#include <stdlib.h>
#include <stdio.h>
#include <stdarg.h>
#include <inttypes.h>
#include <string.h>
#include <assert.h>
#include "cutils.h"
#include "libregexp.h"
#include "libregexp-opcode.h"

Include dependency graph for libregexp.c:

Data Structures
struct	REParseState

struct	REOpCode

struct	REExecState

struct	REExecContext

union	REParseState.u

Macros
#define	DEF(id, size) REOP_ ## id,

#define	CAPTURE_COUNT_MAX 255

#define	STACK_SIZE_MAX 255

#define	CP_LS 0x2028

#define	CP_PS 0x2029

#define	TMP_BUF_SIZE 128

#define	DEF(id, size) { size },

#define	RE_HEADER_FLAGS 0

#define	RE_HEADER_CAPTURE_COUNT 1

#define	RE_HEADER_STACK_SIZE 2

#define	RE_HEADER_LEN 7

#define	CLASS_RANGE_BASE 0x40000000

#define	GET_CHAR(c, cptr, cbuf_end)

#define	PEEK_CHAR(c, cptr, cbuf_end)

#define	PEEK_PREV_CHAR(c, cptr, cbuf_start)

#define	GET_PREV_CHAR(c, cptr, cbuf_start)

#define	PREV_CHAR(cptr, cbuf_start)

Typedefs
typedef uintptr_t	StackInt

Enumerations
enum	REOPCodeEnum { REOP_COUNT }

enum	CharRangeEnum { CHAR_RANGE_d , CHAR_RANGE_D , CHAR_RANGE_s , CHAR_RANGE_S , CHAR_RANGE_w , CHAR_RANGE_W }

enum	REExecStateEnum { RE_EXEC_STATE_SPLIT , RE_EXEC_STATE_LOOKAHEAD , RE_EXEC_STATE_NEGATIVE_LOOKAHEAD , RE_EXEC_STATE_GREEDY_QUANT }

Functions
static int	is_digit (int c)

static int	dbuf_insert (DynBuf *s, int pos, int len)

static uint32_t	lre_canonicalize (uint32_t c, BOOL is_utf16)

BOOL	lre_is_space (int c)

static int	cr_init_char_range (REParseState s, CharRange cr, uint32_t c)

static int	cr_canonicalize (CharRange *cr)

static void	re_emit_op (REParseState *s, int op)

static int	re_emit_op_u32 (REParseState *s, int op, uint32_t val)

static int	re_emit_goto (REParseState *s, int op, uint32_t val)

static void	re_emit_op_u8 (REParseState *s, int op, uint32_t val)

static void	re_emit_op_u16 (REParseState *s, int op, uint32_t val)

static int	__attribute__ ((format(printf, 2, 3)))

static int	re_parse_out_of_memory (REParseState *s)

static int	parse_digits (const uint8_t **pp, BOOL allow_overflow)

static int	re_parse_expect (REParseState s, const uint8_t *pp, int c)

int	lre_parse_escape (const uint8_t **pp, int allow_utf16)

static BOOL	is_unicode_char (int c)

static int	parse_unicode_property (REParseState s, CharRange cr, const uint8_t **pp, BOOL is_inv)

static int	get_class_atom (REParseState s, CharRange cr, const uint8_t **pp, BOOL inclass)

static int	re_emit_range (REParseState s, const CharRange cr)

static int	re_parse_char_class (REParseState s, const uint8_t *pp)

static int	re_check_advance (const uint8_t *bc_buf, int bc_buf_len)

static int	re_is_simple_quantifier (const uint8_t *bc_buf, int bc_buf_len)

static int	re_parse_group_name (char buf, int buf_size, const uint8_t *pp, BOOL is_utf16)

static int	re_parse_captures (REParseState s, int phas_named_captures, const char *capture_name)

static int	re_count_captures (REParseState *s)

static BOOL	re_has_named_captures (REParseState *s)

static int	find_group_name (REParseState s, const char name)

static int	re_parse_disjunction (REParseState *s, BOOL is_backward_dir)

static int	re_parse_term (REParseState *s, BOOL is_backward_dir)

static int	re_parse_alternative (REParseState *s, BOOL is_backward_dir)

static int	compute_stack_size (const uint8_t *bc_buf, int bc_buf_len)

uint8_t *	lre_compile (int plen, char error_msg, int error_msg_size, const char buf, size_t buf_len, int re_flags, void opaque)

static BOOL	is_line_terminator (uint32_t c)

static BOOL	is_word_char (uint32_t c)

static int	push_state (REExecContext s, uint8_t capture, StackInt stack, size_t stack_len, const uint8_t pc, const uint8_t cptr, REExecStateEnum type, size_t count)

static intptr_t	lre_exec_backtrack (REExecContext s, uint8_t capture, StackInt stack, int stack_len, const uint8_t pc, const uint8_t cptr, BOOL no_recurse)

int	lre_exec (uint8_t *capture, const uint8_t bc_buf, const uint8_t cbuf, int cindex, int clen, int cbuf_type, void opaque)

int	lre_get_capture_count (const uint8_t *bc_buf)

int	lre_get_flags (const uint8_t *bc_buf)

const char *	lre_get_groupnames (const uint8_t *bc_buf)

