stem_UTF_8_indonesian.c File Reference

#include "stem_UTF_8_indonesian.h"
#include <stddef.h>
#include "snowball_runtime.h"

Include dependency graph for stem_UTF_8_indonesian.c:

Go to the source code of this file.

Data Structures
struct	SN_local

Typedefs
typedef struct SN_local	SN_local

Functions
int	indonesian_UTF_8_stem (struct SN_env *z)
static int	r_remove_suffix (struct SN_env *z)
static int	r_remove_second_order_prefix (struct SN_env *z)
static int	r_remove_first_order_prefix (struct SN_env *z)
static int	r_remove_possessive_pronoun (struct SN_env *z)
static int	r_remove_particle (struct SN_env *z)
struct SN_env *	indonesian_UTF_8_create_env (void)
void	indonesian_UTF_8_close_env (struct SN_env *z)

Variables
static const symbol	s_0 [] = { 's' }
static const symbol	s_1 [] = { 's' }
static const symbol	s_2 [] = { 'p' }
static const symbol	s_3 [] = { 'p' }
static const symbol	s_4 [] = { 'a', 'j', 'a', 'r' }
static const symbol	s_5 [] = { 'a', 'j', 'a', 'r' }
static const symbol	s_6 [] = { 'e', 'r' }
static const symbol	s_0_0 [3] = { 'k', 'a', 'h' }
static const symbol	s_0_1 [3] = { 'l', 'a', 'h' }
static const symbol	s_0_2 [3] = { 'p', 'u', 'n' }
static const struct among	a_0 [3]
static const symbol	s_1_0 [3] = { 'n', 'y', 'a' }
static const symbol	s_1_1 [2] = { 'k', 'u' }
static const symbol	s_1_2 [2] = { 'm', 'u' }
static const struct among	a_1 [3]
static const symbol	s_2_0 [1] = { 'i' }
static const symbol	s_2_1 [2] = { 'a', 'n' }
static const struct among	a_2 [2]
static const symbol	s_3_0 [2] = { 'd', 'i' }
static const symbol	s_3_1 [2] = { 'k', 'e' }
static const symbol	s_3_2 [2] = { 'm', 'e' }
static const symbol	s_3_3 [3] = { 'm', 'e', 'm' }
static const symbol	s_3_4 [3] = { 'm', 'e', 'n' }
static const symbol	s_3_5 [4] = { 'm', 'e', 'n', 'g' }
static const symbol	s_3_6 [3] = { 'p', 'e', 'm' }
static const symbol	s_3_7 [3] = { 'p', 'e', 'n' }
static const symbol	s_3_8 [4] = { 'p', 'e', 'n', 'g' }
static const symbol	s_3_9 [3] = { 't', 'e', 'r' }
static const struct among	a_3 [10]
static const symbol	s_4_0 [2] = { 'b', 'e' }
static const symbol	s_4_1 [2] = { 'p', 'e' }
static const struct among	a_4 [2]
static const unsigned char	g_vowel [] = { 17, 65, 16 }

Typedef Documentation

SN_local

typedef struct SN_local SN_local

Definition at line 15 of file stem_UTF_8_indonesian.c.

Function Documentation

indonesian_UTF_8_close_env()

void indonesian_UTF_8_close_env

(

struct SN_env *

z

)

Definition at line 465 of file stem_UTF_8_indonesian.c.

                                                          {
    SN_delete_env(z);
}

References SN_delete_env().

indonesian_UTF_8_create_env()

struct SN_env * indonesian_UTF_8_create_env

(

void

)

Definition at line 456 of file stem_UTF_8_indonesian.c.

                                                         {
    struct SN_env * z = SN_new_env(sizeof(SN_local));
    if (z) {
        ((SN_local *)z)->i_prefix = 0;
        ((SN_local *)z)->i_measure = 0;
    }
    return z;
}

References SN_new_env().

indonesian_UTF_8_stem()

int indonesian_UTF_8_stem

(

struct SN_env *

z

)

Definition at line 353 of file stem_UTF_8_indonesian.c.

