tokenizer.c File Reference

#include "lexbor/html/tokenizer.h"
#include "lexbor/html/tokenizer/state.h"
#include "lexbor/html/tokenizer/state_rcdata.h"
#include "lexbor/html/tokenizer/state_rawtext.h"
#include "lexbor/html/tokenizer/state_script.h"
#include "lexbor/html/tree.h"
#include "lexbor/html/tag_res.h"

Go to the source code of this file.

Macros
#define	LXB_HTML_TAG_RES_DATA
#define	LXB_HTML_TAG_RES_SHS_DATA
#define	LXB_HTML_TKZ_TEMP_SIZE   (4096 * 4)

Enumerations
enum	{ LXB_HTML_TOKENIZER_OPT_UNDEF = 0x00 , LXB_HTML_TOKENIZER_OPT_TAGS_SELF = 0x01 , LXB_HTML_TOKENIZER_OPT_ATTRS_SELF = 0x02 , LXB_HTML_TOKENIZER_OPT_ATTRS_MRAW_SELF = 0x04 }

Functions
lxb_html_tokenizer_t *	lxb_html_tokenizer_create (void)
lxb_status_t	lxb_html_tokenizer_init (lxb_html_tokenizer_t *tkz)
lxb_status_t	lxb_html_tokenizer_inherit (lxb_html_tokenizer_t *tkz_to, lxb_html_tokenizer_t *tkz_from)
lxb_html_tokenizer_t *	lxb_html_tokenizer_ref (lxb_html_tokenizer_t *tkz)
lxb_html_tokenizer_t *	lxb_html_tokenizer_unref (lxb_html_tokenizer_t *tkz)
void	lxb_html_tokenizer_clean (lxb_html_tokenizer_t *tkz)
lxb_html_tokenizer_t *	lxb_html_tokenizer_destroy (lxb_html_tokenizer_t *tkz)
lxb_status_t	lxb_html_tokenizer_tags_make (lxb_html_tokenizer_t *tkz, size_t table_size)
void	lxb_html_tokenizer_tags_destroy (lxb_html_tokenizer_t *tkz)
lxb_status_t	lxb_html_tokenizer_attrs_make (lxb_html_tokenizer_t *tkz, size_t table_size)
void	lxb_html_tokenizer_attrs_destroy (lxb_html_tokenizer_t *tkz)
lxb_status_t	lxb_html_tokenizer_begin (lxb_html_tokenizer_t *tkz)
lxb_status_t	lxb_html_tokenizer_chunk (lxb_html_tokenizer_t *tkz, const lxb_char_t *data, size_t size)
lxb_status_t	lxb_html_tokenizer_end (lxb_html_tokenizer_t *tkz)
lxb_ns_id_t	lxb_html_tokenizer_current_namespace (lxb_html_tokenizer_t *tkz)
void	lxb_html_tokenizer_set_state_by_tag (lxb_html_tokenizer_t *tkz, bool scripting, lxb_tag_id_t tag_id, lxb_ns_id_t ns)
void	lxb_html_tokenizer_status_set_noi (lxb_html_tokenizer_t *tkz, lxb_status_t status)
void	lxb_html_tokenizer_callback_token_done_set_noi (lxb_html_tokenizer_t *tkz, lxb_html_tokenizer_token_f call_func, void *ctx)
void *	lxb_html_tokenizer_callback_token_done_ctx_noi (lxb_html_tokenizer_t *tkz)
void	lxb_html_tokenizer_state_set_noi (lxb_html_tokenizer_t *tkz, lxb_html_tokenizer_state_f state)
void	lxb_html_tokenizer_tmp_tag_id_set_noi (lxb_html_tokenizer_t *tkz, lxb_tag_id_t tag_id)
lxb_html_tree_t *	lxb_html_tokenizer_tree_noi (lxb_html_tokenizer_t *tkz)
void	lxb_html_tokenizer_tree_set_noi (lxb_html_tokenizer_t *tkz, lxb_html_tree_t *tree)
lexbor_mraw_t *	lxb_html_tokenizer_mraw_noi (lxb_html_tokenizer_t *tkz)
lexbor_hash_t *	lxb_html_tokenizer_tags_noi (lxb_html_tokenizer_t *tkz)

Variables
const lxb_char_t *	lxb_html_tokenizer_eof = (const lxb_char_t *) "\x00"

Macro Definition Documentation

LXB_HTML_TAG_RES_DATA

#define LXB_HTML_TAG_RES_DATA

Definition at line 14 of file tokenizer.c.

LXB_HTML_TAG_RES_SHS_DATA

#define LXB_HTML_TAG_RES_SHS_DATA

Definition at line 15 of file tokenizer.c.

LXB_HTML_TKZ_TEMP_SIZE

#define LXB_HTML_TKZ_TEMP_SIZE   (4096 * 4)

Definition at line 19 of file tokenizer.c.

Enumeration Type Documentation

anonymous enum

anonymous enum

Enumerator
LXB_HTML_TOKENIZER_OPT_UNDEF
LXB_HTML_TOKENIZER_OPT_TAGS_SELF
LXB_HTML_TOKENIZER_OPT_ATTRS_SELF
LXB_HTML_TOKENIZER_OPT_ATTRS_MRAW_SELF

Definition at line 22 of file tokenizer.c.