Variables
static const REOpCode	reopcode_info [REOP_COUNT]

static const uint16_t	char_range_d []

static const uint16_t	char_range_s []

uint32_t const	lre_id_start_table_ascii [4]

uint32_t const	lre_id_continue_table_ascii [4]

static const uint16_t	char_range_w []

static const uint16_t *	char_range_table []

Data Structure Documentation

◆ REParseState

struct REParseState

Collaboration diagram for REParseState:

Data Fields
DynBuf	byte_code
const uint8_t *	buf_ptr
const uint8_t *	buf_end
const uint8_t *	buf_start
int	re_flags
BOOL	is_utf16
BOOL	ignore_case
BOOL	dotall
int	capture_count
int	total_capture_count
int	has_named_captures
void *	opaque
DynBuf	group_names
union REParseState	u

◆ REOpCode

struct REOpCode

Data Fields
uint8_t	size

◆ REExecState

struct REExecState

Data Fields
REExecStateEnum	type: 8
uint8_t	stack_len
size_t	count
const uint8_t *	cptr
const uint8_t *	pc
void *	buf[0]

◆ REExecContext

struct REExecContext

Data Fields
const uint8_t *	cbuf
const uint8_t *	cbuf_end
int	cbuf_type
int	capture_count
int	stack_size_max
BOOL	multi_line
BOOL	ignore_case
BOOL	is_utf16
void *	opaque
size_t	state_size
uint8_t *	state_stack
size_t	state_stack_size
size_t	state_stack_len

◆ REParseState.u

union REParseState.u

Data Fields
char	error_msg[TMP_BUF_SIZE]
char	tmp_buf[TMP_BUF_SIZE]

Macro Definition Documentation

◆ DEF [1/2]

#define DEF	(	id,
		size
	)	REOP_ ## id,

◆ CAPTURE_COUNT_MAX

#define CAPTURE_COUNT_MAX 255

◆ STACK_SIZE_MAX

#define STACK_SIZE_MAX 255

◆ CP_LS

#define CP_LS 0x2028

◆ CP_PS

#define CP_PS 0x2029

◆ TMP_BUF_SIZE

#define TMP_BUF_SIZE 128

◆ DEF [2/2]

#define DEF	(	id,
		size
	)	{ size },

◆ RE_HEADER_FLAGS

#define RE_HEADER_FLAGS 0

◆ RE_HEADER_CAPTURE_COUNT

#define RE_HEADER_CAPTURE_COUNT 1

◆ RE_HEADER_STACK_SIZE

#define RE_HEADER_STACK_SIZE 2

◆ RE_HEADER_LEN

#define RE_HEADER_LEN 7

◆ CLASS_RANGE_BASE

#define CLASS_RANGE_BASE 0x40000000

◆ GET_CHAR

#define GET_CHAR	(	c,
		cptr,
		cbuf_end
	)

Value:

    do {                                                                \
        if (cbuf_type == 0) {                                           \
            c = *cptr++;                                                \
        } else {                                                        \
            uint32_t __c1;                                              \
            c = *(uint16_t *)cptr;                                      \
            cptr += 2;                                                  \
            if (c >= 0xd800 && c < 0xdc00 &&                            \
                cbuf_type == 2 && cptr < cbuf_end) {                    \
                __c1 = *(uint16_t *)cptr;                               \
                if (__c1 >= 0xdc00 && __c1 < 0xe000) {                  \
                    c = (((c & 0x3ff) << 10) | (__c1 & 0x3ff)) + 0x10000; \
                    cptr += 2;                                          \
                }                                                       \
            }                                                           \
        }                                                               \
    } while (0)

◆ PEEK_CHAR

#define PEEK_CHAR	(	c,
		cptr,
		cbuf_end
	)

