PostgreSQL Source Code  git master
stem_UTF_8_danish.c File Reference
#include "header.h"
Include dependency graph for stem_UTF_8_danish.c:

Go to the source code of this file.

Functions

int danish_UTF_8_stem (struct SN_env *z)
 
static int r_undouble (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_envdanish_UTF_8_create_env (void)
 
void danish_UTF_8_close_env (struct SN_env *z)
 

Variables

static const symbol s_0_0 [3] = { 'h', 'e', 'd' }
 
static const symbol s_0_1 [5] = { 'e', 't', 'h', 'e', 'd' }
 
static const symbol s_0_2 [4] = { 'e', 'r', 'e', 'd' }
 
static const symbol s_0_3 [1] = { 'e' }
 
static const symbol s_0_4 [5] = { 'e', 'r', 'e', 'd', 'e' }
 
static const symbol s_0_5 [4] = { 'e', 'n', 'd', 'e' }
 
static const symbol s_0_6 [6] = { 'e', 'r', 'e', 'n', 'd', 'e' }
 
static const symbol s_0_7 [3] = { 'e', 'n', 'e' }
 
static const symbol s_0_8 [4] = { 'e', 'r', 'n', 'e' }
 
static const symbol s_0_9 [3] = { 'e', 'r', 'e' }
 
static const symbol s_0_10 [2] = { 'e', 'n' }
 
static const symbol s_0_11 [5] = { 'h', 'e', 'd', 'e', 'n' }
 
static const symbol s_0_12 [4] = { 'e', 'r', 'e', 'n' }
 
static const symbol s_0_13 [2] = { 'e', 'r' }
 
static const symbol s_0_14 [5] = { 'h', 'e', 'd', 'e', 'r' }
 
static const symbol s_0_15 [4] = { 'e', 'r', 'e', 'r' }
 
static const symbol s_0_16 [1] = { 's' }
 
static const symbol s_0_17 [4] = { 'h', 'e', 'd', 's' }
 
static const symbol s_0_18 [2] = { 'e', 's' }
 
static const symbol s_0_19 [5] = { 'e', 'n', 'd', 'e', 's' }
 
static const symbol s_0_20 [7] = { 'e', 'r', 'e', 'n', 'd', 'e', 's' }
 
static const symbol s_0_21 [4] = { 'e', 'n', 'e', 's' }
 
static const symbol s_0_22 [5] = { 'e', 'r', 'n', 'e', 's' }
 
static const symbol s_0_23 [4] = { 'e', 'r', 'e', 's' }
 
static const symbol s_0_24 [3] = { 'e', 'n', 's' }
 
static const symbol s_0_25 [6] = { 'h', 'e', 'd', 'e', 'n', 's' }
 
static const symbol s_0_26 [5] = { 'e', 'r', 'e', 'n', 's' }
 
static const symbol s_0_27 [3] = { 'e', 'r', 's' }
 
static const symbol s_0_28 [3] = { 'e', 't', 's' }
 
static const symbol s_0_29 [5] = { 'e', 'r', 'e', 't', 's' }
 
static const symbol s_0_30 [2] = { 'e', 't' }
 
static const symbol s_0_31 [4] = { 'e', 'r', 'e', 't' }
 
static const struct among a_0 [32]
 
static const symbol s_1_0 [2] = { 'g', 'd' }
 
static const symbol s_1_1 [2] = { 'd', 't' }
 
static const symbol s_1_2 [2] = { 'g', 't' }
 
static const symbol s_1_3 [2] = { 'k', 't' }
 
static const struct among a_1 [4]
 
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 [4] = { 'e', 'l', 'i', 'g' }
 
static const symbol s_2_3 [3] = { 'e', 'l', 's' }
 
static const symbol s_2_4 [5] = { 'l', 0xC3, 0xB8, 's', 't' }
 
static const struct among a_2 [5]
 
static const unsigned char g_c [] = { 119, 223, 119, 1 }
 
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 [] = { 239, 254, 42, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 16 }
 
static const symbol s_0 [] = { 's', 't' }
 
static const symbol s_1 [] = { 'i', 'g' }
 
static const symbol s_2 [] = { 'l', 0xC3, 0xB8, 's' }
 

Function Documentation

◆ danish_UTF_8_close_env()

void danish_UTF_8_close_env ( struct SN_env z)

Definition at line 317 of file stem_UTF_8_danish.c.

