utilities.c File Reference

#include "snowball_runtime.h"

Include dependency graph for utilities.c:

Go to the source code of this file.

Macros
#define	SNOWBALL_RETURN_OK   return 0
#define	SNOWBALL_RETURN_OR_THROW(R, E)   return R
#define	SNOWBALL_PROPAGATE_ERR(F)
#define	CREATE_SIZE   1
#define	REPLACE_S(Z, B, K, SIZE, S)    SNOWBALL_PROPAGATE_ERR(replace_s(Z, B, K, SIZE, S))
#define	SLICE_CHECK(Z)   SNOWBALL_PROPAGATE_ERR(slice_check(Z))

Functions
symbol *	create_s (void)
void	lose_s (symbol *p)
int	skip_utf8 (const symbol *p, int c, int limit, int n)
int	skip_b_utf8 (const symbol *p, int c, int limit, int n)
static int	get_utf8 (const symbol *p, int c, int l, int *slot)
static int	get_b_utf8 (const symbol *p, int c, int lb, int *slot)
int	in_grouping_U (struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
int	in_grouping_b_U (struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
int	out_grouping_U (struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
int	out_grouping_b_U (struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
int	in_grouping (struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
int	in_grouping_b (struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
int	out_grouping (struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
int	out_grouping_b (struct SN_env *z, const unsigned char *s, int min, int max, int repeat)
int	eq_s (struct SN_env *z, int s_size, const symbol *s)
int	eq_s_b (struct SN_env *z, int s_size, const symbol *s)
int	eq_v (struct SN_env *z, const symbol *p)
int	eq_v_b (struct SN_env *z, const symbol *p)
int	find_among (struct SN_env *z, const struct among *v, int v_size, int(*call_among_func)(struct SN_env *))
int	find_among_b (struct SN_env *z, const struct among *v, int v_size, int(*call_among_func)(struct SN_env *))
static int	increase_size (symbol **p, int n)
SNOWBALL_ERR	replace_s (struct SN_env *z, int c_bra, int c_ket, int s_size, const symbol *s)
static SNOWBALL_ERR	slice_check (struct SN_env *z)
SNOWBALL_ERR	slice_from_s (struct SN_env *z, int s_size, const symbol *s)
SNOWBALL_ERR	slice_from_v (struct SN_env *z, const symbol *p)
SNOWBALL_ERR	slice_del (struct SN_env *z)
SNOWBALL_ERR	insert_s (struct SN_env *z, int bra, int ket, int s_size, const symbol *s)
SNOWBALL_ERR	insert_v (struct SN_env *z, int bra, int ket, const symbol *p)
SNOWBALL_ERR	slice_to (struct SN_env *z, symbol **p)
SNOWBALL_ERR	assign_to (struct SN_env *z, symbol **p)
int	len_utf8 (const symbol *p)

Macro Definition Documentation

CREATE_SIZE

#define CREATE_SIZE   1

Definition at line 18 of file utilities.c.

REPLACE_S

#define REPLACE_S	(		Z,
			B,
			K,
			SIZE,
			S
	)		SNOWBALL_PROPAGATE_ERR(replace_s(Z, B, K, SIZE, S))

Definition at line 411 of file utilities.c.

SLICE_CHECK

#define SLICE_CHECK

(

Z

)

SNOWBALL_PROPAGATE_ERR(slice_check(Z))

Definition at line 430 of file utilities.c.

SNOWBALL_PROPAGATE_ERR

#define SNOWBALL_PROPAGATE_ERR

(

F

)

Value:

        do { \
        int snowball_err = F; \
        if (snowball_err < 0) return snowball_err; \
    } while (0)

Definition at line 12 of file utilities.c.

SNOWBALL_RETURN_OK

#define SNOWBALL_RETURN_OK   return 0

Definition at line 10 of file utilities.c.

SNOWBALL_RETURN_OR_THROW

#define SNOWBALL_RETURN_OR_THROW	(		R,
			E
	)		return R

Definition at line 11 of file utilities.c.

Function Documentation

assign_to()

SNOWBALL_ERR assign_to	(	struct SN_env *	z,
		symbol **	p
	)

Definition at line 491 of file utilities.c.

                                                              {
    int len = z->l;
    if (CAPACITY(*p) < len) {
        SNOWBALL_PROPAGATE_ERR(increase_size(p, len));
    }
    memmove(*p, z->p, len * sizeof(symbol));
    SET_SIZE(*p, len);
    SNOWBALL_RETURN_OK;
}

References CAPACITY, increase_size(), SN_env::l, len, SN_env::p, SET_SIZE, SNOWBALL_PROPAGATE_ERR, and SNOWBALL_RETURN_OK.

create_s()

symbol * create_s

(

void

)

Definition at line 20 of file utilities.c.

                               {
    symbol * p;
    void * mem = malloc(HEAD + (CREATE_SIZE + 1) * sizeof(symbol));
    if (mem == NULL)
        SNOWBALL_RETURN_OR_THROW(NULL, std::bad_alloc());
    p = (symbol *) (HEAD + (char *) mem);
    CAPACITY(p) = CREATE_SIZE;
    SET_SIZE(p, 0);
    return p;
}

References CAPACITY, CREATE_SIZE, HEAD, malloc, SET_SIZE, and SNOWBALL_RETURN_OR_THROW.

