tokenizer.h File Reference

#include "lexbor/core/sbst.h"
#include "lexbor/core/array_obj.h"
#include "lexbor/html/base.h"
#include "lexbor/html/token.h"
#include "lexbor/tag/tag.h"
#include "lexbor/ns/ns.h"
#include "lexbor/html/tokenizer/error.h"

Go to the source code of this file.

Data Structures
struct	lxb_html_tokenizer

Typedefs
typedef const lxb_char_t *(*	lxb_html_tokenizer_state_f) (lxb_html_tokenizer_t *tkz, const lxb_char_t *data, const lxb_char_t *end)
typedef lxb_html_token_t *(*	lxb_html_tokenizer_token_f) (lxb_html_tokenizer_t *tkz, lxb_html_token_t *token, void *ctx)

Functions
LXB_API lxb_html_tokenizer_t *	lxb_html_tokenizer_create (void)
LXB_API lxb_status_t	lxb_html_tokenizer_init (lxb_html_tokenizer_t *tkz)
LXB_API lxb_status_t	lxb_html_tokenizer_inherit (lxb_html_tokenizer_t *tkz_to, lxb_html_tokenizer_t *tkz_from)
LXB_API lxb_html_tokenizer_t *	lxb_html_tokenizer_ref (lxb_html_tokenizer_t *tkz)
LXB_API lxb_html_tokenizer_t *	lxb_html_tokenizer_unref (lxb_html_tokenizer_t *tkz)
LXB_API void	lxb_html_tokenizer_clean (lxb_html_tokenizer_t *tkz)
LXB_API lxb_html_tokenizer_t *	lxb_html_tokenizer_destroy (lxb_html_tokenizer_t *tkz)
LXB_API lxb_status_t	lxb_html_tokenizer_tags_make (lxb_html_tokenizer_t *tkz, size_t table_size)
LXB_API void	lxb_html_tokenizer_tags_destroy (lxb_html_tokenizer_t *tkz)
LXB_API lxb_status_t	lxb_html_tokenizer_attrs_make (lxb_html_tokenizer_t *tkz, size_t table_size)
LXB_API void	lxb_html_tokenizer_attrs_destroy (lxb_html_tokenizer_t *tkz)
LXB_API lxb_status_t	lxb_html_tokenizer_begin (lxb_html_tokenizer_t *tkz)
LXB_API lxb_status_t	lxb_html_tokenizer_chunk (lxb_html_tokenizer_t *tkz, const lxb_char_t *data, size_t size)
LXB_API lxb_status_t	lxb_html_tokenizer_end (lxb_html_tokenizer_t *tkz)
LXB_API const lxb_char_t *	lxb_html_tokenizer_change_incoming (lxb_html_tokenizer_t *tkz, const lxb_char_t *pos)
LXB_API lxb_ns_id_t	lxb_html_tokenizer_current_namespace (lxb_html_tokenizer_t *tkz)
LXB_API 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)
lxb_inline void	lxb_html_tokenizer_status_set (lxb_html_tokenizer_t *tkz, lxb_status_t status)
lxb_inline void	lxb_html_tokenizer_tags_set (lxb_html_tokenizer_t *tkz, lexbor_hash_t *tags)
lxb_inline lexbor_hash_t *	lxb_html_tokenizer_tags (lxb_html_tokenizer_t *tkz)
lxb_inline void	lxb_html_tokenizer_attrs_set (lxb_html_tokenizer_t *tkz, lexbor_hash_t *attrs)
lxb_inline lexbor_hash_t *	lxb_html_tokenizer_attrs (lxb_html_tokenizer_t *tkz)
lxb_inline void	lxb_html_tokenizer_attrs_mraw_set (lxb_html_tokenizer_t *tkz, lexbor_mraw_t *mraw)
lxb_inline lexbor_mraw_t *	lxb_html_tokenizer_attrs_mraw (lxb_html_tokenizer_t *tkz)
lxb_inline void	lxb_html_tokenizer_callback_token_done_set (lxb_html_tokenizer_t *tkz, lxb_html_tokenizer_token_f call_func, void *ctx)
lxb_inline void *	lxb_html_tokenizer_callback_token_done_ctx (lxb_html_tokenizer_t *tkz)
lxb_inline void	lxb_html_tokenizer_state_set (lxb_html_tokenizer_t *tkz, lxb_html_tokenizer_state_f state)
lxb_inline void	lxb_html_tokenizer_tmp_tag_id_set (lxb_html_tokenizer_t *tkz, lxb_tag_id_t tag_id)
lxb_inline lxb_html_tree_t *	lxb_html_tokenizer_tree (lxb_html_tokenizer_t *tkz)
lxb_inline void	lxb_html_tokenizer_tree_set (lxb_html_tokenizer_t *tkz, lxb_html_tree_t *tree)
lxb_inline lexbor_mraw_t *	lxb_html_tokenizer_mraw (lxb_html_tokenizer_t *tkz)
lxb_inline lxb_status_t	lxb_html_tokenizer_temp_realloc (lxb_html_tokenizer_t *tkz, size_t size)
lxb_inline lxb_status_t	lxb_html_tokenizer_temp_append_data (lxb_html_tokenizer_t *tkz, const lxb_char_t *data)
lxb_inline lxb_status_t	lxb_html_tokenizer_temp_append (lxb_html_tokenizer_t *tkz, const lxb_char_t *data, size_t size)
LXB_API void	lxb_html_tokenizer_status_set_noi (lxb_html_tokenizer_t *tkz, lxb_status_t status)
LXB_API void	lxb_html_tokenizer_callback_token_done_set_noi (lxb_html_tokenizer_t *tkz, lxb_html_tokenizer_token_f call_func, void *ctx)
LXB_API void *	lxb_html_tokenizer_callback_token_done_ctx_noi (lxb_html_tokenizer_t *tkz)
LXB_API void	lxb_html_tokenizer_state_set_noi (lxb_html_tokenizer_t *tkz, lxb_html_tokenizer_state_f state)
LXB_API void	lxb_html_tokenizer_tmp_tag_id_set_noi (lxb_html_tokenizer_t *tkz, lxb_tag_id_t tag_id)
LXB_API lxb_html_tree_t *	lxb_html_tokenizer_tree_noi (lxb_html_tokenizer_t *tkz)
LXB_API void	lxb_html_tokenizer_tree_set_noi (lxb_html_tokenizer_t *tkz, lxb_html_tree_t *tree)
LXB_API lexbor_mraw_t *	lxb_html_tokenizer_mraw_noi (lxb_html_tokenizer_t *tkz)
LXB_API lexbor_hash_t *	lxb_html_tokenizer_tags_noi (lxb_html_tokenizer_t *tkz)

