n' }
 
static const symbol s_0_17 [2] = { 'a', 'r' }
 
static const symbol s_0_18 [2] = { 'e', 'r' }
 
static const symbol s_0_19 [5] = { 'h', 'e', 't', 'e', 'r' }
 
static const symbol s_0_20 [2] = { 'o', 'r' }
 
static const symbol s_0_21 [1] = { 's' }
 
static const symbol s_0_22 [2] = { 'a', 's' }
 
static const symbol s_0_23 [5] = { 'a', 'r', 'n', 'a', 's' }
 
static const symbol s_0_24 [5] = { 'e', 'r', 'n', 'a', 's' }
 
static const symbol s_0_25 [5] = { 'o', 'r', 'n', 'a', 's' }
 
static const symbol s_0_26 [2] = { 'e', 's' }
 
static const symbol s_0_27 [4] = { 'a', 'd', 'e', 's' }
 
static const symbol s_0_28 [5] = { 'a', 'n', 'd', 'e', 's' }
 
static const symbol s_0_29 [3] = { 'e', 'n', 's' }
 
static const symbol s_0_30 [5] = { 'a', 'r', 'e', 'n', 's' }
 
static const symbol s_0_31 [6] = { 'h', 'e', 't', 'e', 'n', 's' }
 
static const symbol s_0_32 [4] = { 'e', 'r', 'n', 's' }
 
static const symbol s_0_33 [2] = { 'a', 't' }
 
static const symbol s_0_34 [5] = { 'a', 'n', 'd', 'e', 't' }
 
static const symbol s_0_35 [3] = { 'h', 'e', 't' }
 
static const symbol s_0_36 [3] = { 'a', 's', 't' }
 
static const struct among a_0 [37]
 
static const symbol s_1_0 [2] = { 'd', 'd' }
 
static const symbol s_1_1 [2] = { 'g', 'd' }
 
static const symbol s_1_2 [2] = { 'n', 'n' }
 
static const symbol s_1_3 [2] = { 'd', 't' }
 
static const symbol s_1_4 [2] = { 'g', 't' }
 
static const symbol s_1_5 [2] = { 'k', 't' }
 
static const symbol s_1_6 [2] = { 't', 't' }
 
static const struct among a_1 [7]
 
static const symbol s_2_0 [2] = { 'i', 'g' }
 
static const symbol s_2_1 [3] = { 'l', 'i', 'g' }
 
static const symbol s_2_2 [3] = { 'e', 'l', 's' }
 
static const symbol s_2_3 [5] = { 'f', 'u', 'l', 'l', 't' }
 
static const symbol s_2_4 [5] = { 'l', 0xC3, 0xB6, 's', 't' }
 
static const struct among a_2 [5]
 
static const unsigned char g_v [] = { 17, 65, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 24, 0, 32 }
 
static const unsigned char g_s_ending [] = { 119, 127, 149 }
 
static const symbol s_0 [] = { 'l', 0xC3, 0xB6, 's' }
 
static const symbol s_1 [] = { 'f', 'u', 'l', 'l' }
 

Function Documentation

◆ r_consonant_pair()

static int r_consonant_pair ( struct SN_env z)
static

Definition at line 201 of file stem_UTF_8_swedish.c.

201  {
202 
203  { int mlimit1;
204  if (z->c < z->I[1]) return 0;
205  mlimit1 = z->lb; z->lb = z->I[1];
206  { int m2 = z->l - z->c; (void)m2;
207  if (z->c - 1 <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1064976 >> (z->p[z->c - 1] & 0x1f)) & 1)) { z->lb = mlimit1; return 0; }
208  if (!(find_among_b(z, a_1, 7))) { z->lb = mlimit1; return 0; }
209  z->c = z->l - m2;
210  z->ket = z->c;
211  { int ret = skip_b_utf8(z->p, z->c, z->lb, 1);
212  if (ret < 0) { z->lb = mlimit1; return 0; }
213  z->c = ret;
214  }
215  z->bra = z->c;
216  { int ret = slice_del(z);
217  if (ret < 0) return ret;
218  }
219  }
220  z->lb = mlimit1;
221  }
222  return 1;
223 }
static const struct among a_1[7]
int lb
Definition: api.h:16
symbol * p
Definition: api.h:15
int * I
Definition: api.h:18
int ket
Definition: api.h:16
int c
Definition: api.h:16
int bra
Definition: api.h:16
int l
Definition: api.h:16
int find_among_b(struct SN_env *z, const struct among *v, int v_size)
Definition: utilities.c:298
int slice_del(struct SN_env *z)
Definition: utilities.c:431
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 swedish_UTF_8_stem().