317 { SN_close_env(z, 1); }
void SN_close_env(struct SN_env *z, int S_size)
Definition: api.c:34

References SN_close_env().

◆ danish_UTF_8_create_env()

struct SN_env * danish_UTF_8_create_env ( void  )

Definition at line 315 of file stem_UTF_8_danish.c.

315 { return SN_create_env(1, 2); }
struct SN_env * SN_create_env(int S_size, int I_size)
Definition: api.c:3

References SN_create_env().

◆ danish_UTF_8_stem()

int danish_UTF_8_stem ( struct SN_env z)

Definition at line 278 of file stem_UTF_8_danish.c.

278  {
279  { int c1 = z->c;
280  { int ret = r_mark_regions(z);
281  if (ret < 0) return ret;
282  }
283  z->c = c1;
284  }
285  z->lb = z->c; z->c = z->l;
286 
287  { int m2 = z->l - z->c; (void)m2;
288  { int ret = r_main_suffix(z);
289  if (ret < 0) return ret;
290  }
291  z->c = z->l - m2;
292  }
293  { int m3 = z->l - z->c; (void)m3;
294  { int ret = r_consonant_pair(z);
295  if (ret < 0) return ret;
296  }
297  z->c = z->l - m3;
298  }
299  { int m4 = z->l - z->c; (void)m4;
300  { int ret = r_other_suffix(z);
301  if (ret < 0) return ret;
302  }
303  z->c = z->l - m4;
304  }
305  { int m5 = z->l - z->c; (void)m5;
306  { int ret = r_undouble(z);
307  if (ret < 0) return ret;
308  }
309  z->c = z->l - m5;
310  }
311  z->c = z->lb;
312  return 1;
313 }
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_undouble(struct SN_env *z)
static int r_mark_regions(struct SN_env *z)
int lb
Definition: api.h:16
int c
Definition: api.h:16
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(), r_other_suffix(), and r_undouble().

◆ r_consonant_pair()

static int r_consonant_pair ( struct SN_env z)
static

Definition at line 189 of file stem_UTF_8_danish.c.

189  {
190  { int m_test1 = z->l - z->c;
191 
192  { int mlimit2;
193  if (z->c < z->I[1]) return 0;
194  mlimit2 = z->lb; z->lb = z->I[1];
195  z->ket = z->c;
196  if (z->c - 1 <= z->lb || (z->p[z->c - 1] != 100 && z->p[z->c - 1] != 116)) { z->lb = mlimit2; return 0; }
197  if (!(find_among_b(z, a_1, 4))) { z->lb = mlimit2; return 0; }
198  z->bra = z->c;
199  z->lb = mlimit2;
200  }
201  z->c = z->l - m_test1;
202  }
203  { int ret = skip_b_utf8(z->p, z->c, z->lb, 1);
204  if (ret < 0) return 0;
205  z->c = ret;
206  }
207  z->bra = z->c;
208  { int ret = slice_del(z);
209  if (ret < 0) return ret;
210  }
211  return 1;
212 }
static const struct among a_1[4]
symbol * p
Definition: api.h:15
int * I
Definition: api.h:18
int ket
Definition: api.h:16
int bra
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 danish_UTF_8_stem(), and r_other_suffix().

◆ r_main_suffix()

static int r_main_suffix ( struct SN_env z)
static

Definition at line 160 of file stem_UTF_8_danish.c.

160  {
161  int among_var;
162 
163  { int mlimit1;
164  if (z->c < z->I[1]) return 0;
165  mlimit1 = z->lb; z->lb = z->I[1];
166  z->ket = z->c;
167  if (z->c <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1851440 >> (z->p[z->c - 1] & 0x1f)) & 1)) { z->lb = mlimit1; return 0; }
168  among_var = find_among_b(z, a_0, 32);
169  if (!(among_var)) { z->lb = mlimit1; return 0; }
170  z->bra = z->c;
171  z->lb = mlimit1;
172  }
173  switch (among_var) {
174  case 1:
175  { int ret = slice_del(z);
176  if (ret < 0) return ret;
177  }
178  break;
179  case 2:
180  if (in_grouping_b_U(z, g_s_ending, 97, 229, 0)) return 0;
181  { int ret = slice_del(z);
182  if (ret < 0) return ret;
183  }
184  break;
185  }
186  return 1;
187 }
static const struct among a_0[32]
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 danish_UTF_8_stem().

