libunicode.c File Reference

#include <stdlib.h>
#include <stdio.h>
#include <stdarg.h>
#include <string.h>
#include <assert.h>
#include "cutils.h"
#include "libunicode.h"
#include "libunicode-table.h"

Include dependency graph for libunicode.c:

Macros
#define	UNICODE_INDEX_BLOCK_LEN   32
#define	CASE_U   (1 << 0)
#define	CASE_L   (1 << 1)
#define	CASE_F   (1 << 2)
#define	MR(x)   (1 << RUN_TYPE_ ## x)
#define	UNICODE_DECOMP_LEN_MAX   18
#define	M(id)   (1U << UNICODE_GC_ ## id)
#define	POP_STACK_LEN_MAX   4
#define	S   UNICODE_C_SPACE
#define	D   UNICODE_C_DIGIT
#define	X   UNICODE_C_XDIGIT
#define	U   UNICODE_C_UPPER
#define	L   UNICODE_C_LOWER
#define	_   UNICODE_C_UNDER
#define	d   UNICODE_C_DOLLAR
#define	SEQ_MAX_LEN   16

Enumerations
enum	{   RUN_TYPE_U , RUN_TYPE_L , RUN_TYPE_UF , RUN_TYPE_LF ,   RUN_TYPE_UL , RUN_TYPE_LSU , RUN_TYPE_U2L_399_EXT2 , RUN_TYPE_UF_D20 ,   RUN_TYPE_UF_D1_EXT , RUN_TYPE_U_EXT , RUN_TYPE_LF_EXT , RUN_TYPE_UF_EXT2 ,   RUN_TYPE_LF_EXT2 , RUN_TYPE_UF_EXT3 }
enum	DecompTypeEnum {   DECOMP_TYPE_C1 , DECOMP_TYPE_L1 , DECOMP_TYPE_L2 , DECOMP_TYPE_L3 ,   DECOMP_TYPE_L4 , DECOMP_TYPE_L5 , DECOMP_TYPE_L6 , DECOMP_TYPE_L7 ,   DECOMP_TYPE_LL1 , DECOMP_TYPE_LL2 , DECOMP_TYPE_S1 , DECOMP_TYPE_S2 ,   DECOMP_TYPE_S3 , DECOMP_TYPE_S4 , DECOMP_TYPE_S5 , DECOMP_TYPE_I1 ,   DECOMP_TYPE_I2_0 , DECOMP_TYPE_I2_1 , DECOMP_TYPE_I3_1 , DECOMP_TYPE_I3_2 ,   DECOMP_TYPE_I4_1 , DECOMP_TYPE_I4_2 , DECOMP_TYPE_B1 , DECOMP_TYPE_B2 ,   DECOMP_TYPE_B3 , DECOMP_TYPE_B4 , DECOMP_TYPE_B5 , DECOMP_TYPE_B6 ,   DECOMP_TYPE_B7 , DECOMP_TYPE_B8 , DECOMP_TYPE_B18 , DECOMP_TYPE_LS2 ,   DECOMP_TYPE_PAT3 , DECOMP_TYPE_S2_UL , DECOMP_TYPE_LS2_UL }
enum	PropOPEnum {   POP_GC , POP_PROP , POP_CASE , POP_UNION ,   POP_INTER , POP_XOR , POP_INVERT , POP_END }

Functions
int	lre_case_conv (uint32_t *res, uint32_t c, int conv_type)
int	lre_canonicalize (uint32_t c, BOOL is_unicode)
BOOL	lre_is_cased (uint32_t c)
BOOL	lre_is_case_ignorable (uint32_t c)
void	cr_init (CharRange *cr, void *mem_opaque, DynBufReallocFunc *realloc_func)
void	cr_free (CharRange *cr)
int	cr_realloc (CharRange *cr, int size)
int	cr_copy (CharRange *cr, const CharRange *cr1)
int	cr_op (CharRange *cr, const uint32_t *a_pt, int a_len, const uint32_t *b_pt, int b_len, int op)
int	cr_op1 (CharRange *cr, const uint32_t *b_pt, int b_len, int op)
int	cr_invert (CharRange *cr)
int	cr_regexp_canonicalize (CharRange *cr, BOOL is_unicode)
BOOL	lre_is_id_start (uint32_t c)
BOOL	lre_is_id_continue (uint32_t c)
int	unicode_normalize (uint32_t **pdst, const uint32_t *src, int src_len, UnicodeNormalizationEnum n_type, void *opaque, DynBufReallocFunc *realloc_func)
int	unicode_script (CharRange *cr, const char *script_name, BOOL is_ext)
int	unicode_general_category (CharRange *cr, const char *gc_name)
int	unicode_prop (CharRange *cr, const char *prop_name)
BOOL	lre_is_space_non_ascii (uint32_t c)
int	unicode_sequence_prop (const char *prop_name, UnicodeSequencePropCB *cb, void *opaque, CharRange *cr)

Variables
uint8_t const	lre_ctype_bits [256]

Macro Definition Documentation

_

#define _   UNICODE_C_UNDER

CASE_F

#define CASE_F   (1 << 2)

CASE_L

#define CASE_L   (1 << 1)

CASE_U

#define CASE_U   (1 << 0)

D

#define D   UNICODE_C_DIGIT

d

#define d   UNICODE_C_DOLLAR

L

#define L   UNICODE_C_LOWER

M

#define M

(

id

)