◆ r_main_suffix()

static int r_main_suffix ( struct SN_env z)
static

Definition at line 172 of file stem_UTF_8_swedish.c.

172  {
173  int among_var;
174 
175  { int mlimit1;
176  if (z->c < z->I[1]) return 0;
177  mlimit1 = z->lb; z->lb = z->I[1];
178  z->ket = z->c;
179  if (z->c <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1851442 >> (z->p[z->c - 1] & 0x1f)) & 1)) { z->lb = mlimit1; return 0; }
180  among_var = find_among_b(z, a_0, 37);
181  if (!(among_var)) { z->lb = mlimit1; return 0; }
182  z->bra = z->c;
183  z->lb = mlimit1;
184  }
185  switch (among_var) {
186  case 1:
187  { int ret = slice_del(z);
188  if (ret < 0) return ret;
189  }
190  break;
191  case 2:
192  if (in_grouping_b_U(z, g_s_ending, 98, 121, 0)) return 0;
193  { int ret = slice_del(z);
194  if (ret < 0) return ret;
195  }
196  break;
197  }
198  return 1;
199 }
static const struct among a_0[37]
static const unsigned char g_s_ending[]
int in_grouping_b_U(struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
Definition: utilities.c:129

References a_0, SN_env::bra, SN_env::c, find_among_b(), g_s_ending, SN_env::I, in_grouping_b_U(), SN_env::ket, SN_env::lb, SN_env::p, and slice_del().

Referenced by swedish_UTF_8_stem().

◆ r_mark_regions()

static int r_mark_regions ( struct SN_env z)
static

Definition at line 148 of file stem_UTF_8_swedish.c.

148  {
149  z->I[1] = z->l;
150  { int c_test1 = z->c;
151  { int ret = skip_utf8(z->p, z->c, z->l, 3);
152  if (ret < 0) return 0;
153  z->c = ret;
154  }
155  z->I[0] = z->c;
156  z->c = c_test1;
157  }
158  if (out_grouping_U(z, g_v, 97, 246, 1) < 0) return 0;
159  {
160  int ret = in_grouping_U(z, g_v, 97, 246, 1);
161  if (ret < 0) return 0;
162  z->c += ret;
163  }
164  z->I[1] = z->c;
165 
166  if (!(z->I[1] < z->I[0])) goto lab0;
167  z->I[1] = z->I[0];
168 lab0:
169  return 1;
170 }
static const unsigned char g_v[]
int out_grouping_U(struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
Definition: utilities.c:141
int in_grouping_U(struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
Definition: utilities.c:117
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 swedish_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_swedish.c.

225  {
226  int among_var;
227 
228  { int mlimit1;
229  if (z->c < z->I[1]) return 0;
230  mlimit1 = z->lb; z->lb = z->I[1];
231  z->ket = z->c;
232  if (z->c - 1 <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1572992 >> (z->p[z->c - 1] & 0x1f)) & 1)) { z->lb = mlimit1; return 0; }
233  among_var = find_among_b(z, a_2, 5);
234  if (!(among_var)) { z->lb = mlimit1; return 0; }
235  z->bra = z->c;
236  switch (among_var) {
237  case 1:
238  { int ret = slice_del(z);
239  if (ret < 0) return ret;
240  }
241  break;
242  case 2:
243  { int ret = slice_from_s(z, 4, s_0);
244  if (ret < 0) return ret;
245  }
246  break;
247  case 3:
248  { int ret = slice_from_s(z, 4, s_1);
249  if (ret < 0) return ret;
250  }
251  break;
252  }
253  z->lb = mlimit1;
254  }
255  return 1;
256 }
static const struct among a_2[5]
static const symbol s_1[]
static const symbol s_0[]
int slice_from_s(struct SN_env *z, int s_size, const symbol *s)
Definition: utilities.c:422

References a_2, SN_env::bra, SN_env::c, find_among_b(), SN_env::I, SN_env::ket, SN_env::lb, SN_env::p, s_0, s_1, slice_del(), and slice_from_s().