◆ r_mark_regions()

static int r_mark_regions ( struct SN_env z)
static

Definition at line 136 of file stem_UTF_8_danish.c.

136  {
137  z->I[1] = z->l;
138  { int c_test1 = z->c;
139  { int ret = skip_utf8(z->p, z->c, z->l, 3);
140  if (ret < 0) return 0;
141  z->c = ret;
142  }
143  z->I[0] = z->c;
144  z->c = c_test1;
145  }
146  if (out_grouping_U(z, g_v, 97, 248, 1) < 0) return 0;
147  {
148  int ret = in_grouping_U(z, g_v, 97, 248, 1);
149  if (ret < 0) return 0;
150  z->c += ret;
151  }
152  z->I[1] = z->c;
153 
154  if (!(z->I[1] < z->I[0])) goto lab0;
155  z->I[1] = z->I[0];
156 lab0:
157  return 1;
158 }
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 danish_UTF_8_stem().

◆ r_other_suffix()

static int r_other_suffix ( struct SN_env z)
static

Definition at line 214 of file stem_UTF_8_danish.c.

214  {
215  int among_var;
216  { int m1 = z->l - z->c; (void)m1;
217  z->ket = z->c;
218  if (!(eq_s_b(z, 2, s_0))) goto lab0;
219  z->bra = z->c;
220  if (!(eq_s_b(z, 2, s_1))) goto lab0;
221  { int ret = slice_del(z);
222  if (ret < 0) return ret;
223  }
224  lab0:
225  z->c = z->l - m1;
226  }
227 
228  { int mlimit2;
229  if (z->c < z->I[1]) return 0;
230  mlimit2 = 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 = mlimit2; return 0; }
233  among_var = find_among_b(z, a_2, 5);
234  if (!(among_var)) { z->lb = mlimit2; return 0; }
235  z->bra = z->c;
236  z->lb = mlimit2;
237  }
238  switch (among_var) {
239  case 1:
240  { int ret = slice_del(z);
241  if (ret < 0) return ret;
242  }
243  { int m3 = z->l - z->c; (void)m3;
244  { int ret = r_consonant_pair(z);
245  if (ret < 0) return ret;
246  }
247  z->c = z->l - m3;
248  }
249  break;
250  case 2:
251  { int ret = slice_from_s(z, 4, s_2);
252  if (ret < 0) return ret;
253  }
254  break;
255  }
256  return 1;
257 }
static const struct among a_2[5]
static const symbol s_2[]
static const symbol s_1[]
static const symbol s_0[]
int eq_s_b(struct SN_env *z, int s_size, const symbol *s)
Definition: utilities.c:220
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, eq_s_b(), find_among_b(), SN_env::I, SN_env::ket, SN_env::l, SN_env::lb, SN_env::p, r_consonant_pair(), s_0, s_1, s_2, slice_del(), and slice_from_s().

Referenced by danish_UTF_8_stem().

◆ r_undouble()

static int r_undouble ( struct SN_env z)
static

Definition at line 259 of file stem_UTF_8_danish.c.

259  {
260 
261  { int mlimit1;
262  if (z->c < z->I[1]) return 0;
263  mlimit1 = z->lb; z->lb = z->I[1];
264  z->ket = z->c;
265  if (in_grouping_b_U(z, g_c, 98, 122, 0)) { z->lb = mlimit1; return 0; }
266  z->bra = z->c;
267  z->S[0] = slice_to(z, z->S[0]);
268  if (z->S[0] == 0) return -1;
269  z->lb = mlimit1;
270  }
271  if (!(eq_v_b(z, z->S[0]))) return 0;
272  { int ret = slice_del(z);
273  if (ret < 0) return ret;
274  }
275  return 1;
276 }
static const unsigned char g_c[]
symbol ** S
Definition: api.h:17
int eq_v_b(struct SN_env *z, const symbol *p)
Definition: utilities.c:229
symbol * slice_to(struct SN_env *z, symbol *p)
Definition: utilities.c:448

References SN_env::bra, SN_env::c, eq_v_b(), g_c, SN_env::I, in_grouping_b_U(), SN_env::ket, SN_env::lb, SN_env::S, slice_del(), and slice_to().