Referenced by danish_ISO_8859_1_create_env(), danish_UTF_8_create_env(), dutch_ISO_8859_1_create_env(), dutch_UTF_8_create_env(), finnish_ISO_8859_1_create_env(), finnish_UTF_8_create_env(), and SN_new_env().

eq_s()

int eq_s	(	struct SN_env *	z,
		int	s_size,
		const symbol *	s
	)

Definition at line 231 of file utilities.c.

                                                                 {
    if (z->l - z->c < s_size || memcmp(z->p + z->c, s, s_size * sizeof(symbol)) != 0) return 0;
    z->c += s_size; return 1;
}

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

Referenced by eq_v(), r_elisions(), r_initial_apostrophe(), r_Lose_infix(), r_Lose_prefix(), r_mark_regions(), r_measure(), r_Prefix_Step2(), r_prelude(), r_remove_pronoun_prefixes(), r_remove_question_prefixes(), r_remove_second_order_prefix(), and russian_UTF_8_stem().

eq_s_b()

int eq_s_b	(	struct SN_env *	z,
		int	s_size,
		const symbol *	s
	)

Definition at line 236 of file utilities.c.

                                                                   {
    if (z->c - z->lb < s_size || memcmp(z->p + z->c - s_size, s, s_size * sizeof(symbol)) != 0) return 0;
    z->c -= s_size; return 1;
}

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

Referenced by eq_v_b(), french_UTF_8_stem(), r_adjectival(), r_append_U_to_stems_ending_with_d_or_g(), r_C(), r_case_ending(), r_check_vowel_harmony(), r_en_ending(), r_final_apostrophe(), r_fix_ending(), r_fix_gd(), r_is_reserved_word(), r_main_suffix(), r_mark_ki(), r_mark_yken(), r_other_endings(), r_other_suffix(), r_perfective_gerund(), r_possessive(), r_remove_category_1(), r_remove_category_2(), r_remove_tense_suffix(), r_remove_um(), r_remove_vetrumai_urupukal(), r_residual_suffix(), r_shortv(), r_standard_suffix(), r_step_0(), r_Step_1(), r_Step_1c(), r_Step_2(), r_Step_3(), r_Step_4(), r_step_5a(), r_step_5b(), r_step_5c(), r_step_5d(), r_step_5e(), r_step_5f(), r_step_5i(), r_step_5k(), r_step_s3(), r_step_s8(), r_Suffix_All_alef_maqsura(), r_Suffix_Noun_Step1b(), r_Suffix_Noun_Step2b(), r_Suffix_Noun_Step2c1(), r_Suffix_Noun_Step2c2(), r_Suffix_Noun_Step3(), r_t_plural(), r_tidy_up(), r_un_accent(), r_V(), r_verb(), r_VX(), and russian_UTF_8_stem().

eq_v()

int eq_v	(	struct SN_env *	z,
		const symbol *	p
	)

Definition at line 241 of file utilities.c.

                                                     {
    return eq_s(z, SIZE(p), p);
}

References eq_s(), and SIZE.

eq_v_b()

int eq_v_b	(	struct SN_env *	z,
		const symbol *	p
	)

Definition at line 245 of file utilities.c.

                                                       {
    return eq_s_b(z, SIZE(p), p);
}

References eq_s_b(), and SIZE.

Referenced by r_tidy(), and r_undouble().

find_among()

int find_among	(	struct SN_env *	z,
		const struct among *	v,
		int	v_size,
		int(*)(struct SN_env *)	call_among_func
	)

Definition at line 249 of file utilities.c.

                                                              {
 
    int i = 0;
    int j = v_size;
 
    int c = z->c; int l = z->l;
    const symbol * q = z->p + c;
 
    const struct among * w;
 
    int common_i = 0;
    int common_j = 0;
 
    int first_key_inspected = 0;
 
    while (1) {
        int k = i + ((j - i) >> 1);
        int diff = 0;
        int common = common_i < common_j ? common_i : common_j; /* smaller */
        w = v + k;
        {
            int i2; for (i2 = common; i2 < w->s_size; i2++) {
                if (c + common == l) { diff = -1; break; }
                diff = q[common] - w->s[i2];
                if (diff != 0) break;
                common++;
            }
        }
        if (diff < 0) {
            j = k;
            common_j = common;
        } else {
            i = k;
            common_i = common;
        }
        if (j - i <= 1) {
            if (i > 0) break; /* v->s has been inspected */
            if (j == i) break; /* only one item in v */
 
            /* - but now we need to go round once more to get
               v->s inspected. This looks messy, but is actually
               the optimal approach.  */
 
            if (first_key_inspected) break;
            first_key_inspected = 1;
        }
    }
    w = v + i;
    while (1) {
        if (common_i >= w->s_size) {
            z->c = c + w->s_size;
            if (!w->function) return w->result;
            z->af = w->function;
            if (call_among_func(z)) {
                z->c = c + w->s_size;
                return w->result;
            }
        }
        if (!w->substring_i) return 0;
        w += w->substring_i;
    }
}

References SN_env::af, SN_env::c, among::function, i, j, SN_env::l, SN_env::p, among::result, among::s, among::s_size, and among::substring_i.