Value:

    do {                                         \
        if (cbuf_type == 0) {                    \
            c = cptr[0];                         \
        } else {                                 \
            uint32_t __c1;                                              \
            c = ((uint16_t *)cptr)[0];                                  \
            if (c >= 0xd800 && c < 0xdc00 &&                            \
                cbuf_type == 2 && (cptr + 2) < cbuf_end) {              \
                __c1 = ((uint16_t *)cptr)[1];                           \
                if (__c1 >= 0xdc00 && __c1 < 0xe000) {                  \
                    c = (((c & 0x3ff) << 10) | (__c1 & 0x3ff)) + 0x10000; \
                }                                                       \
            }                                                           \
        }                                        \
    } while (0)

◆ PEEK_PREV_CHAR

#define PEEK_PREV_CHAR	(	c,
		cptr,
		cbuf_start
	)

Value:

    do {                                         \
        if (cbuf_type == 0) {                    \
            c = cptr[-1];                        \
        } else {                                 \
            uint32_t __c1;                                              \
            c = ((uint16_t *)cptr)[-1];                                 \
            if (c >= 0xdc00 && c < 0xe000 &&                            \
                cbuf_type == 2 && (cptr - 4) >= cbuf_start) {              \
                __c1 = ((uint16_t *)cptr)[-2];                          \
                if (__c1 >= 0xd800 && __c1 < 0xdc00 ) {                 \
                    c = (((__c1 & 0x3ff) << 10) | (c & 0x3ff)) + 0x10000; \
                }                                                       \
            }                                                           \
        }                                                               \
    } while (0)

◆ GET_PREV_CHAR

#define GET_PREV_CHAR	(	c,
		cptr,
		cbuf_start
	)

Value:

    do {                                         \
        if (cbuf_type == 0) {                    \
            cptr--;                              \
            c = cptr[0];                         \
        } else {                                 \
            uint32_t __c1;                                              \
            cptr -= 2;                                                  \
            c = ((uint16_t *)cptr)[0];                                 \
            if (c >= 0xdc00 && c < 0xe000 &&                            \
                cbuf_type == 2 && cptr > cbuf_start) {                  \
                __c1 = ((uint16_t *)cptr)[-1];                          \
                if (__c1 >= 0xd800 && __c1 < 0xdc00 ) {                 \
                    cptr -= 2;                                          \
                    c = (((__c1 & 0x3ff) << 10) | (c & 0x3ff)) + 0x10000; \
                }                                                       \
            }                                                           \
        }                                                               \
    } while (0)

◆ PREV_CHAR

#define PREV_CHAR	(	cptr,
		cbuf_start
	)

Value:

    do {                                  \
        if (cbuf_type == 0) {             \
            cptr--;                       \
        } else {                          \
            cptr -= 2;                          \
            if (cbuf_type == 2) {                                       \
                c = ((uint16_t *)cptr)[0];                              \
                if (c >= 0xdc00 && c < 0xe000 && cptr > cbuf_start) {   \
                    c = ((uint16_t *)cptr)[-1];                         \
                    if (c >= 0xd800 && c < 0xdc00)                      \
                        cptr -= 2;                                      \
                }                                                       \
            }                                                           \
        }                                                               \
    } while (0)

Typedef Documentation

◆ StackInt

typedef uintptr_t StackInt

Enumeration Type Documentation

◆ REOPCodeEnum

enum REOPCodeEnum

Enumerator
REOP_COUNT

◆ CharRangeEnum

enum CharRangeEnum

Enumerator
CHAR_RANGE_d
CHAR_RANGE_D
CHAR_RANGE_s
CHAR_RANGE_S
CHAR_RANGE_w
CHAR_RANGE_W

◆ REExecStateEnum

enum REExecStateEnum

Enumerator
RE_EXEC_STATE_SPLIT
RE_EXEC_STATE_LOOKAHEAD
RE_EXEC_STATE_NEGATIVE_LOOKAHEAD
RE_EXEC_STATE_GREEDY_QUANT

Function Documentation

◆ is_digit()

static int is_digit ( int c )

static

Here is the caller graph for this function:

◆ dbuf_insert()

static int dbuf_insert	(	DynBuf *	s,
		int	pos,
		int	len
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ lre_canonicalize()

static uint32_t lre_canonicalize	(	uint32_t	c,
		BOOL	is_utf16
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ lre_is_space()

BOOL lre_is_space ( int c )

Here is the caller graph for this function:

◆ cr_init_char_range()

static int cr_init_char_range	(	REParseState *	s,
		CharRange *	cr,
		uint32_t	c
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ cr_canonicalize()

static int cr_canonicalize ( CharRange * cr )