Referenced by danish_UTF_8_stem().

Variable Documentation

◆ a_0

const struct among a_0[32]
static

Definition at line 60 of file stem_UTF_8_danish.c.

Referenced by r_main_suffix().

◆ a_1

const struct among a_1[4]
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}
}
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]

Definition at line 101 of file stem_UTF_8_danish.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},
{ 4, s_2_2, 1, 1, 0},
{ 3, s_2_3, -1, 1, 0},
{ 5, s_2_4, -1, 2, 0}
}
static const symbol s_2_3[3]
static const symbol s_2_0[2]
static const symbol s_2_1[3]
static const symbol s_2_2[4]
static const symbol s_2_4[5]

Definition at line 115 of file stem_UTF_8_danish.c.

Referenced by r_other_suffix().

◆ g_c

const unsigned char g_c[] = { 119, 223, 119, 1 }
static

Definition at line 126 of file stem_UTF_8_danish.c.

Referenced by r_undouble().

◆ g_s_ending

const unsigned char g_s_ending[] = { 239, 254, 42, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 16 }
static

Definition at line 130 of file stem_UTF_8_danish.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 128 of file stem_UTF_8_danish.c.

Referenced by r_mark_regions().

◆ s_0

const symbol s_0[] = { 's', 't' }
static

Definition at line 132 of file stem_UTF_8_danish.c.

Referenced by r_other_suffix().

◆ s_0_0

const symbol s_0_0[3] = { 'h', 'e', 'd' }
static

Definition at line 29 of file stem_UTF_8_danish.c.

◆ s_0_1

const symbol s_0_1[5] = { 'e', 't', 'h', 'e', 'd' }
static

Definition at line 30 of file stem_UTF_8_danish.c.

◆ s_0_10

const symbol s_0_10[2] = { 'e', 'n' }
static

Definition at line 39 of file stem_UTF_8_danish.c.

◆ s_0_11

const symbol s_0_11[5] = { 'h', 'e', 'd', 'e', 'n' }
static

Definition at line 40 of file stem_UTF_8_danish.c.

◆ s_0_12

const symbol s_0_12[4] = { 'e', 'r', 'e', 'n' }
static

Definition at line 41 of file stem_UTF_8_danish.c.

◆ s_0_13

const symbol s_0_13[2] = { 'e', 'r' }
static

Definition at line 42 of file stem_UTF_8_danish.c.

◆ s_0_14

const symbol s_0_14[5] = { 'h', 'e', 'd', 'e', 'r' }
static

Definition at line 43 of file stem_UTF_8_danish.c.

◆ s_0_15

const symbol s_0_15[4] = { 'e', 'r', 'e', 'r' }
static

Definition at line 44 of file stem_UTF_8_danish.c.

◆ s_0_16

const symbol s_0_16[1] = { 's' }
static

Definition at line 45 of file stem_UTF_8_danish.c.

◆ s_0_17

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

Definition at line 46 of file stem_UTF_8_danish.c.

◆ s_0_18

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

Definition at line 47 of file stem_UTF_8_danish.c.

◆ s_0_19

const symbol s_0_19[5] = { 'e', 'n', 'd', 'e', 's' }
static

Definition at line 48 of file stem_UTF_8_danish.c.

◆ s_0_2

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

Definition at line 31 of file stem_UTF_8_danish.c.

◆ s_0_20

const symbol s_0_20[7] = { 'e', 'r', 'e', 'n', 'd', 'e', 's' }
static

Definition at line 49 of file stem_UTF_8_danish.c.

◆ s_0_21

const symbol s_0_21[4] = { 'e', 'n', 'e', 's' }
static

Definition at line 50 of file stem_UTF_8_danish.c.

