stem_UTF_8_danish.c File Reference

#include "stem_UTF_8_danish.h"
#include <stddef.h>
#include "snowball_runtime.h"

Include dependency graph for stem_UTF_8_danish.c:

Go to the source code of this file.

Data Structures
struct	SN_local

Typedefs
typedef struct SN_local	SN_local

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_env *	danish_UTF_8_create_env (void)
void	danish_UTF_8_close_env (struct SN_env *z)

Variables
static const symbol	s_0 [] = { 's', 't' }
static const symbol	s_1 [] = { 'i', 'g' }
static const symbol	s_2 [] = { 'l', 0xC3, 0xB8, 's' }
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 }

Typedef Documentation

SN_local

typedef struct SN_local SN_local

Definition at line 15 of file stem_UTF_8_danish.c.

Function Documentation

danish_UTF_8_close_env()

void danish_UTF_8_close_env

(

struct SN_env *

z

)

Definition at line 354 of file stem_UTF_8_danish.c.

                                                      {
    if (z) {
        lose_s(((SN_local *)z)->s_ch);
    }
    SN_delete_env(z);
}

References lose_s(), and SN_delete_env().

Referenced by danish_UTF_8_create_env().

danish_UTF_8_create_env()

struct SN_env * danish_UTF_8_create_env

(

void

)

Definition at line 340 of file stem_UTF_8_danish.c.

                                                     {
    struct SN_env * z = SN_new_env(sizeof(SN_local));
    if (z) {
        ((SN_local *)z)->i_p1 = 0;
        ((SN_local *)z)->s_ch = NULL;
 
        if ((((SN_local *)z)->s_ch = create_s()) == NULL) {
            danish_UTF_8_close_env(z);
            return NULL;
        }
    }
    return z;
}

References create_s(), danish_UTF_8_close_env(), and SN_new_env().

danish_UTF_8_stem()

int danish_UTF_8_stem

(

struct SN_env *

z

)

Definition at line 294 of file stem_UTF_8_danish.c.

                                                {
    {
        int v_1 = z->c;
        {
            int ret = r_mark_regions(z);
            if (ret < 0) return ret;
        }
        z->c = v_1;
    }
    z->lb = z->c; z->c = z->l;
    {
        int v_2 = z->l - z->c;
        {
            int ret = r_main_suffix(z);
            if (ret < 0) return ret;
        }
        z->c = z->l - v_2;
    }
    {
        int v_3 = z->l - z->c;
        {
            int ret = r_consonant_pair(z);
            if (ret < 0) return ret;
        }
        z->c = z->l - v_3;
    }
    {
        int v_4 = z->l - z->c;
        {
            int ret = r_other_suffix(z);
            if (ret < 0) return ret;
        }
        z->c = z->l - v_4;
    }
    {
        int v_5 = z->l - z->c;
        {
            int ret = r_undouble(z);
            if (ret < 0) return ret;
        }
        z->c = z->l - v_5;
    }
    z->c = z->lb;
    return 1;
}

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 193 of file stem_UTF_8_danish.c.

                                               {
    {
        int v_1 = z->l - z->c;
        {
            int v_2;
            if (z->c < ((SN_local *)z)->i_p1) return 0;
            v_2 = z->lb; z->lb = ((SN_local *)z)->i_p1;
            z->ket = z->c;
            if (z->c - 1 <= z->lb || (z->p[z->c - 1] != 100 && z->p[z->c - 1] != 116)) { z->lb = v_2; return 0; }
            if (!find_among_b(z, a_1, 4, 0)) { z->lb = v_2; return 0; }
            z->bra = z->c;
            z->lb = v_2;
        }
        z->c = z->l - v_1;
    }
    {
        int ret = skip_b_utf8(z->p, z->c, z->lb, 1);
        if (ret < 0) return 0;
        z->c = ret;
    }
    z->bra = z->c;
    {
        int ret = slice_del(z);
        if (ret < 0) return ret;
    }
    return 1;
}

References a_1, SN_env::bra, SN_env::c, find_among_b(), SN_local::i_p1, 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 162 of file stem_UTF_8_danish.c.