                                                    {
    ((SN_local *)z)->i_measure = 0;
    {
        int v_1 = z->c;
        while (1) {
            int v_2 = z->c;
            {
                int ret = out_grouping_U(z, g_vowel, 97, 117, 1);
                if (ret < 0) goto lab1;
                z->c += ret;
            }
            ((SN_local *)z)->i_measure += 1;
            continue;
        lab1:
            z->c = v_2;
            break;
        }
        z->c = v_1;
    }
    if (((SN_local *)z)->i_measure <= 2) return 0;
    ((SN_local *)z)->i_prefix = 0;
    z->lb = z->c; z->c = z->l;
    {
        int v_3 = z->l - z->c;
        {
            int ret = r_remove_particle(z);
            if (ret < 0) return ret;
        }
        z->c = z->l - v_3;
    }
    if (((SN_local *)z)->i_measure <= 2) return 0;
    {
        int v_4 = z->l - z->c;
        {
            int ret = r_remove_possessive_pronoun(z);
            if (ret < 0) return ret;
        }
        z->c = z->l - v_4;
    }
    z->c = z->lb;
    if (((SN_local *)z)->i_measure <= 2) return 0;
    do {
        int v_5 = z->c;
        {
            int v_6 = z->c;
            {
                int ret = r_remove_first_order_prefix(z);
                if (ret == 0) goto lab2;
                if (ret < 0) return ret;
            }
            {
                int v_7 = z->c;
                {
                    int v_8 = z->c;
                    if (((SN_local *)z)->i_measure <= 2) goto lab3;
                    z->lb = z->c; z->c = z->l;
                    {
                        int ret = r_remove_suffix(z);
                        if (ret == 0) goto lab3;
                        if (ret < 0) return ret;
                    }
                    z->c = z->lb;
                    z->c = v_8;
                }
                if (((SN_local *)z)->i_measure <= 2) goto lab3;
                {
                    int ret = r_remove_second_order_prefix(z);
                    if (ret == 0) goto lab3;
                    if (ret < 0) return ret;
                }
            lab3:
                z->c = v_7;
            }
            z->c = v_6;
        }
        break;
    lab2:
        z->c = v_5;
        {
            int v_9 = z->c;
            {
                int ret = r_remove_second_order_prefix(z);
                if (ret < 0) return ret;
            }
            z->c = v_9;
        }
        {
            int v_10 = z->c;
            if (((SN_local *)z)->i_measure <= 2) goto lab4;
            z->lb = z->c; z->c = z->l;
            {
                int ret = r_remove_suffix(z);
                if (ret == 0) goto lab4;
                if (ret < 0) return ret;
            }
            z->c = z->lb;
        lab4:
            z->c = v_10;
        }
    } while (0);
    return 1;
}

References SN_env::c, g_vowel, SN_env::l, SN_env::lb, out_grouping_U(), r_remove_first_order_prefix(), r_remove_particle(), r_remove_possessive_pronoun(), r_remove_second_order_prefix(), and r_remove_suffix().

r_remove_first_order_prefix()

static int r_remove_first_order_prefix ( struct SN_env *  z )

static

Definition at line 164 of file stem_UTF_8_indonesian.c.