Referenced by swedish_UTF_8_stem().

◆ swedish_UTF_8_close_env()

void swedish_UTF_8_close_env ( struct SN_env z)

Definition at line 291 of file stem_UTF_8_swedish.c.

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

References SN_close_env().

◆ swedish_UTF_8_create_env()

struct SN_env * swedish_UTF_8_create_env ( void  )

Definition at line 289 of file stem_UTF_8_swedish.c.

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

References SN_create_env().

◆ swedish_UTF_8_stem()

int swedish_UTF_8_stem ( struct SN_env z)

Definition at line 258 of file stem_UTF_8_swedish.c.

258  {
259  { int c1 = z->c;
260  { int ret = r_mark_regions(z);
261  if (ret < 0) return ret;
262  }
263  z->c = c1;
264  }
265  z->lb = z->c; z->c = z->l;
266 
267  { int m2 = z->l - z->c; (void)m2;
268  { int ret = r_main_suffix(z);
269  if (ret < 0) return ret;
270  }
271  z->c = z->l - m2;
272  }
273  { int m3 = z->l - z->c; (void)m3;
274  { int ret = r_consonant_pair(z);
275  if (ret < 0) return ret;
276  }
277  z->c = z->l - m3;
278  }
279  { int m4 = z->l - z->c; (void)m4;
280  { int ret = r_other_suffix(z);
281  if (ret < 0) return ret;
282  }
283  z->c = z->l - m4;
284  }
285  z->c = z->lb;
286  return 1;
287 }
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)

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

Variable Documentation

◆ a_0

const struct among a_0[37]
static

Definition at line 64 of file stem_UTF_8_swedish.c.

Referenced by r_main_suffix().

◆ a_1

const struct among a_1[7]
static
Initial value:
=
{
{ 2, s_1_0, -1, -1, 0},
{ 2, s_1_1, -1, -1, 0},
{ 2, s_1_2, -1, -1, 0},
{ 2, s_1_3, -1, -1, 0},
{ 2, s_1_4, -1, -1, 0},
{ 2, s_1_5, -1, -1, 0},
{ 2, s_1_6, -1, -1, 0}
}
static const symbol s_1_5[2]
static const symbol s_1_6[2]
static const symbol s_1_2[2]
static const symbol s_1_3[2]
static const symbol s_1_0[2]
static const symbol s_1_1[2]
static const symbol s_1_4[2]

Definition at line 113 of file stem_UTF_8_swedish.c.

Referenced by r_consonant_pair().

◆ a_2

const struct among a_2[5]
static
Initial value:
=
{
{ 2, s_2_0, -1, 1, 0},
{ 3, s_2_1, 0, 1, 0},
{ 3, s_2_2, -1, 1, 0},
{ 5, s_2_3, -1, 3, 0},
{ 5, s_2_4, -1, 2, 0}
}
static const symbol s_2_0[2]
static const symbol s_2_1[3]
static const symbol s_2_4[5]
static const symbol s_2_3[5]
static const symbol s_2_2[3]

Definition at line 130 of file stem_UTF_8_swedish.c.

Referenced by r_other_suffix().

◆ g_s_ending

const unsigned char g_s_ending[] = { 119, 127, 149 }
static

Definition at line 143 of file stem_UTF_8_swedish.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, 24, 0, 32 }
static

Definition at line 141 of file stem_UTF_8_swedish.c.

Referenced by r_mark_regions().

◆ s_0

const symbol s_0[] = { 'l', 0xC3, 0xB6, 's' }
static

Definition at line 145 of file stem_UTF_8_swedish.c.

Referenced by r_other_suffix().