                                            {
    int among_var;
    {
        int v_1;
        if (z->c < ((SN_local *)z)->i_p1) return 0;
        v_1 = z->lb; z->lb = ((SN_local *)z)->i_p1;
        z->ket = z->c;
        if (z->c <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1851440 >> (z->p[z->c - 1] & 0x1f)) & 1)) { z->lb = v_1; return 0; }
        among_var = find_among_b(z, a_0, 32, 0);
        if (!among_var) { z->lb = v_1; return 0; }
        z->bra = z->c;
        z->lb = v_1;
    }
    switch (among_var) {
        case 1:
            {
                int ret = slice_del(z);
                if (ret < 0) return ret;
            }
            break;
        case 2:
            if (in_grouping_b_U(z, g_s_ending, 97, 229, 0)) return 0;
            {
                int ret = slice_del(z);
                if (ret < 0) return ret;
            }
            break;
    }
    return 1;
}

References a_0, SN_env::c, find_among_b(), g_s_ending, SN_local::i_p1, in_grouping_b_U(), 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 132 of file stem_UTF_8_danish.c.

                                             {
    int i_x;
    ((SN_local *)z)->i_p1 = z->l;
    {
        int v_1 = z->c;
        {
            int ret = skip_utf8(z->p, z->c, z->l, 3);
            if (ret < 0) return 0;
            z->c = ret;
        }
        i_x = z->c;
        z->c = v_1;
    }
    {
        int ret = out_grouping_U(z, g_v, 97, 248, 1);
        if (ret < 0) return 0;
        z->c += ret;
    }
    {
        int ret = in_grouping_U(z, g_v, 97, 248, 1);
        if (ret < 0) return 0;
        z->c += ret;
    }
    ((SN_local *)z)->i_p1 = z->c;
    if (((SN_local *)z)->i_p1 >= i_x) goto lab0;
    ((SN_local *)z)->i_p1 = i_x;
lab0:
    return 1;
}

References SN_env::c, g_v, 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 221 of file stem_UTF_8_danish.c.

                                             {
    int among_var;
    {
        int v_1 = z->l - z->c;
        z->ket = z->c;
        if (!(eq_s_b(z, 2, s_0))) goto lab0;
        z->bra = z->c;
        if (!(eq_s_b(z, 2, s_1))) goto lab0;
        {
            int ret = slice_del(z);
            if (ret < 0) return ret;
        }
    lab0:
        z->c = z->l - v_1;
    }
    {
        int v_2;
        if (z->c < ((SN_local *)z)->i_p1) return 0;
        v_2 = z->lb; z->lb = ((SN_local *)z)->i_p1;
        z->ket = z->c;
        if (z->c - 1 <= z->lb || z->p[z->c - 1] >> 5 != 3 || !((1572992 >> (z->p[z->c - 1] & 0x1f)) & 1)) { z->lb = v_2; return 0; }
        among_var = find_among_b(z, a_2, 5, 0);
        if (!among_var) { z->lb = v_2; return 0; }
        z->bra = z->c;
        z->lb = v_2;
    }
    switch (among_var) {
        case 1:
            {
                int ret = slice_del(z);
                if (ret < 0) return ret;
            }
            {
                int v_3 = z->l - z->c;
                {
                    int ret = r_consonant_pair(z);
                    if (ret < 0) return ret;
                }
                z->c = z->l - v_3;
            }
            break;
        case 2:
            {
                int ret = slice_from_s(z, 4, s_2);
                if (ret < 0) return ret;
            }
            break;
    }
    return 1;
}

References a_2, SN_env::bra, SN_env::c, eq_s_b(), find_among_b(), SN_local::i_p1, SN_env::ket, SN_env::l, 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 272 of file stem_UTF_8_danish.c.