                                                          {
    int among_var;
    z->bra = z->c;
    if (z->c + 1 >= z->l || (z->p[z->c + 1] != 105 && z->p[z->c + 1] != 101)) return 0;
    among_var = find_among(z, a_3, 10, 0);
    if (!among_var) return 0;
    z->ket = z->c;
    switch (among_var) {
        case 1:
            {
                int ret = slice_del(z);
                if (ret < 0) return ret;
            }
            ((SN_local *)z)->i_prefix = 1;
            ((SN_local *)z)->i_measure -= 1;
            break;
        case 2:
            do {
                int v_1 = z->c;
                if (z->c == z->l || z->p[z->c] != 'y') goto lab0;
                z->c++;
                {
                    int v_2 = z->c;
                    if (in_grouping_U(z, g_vowel, 97, 117, 0)) goto lab0;
                    z->c = v_2;
                }
                z->ket = z->c;
                {
                    int ret = slice_from_s(z, 1, s_0);
                    if (ret < 0) return ret;
                }
                ((SN_local *)z)->i_prefix = 1;
                ((SN_local *)z)->i_measure -= 1;
                break;
            lab0:
                z->c = v_1;
                {
                    int ret = slice_del(z);
                    if (ret < 0) return ret;
                }
                ((SN_local *)z)->i_prefix = 1;
                ((SN_local *)z)->i_measure -= 1;
            } while (0);
            break;
        case 3:
            {
                int ret = slice_del(z);
                if (ret < 0) return ret;
            }
            ((SN_local *)z)->i_prefix = 3;
            ((SN_local *)z)->i_measure -= 1;
            break;
        case 4:
            do {
                int v_3 = z->c;
                if (z->c == z->l || z->p[z->c] != 'y') goto lab1;
                z->c++;
                {
                    int v_4 = z->c;
                    if (in_grouping_U(z, g_vowel, 97, 117, 0)) goto lab1;
                    z->c = v_4;
                }
                z->ket = z->c;
                {
                    int ret = slice_from_s(z, 1, s_1);
                    if (ret < 0) return ret;
                }
                ((SN_local *)z)->i_prefix = 3;
                ((SN_local *)z)->i_measure -= 1;
                break;
            lab1:
                z->c = v_3;
                {
                    int ret = slice_del(z);
                    if (ret < 0) return ret;
                }
                ((SN_local *)z)->i_prefix = 3;
                ((SN_local *)z)->i_measure -= 1;
            } while (0);
            break;
        case 5:
            ((SN_local *)z)->i_prefix = 1;
            ((SN_local *)z)->i_measure -= 1;
            do {
                int v_5 = z->c;
                {
                    int v_6 = z->c;
                    if (in_grouping_U(z, g_vowel, 97, 117, 0)) goto lab2;
                    z->c = v_6;
                    {
                        int ret = slice_from_s(z, 1, s_2);
                        if (ret < 0) return ret;
                    }
                }
                break;
            lab2:
                z->c = v_5;
                {
                    int ret = slice_del(z);
                    if (ret < 0) return ret;
                }
            } while (0);
            break;
        case 6:
            ((SN_local *)z)->i_prefix = 3;
            ((SN_local *)z)->i_measure -= 1;
            do {
                int v_7 = z->c;
                {
                    int v_8 = z->c;
                    if (in_grouping_U(z, g_vowel, 97, 117, 0)) goto lab3;
                    z->c = v_8;
                    {
                        int ret = slice_from_s(z, 1, s_3);
                        if (ret < 0) return ret;
                    }
                }
                break;
            lab3:
                z->c = v_7;
                {
                    int ret = slice_del(z);
                    if (ret < 0) return ret;
                }
            } while (0);
            break;
    }
    return 1;
}

References a_3, SN_env::bra, SN_env::c, find_among(), g_vowel, in_grouping_U(), SN_env::ket, SN_env::l, SN_env::p, s_0, s_1, s_2, s_3, slice_del(), and slice_from_s().

Referenced by indonesian_UTF_8_stem().

r_remove_particle()

static int r_remove_particle ( struct SN_env *  z )

static

Definition at line 96 of file stem_UTF_8_indonesian.c.

                                                {
    z->ket = z->c;
    if (z->c - 2 <= z->lb || (z->p[z->c - 1] != 104 && z->p[z->c - 1] != 110)) return 0;
    if (!find_among_b(z, a_0, 3, 0)) return 0;
    z->bra = z->c;
    {
        int ret = slice_del(z);
        if (ret < 0) return ret;
    }
    ((SN_local *)z)->i_measure -= 1;
    return 1;
}

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

Referenced by indonesian_UTF_8_stem().

r_remove_possessive_pronoun()

static int r_remove_possessive_pronoun ( struct SN_env *  z )

static

Definition at line 109 of file stem_UTF_8_indonesian.c.

                                                          {
    z->ket = z->c;
    if (z->c - 1 <= z->lb || (z->p[z->c - 1] != 97 && z->p[z->c - 1] != 117)) return 0;
    if (!find_among_b(z, a_1, 3, 0)) return 0;
    z->bra = z->c;
    {
        int ret = slice_del(z);
        if (ret < 0) return ret;
    }
    ((SN_local *)z)->i_measure -= 1;
    return 1;
}

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

Referenced by indonesian_UTF_8_stem().

r_remove_second_order_prefix()

static int r_remove_second_order_prefix ( struct SN_env *  z )

static

Definition at line 294 of file stem_UTF_8_indonesian.c.