Referenced by r_canonical_form(), r_Checks1(), r_cleaning(), r_cyr_to_lat(), r_exception1(), r_fix_va_start(), r_initial_apostrophe(), r_initial_morph(), r_Lose_infix(), r_Lose_prefix(), r_mark_regions(), r_norm(), r_Normalize_post(), r_Normalize_pre(), r_postlude(), r_Prefix_Step1(), r_Prefix_Step2(), r_Prefix_Step3_Verb(), r_Prefix_Step3a_Noun(), r_Prefix_Step3b_Noun(), r_Prefix_Step4_Verb(), r_prelude(), r_remove_first_order_prefix(), r_remove_pronoun_prefixes(), r_remove_question_prefixes(), r_remove_second_order_prefix(), and r_verb_exceptions().

find_among_b()

int find_among_b	(	struct SN_env *	z,
		const struct among *	v,
		int	v_size,
		int(*)(struct SN_env *)	call_among_func
	)

Definition at line 315 of file utilities.c.

                                                                {
 
    int i = 0;
    int j = v_size;
 
    int c = z->c; int lb = z->lb;
    const symbol * q = z->p + c - 1;
 
    const struct among * w;
 
    int common_i = 0;
    int common_j = 0;
 
    int first_key_inspected = 0;
 
    while (1) {
        int k = i + ((j - i) >> 1);
        int diff = 0;
        int common = common_i < common_j ? common_i : common_j;
        w = v + k;
        {
            int i2; for (i2 = w->s_size - 1 - common; i2 >= 0; i2--) {
                if (c - common == lb) { diff = -1; break; }
                diff = q[- common] - w->s[i2];
                if (diff != 0) break;
                common++;
            }
        }
        if (diff < 0) { j = k; common_j = common; }
                 else { i = k; common_i = common; }
        if (j - i <= 1) {
            if (i > 0) break;
            if (j == i) break;
            if (first_key_inspected) break;
            first_key_inspected = 1;
        }
    }
    w = v + i;
    while (1) {
        if (common_i >= w->s_size) {
            z->c = c - w->s_size;
            if (!w->function) return w->result;
            z->af = w->function;
            if (call_among_func(z)) {
                z->c = c - w->s_size;
                return w->result;
            }
        }
        if (!w->substring_i) return 0;
        w += w->substring_i;
    }
}

References SN_env::af, SN_env::c, among::function, i, j, SN_env::lb, SN_env::p, among::result, among::s, among::s_size, and among::substring_i.

Referenced by hindi_UTF_8_stem(), r_aditzak(), r_adjectival(), r_adjective(), r_adjetiboak(), r_attached_pronoun(), r_case(), r_case_ending(), r_case_other(), r_case_special(), r_combo_suffix(), r_consonant_pair(), r_correlative(), r_degrees(), r_deriv(), r_derivational(), r_double(), r_emphasis(), r_ending(), r_et_condition(), r_factive(), r_final_apostrophe(), r_fix_chdz(), r_fix_conflicts(), r_fix_ending(), r_i_verb_suffix(), r_instrum(), r_izenak(), r_lengthen_V(), r_LONG(), r_LONGV(), r_main_suffix(), r_mark_cAsInA(), r_mark_DA(), r_mark_DAn(), r_mark_DUr(), r_mark_lAr(), r_mark_lArI(), r_mark_nA(), r_mark_ncA(), r_mark_ndA(), r_mark_ndAn(), r_mark_nU(), r_mark_nUn(), r_mark_nUz(), r_mark_possessives(), r_mark_sUn(), r_mark_sUnUz(), r_mark_yDU(), r_mark_ylA(), r_mark_ymUs_(), r_mark_ysA(), r_mark_yUm(), r_mark_yUz(), r_merged_numeral(), r_normalize_consonant(), r_Normalize_post(), r_noun(), r_noun_sfx(), r_nu(), r_other_endings(), r_other_suffix(), r_owned(), r_particle_etc(), r_perfective_gerund(), r_plur_owner(), r_plural(), r_plural_three_first_cases(), r_possessive(), r_post_process_last_consonants(), r_pronoun(), r_reflexive(), r_remove_category_1(), r_remove_category_2(), r_remove_category_3(), r_remove_command_suffixes(), r_remove_common_word_endings(), r_remove_endings(), r_remove_particle(), r_remove_plural_suffix(), r_remove_possessive_pronoun(), r_remove_question_suffixes(), r_remove_suffix(), r_remove_tense_suffix(), r_remove_vetrumai_urupukal(), r_residual_form(), r_residual_suffix(), r_sing_owner(), r_special_noun_endings(), r_standard_suffix(), r_step1(), r_step2(), r_step_0(), r_Step_1(), r_step_1(), r_Step_1a(), r_Step_1b(), r_Step_1c(), r_Step_2(), r_step_2a(), r_step_2b(), r_step_2c(), r_step_2d(), r_Step_3(), r_step_3(), r_Step_4(), r_step_4(), r_Step_5(), r_step_5a(), r_step_5b(), r_step_5c(), r_step_5d(), r_step_5e(), r_step_5f(), r_step_5g(), r_step_5h(), r_step_5i(), r_step_5j(), r_step_5k(), r_step_5l(), r_step_5m(), r_Step_6(), r_step_6(), r_Step_7(), r_step_7(), r_step_s1(), r_step_s10(), r_step_s2(), r_step_s3(), r_step_s4(), r_step_s5(), r_step_s6(), r_step_s7(), r_step_s8(), r_step_s9(), r_Suffix_Noun_Step1a(), r_Suffix_Noun_Step2a(), r_Suffix_Verb_Step1(), r_Suffix_Verb_Step2a(), r_Suffix_Verb_Step2b(), r_Suffix_Verb_Step2c(), r_t_plural(), r_tidy_up(), r_tolower(), r_ujn_suffix(), r_un_double(), r_undouble(), r_undouble_kpt(), r_uninflected(), r_v_ending(), r_verb(), r_verb_sfx(), r_verb_suffix(), r_vowel_suffix(), and r_y_verb_suffix().