Function Documentation

lxb_html_tokenizer_attrs_destroy()

void lxb_html_tokenizer_attrs_destroy

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 267 of file tokenizer.c.

lxb_html_tokenizer_attrs_make()

lxb_status_t lxb_html_tokenizer_attrs_make	(	lxb_html_tokenizer_t *	tkz,
		size_t	table_size )

Definition at line 259 of file tokenizer.c.

lxb_html_tokenizer_begin()

lxb_status_t lxb_html_tokenizer_begin

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 273 of file tokenizer.c.

lxb_html_tokenizer_callback_token_done_ctx_noi()

void * lxb_html_tokenizer_callback_token_done_ctx_noi

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 479 of file tokenizer.c.

lxb_html_tokenizer_callback_token_done_set_noi()

void lxb_html_tokenizer_callback_token_done_set_noi	(	lxb_html_tokenizer_t *	tkz,
		lxb_html_tokenizer_token_f	call_func,
		void *	ctx )

Definition at line 471 of file tokenizer.c.

lxb_html_tokenizer_chunk()

lxb_status_t lxb_html_tokenizer_chunk	(	lxb_html_tokenizer_t *	tkz,
		const lxb_char_t *	data,
		size_t	size )

Definition at line 308 of file tokenizer.c.

lxb_html_tokenizer_clean()

void lxb_html_tokenizer_clean

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 199 of file tokenizer.c.

lxb_html_tokenizer_create()

lxb_html_tokenizer_t * lxb_html_tokenizer_create

(

void

)

Definition at line 39 of file tokenizer.c.

lxb_html_tokenizer_current_namespace()

lxb_ns_id_t lxb_html_tokenizer_current_namespace

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 389 of file tokenizer.c.

lxb_html_tokenizer_destroy()

lxb_html_tokenizer_t * lxb_html_tokenizer_destroy

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 219 of file tokenizer.c.

lxb_html_tokenizer_end()

lxb_status_t lxb_html_tokenizer_end

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 344 of file tokenizer.c.

lxb_html_tokenizer_inherit()

lxb_status_t lxb_html_tokenizer_inherit	(	lxb_html_tokenizer_t *	tkz_to,
		lxb_html_tokenizer_t *	tkz_from )

Definition at line 117 of file tokenizer.c.

lxb_html_tokenizer_init()

lxb_status_t lxb_html_tokenizer_init

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 45 of file tokenizer.c.

lxb_html_tokenizer_mraw_noi()

lexbor_mraw_t * lxb_html_tokenizer_mraw_noi

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 512 of file tokenizer.c.

lxb_html_tokenizer_ref()

lxb_html_tokenizer_t * lxb_html_tokenizer_ref

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 163 of file tokenizer.c.

lxb_html_tokenizer_set_state_by_tag()

void lxb_html_tokenizer_set_state_by_tag	(	lxb_html_tokenizer_t *	tkz,
		bool	scripting,
		lxb_tag_id_t	tag_id,
		lxb_ns_id_t	ns )

Definition at line 405 of file tokenizer.c.

lxb_html_tokenizer_state_set_noi()

void lxb_html_tokenizer_state_set_noi	(	lxb_html_tokenizer_t *	tkz,
		lxb_html_tokenizer_state_f	state )

Definition at line 485 of file tokenizer.c.

lxb_html_tokenizer_status_set_noi()

void lxb_html_tokenizer_status_set_noi	(	lxb_html_tokenizer_t *	tkz,
		lxb_status_t	status )

Definition at line 464 of file tokenizer.c.

lxb_html_tokenizer_tags_destroy()

void lxb_html_tokenizer_tags_destroy

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 253 of file tokenizer.c.

lxb_html_tokenizer_tags_make()

lxb_status_t lxb_html_tokenizer_tags_make	(	lxb_html_tokenizer_t *	tkz,
		size_t	table_size )

Definition at line 246 of file tokenizer.c.

lxb_html_tokenizer_tags_noi()

lexbor_hash_t * lxb_html_tokenizer_tags_noi

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 518 of file tokenizer.c.

lxb_html_tokenizer_tmp_tag_id_set_noi()

void lxb_html_tokenizer_tmp_tag_id_set_noi	(	lxb_html_tokenizer_t *	tkz,
		lxb_tag_id_t	tag_id )

Definition at line 492 of file tokenizer.c.

lxb_html_tokenizer_tree_noi()

lxb_html_tree_t * lxb_html_tokenizer_tree_noi

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 499 of file tokenizer.c.

lxb_html_tokenizer_tree_set_noi()

void lxb_html_tokenizer_tree_set_noi	(	lxb_html_tokenizer_t *	tkz,
		lxb_html_tree_t *	tree )

Definition at line 505 of file tokenizer.c.

lxb_html_tokenizer_unref()

lxb_html_tokenizer_t * lxb_html_tokenizer_unref

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 179 of file tokenizer.c.

Variable Documentation

lxb_html_tokenizer_eof

const lxb_char_t* lxb_html_tokenizer_eof = (const lxb_char_t *) "\x00"

Definition at line 30 of file tokenizer.c.