stem_UTF_8_indonesian.c File Reference

#include "header.h"

Include dependency graph for stem_UTF_8_indonesian.c:

Go to the source code of this file.

Functions
int	indonesian_UTF_8_stem (struct SN_env *z)
static int	r_VOWEL (struct SN_env *z)
static int	r_SUFFIX_I_OK (struct SN_env *z)
static int	r_SUFFIX_AN_OK (struct SN_env *z)
static int	r_SUFFIX_KAN_OK (struct SN_env *z)
static int	r_KER (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_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 symbol	s_2_2 [3] = { 'k', 'a', 'n' }
static const struct among	a_2 [3]
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 [4] = { 'm', 'e', 'n', 'y' }
static const symbol	s_3_7 [3] = { 'p', 'e', 'm' }
static const symbol	s_3_8 [3] = { 'p', 'e', 'n' }
static const symbol	s_3_9 [4] = { 'p', 'e', 'n', 'g' }
static const symbol	s_3_10 [4] = { 'p', 'e', 'n', 'y' }
static const symbol	s_3_11 [3] = { 't', 'e', 'r' }
static const struct among	a_3 [12]
static const symbol	s_4_0 [2] = { 'b', 'e' }
static const symbol	s_4_1 [7] = { 'b', 'e', 'l', 'a', 'j', 'a', 'r' }
static const symbol	s_4_2 [3] = { 'b', 'e', 'r' }
static const symbol	s_4_3 [2] = { 'p', 'e' }
static const symbol	s_4_4 [7] = { 'p', 'e', 'l', 'a', 'j', 'a', 'r' }
static const symbol	s_4_5 [3] = { 'p', 'e', 'r' }
static const struct among	a_4 [6]
static const unsigned char	g_vowel [] = { 17, 65, 16 }
static const symbol	s_0 [] = { 'e', 'r' }
static const symbol	s_1 [] = { 's' }
static const symbol	s_2 [] = { 's' }
static const symbol	s_3 [] = { 'p' }
static const symbol	s_4 [] = { 'p' }
static const symbol	s_5 [] = { 'a', 'j', 'a', 'r' }
static const symbol	s_6 [] = { 'a', 'j', 'a', 'r' }

Function Documentation

indonesian_UTF_8_close_env()

void indonesian_UTF_8_close_env

(

struct SN_env *

z

)

Definition at line 406 of file stem_UTF_8_indonesian.c.

{ SN_close_env(z, 0); }

References SN_close_env().

indonesian_UTF_8_create_env()

struct SN_env * indonesian_UTF_8_create_env

(

void

)

Definition at line 404 of file stem_UTF_8_indonesian.c.

{ return SN_create_env(0, 2); }

References SN_create_env().

indonesian_UTF_8_stem()

int indonesian_UTF_8_stem

(

struct SN_env *

z

)

Definition at line 312 of file stem_UTF_8_indonesian.c.

                                                    {
    z->I[1] = 0;
    {   int c1 = z->c;
        while(1) {
            int c2 = z->c;
 
            {
                int ret = out_grouping_U(z, g_vowel, 97, 117, 1);
                if (ret < 0) goto lab1;
                z->c += ret;
            }
            z->I[1] += 1;
            continue;
        lab1:
            z->c = c2;
            break;
        }
        z->c = c1;
    }
    if (z->I[1] <= 2) return 0;
    z->I[0] = 0;
    z->lb = z->c; z->c = z->l;
 
    {   int m3 = z->l - z->c; (void)m3;
        {   int ret = r_remove_particle(z);
            if (ret < 0) return ret;
        }
        z->c = z->l - m3;
    }
    if (z->I[1] <= 2) return 0;
    {   int m4 = z->l - z->c; (void)m4;
        {   int ret = r_remove_possessive_pronoun(z);
            if (ret < 0) return ret;
        }
        z->c = z->l - m4;
    }
    z->c = z->lb;
    if (z->I[1] <= 2) return 0;
    {   int c5 = z->c;
        {   int c_test6 = z->c;
            {   int ret = r_remove_first_order_prefix(z);
                if (ret == 0) goto lab3;
                if (ret < 0) return ret;
            }
            {   int c7 = z->c;
                {   int c_test8 = z->c;
                    if (z->I[1] <= 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;
                    z->c = c_test8;
                }
                if (z->I[1] <= 2) goto lab4;
                {   int ret = r_remove_second_order_prefix(z);
                    if (ret == 0) goto lab4;
                    if (ret < 0) return ret;
                }
            lab4:
                z->c = c7;
            }
            z->c = c_test6;
        }
        goto lab2;
    lab3:
        z->c = c5;
        {   int c9 = z->c;
            {   int ret = r_remove_second_order_prefix(z);
                if (ret < 0) return ret;
            }
            z->c = c9;
        }
        {   int c10 = z->c;
            if (z->I[1] <= 2) goto lab5;
            z->lb = z->c; z->c = z->l;
 