                                                           {
    int among_var;
    z->bra = z->c;
    if (z->c + 1 >= z->l || z->p[z->c + 1] != 101) return 0;
    among_var = find_among(z, a_4, 2, 0);
    if (!among_var) return 0;
    switch (among_var) {
        case 1:
            do {
                int v_1 = z->c;
                if (z->c == z->l || z->p[z->c] != 'r') goto lab0;
                z->c++;
                z->ket = z->c;
                ((SN_local *)z)->i_prefix = 2;
                break;
            lab0:
                z->c = v_1;
                if (z->c == z->l || z->p[z->c] != 'l') goto lab1;
                z->c++;
                z->ket = z->c;
                if (!(eq_s(z, 4, s_4))) goto lab1;
                break;
            lab1:
                z->c = v_1;
                z->ket = z->c;
                ((SN_local *)z)->i_prefix = 2;
            } while (0);
            break;
        case 2:
            do {
                int v_2 = z->c;
                if (z->c == z->l || z->p[z->c] != 'r') goto lab2;
                z->c++;
                z->ket = z->c;
                break;
            lab2:
                z->c = v_2;
                if (z->c == z->l || z->p[z->c] != 'l') goto lab3;
                z->c++;
                z->ket = z->c;
                if (!(eq_s(z, 4, s_5))) goto lab3;
                break;
            lab3:
                z->c = v_2;
                z->ket = z->c;
                if (out_grouping_U(z, g_vowel, 97, 117, 0)) return 0;
                if (!(eq_s(z, 2, s_6))) return 0;
            } while (0);
            ((SN_local *)z)->i_prefix = 4;
            break;
    }
    ((SN_local *)z)->i_measure -= 1;
    {
        int ret = slice_del(z);
        if (ret < 0) return ret;
    }
    return 1;
}

References a_4, SN_env::bra, SN_env::c, eq_s(), find_among(), g_vowel, SN_env::ket, SN_env::l, out_grouping_U(), SN_env::p, s_4, s_5, s_6, and slice_del().

Referenced by indonesian_UTF_8_stem().

r_remove_suffix()

static int r_remove_suffix ( struct SN_env *  z )

static

Definition at line 122 of file stem_UTF_8_indonesian.c.

                                              {
    int among_var;
    z->ket = z->c;
    if (z->c <= z->lb || (z->p[z->c - 1] != 105 && z->p[z->c - 1] != 110)) return 0;
    among_var = find_among_b(z, a_2, 2, 0);
    if (!among_var) return 0;
    z->bra = z->c;
    switch (among_var) {
        case 1:
            do {
                int v_1 = z->l - z->c;
                if (((SN_local *)z)->i_prefix == 3) goto lab0;
                if (((SN_local *)z)->i_prefix == 2) goto lab0;
                if (z->c <= z->lb || z->p[z->c - 1] != 'k') goto lab0;
                z->c--;
                z->bra = z->c;
                break;
            lab0:
                z->c = z->l - v_1;
                if (((SN_local *)z)->i_prefix == 1) return 0;
            } while (0);
            break;
        case 2:
            if (((SN_local *)z)->i_prefix > 2) return 0;
            {
                int v_2 = z->l - z->c;
                if (z->c <= z->lb || z->p[z->c - 1] != 's') goto lab1;
                z->c--;
                return 0;
            lab1:
                z->c = z->l - v_2;
            }
            break;
    }
    {
        int ret = slice_del(z);
        if (ret < 0) return ret;
    }
    ((SN_local *)z)->i_measure -= 1;
    return 1;
}

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

Referenced by indonesian_UTF_8_stem().

Variable Documentation

a_0

const struct among a_0[3]

static

Initial value:

= {
{ 3, s_0_0, 0, 1, 0},
{ 3, s_0_1, 0, 1, 0},
{ 3, s_0_2, 0, 1, 0}
}

Definition at line 42 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_particle().

a_1

const struct among a_1[3]

static

Initial value:

= {
{ 3, s_1_0, 0, 1, 0},
{ 2, s_1_1, 0, 1, 0},
{ 2, s_1_2, 0, 1, 0}
}

Definition at line 51 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_possessive_pronoun().

a_2

const struct among a_2[2]

static

Initial value:

= {
{ 1, s_2_0, 0, 2, 0},
{ 2, s_2_1, 0, 1, 0}
}

Definition at line 59 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_suffix().

a_3

const struct among a_3[10]

static

Initial value:

= {
{ 2, s_3_0, 0, 1, 0},
{ 2, s_3_1, 0, 3, 0},
{ 2, s_3_2, 0, 1, 0},
{ 3, s_3_3, -1, 5, 0},
{ 3, s_3_4, -2, 2, 0},
{ 4, s_3_5, -1, 1, 0},
{ 3, s_3_6, 0, 6, 0},
{ 3, s_3_7, 0, 4, 0},
{ 4, s_3_8, -1, 3, 0},
{ 3, s_3_9, 0, 1, 0}
}

Definition at line 74 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_first_order_prefix().

a_4

const struct among a_4[2]

static

Initial value:

= {
{ 2, s_4_0, 0, 2, 0},
{ 2, s_4_1, 0, 1, 0}
}

Definition at line 89 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_second_order_prefix().

g_vowel

const unsigned char g_vowel[] = { 17, 65, 16 }

static

Definition at line 94 of file stem_UTF_8_indonesian.c.