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_emit_op()

static void re_emit_op	(	REParseState *	s,
		int	op
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_emit_op_u32()

static int re_emit_op_u32	(	REParseState *	s,
		int	op,
		uint32_t	val
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_emit_goto()

static int re_emit_goto	(	REParseState *	s,
		int	op,
		uint32_t	val
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_emit_op_u8()

static void re_emit_op_u8	(	REParseState *	s,
		int	op,
		uint32_t	val
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_emit_op_u16()

static void re_emit_op_u16	(	REParseState *	s,
		int	op,
		uint32_t	val
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ attribute()

static int __attribute__ ( (format(printf, 2, 3)) )

static

◆ re_parse_out_of_memory()

static int re_parse_out_of_memory ( REParseState * s )

static

Here is the caller graph for this function:

◆ parse_digits()

static int parse_digits	(	const uint8_t **	pp,
		BOOL	allow_overflow
	)

static

Here is the caller graph for this function:

◆ re_parse_expect()

static int re_parse_expect	(	REParseState *	s,
		const uint8_t **	pp,
		int	c
	)

static

Here is the caller graph for this function:

◆ lre_parse_escape()

int lre_parse_escape	(	const uint8_t **	pp,
		int	allow_utf16
	)

Here is the call graph for this function:

Here is the caller graph for this function:

◆ is_unicode_char()

static BOOL is_unicode_char ( int c )

static

Here is the caller graph for this function:

◆ parse_unicode_property()

static int parse_unicode_property	(	REParseState *	s,
		CharRange *	cr,
		const uint8_t **	pp,
		BOOL	is_inv
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ get_class_atom()

static int get_class_atom	(	REParseState *	s,
		CharRange *	cr,
		const uint8_t **	pp,
		BOOL	inclass
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_emit_range()

static int re_emit_range	(	REParseState *	s,
		const CharRange *	cr
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_parse_char_class()

static int re_parse_char_class	(	REParseState *	s,
		const uint8_t **	pp
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_check_advance()

static int re_check_advance	(	const uint8_t *	bc_buf,
		int	bc_buf_len
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_is_simple_quantifier()

static int re_is_simple_quantifier	(	const uint8_t *	bc_buf,
		int	bc_buf_len
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_parse_group_name()

static int re_parse_group_name	(	char *	buf,
		int	buf_size,
		const uint8_t **	pp,
		BOOL	is_utf16
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_parse_captures()

static int re_parse_captures	(	REParseState *	s,
		int *	phas_named_captures,
		const char *	capture_name
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_count_captures()

static int re_count_captures ( REParseState * s )

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_has_named_captures()

static BOOL re_has_named_captures ( REParseState * s )

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ find_group_name()

static int find_group_name	(	REParseState *	s,
		const char *	name
	)

static

Here is the caller graph for this function:

◆ re_parse_disjunction()

static int re_parse_disjunction	(	REParseState *	s,
		BOOL	is_backward_dir
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_parse_term()

static int re_parse_term	(	REParseState *	s,
		BOOL	is_backward_dir
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ re_parse_alternative()

static int re_parse_alternative	(	REParseState *	s,
		BOOL	is_backward_dir
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ compute_stack_size()

static int compute_stack_size	(	const uint8_t *	bc_buf,
		int	bc_buf_len
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ lre_compile()

uint8_t* lre_compile	(	int *	plen,
		char *	error_msg,
		int	error_msg_size,
		const char *	buf,
		size_t	buf_len,
		int	re_flags,
		void *	opaque
	)

Here is the call graph for this function:

Here is the caller graph for this function:

◆ is_line_terminator()

static BOOL is_line_terminator ( uint32_t c )

static

Here is the caller graph for this function:

◆ is_word_char()

static BOOL is_word_char ( uint32_t c )

static

Here is the caller graph for this function:

◆ push_state()

static int push_state	(	REExecContext *	s,
		uint8_t **	capture,
		StackInt *	stack,
		size_t	stack_len,
		const uint8_t *	pc,
		const uint8_t *	cptr,
		REExecStateEnum	type,
		size_t	count
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ lre_exec_backtrack()

static intptr_t lre_exec_backtrack	(	REExecContext *	s,
		uint8_t **	capture,
		StackInt *	stack,
		int	stack_len,
		const uint8_t *	pc,
		const uint8_t *	cptr,
		BOOL	no_recurse
	)

static

Here is the call graph for this function:

Here is the caller graph for this function:

◆ lre_exec()

int lre_exec	(	uint8_t **	capture,
		const uint8_t *	bc_buf,
		const uint8_t *	cbuf,
		int	cindex,
		int	clen,
		int	cbuf_type,
		void *	opaque
	)