get_b_utf8()

static int get_b_utf8 ( const symbol *  p, int  c, int  lb, int *  slot  )

static

Definition at line 109 of file utilities.c.

                                                                   {
    int a, b;
    if (c <= lb) return 0;
    b = p[--c];
    if (b < 0x80 || c == lb) {   /* 1000 0000 */
        *slot = b;
        return 1;
    }
    a = b & 0x3F;
    b = p[--c];
    if (b >= 0xC0 || c == lb) {   /* 1100 0000 */
        *slot = (b & 0x1F) << 6 | a;
        return 2;
    }
    a |= (b & 0x3F) << 6;
    b = p[--c];
    if (b >= 0xE0 || c == lb) {   /* 1110 0000 */
        *slot = (b & 0xF) << 12 | a;
        return 3;
    }
    *slot = (p[--c] & 0x7) << 18 | (b & 0x3F) << 12 | a;
    return 4;
}

References a, and b.

Referenced by in_grouping_b_U(), and out_grouping_b_U().

get_utf8()

static int get_utf8 ( const symbol *  p, int  c, int  l, int *  slot  )

static

Definition at line 87 of file utilities.c.

                                                                {
    int b0, b1, b2;
    if (c >= l) return 0;
    b0 = p[c++];
    if (b0 < 0xC0 || c == l) {   /* 1100 0000 */
        *slot = b0;
        return 1;
    }
    b1 = p[c++] & 0x3F;
    if (b0 < 0xE0 || c == l) {   /* 1110 0000 */
        *slot = (b0 & 0x1F) << 6 | b1;
        return 2;
    }
    b2 = p[c++] & 0x3F;
    if (b0 < 0xF0 || c == l) {   /* 1111 0000 */
        *slot = (b0 & 0xF) << 12 | b1 << 6 | b2;
        return 3;
    }
    *slot = (b0 & 0x7) << 18 | b1 << 12 | b2 << 6 | (p[c] & 0x3F);
    return 4;
}

Referenced by in_grouping_U(), and out_grouping_U().

in_grouping()

int in_grouping	(	struct SN_env *	z,
		const unsigned char *	s,
		int	min,
		int	max,
		int	repeat
	)

Definition at line 183 of file utilities.c.

                                                                                                 {
    do {
        int ch;
        if (z->c >= z->l) return -1;
        ch = z->p[z->c];
        if (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0)
            return 1;
        z->c++;
    } while (repeat);
    return 0;
}

References SN_env::c, SN_env::l, SN_env::p, and repeat().

Referenced by porter_ISO_8859_1_stem(), r_elisions(), r_Lose_infix(), r_Lose_prefix(), r_mark_regions(), r_measure(), r_prelude(), and r_remove_first_order_prefix().

in_grouping_b()

int in_grouping_b	(	struct SN_env *	z,
		const unsigned char *	s,
		int	min,
		int	max,
		int	repeat
	)

Definition at line 195 of file utilities.c.

                                                                                                   {
    do {
        int ch;
        if (z->c <= z->lb) return -1;
        ch = z->p[z->c - 1];
        if (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0)
            return 1;
        z->c--;
    } while (repeat);
    return 0;
}

References SN_env::c, SN_env::lb, SN_env::p, and repeat().

Referenced by r_case_ending(), r_et_condition(), r_lengthen_V(), r_main_suffix(), r_other_suffix(), r_particle_etc(), r_shortv(), r_standard_suffix(), r_Step_1b(), r_Step_2(), r_t_plural(), r_tidy(), r_undouble(), r_V(), r_VI(), r_vowel_suffix(), and r_VX().

in_grouping_b_U()

int in_grouping_b_U	(	struct SN_env *	z,
		const unsigned char *	s,
		int	min,
		int	max,
		int	repeat
	)

Definition at line 145 of file utilities.c.

                                                                                                     {
    do {
        int ch;
        int w = get_b_utf8(z->p, z->c, z->lb, & ch);
        if (!w) return -1;
        if (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0)
            return w;
        z->c -= w;
    } while (repeat);
    return 0;
}

References SN_env::c, get_b_utf8(), SN_env::lb, SN_env::p, and repeat().