            {   int ret = r_remove_suffix(z);
                if (ret == 0) goto lab5;
                if (ret < 0) return ret;
            }
            z->c = z->lb;
        lab5:
            z->c = c10;
        }
    }
lab2:
    return 1;
}

References SN_env::c, g_vowel, SN_env::I, 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_KER()

static int r_KER ( struct SN_env *  z )

static

Definition at line 187 of file stem_UTF_8_indonesian.c.

                                    {
    if (out_grouping_U(z, g_vowel, 97, 117, 0)) return 0;
    if (!(eq_s(z, 2, s_0))) return 0;
    return 1;
}

References eq_s(), g_vowel, out_grouping_U(), and s_0.

r_remove_first_order_prefix()

static int r_remove_first_order_prefix ( struct SN_env *  z )

static

Definition at line 193 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, 12);
    if (!among_var) return 0;
    z->ket = z->c;
    switch (among_var) {
        case 1:
            {   int ret = slice_del(z);
                if (ret < 0) return ret;
            }
            z->I[0] = 1;
            z->I[1] -= 1;
            break;
        case 2:
            {   int ret = slice_del(z);
                if (ret < 0) return ret;
            }
            z->I[0] = 3;
            z->I[1] -= 1;
            break;
        case 3:
            z->I[0] = 1;
            {   int ret = slice_from_s(z, 1, s_1);
                if (ret < 0) return ret;
            }
            z->I[1] -= 1;
            break;
        case 4:
            z->I[0] = 3;
            {   int ret = slice_from_s(z, 1, s_2);
                if (ret < 0) return ret;
            }
            z->I[1] -= 1;
            break;
        case 5:
            z->I[0] = 1;
            z->I[1] -= 1;
            {   int c1 = z->c;
                {   int c2 = z->c;
                    if (in_grouping_U(z, g_vowel, 97, 117, 0)) goto lab1;
                    z->c = c2;
                    {   int ret = slice_from_s(z, 1, s_3);
                        if (ret < 0) return ret;
                    }
                }
                goto lab0;
            lab1:
                z->c = c1;
                {   int ret = slice_del(z);
                    if (ret < 0) return ret;
                }
            }
        lab0:
            break;
        case 6:
            z->I[0] = 3;
            z->I[1] -= 1;
            {   int c3 = z->c;
                {   int c4 = z->c;
                    if (in_grouping_U(z, g_vowel, 97, 117, 0)) goto lab3;
                    z->c = c4;
                    {   int ret = slice_from_s(z, 1, s_4);
                        if (ret < 0) return ret;
                    }
                }
                goto lab2;
            lab3:
                z->c = c3;
                {   int ret = slice_del(z);
                    if (ret < 0) return ret;
                }
            }
        lab2:
            break;
    }
    return 1;
}

References a_3, SN_env::bra, SN_env::c, find_among(), g_vowel, SN_env::I, in_grouping_U(), SN_env::ket, SN_env::l, SN_env::p, s_1, s_2, s_3, s_4, 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 123 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)) return 0;
    z->bra = z->c;
    {   int ret = slice_del(z);
        if (ret < 0) return ret;
    }
    z->I[1] -= 1;
    return 1;
}

References a_0, 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 indonesian_UTF_8_stem().

r_remove_possessive_pronoun()

static int r_remove_possessive_pronoun ( struct SN_env *  z )

static

Definition at line 135 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)) return 0;
    z->bra = z->c;
    {   int ret = slice_del(z);
        if (ret < 0) return ret;
    }
    z->I[1] -= 1;
    return 1;
}

References a_1, 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 indonesian_UTF_8_stem().