Here is the call graph for this function:

Here is the caller graph for this function:

◆ lre_get_capture_count()

int lre_get_capture_count ( const uint8_t * bc_buf )

Here is the caller graph for this function:

◆ lre_get_flags()

int lre_get_flags ( const uint8_t * bc_buf )

Here is the caller graph for this function:

◆ lre_get_groupnames()

const char* lre_get_groupnames ( const uint8_t * bc_buf )

Here is the call graph for this function:

Here is the caller graph for this function:

Variable Documentation

◆ reopcode_info

const REOpCode reopcode_info[REOP_COUNT]

static

Initial value:

= {
 
#define DEF(id, size) 
 
}

◆ char_range_d

const uint16_t char_range_d[]

static

Initial value:

= {
    1,
    0x0030, 0x0039 + 1,
}

◆ char_range_s

const uint16_t char_range_s[]

static

Initial value:

= {
    10,
    0x0009, 0x000D + 1,
    0x0020, 0x0020 + 1,
    0x00A0, 0x00A0 + 1,
    0x1680, 0x1680 + 1,
    0x2000, 0x200A + 1,
    
    
    0x2028, 0x2029 + 1,
    0x202F, 0x202F + 1,
    0x205F, 0x205F + 1,
    0x3000, 0x3000 + 1,
    
    0xFEFF, 0xFEFF + 1,
}

◆ lre_id_start_table_ascii

uint32_t const lre_id_start_table_ascii[4]

Initial value:

= {
    
    0x00000000, 0x00000010, 0x87FFFFFE, 0x07FFFFFE
}

◆ lre_id_continue_table_ascii

uint32_t const lre_id_continue_table_ascii[4]

Initial value:

= {
    
    0x00000000, 0x03FF0010, 0x87FFFFFE, 0x07FFFFFE
}

◆ char_range_w

const uint16_t char_range_w[]

static

Initial value:

= {
    4,
    0x0030, 0x0039 + 1,
    0x0041, 0x005A + 1,
    0x005F, 0x005F + 1,
    0x0061, 0x007A + 1,
}

◆ char_range_table

const uint16_t* char_range_table[]

static

Initial value:

= {
    char_range_d,
    char_range_s,
    char_range_w,
}

Data Structures

Macros

Typedefs

Enumerations

Functions

Variables

Data Structure Documentation

◆ REParseState

◆ REOpCode

◆ REExecState

◆ REExecContext

◆ REParseState.u

Macro Definition Documentation

◆ DEF [1/2]

◆ CAPTURE_COUNT_MAX

◆ STACK_SIZE_MAX

◆ CP_LS

◆ CP_PS

◆ TMP_BUF_SIZE

◆ DEF [2/2]

◆ RE_HEADER_FLAGS

◆ RE_HEADER_CAPTURE_COUNT

◆ RE_HEADER_STACK_SIZE

◆ RE_HEADER_LEN

◆ CLASS_RANGE_BASE

◆ GET_CHAR

◆ PEEK_CHAR

◆ PEEK_PREV_CHAR

◆ GET_PREV_CHAR

◆ PREV_CHAR

Typedef Documentation

◆ StackInt

Enumeration Type Documentation

◆ REOPCodeEnum

◆ CharRangeEnum

◆ REExecStateEnum

Function Documentation

◆ is_digit()

◆ dbuf_insert()

◆ lre_canonicalize()

◆ lre_is_space()

◆ cr_init_char_range()

◆ cr_canonicalize()

◆ re_emit_op()

◆ re_emit_op_u32()

◆ re_emit_goto()

◆ re_emit_op_u8()

◆ re_emit_op_u16()

◆ __attribute__()

◆ re_parse_out_of_memory()

◆ parse_digits()

◆ re_parse_expect()

◆ lre_parse_escape()

◆ is_unicode_char()

◆ parse_unicode_property()

◆ get_class_atom()

◆ re_emit_range()

◆ re_parse_char_class()

◆ re_check_advance()

◆ re_is_simple_quantifier()

◆ re_parse_group_name()

◆ re_parse_captures()

◆ re_count_captures()

◆ re_has_named_captures()

◆ find_group_name()

◆ re_parse_disjunction()

◆ re_parse_term()

◆ re_parse_alternative()

◆ compute_stack_size()

◆ lre_compile()

◆ is_line_terminator()

◆ is_word_char()

◆ push_state()

◆ lre_exec_backtrack()

◆ lre_exec()

◆ lre_get_capture_count()

◆ lre_get_flags()

◆ lre_get_groupnames()

Variable Documentation

◆ reopcode_info

◆ attribute()