Referenced by r_case_ending(), r_CONSONANT(), r_correlative(), r_degrees(), r_emphasis(), r_et_condition(), r_i_plural(), r_lengthen_V(), r_main_suffix(), r_mark_sU(), r_mark_suffix_with_optional_n_consonant(), r_mark_suffix_with_optional_s_consonant(), r_mark_suffix_with_optional_U_vowel(), r_mark_suffix_with_optional_y_consonant(), r_mark_yU(), r_other_suffix(), r_particle_etc(), r_plural_three_first_cases(), r_shortv(), r_standard_suffix(), r_Step_1b(), r_Step_2(), r_step_3(), r_step_4(), r_step_5b(), r_step_5c(), r_t_plural(), r_tidy(), r_undouble(), r_undouble_kpt(), r_V(), r_verb(), r_VI(), r_vowel_suffix(), and r_VX().

in_grouping_U()

int in_grouping_U	(	struct SN_env *	z,
		const unsigned char *	s,
		int	min,
		int	max,
		int	repeat
	)

Definition at line 133 of file utilities.c.

                                                                                                   {
    do {
        int ch;
        int w = get_utf8(z->p, z->c, z->l, & ch);
        if (!w) return -1;
        if (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0)
            return w;
        z->c += w;
    } while (repeat);
    return 0;
}

References SN_env::c, get_utf8(), SN_env::l, SN_env::p, and repeat().

Referenced by lithuanian_UTF_8_stem(), porter_UTF_8_stem(), r_elisions(), r_Lose_infix(), r_Lose_prefix(), r_mark_regions(), r_measure(), r_prelude(), and r_remove_first_order_prefix().

increase_size()

static int increase_size ( symbol **  p, int  n  )

static

Definition at line 373 of file utilities.c.

                                             {
    int new_size = n + 20;
    void * mem = realloc((char *) *p - HEAD,
                         HEAD + (new_size + 1) * sizeof(symbol));
    symbol * q;
    if (mem == NULL) return -1;
    q = (symbol *) (HEAD + (char *)mem);
    CAPACITY(q) = new_size;
    *p = q;
    return 0;
}

References CAPACITY, HEAD, and realloc.

Referenced by assign_to(), replace_s(), and slice_to().

insert_s()

SNOWBALL_ERR insert_s	(	struct SN_env *	z,
		int	bra,
		int	ket,
		int	s_size,
		const symbol *	s
	)

Definition at line 464 of file utilities.c.

                                                                                                {
    REPLACE_S(z, bra, ket, s_size, s);
    if (bra <= z->ket) {
        int adjustment = s_size - (ket - bra);
        z->ket += adjustment;
        if (bra <= z->bra) z->bra += adjustment;
    }
    SNOWBALL_RETURN_OK;
}

References SN_env::bra, SN_env::ket, REPLACE_S, among::s, among::s_size, and SNOWBALL_RETURN_OK.

Referenced by insert_v(), r_Step_1b(), r_Step_2(), r_step_2a(), and r_Step_3().

insert_v()

SNOWBALL_ERR insert_v	(	struct SN_env *	z,
		int	bra,
		int	ket,
		const symbol *	p
	)

Definition at line 474 of file utilities.c.

                                                                                    {
    return insert_s(z, bra, ket, SIZE(p), p);
}

References insert_s(), and SIZE.

Referenced by r_lengthen_V().

len_utf8()

int len_utf8

(

const symbol *

p

)

Definition at line 501 of file utilities.c.

                                      {
    int size = SIZE(p);
    int len = 0;
    while (size--) {
        symbol b = *p++;
        if (b >= 0xC0 || b < 0x80) ++len;
    }
    return len;
}

References b, len, and SIZE.

Referenced by lithuanian_UTF_8_stem(), r_Checks1(), r_fix_ending(), r_has_min_length(), r_Prefix_Step1(), r_Prefix_Step2(), r_Prefix_Step3_Verb(), r_Prefix_Step3a_Noun(), r_Prefix_Step3b_Noun(), r_Prefix_Step4_Verb(), r_remove_vetrumai_urupukal(), r_Suffix_Noun_Step1a(), r_Suffix_Noun_Step1b(), r_Suffix_Noun_Step2a(), r_Suffix_Noun_Step2b(), r_Suffix_Noun_Step2c1(), r_Suffix_Noun_Step2c2(), r_Suffix_Noun_Step3(), r_Suffix_Verb_Step1(), r_Suffix_Verb_Step2a(), r_Suffix_Verb_Step2b(), and r_Suffix_Verb_Step2c().

lose_s()

void lose_s

(

symbol *

p

)

Definition at line 31 of file utilities.c.

                               {
    if (p == NULL) return;
    free((char *) p - HEAD);
}

References free, and HEAD.

Referenced by danish_ISO_8859_1_close_env(), danish_UTF_8_close_env(), dutch_ISO_8859_1_close_env(), dutch_UTF_8_close_env(), finnish_ISO_8859_1_close_env(), finnish_UTF_8_close_env(), and SN_delete_env().

out_grouping()

int out_grouping	(	struct SN_env *	z,
		const unsigned char *	s,
		int	min,
		int	max,
		int	repeat
	)

Definition at line 207 of file utilities.c.

                                                                                                  {
    do {
        int ch;
        if (z->c >= z->l) return -1;
        ch = z->p[z->c];
        if (!(ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0))
            return 1;
        z->c++;
    } while (repeat);
    return 0;
}

References SN_env::c, SN_env::l, SN_env::p, and repeat().