r_remove_second_order_prefix()

static int r_remove_second_order_prefix ( struct SN_env *  z )

static

Definition at line 273 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, 6);
    if (!among_var) return 0;
    z->ket = z->c;
    switch (among_var) {
        case 1:
            {   int ret = slice_del(z);
                if (ret < 0) return ret;
            }
            z->I[0] = 2;
            z->I[1] -= 1;
            break;
        case 2:
            {   int ret = slice_from_s(z, 4, s_5);
                if (ret < 0) return ret;
            }
            z->I[1] -= 1;
            break;
        case 3:
            {   int ret = slice_del(z);
                if (ret < 0) return ret;
            }
            z->I[0] = 4;
            z->I[1] -= 1;
            break;
        case 4:
            {   int ret = slice_from_s(z, 4, s_6);
                if (ret < 0) return ret;
            }
            z->I[0] = 4;
            z->I[1] -= 1;
            break;
    }
    return 1;
}

References a_4, SN_env::bra, SN_env::c, find_among(), SN_env::I, SN_env::ket, SN_env::l, SN_env::p, s_5, s_6, slice_del(), and slice_from_s().

Referenced by indonesian_UTF_8_stem().

r_remove_suffix()

static int r_remove_suffix ( struct SN_env *  z )

static

Definition at line 170 of file stem_UTF_8_indonesian.c.

                                              {
    z->ket = z->c;
    if (z->c <= z->lb || (z->p[z->c - 1] != 105 && z->p[z->c - 1] != 110)) return 0;
    if (!find_among_b(z, a_2, 3)) return 0;
    z->bra = z->c;
    {   int ret = slice_del(z);
        if (ret < 0) return ret;
    }
    z->I[1] -= 1;
    return 1;
}

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 indonesian_UTF_8_stem().

r_SUFFIX_AN_OK()

static int r_SUFFIX_AN_OK ( struct SN_env *  z )

static

Definition at line 154 of file stem_UTF_8_indonesian.c.

                                             {
    return z->I[0] != 1;
}

References SN_env::I.

r_SUFFIX_I_OK()

static int r_SUFFIX_I_OK ( struct SN_env *  z )

static

Definition at line 158 of file stem_UTF_8_indonesian.c.

                                            {
    if (z->I[0] > 2) return 0;
    {   int m1 = z->l - z->c; (void)m1;
        if (z->c <= z->lb || z->p[z->c - 1] != 's') goto lab0;
        z->c--;
        return 0;
    lab0:
        z->c = z->l - m1;
    }
    return 1;
}

References SN_env::c, SN_env::I, SN_env::l, SN_env::lb, and SN_env::p.

r_SUFFIX_KAN_OK()

static int r_SUFFIX_KAN_OK ( struct SN_env *  z )

static

Definition at line 147 of file stem_UTF_8_indonesian.c.

                                              {
 
    if (z->I[0] == 3) return 0;
    if (z->I[0] == 2) return 0;
    return 1;
}

References SN_env::I.

r_VOWEL()

static int r_VOWEL ( struct SN_env *  z )

static

Definition at line 182 of file stem_UTF_8_indonesian.c.

                                      {
    if (in_grouping_U(z, g_vowel, 97, 117, 0)) return 0;
    return 1;
}

References g_vowel, and in_grouping_U().

Variable Documentation

a_0

const struct among a_0[3]

static

Initial value:

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

Definition at line 38 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, -1, 1, 0},
{ 2, s_1_1, -1, 1, 0},
{ 2, s_1_2, -1, 1, 0}
}

Definition at line 49 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_possessive_pronoun().

a_2

const struct among a_2[3]

static

Initial value:

=
{
{ 1, s_2_0, -1, 1, r_SUFFIX_I_OK},
{ 2, s_2_1, -1, 1, r_SUFFIX_AN_OK},
{ 3, s_2_2, 1, 1, r_SUFFIX_KAN_OK}
}

Definition at line 60 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_suffix().

a_3

const struct among a_3[12]

static

Initial value:

=
{
{ 2, s_3_0, -1, 1, 0},
{ 2, s_3_1, -1, 2, 0},
{ 2, s_3_2, -1, 1, 0},
{ 3, s_3_3, 2, 5, 0},
{ 3, s_3_4, 2, 1, 0},
{ 4, s_3_5, 4, 1, 0},
{ 4, s_3_6, 4, 3, r_VOWEL},
{ 3, s_3_7, -1, 6, 0},
{ 3, s_3_8, -1, 2, 0},
{ 4, s_3_9, 8, 2, 0},
{ 4, s_3_10, 8, 4, r_VOWEL},
{ 3, s_3_11, -1, 1, 0}
}

Definition at line 80 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_first_order_prefix().

a_4

const struct among a_4[6]

static

Initial value:

=
{
{ 2, s_4_0, -1, 3, r_KER},
{ 7, s_4_1, 0, 4, 0},
{ 3, s_4_2, 0, 3, 0},
{ 2, s_4_3, -1, 1, 0},
{ 7, s_4_4, 3, 2, 0},
{ 3, s_4_5, 3, 1, 0}
}

Definition at line 103 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 113 of file stem_UTF_8_indonesian.c.