(1U << UNICODE_GC_ ## id)

MR

#define MR

(

x

)

(1 << RUN_TYPE_ ## x)

POP_STACK_LEN_MAX

#define POP_STACK_LEN_MAX   4

S

#define S   UNICODE_C_SPACE

SEQ_MAX_LEN

#define SEQ_MAX_LEN   16

U

#define U   UNICODE_C_UPPER

UNICODE_DECOMP_LEN_MAX

#define UNICODE_DECOMP_LEN_MAX   18

UNICODE_INDEX_BLOCK_LEN

#define UNICODE_INDEX_BLOCK_LEN   32

X

#define X   UNICODE_C_XDIGIT

Enumeration Type Documentation

anonymous enum

anonymous enum

Enumerator
RUN_TYPE_U
RUN_TYPE_L
RUN_TYPE_UF
RUN_TYPE_LF
RUN_TYPE_UL
RUN_TYPE_LSU
RUN_TYPE_U2L_399_EXT2
RUN_TYPE_UF_D20
RUN_TYPE_UF_D1_EXT
RUN_TYPE_U_EXT
RUN_TYPE_LF_EXT
RUN_TYPE_UF_EXT2
RUN_TYPE_LF_EXT2
RUN_TYPE_UF_EXT3

DecompTypeEnum

enum DecompTypeEnum

Enumerator
DECOMP_TYPE_C1
DECOMP_TYPE_L1
DECOMP_TYPE_L2
DECOMP_TYPE_L3
DECOMP_TYPE_L4
DECOMP_TYPE_L5
DECOMP_TYPE_L6
DECOMP_TYPE_L7
DECOMP_TYPE_LL1
DECOMP_TYPE_LL2
DECOMP_TYPE_S1
DECOMP_TYPE_S2
DECOMP_TYPE_S3
DECOMP_TYPE_S4
DECOMP_TYPE_S5
DECOMP_TYPE_I1
DECOMP_TYPE_I2_0
DECOMP_TYPE_I2_1
DECOMP_TYPE_I3_1
DECOMP_TYPE_I3_2
DECOMP_TYPE_I4_1
DECOMP_TYPE_I4_2
DECOMP_TYPE_B1
DECOMP_TYPE_B2
DECOMP_TYPE_B3
DECOMP_TYPE_B4
DECOMP_TYPE_B5
DECOMP_TYPE_B6
DECOMP_TYPE_B7
DECOMP_TYPE_B8
DECOMP_TYPE_B18
DECOMP_TYPE_LS2
DECOMP_TYPE_PAT3
DECOMP_TYPE_S2_UL
DECOMP_TYPE_LS2_UL

PropOPEnum

enum PropOPEnum

Enumerator
POP_GC
POP_PROP
POP_CASE
POP_UNION
POP_INTER
POP_XOR
POP_INVERT
POP_END

Function Documentation

cr_copy()

int cr_copy	(	CharRange *	cr,
		const CharRange *	cr1
	)

cr_free()

void cr_free

(

CharRange *

cr

)

cr_init()

void cr_init	(	CharRange *	cr,
		void *	mem_opaque,
		DynBufReallocFunc *	realloc_func
	)

cr_invert()

int cr_invert

(

CharRange *

cr

)

cr_op()

int cr_op	(	CharRange *	cr,
		const uint32_t *	a_pt,
		int	a_len,
		const uint32_t *	b_pt,
		int	b_len,
		int	op
	)

cr_op1()

int cr_op1	(	CharRange *	cr,
		const uint32_t *	b_pt,
		int	b_len,
		int	op
	)

cr_realloc()

int cr_realloc	(	CharRange *	cr,
		int	size
	)

cr_regexp_canonicalize()

int cr_regexp_canonicalize	(	CharRange *	cr,
		BOOL	is_unicode
	)

lre_canonicalize()

int lre_canonicalize	(	uint32_t	c,
		BOOL	is_unicode
	)

lre_case_conv()

int lre_case_conv	(	uint32_t *	res,
		uint32_t	c,
		int	conv_type
	)

lre_is_case_ignorable()

BOOL lre_is_case_ignorable

(

uint32_t

c

)

lre_is_cased()

BOOL lre_is_cased

(

uint32_t

c

)

lre_is_id_continue()

BOOL lre_is_id_continue

(

uint32_t

c

)

lre_is_id_start()

BOOL lre_is_id_start

(

uint32_t

c

)

lre_is_space_non_ascii()

BOOL lre_is_space_non_ascii

(

uint32_t

c

)

unicode_general_category()

int unicode_general_category	(	CharRange *	cr,
		const char *	gc_name
	)

unicode_normalize()

int unicode_normalize	(	uint32_t **	pdst,
		const uint32_t *	src,
		int	src_len,
		UnicodeNormalizationEnum	n_type,
		void *	opaque,
		DynBufReallocFunc *	realloc_func
	)

unicode_prop()

int unicode_prop	(	CharRange *	cr,
		const char *	prop_name
	)

unicode_script()

int unicode_script	(	CharRange *	cr,
		const char *	script_name,
		BOOL	is_ext
	)

unicode_sequence_prop()

int unicode_sequence_prop	(	const char *	prop_name,
		UnicodeSequencePropCB *	cb,
		void *	opaque,
		CharRange *	cr
	)

Variable Documentation

lre_ctype_bits

uint8_t const lre_ctype_bits[256]