Referenced by indonesian_UTF_8_stem(), r_remove_first_order_prefix(), and r_remove_second_order_prefix().

s_0

const symbol s_0[] = { 's' }

static

Definition at line 31 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_first_order_prefix().

s_0_0

const symbol s_0_0[3] = { 'k', 'a', 'h' }

static

Definition at line 39 of file stem_UTF_8_indonesian.c.

s_0_1

const symbol s_0_1[3] = { 'l', 'a', 'h' }

static

Definition at line 40 of file stem_UTF_8_indonesian.c.

s_0_2

const symbol s_0_2[3] = { 'p', 'u', 'n' }

static

Definition at line 41 of file stem_UTF_8_indonesian.c.

s_1

const symbol s_1[] = { 's' }

static

Definition at line 32 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_first_order_prefix().

s_1_0

const symbol s_1_0[3] = { 'n', 'y', 'a' }

static

Definition at line 48 of file stem_UTF_8_indonesian.c.

s_1_1

const symbol s_1_1[2] = { 'k', 'u' }

static

Definition at line 49 of file stem_UTF_8_indonesian.c.

s_1_2

const symbol s_1_2[2] = { 'm', 'u' }

static

Definition at line 50 of file stem_UTF_8_indonesian.c.

s_2

const symbol s_2[] = { 'p' }

static

Definition at line 33 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_first_order_prefix().

s_2_0

const symbol s_2_0[1] = { 'i' }

static

Definition at line 57 of file stem_UTF_8_indonesian.c.

s_2_1

const symbol s_2_1[2] = { 'a', 'n' }

static

Definition at line 58 of file stem_UTF_8_indonesian.c.

s_3

const symbol s_3[] = { 'p' }

static

Definition at line 34 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_first_order_prefix().

s_3_0

const symbol s_3_0[2] = { 'd', 'i' }

static

Definition at line 64 of file stem_UTF_8_indonesian.c.

s_3_1

const symbol s_3_1[2] = { 'k', 'e' }

static

Definition at line 65 of file stem_UTF_8_indonesian.c.

s_3_2

const symbol s_3_2[2] = { 'm', 'e' }

static

Definition at line 66 of file stem_UTF_8_indonesian.c.

s_3_3

const symbol s_3_3[3] = { 'm', 'e', 'm' }

static

Definition at line 67 of file stem_UTF_8_indonesian.c.

s_3_4

const symbol s_3_4[3] = { 'm', 'e', 'n' }

static

Definition at line 68 of file stem_UTF_8_indonesian.c.

s_3_5

const symbol s_3_5[4] = { 'm', 'e', 'n', 'g' }

static

Definition at line 69 of file stem_UTF_8_indonesian.c.

s_3_6

const symbol s_3_6[3] = { 'p', 'e', 'm' }

static

Definition at line 70 of file stem_UTF_8_indonesian.c.

s_3_7

const symbol s_3_7[3] = { 'p', 'e', 'n' }

static

Definition at line 71 of file stem_UTF_8_indonesian.c.

s_3_8

const symbol s_3_8[4] = { 'p', 'e', 'n', 'g' }

static

Definition at line 72 of file stem_UTF_8_indonesian.c.

s_3_9

const symbol s_3_9[3] = { 't', 'e', 'r' }

static

Definition at line 73 of file stem_UTF_8_indonesian.c.

s_4

const symbol s_4[] = { 'a', 'j', 'a', 'r' }

static

Definition at line 35 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_second_order_prefix().

s_4_0

const symbol s_4_0[2] = { 'b', 'e' }

static

Definition at line 87 of file stem_UTF_8_indonesian.c.

s_4_1

const symbol s_4_1[2] = { 'p', 'e' }

static

Definition at line 88 of file stem_UTF_8_indonesian.c.

s_5

const symbol s_5[] = { 'a', 'j', 'a', 'r' }

static

Definition at line 36 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_second_order_prefix().

s_6

const symbol s_6[] = { 'e', 'r' }

static

Definition at line 37 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_second_order_prefix().