◆ s_0_0

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

Definition at line 28 of file stem_UTF_8_swedish.c.

◆ s_0_1

const symbol s_0_1[4] = { 'a', 'r', 'n', 'a' }
static

Definition at line 29 of file stem_UTF_8_swedish.c.

◆ s_0_10

const symbol s_0_10[3] = { 'a', 'r', 'e' }
static

Definition at line 38 of file stem_UTF_8_swedish.c.

◆ s_0_11

const symbol s_0_11[4] = { 'a', 's', 't', 'e' }
static

Definition at line 39 of file stem_UTF_8_swedish.c.

◆ s_0_12

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

Definition at line 40 of file stem_UTF_8_swedish.c.

◆ s_0_13

const symbol s_0_13[5] = { 'a', 'n', 'd', 'e', 'n' }
static

Definition at line 41 of file stem_UTF_8_swedish.c.

◆ s_0_14

const symbol s_0_14[4] = { 'a', 'r', 'e', 'n' }
static

Definition at line 42 of file stem_UTF_8_swedish.c.

◆ s_0_15

const symbol s_0_15[5] = { 'h', 'e', 't', 'e', 'n' }
static

Definition at line 43 of file stem_UTF_8_swedish.c.

◆ s_0_16

const symbol s_0_16[3] = { 'e', 'r', 'n' }
static

Definition at line 44 of file stem_UTF_8_swedish.c.

◆ s_0_17

const symbol s_0_17[2] = { 'a', 'r' }
static

Definition at line 45 of file stem_UTF_8_swedish.c.

◆ s_0_18

const symbol s_0_18[2] = { 'e', 'r' }
static

Definition at line 46 of file stem_UTF_8_swedish.c.

◆ s_0_19

const symbol s_0_19[5] = { 'h', 'e', 't', 'e', 'r' }
static

Definition at line 47 of file stem_UTF_8_swedish.c.

◆ s_0_2

const symbol s_0_2[4] = { 'e', 'r', 'n', 'a' }
static

Definition at line 30 of file stem_UTF_8_swedish.c.

◆ s_0_20

const symbol s_0_20[2] = { 'o', 'r' }
static

Definition at line 48 of file stem_UTF_8_swedish.c.

◆ s_0_21

const symbol s_0_21[1] = { 's' }
static

Definition at line 49 of file stem_UTF_8_swedish.c.

◆ s_0_22

const symbol s_0_22[2] = { 'a', 's' }
static

Definition at line 50 of file stem_UTF_8_swedish.c.

◆ s_0_23

const symbol s_0_23[5] = { 'a', 'r', 'n', 'a', 's' }
static

Definition at line 51 of file stem_UTF_8_swedish.c.

◆ s_0_24

const symbol s_0_24[5] = { 'e', 'r', 'n', 'a', 's' }
static

Definition at line 52 of file stem_UTF_8_swedish.c.

◆ s_0_25

const symbol s_0_25[5] = { 'o', 'r', 'n', 'a', 's' }
static

Definition at line 53 of file stem_UTF_8_swedish.c.

◆ s_0_26

const symbol s_0_26[2] = { 'e', 's' }
static

Definition at line 54 of file stem_UTF_8_swedish.c.

◆ s_0_27

const symbol s_0_27[4] = { 'a', 'd', 'e', 's' }
static

Definition at line 55 of file stem_UTF_8_swedish.c.

◆ s_0_28

const symbol s_0_28[5] = { 'a', 'n', 'd', 'e', 's' }
static

Definition at line 56 of file stem_UTF_8_swedish.c.

◆ s_0_29

const symbol s_0_29[3] = { 'e', 'n', 's' }
static

Definition at line 57 of file stem_UTF_8_swedish.c.

◆ s_0_3

const symbol s_0_3[7] = { 'h', 'e', 't', 'e', 'r', 'n', 'a' }
static

Definition at line 31 of file stem_UTF_8_swedish.c.

◆ s_0_30

const symbol s_0_30[5] = { 'a', 'r', 'e', 'n', 's' }
static

Definition at line 58 of file stem_UTF_8_swedish.c.