◆ s_0_22

const symbol s_0_22[5] = { 'e', 'r', 'n', 'e', 's' }
static

Definition at line 51 of file stem_UTF_8_danish.c.

◆ s_0_23

const symbol s_0_23[4] = { 'e', 'r', 'e', 's' }
static

Definition at line 52 of file stem_UTF_8_danish.c.

◆ s_0_24

const symbol s_0_24[3] = { 'e', 'n', 's' }
static

Definition at line 53 of file stem_UTF_8_danish.c.

◆ s_0_25

const symbol s_0_25[6] = { 'h', 'e', 'd', 'e', 'n', 's' }
static

Definition at line 54 of file stem_UTF_8_danish.c.

◆ s_0_26

const symbol s_0_26[5] = { 'e', 'r', 'e', 'n', 's' }
static

Definition at line 55 of file stem_UTF_8_danish.c.

◆ s_0_27

const symbol s_0_27[3] = { 'e', 'r', 's' }
static

Definition at line 56 of file stem_UTF_8_danish.c.

◆ s_0_28

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

Definition at line 57 of file stem_UTF_8_danish.c.

◆ s_0_29

const symbol s_0_29[5] = { 'e', 'r', 'e', 't', 's' }
static

Definition at line 58 of file stem_UTF_8_danish.c.

◆ s_0_3

const symbol s_0_3[1] = { 'e' }
static

Definition at line 32 of file stem_UTF_8_danish.c.

◆ s_0_30

const symbol s_0_30[2] = { 'e', 't' }
static

Definition at line 59 of file stem_UTF_8_danish.c.

◆ s_0_31

const symbol s_0_31[4] = { 'e', 'r', 'e', 't' }
static

Definition at line 60 of file stem_UTF_8_danish.c.

◆ s_0_4

const symbol s_0_4[5] = { 'e', 'r', 'e', 'd', 'e' }
static

Definition at line 33 of file stem_UTF_8_danish.c.

◆ s_0_5

const symbol s_0_5[4] = { 'e', 'n', 'd', 'e' }
static

Definition at line 34 of file stem_UTF_8_danish.c.

◆ s_0_6

const symbol s_0_6[6] = { 'e', 'r', 'e', 'n', 'd', 'e' }
static

Definition at line 35 of file stem_UTF_8_danish.c.

◆ s_0_7

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

Definition at line 36 of file stem_UTF_8_danish.c.

◆ s_0_8

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

Definition at line 37 of file stem_UTF_8_danish.c.

◆ s_0_9

const symbol s_0_9[3] = { 'e', 'r', 'e' }
static

Definition at line 38 of file stem_UTF_8_danish.c.

◆ s_1

const symbol s_1[] = { 'i', 'g' }
static

Definition at line 133 of file stem_UTF_8_danish.c.

Referenced by r_other_suffix().

◆ s_1_0

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

Definition at line 98 of file stem_UTF_8_danish.c.

◆ s_1_1

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

Definition at line 99 of file stem_UTF_8_danish.c.

◆ s_1_2

const symbol s_1_2[2] = { 'g', 't' }
static

Definition at line 100 of file stem_UTF_8_danish.c.

◆ s_1_3

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

Definition at line 101 of file stem_UTF_8_danish.c.

◆ s_2

const symbol s_2[] = { 'l', 0xC3, 0xB8, 's' }
static

Definition at line 134 of file stem_UTF_8_danish.c.

Referenced by r_other_suffix().

◆ s_2_0

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

Definition at line 111 of file stem_UTF_8_danish.c.

◆ s_2_1

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

Definition at line 112 of file stem_UTF_8_danish.c.

◆ s_2_2

const symbol s_2_2[4] = { 'e', 'l', 'i', 'g' }
static

Definition at line 113 of file stem_UTF_8_danish.c.

◆ s_2_3

const symbol s_2_3[3] = { 'e', 'l', 's' }
static

Definition at line 114 of file stem_UTF_8_danish.c.

◆ s_2_4

const symbol s_2_4[5] = { 'l', 0xC3, 0xB8, 's', 't' }
static

Definition at line 115 of file stem_UTF_8_danish.c.