Referenced by indonesian_ISO_8859_1_stem(), porter_ISO_8859_1_stem(), r_mark_regions(), r_measure(), r_prelude(), and r_remove_second_order_prefix().

out_grouping_b()

int out_grouping_b	(	struct SN_env *	z,
		const unsigned char *	s,
		int	min,
		int	max,
		int	repeat
	)

Definition at line 219 of file utilities.c.

                                                                                                    {
    do {
        int ch;
        if (z->c <= z->lb) return -1;
        ch = z->p[z->c - 1];
        if (!(ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0))
            return 1;
        z->c--;
    } while (repeat);
    return 0;
}

References SN_env::c, SN_env::lb, SN_env::p, and repeat().

Referenced by r_C(), r_e_ending(), r_en_ending(), r_et_condition(), r_i_verb_suffix(), r_lengthen_V(), r_main_suffix(), r_residual_suffix(), r_shortv(), r_standard_suffix(), r_Step_1a(), r_Step_1b(), r_Step_1c(), and r_un_accent().

out_grouping_b_U()

int out_grouping_b_U	(	struct SN_env *	z,
		const unsigned char *	s,
		int	min,
		int	max,
		int	repeat
	)

Definition at line 169 of file utilities.c.

                                                                                                      {
    do {
        int ch;
        int w = get_b_utf8(z->p, z->c, z->lb, & ch);
        if (!w) return -1;
        if (!(ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0))
            return w;
        z->c -= w;
    } while (repeat);
    return 0;
}

References SN_env::c, get_b_utf8(), SN_env::lb, SN_env::p, and repeat().

Referenced by r_append_U_to_stems_ending_with_d_or_g(), r_C(), r_check_vowel_harmony(), r_e_ending(), r_en_ending(), r_et_condition(), r_i_verb_suffix(), r_lengthen_V(), r_long_word(), r_main_suffix(), r_mark_suffix_with_optional_U_vowel(), r_residual_suffix(), r_shortv(), r_standard_suffix(), r_Step_1a(), r_Step_1b(), r_Step_1c(), r_un_accent(), and r_verb_suffix().

out_grouping_U()

int out_grouping_U	(	struct SN_env *	z,
		const unsigned char *	s,
		int	min,
		int	max,
		int	repeat
	)

Definition at line 157 of file utilities.c.

                                                                                                    {
    do {
        int ch;
        int w = get_utf8(z->p, z->c, z->l, & ch);
        if (!w) return -1;
        if (!(ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0))
            return w;
        z->c += w;
    } while (repeat);
    return 0;
}

References SN_env::c, get_utf8(), SN_env::l, SN_env::p, and repeat().

Referenced by indonesian_UTF_8_stem(), lithuanian_UTF_8_stem(), porter_UTF_8_stem(), r_mark_regions(), r_measure(), r_more_than_one_syllable_word(), r_prelude(), and r_remove_second_order_prefix().

replace_s()

SNOWBALL_ERR replace_s	(	struct SN_env *	z,
		int	c_bra,
		int	c_ket,
		int	s_size,
		const symbol *	s
	)

Definition at line 389 of file utilities.c.

{
    int adjustment = s_size - (c_ket - c_bra);
    if (adjustment != 0) {
        int len = SIZE(z->p);
        if (adjustment + len > CAPACITY(z->p)) {
            SNOWBALL_PROPAGATE_ERR(increase_size(&z->p, adjustment + len));
        }
        memmove(z->p + c_ket + adjustment,
                z->p + c_ket,
                (len - c_ket) * sizeof(symbol));
        SET_SIZE(z->p, adjustment + len);
        z->l += adjustment;
        if (z->c >= c_ket)
            z->c += adjustment;
        else if (z->c > c_bra)
            z->c = c_bra;
    }
    if (s_size) memmove(z->p + c_bra, s, s_size * sizeof(symbol));
    SNOWBALL_RETURN_OK;
}

References SN_env::c, CAPACITY, increase_size(), SN_env::l, len, SN_env::p, among::s, among::s_size, SET_SIZE, SIZE, SNOWBALL_PROPAGATE_ERR, and SNOWBALL_RETURN_OK.

Referenced by SN_set_current().

skip_b_utf8()

int skip_b_utf8	(	const symbol *	p,
		int	c,
		int	limit,
		int	n
	)

Definition at line 68 of file utilities.c.

                                                                  {
    int b;
    if (n < 0) return -1;
    for (; n > 0; n--) {
        if (c <= limit) return -1;
        b = p[--c];
        if (b >= 0x80) {   /* 1000 0000 */
            while (c > limit) {
                b = p[c];
                if (b >= 0xC0) break; /* 1100 0000 */
                c--;
            }
        }
    }
    return c;
}

References b.

Referenced by arabic_UTF_8_stem(), r_case_ending(), r_consonant_pair(), r_emphasis(), r_lengthen_V(), r_long_word(), r_mark_suffix_with_optional_n_consonant(), r_mark_suffix_with_optional_s_consonant(), r_mark_suffix_with_optional_U_vowel(), r_mark_suffix_with_optional_y_consonant(), r_plural_three_first_cases(), r_standard_suffix(), r_Step_1a(), r_Step_1b(), r_tidy(), r_tolower(), r_un_double(), r_undouble(), r_verb_suffix(), and r_VX().

skip_utf8()

int skip_utf8	(	const symbol *	p,
		int	c,
		int	limit,
		int	n
	)

Definition at line 43 of file utilities.c.