◆ s_0_31

const symbol s_0_31[6] = { 'h', 'e', 't', 'e', 'n', 's' }
static

Definition at line 59 of file stem_UTF_8_swedish.c.

◆ s_0_32

const symbol s_0_32[4] = { 'e', 'r', 'n', 's' }
static

Definition at line 60 of file stem_UTF_8_swedish.c.

◆ s_0_33

const symbol s_0_33[2] = { 'a', 't' }
static

Definition at line 61 of file stem_UTF_8_swedish.c.

◆ s_0_34

const symbol s_0_34[5] = { 'a', 'n', 'd', 'e', 't' }
static

Definition at line 62 of file stem_UTF_8_swedish.c.

◆ s_0_35

const symbol s_0_35[3] = { 'h', 'e', 't' }
static

Definition at line 63 of file stem_UTF_8_swedish.c.

◆ s_0_36

const symbol s_0_36[3] = { 'a', 's', 't' }
static

Definition at line 64 of file stem_UTF_8_swedish.c.

◆ s_0_4

const symbol s_0_4[4] = { 'o', 'r', 'n', 'a' }
static

Definition at line 32 of file stem_UTF_8_swedish.c.

◆ s_0_5

const symbol s_0_5[2] = { 'a', 'd' }
static

Definition at line 33 of file stem_UTF_8_swedish.c.

◆ s_0_6

const symbol s_0_6[1] = { 'e' }
static

Definition at line 34 of file stem_UTF_8_swedish.c.

◆ s_0_7

const symbol s_0_7[3] = { 'a', 'd', 'e' }
static

Definition at line 35 of file stem_UTF_8_swedish.c.

◆ s_0_8

const symbol s_0_8[4] = { 'a', 'n', 'd', 'e' }
static

Definition at line 36 of file stem_UTF_8_swedish.c.

◆ s_0_9

const symbol s_0_9[4] = { 'a', 'r', 'n', 'e' }
static

Definition at line 37 of file stem_UTF_8_swedish.c.

◆ s_1

const symbol s_1[] = { 'f', 'u', 'l', 'l' }
static

Definition at line 146 of file stem_UTF_8_swedish.c.

Referenced by r_other_suffix().

◆ s_1_0

const symbol s_1_0[2] = { 'd', 'd' }
static

Definition at line 107 of file stem_UTF_8_swedish.c.

◆ s_1_1

const symbol s_1_1[2] = { 'g', 'd' }
static

Definition at line 108 of file stem_UTF_8_swedish.c.

◆ s_1_2

const symbol s_1_2[2] = { 'n', 'n' }
static

Definition at line 109 of file stem_UTF_8_swedish.c.

◆ s_1_3

const symbol s_1_3[2] = { 'd', 't' }
static

Definition at line 110 of file stem_UTF_8_swedish.c.

◆ s_1_4

const symbol s_1_4[2] = { 'g', 't' }
static

Definition at line 111 of file stem_UTF_8_swedish.c.

◆ s_1_5

const symbol s_1_5[2] = { 'k', 't' }
static

Definition at line 112 of file stem_UTF_8_swedish.c.

◆ s_1_6

const symbol s_1_6[2] = { 't', 't' }
static

Definition at line 113 of file stem_UTF_8_swedish.c.

◆ s_2_0

const symbol s_2_0[2] = { 'i', 'g' }
static

Definition at line 126 of file stem_UTF_8_swedish.c.

◆ s_2_1

const symbol s_2_1[3] = { 'l', 'i', 'g' }
static

Definition at line 127 of file stem_UTF_8_swedish.c.

◆ s_2_2

const symbol s_2_2[3] = { 'e', 'l', 's' }
static

Definition at line 128 of file stem_UTF_8_swedish.c.

◆ s_2_3

const symbol s_2_3[5] = { 'f', 'u', 'l', 'l', 't' }
static

Definition at line 129 of file stem_UTF_8_swedish.c.

◆ s_2_4

const symbol s_2_4[5] = { 'l', 0xC3, 0xB6, 's', 't' }
static

Definition at line 130 of file stem_UTF_8_swedish.c.