                                         {
    {
        int v_1;
        if (z->c < ((SN_local *)z)->i_p1) return 0;
        v_1 = z->lb; z->lb = ((SN_local *)z)->i_p1;
        z->ket = z->c;
        if (in_grouping_b_U(z, g_c, 98, 122, 0)) { z->lb = v_1; return 0; }
        z->bra = z->c;
        {
            int ret = slice_to(z, &((SN_local *)z)->s_ch);
            if (ret < 0) return ret;
        }
        z->lb = v_1;
    }
    if (!(eq_v_b(z, ((SN_local *)z)->s_ch))) return 0;
    {
        int ret = slice_del(z);
        if (ret < 0) return ret;
    }
    return 1;
}

References SN_env::c, eq_v_b(), g_c, SN_local::i_p1, in_grouping_b_U(), 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 67 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, 0, -1, 0},
{ 2, s_1_1, 0, -1, 0},
{ 2, s_1_2, 0, -1, 0},
{ 2, s_1_3, 0, -1, 0}
}

Definition at line 106 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, 0, 1, 0},
{ 3, s_2_1, -1, 1, 0},
{ 4, s_2_2, -1, 1, 0},
{ 3, s_2_3, 0, 1, 0},
{ 5, s_2_4, 0, 2, 0}
}

Definition at line 118 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 31 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 35 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 36 of file stem_UTF_8_danish.c.

s_0_10

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

static

Definition at line 45 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 46 of file stem_UTF_8_danish.c.

s_0_12

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

static

Definition at line 47 of file stem_UTF_8_danish.c.

s_0_13

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

static

Definition at line 48 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 49 of file stem_UTF_8_danish.c.

s_0_15

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

static

Definition at line 50 of file stem_UTF_8_danish.c.

s_0_16

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

static

Definition at line 51 of file stem_UTF_8_danish.c.

s_0_17

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

static

Definition at line 52 of file stem_UTF_8_danish.c.

s_0_18

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

static

Definition at line 53 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 54 of file stem_UTF_8_danish.c.

s_0_2

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

static

Definition at line 37 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 55 of file stem_UTF_8_danish.c.

s_0_21

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

static

Definition at line 56 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 57 of file stem_UTF_8_danish.c.

s_0_23

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

static

Definition at line 58 of file stem_UTF_8_danish.c.

s_0_24

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

static

Definition at line 59 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 60 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 61 of file stem_UTF_8_danish.c.

s_0_27

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

static

Definition at line 62 of file stem_UTF_8_danish.c.

s_0_28

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

static

Definition at line 63 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 64 of file stem_UTF_8_danish.c.

s_0_3

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

static

Definition at line 38 of file stem_UTF_8_danish.c.

s_0_30

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

static

Definition at line 65 of file stem_UTF_8_danish.c.

s_0_31

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

static

Definition at line 66 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 39 of file stem_UTF_8_danish.c.

s_0_5

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

static

Definition at line 40 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 41 of file stem_UTF_8_danish.c.

s_0_7

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

static

Definition at line 42 of file stem_UTF_8_danish.c.

s_0_8

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

static

Definition at line 43 of file stem_UTF_8_danish.c.

s_0_9

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

static

Definition at line 44 of file stem_UTF_8_danish.c.

s_1

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

static

Definition at line 32 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 102 of file stem_UTF_8_danish.c.

s_1_1

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

static

Definition at line 103 of file stem_UTF_8_danish.c.

s_1_2

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

static

Definition at line 104 of file stem_UTF_8_danish.c.

s_1_3

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

static

Definition at line 105 of file stem_UTF_8_danish.c.

s_2

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

static

Definition at line 33 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 113 of file stem_UTF_8_danish.c.

s_2_1

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

static

Definition at line 114 of file stem_UTF_8_danish.c.

s_2_2

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

static

Definition at line 115 of file stem_UTF_8_danish.c.

s_2_3

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

static

Definition at line 116 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 117 of file stem_UTF_8_danish.c.