Referenced by indonesian_UTF_8_stem(), r_KER(), r_remove_first_order_prefix(), and r_VOWEL().

s_0

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

static

Definition at line 115 of file stem_UTF_8_indonesian.c.

Referenced by r_KER().

s_0_0

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

static

Definition at line 34 of file stem_UTF_8_indonesian.c.

s_0_1

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

static

Definition at line 35 of file stem_UTF_8_indonesian.c.

s_0_2

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

static

Definition at line 36 of file stem_UTF_8_indonesian.c.

s_1

const symbol s_1[] = { 's' }

static

Definition at line 116 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 45 of file stem_UTF_8_indonesian.c.

s_1_1

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

static

Definition at line 46 of file stem_UTF_8_indonesian.c.

s_1_2

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

static

Definition at line 47 of file stem_UTF_8_indonesian.c.

s_2

const symbol s_2[] = { 's' }

static

Definition at line 117 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 56 of file stem_UTF_8_indonesian.c.

s_2_1

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

static

Definition at line 57 of file stem_UTF_8_indonesian.c.

s_2_2

const symbol s_2_2[3] = { 'k', '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 118 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 67 of file stem_UTF_8_indonesian.c.

s_3_1

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

static

Definition at line 68 of file stem_UTF_8_indonesian.c.

s_3_10

const symbol s_3_10[4] = { 'p', 'e', 'n', 'y' }

static

Definition at line 77 of file stem_UTF_8_indonesian.c.

s_3_11

const symbol s_3_11[3] = { 't', 'e', 'r' }

static

Definition at line 78 of file stem_UTF_8_indonesian.c.

s_3_2

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

static

Definition at line 69 of file stem_UTF_8_indonesian.c.

s_3_3

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

static

Definition at line 70 of file stem_UTF_8_indonesian.c.

s_3_4

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

static

Definition at line 71 of file stem_UTF_8_indonesian.c.

s_3_5

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

static

Definition at line 72 of file stem_UTF_8_indonesian.c.

s_3_6

const symbol s_3_6[4] = { 'm', 'e', 'n', 'y' }

static

Definition at line 73 of file stem_UTF_8_indonesian.c.

s_3_7

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

static

Definition at line 74 of file stem_UTF_8_indonesian.c.

s_3_8

const symbol s_3_8[3] = { 'p', 'e', 'n' }

static

Definition at line 75 of file stem_UTF_8_indonesian.c.

s_3_9

const symbol s_3_9[4] = { 'p', 'e', 'n', 'g' }

static

Definition at line 76 of file stem_UTF_8_indonesian.c.

s_4

const symbol s_4[] = { 'p' }

static

Definition at line 119 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_first_order_prefix().

s_4_0

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

static

Definition at line 96 of file stem_UTF_8_indonesian.c.

s_4_1

const symbol s_4_1[7] = { 'b', 'e', 'l', 'a', 'j', 'a', 'r' }

static

Definition at line 97 of file stem_UTF_8_indonesian.c.

s_4_2

const symbol s_4_2[3] = { 'b', 'e', 'r' }

static

Definition at line 98 of file stem_UTF_8_indonesian.c.

s_4_3

const symbol s_4_3[2] = { 'p', 'e' }

static

Definition at line 99 of file stem_UTF_8_indonesian.c.

s_4_4

const symbol s_4_4[7] = { 'p', 'e', 'l', 'a', 'j', 'a', 'r' }

static

Definition at line 100 of file stem_UTF_8_indonesian.c.

s_4_5

const symbol s_4_5[3] = { 'p', 'e', 'r' }

static

Definition at line 101 of file stem_UTF_8_indonesian.c.

s_5

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

static

Definition at line 120 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_second_order_prefix().

s_6

const symbol s_6[] = { 'a', 'j', 'a', 'r' }

static

Definition at line 121 of file stem_UTF_8_indonesian.c.

Referenced by r_remove_second_order_prefix().