                                                                {
    int b;
    if (n < 0) return -1;
    for (; n > 0; n--) {
        if (c >= limit) return -1;
        b = p[c++];
        if (b >= 0xC0) {   /* 1100 0000 */
            while (c < limit) {
                b = p[c];
                if (b >= 0xC0 || b < 0x80) break;
                /* break unless b is 10------ */
                c++;
            }
        }
    }
    return c;
}

References b.

Referenced by english_UTF_8_stem(), hindi_UTF_8_stem(), polish_UTF_8_stem(), porter_UTF_8_stem(), r_canonical_form(), r_cleaning(), r_cyr_to_lat(), r_Lose_infix(), r_Lose_prefix(), r_mark_regions(), r_norm(), r_Normalize_post(), r_Normalize_pre(), r_postlude(), r_prelude(), r_remove_proper_noun_suffix(), and russian_UTF_8_stem().

slice_check()

static SNOWBALL_ERR slice_check ( struct SN_env *  z )

static

Definition at line 414 of file utilities.c.

                                                   {
 
    if (z->bra < 0 ||
        z->bra > z->ket ||
        z->ket > z->l ||
        z->l > SIZE(z->p)) /* this line could be removed */
    {
#if 0
        fprintf(stderr, "faulty slice operation:\n");
        debug(z, -1, 0);
#endif
        SNOWBALL_RETURN_OR_THROW(-1, std::logic_error("Snowball slice invalid"));
    }
    SNOWBALL_RETURN_OK;
}

References SN_env::bra, debug, fprintf, SN_env::ket, SN_env::l, SN_env::p, SIZE, SNOWBALL_RETURN_OK, and SNOWBALL_RETURN_OR_THROW.

slice_del()

SNOWBALL_ERR slice_del

(

struct SN_env *

z

)

Definition at line 443 of file utilities.c.

                                                 {
    SLICE_CHECK(z);
    {
        int slice_size = z->ket - z->bra;
        if (slice_size != 0) {
            int len = SIZE(z->p);
            memmove(z->p + z->bra,
                    z->p + z->ket,
                    (len - z->ket) * sizeof(symbol));
            SET_SIZE(z->p, len - slice_size);
            z->l -= slice_size;
            if (z->c >= z->ket)
                z->c -= slice_size;
            else if (z->c > z->bra)
                z->c = z->bra;
        }
    }
    z->ket = z->bra;
    SNOWBALL_RETURN_OK;
}

References SN_env::bra, SN_env::c, SN_env::ket, SN_env::l, len, SN_env::p, SET_SIZE, SIZE, SLICE_CHECK, and SNOWBALL_RETURN_OK.

Referenced by hindi_UTF_8_stem(), portuguese_ISO_8859_1_stem(), portuguese_UTF_8_stem(), r_aditzak(), r_adjectival(), r_adjective(), r_adjetiboak(), r_attached_pronoun(), r_case(), r_case_ending(), r_case_other(), r_consonant_pair(), r_correlative(), r_degrees(), r_deriv(), r_derivational(), r_e_ending(), r_elisions(), r_emphasis(), r_en_ending(), r_ending(), r_factive(), r_fix_ending(), r_i_plural(), r_i_verb_suffix(), r_initial_morph(), r_instrum(), r_izenak(), r_Lose_infix(), r_Lose_prefix(), r_main_suffix(), r_Normalize_pre(), r_noun(), r_noun_sfx(), r_nu(), r_other_endings(), r_other_suffix(), r_owned(), r_particle_etc(), r_perfective_gerund(), r_plur_owner(), r_plural(), r_plural_three_first_cases(), r_possessive(), r_postlude(), r_Prefix_Step2(), r_Prefix_Step3a_Noun(), r_Prefix_Step3b_Noun(), r_prelude(), r_pronoun(), r_reflexive(), r_remove_category_1(), r_remove_category_2(), r_remove_category_3(), r_remove_command_suffixes(), r_remove_common_word_endings(), r_remove_endings(), r_remove_first_order_prefix(), r_remove_particle(), r_remove_plural_suffix(), r_remove_possessive_pronoun(), r_remove_pronoun_prefixes(), r_remove_proper_noun_suffix(), r_remove_question_prefixes(), r_remove_second_order_prefix(), r_remove_suffix(), r_remove_tense_suffix(), r_remove_vetrumai_urupukal(), r_residual_form(), r_residual_suffix(), r_sing_owner(), r_standard_suffix(), r_stem_nominal_verb_suffixes(), r_stem_noun_suffixes(), r_stem_suffix_chain_before_ki(), r_step1(), r_step2(), r_step_0(), r_Step_1(), r_Step_1a(), r_Step_1b(), r_Step_1c(), r_Step_2(), r_step_2a(), r_step_2b(), r_step_2c(), r_step_2d(), r_Step_3(), r_step_3(), r_Step_4(), r_step_4(), r_Step_5(), r_Step_5a(), r_step_5a(), r_Step_5b(), r_step_5b(), r_step_5c(), r_step_5d(), r_step_5e(), r_step_5f(), r_step_5g(), r_step_5h(), r_step_5i(), r_step_5j(), r_step_5k(), r_step_5l(), r_step_5m(), r_step_6(), r_step_7(), r_step_s1(), r_step_s10(), r_step_s2(), r_step_s3(), r_step_s4(), r_step_s5(), r_step_s6(), r_step_s7(), r_step_s8(), r_step_s9(), r_Suffix_Noun_Step1a(), r_Suffix_Noun_Step1b(), r_Suffix_Noun_Step2a(), r_Suffix_Noun_Step2b(), r_Suffix_Noun_Step2c1(), r_Suffix_Noun_Step2c2(), r_Suffix_Noun_Step3(), r_Suffix_Verb_Step1(), r_Suffix_Verb_Step2a(), r_Suffix_Verb_Step2b(), r_Suffix_Verb_Step2c(), r_t_plural(), r_tidy(), r_tidy_up(), r_ujn_suffix(), r_un_double(), r_undouble(), r_verb(), r_verb_sfx(), r_verb_suffix(), r_vowel_suffix(), r_y_verb_suffix(), russian_KOI8_R_stem(), and russian_UTF_8_stem().

slice_from_s()

SNOWBALL_ERR slice_from_s	(	struct SN_env *	z,
		int	s_size,
		const symbol *	s
	)

Definition at line 432 of file utilities.c.

                                                                                  {
    SLICE_CHECK(z);
    REPLACE_S(z, z->bra, z->ket, s_size, s);
    z->ket = z->bra + s_size;
    SNOWBALL_RETURN_OK;
}

References SN_env::bra, SN_env::ket, REPLACE_S, among::s, among::s_size, SLICE_CHECK, and SNOWBALL_RETURN_OK.

Referenced by french_ISO_8859_1_stem(), french_UTF_8_stem(), porter_ISO_8859_1_stem(), porter_UTF_8_stem(), r_adjetiboak(), r_append_U_to_stems_ending_with_d_or_g(), r_attached_pronoun(), r_canonical_form(), r_case_other(), r_case_special(), r_cleaning(), r_combo_suffix(), r_cyr_to_lat(), r_deriv(), r_exception1(), r_final_apostrophe(), r_fix_chdz(), r_fix_conflicts(), r_fix_ending(), r_fix_gd(), r_fix_va_start(), r_initial_apostrophe(), r_initial_morph(), r_izenak(), r_lengthen_V(), r_Lose_infix(), r_Lose_prefix(), r_main_suffix(), r_mark_regions(), r_norm(), r_normalize_consonant(), r_Normalize_post(), r_Normalize_pre(), r_other_suffix(), r_owned(), r_plur_owner(), r_plural(), r_plural_three_first_cases(), r_possessive(), r_post_process_last_consonants(), r_postlude(), r_Prefix_Step1(), r_Prefix_Step3_Verb(), r_Prefix_Step3b_Noun(), r_Prefix_Step4_Verb(), r_prelude(), r_remove_common_word_endings(), r_remove_endings(), r_remove_first_order_prefix(), r_remove_plural_suffix(), r_remove_question_suffixes(), r_remove_tense_suffix(), r_remove_um(), r_remove_vetrumai_urupukal(), r_residual_form(), r_residual_suffix(), r_sing_owner(), r_special_noun_endings(), r_standard_suffix(), r_step_0(), r_Step_1(), r_step_1(), r_Step_1a(), r_Step_1b(), r_Step_1c(), r_Step_2(), r_step_2b(), r_step_2c(), r_step_2d(), r_Step_3(), r_step_3(), r_Step_4(), r_step_4(), r_step_5a(), r_step_5b(), r_step_5c(), r_step_5d(), r_step_5e(), r_step_5f(), r_step_5g(), r_step_5h(), r_step_5i(), r_step_5j(), r_step_5k(), r_step_5l(), r_step_5m(), r_Step_6(), r_step_6(), r_Step_7(), r_step_s1(), r_step_s10(), r_step_s2(), r_step_s3(), r_step_s4(), r_step_s5(), r_step_s6(), r_step_s7(), r_step_s8(), r_step_s9(), r_Suffix_All_alef_maqsura(), r_tolower(), r_un_accent(), r_undouble_kpt(), r_v_ending(), r_verb(), r_verb_exceptions(), russian_KOI8_R_stem(), russian_UTF_8_stem(), and slice_from_v().

slice_from_v()

SNOWBALL_ERR slice_from_v	(	struct SN_env *	z,
		const symbol *	p
	)

Definition at line 439 of file utilities.c.

                                                                      {
    return slice_from_s(z, SIZE(p), p);
}

References SIZE, and slice_from_s().

slice_to()

SNOWBALL_ERR slice_to	(	struct SN_env *	z,
		symbol **	p
	)

Definition at line 478 of file utilities.c.

                                                             {
    SLICE_CHECK(z);
    {
        int len = z->ket - z->bra;
        if (CAPACITY(*p) < len) {
            SNOWBALL_PROPAGATE_ERR(increase_size(p, len));
        }
        memmove(*p, z->p + z->bra, len * sizeof(symbol));
        SET_SIZE(*p, len);
    }
    SNOWBALL_RETURN_OK;
}

References SN_env::bra, CAPACITY, increase_size(), SN_env::ket, len, SN_env::p, SET_SIZE, SLICE_CHECK, SNOWBALL_PROPAGATE_ERR, and SNOWBALL_RETURN_OK.

Referenced by r_lengthen_V(), r_tidy(), and r_undouble().