Variables
const lxb_char_t *	lxb_html_tokenizer_eof

Typedef Documentation

lxb_html_tokenizer_state_f

typedef const lxb_char_t *(* lxb_html_tokenizer_state_f) (lxb_html_tokenizer_t *tkz, const lxb_char_t *data, const lxb_char_t *end)

Definition at line 25 of file tokenizer.h.

lxb_html_tokenizer_token_f

typedef lxb_html_token_t *(* lxb_html_tokenizer_token_f) (lxb_html_tokenizer_t *tkz, lxb_html_token_t *token, void *ctx)

Definition at line 29 of file tokenizer.h.

Function Documentation

lxb_html_tokenizer_attrs()

lxb_inline lexbor_hash_t * lxb_html_tokenizer_attrs

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 188 of file tokenizer.h.

lxb_html_tokenizer_attrs_destroy()

LXB_API 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_API 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_attrs_mraw()

lxb_inline lexbor_mraw_t * lxb_html_tokenizer_attrs_mraw

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 201 of file tokenizer.h.

lxb_html_tokenizer_attrs_mraw_set()

lxb_inline void lxb_html_tokenizer_attrs_mraw_set	(	lxb_html_tokenizer_t *	tkz,
		lexbor_mraw_t *	mraw )

Definition at line 194 of file tokenizer.h.

lxb_html_tokenizer_attrs_set()

lxb_inline void lxb_html_tokenizer_attrs_set	(	lxb_html_tokenizer_t *	tkz,
		lexbor_hash_t *	attrs )

Definition at line 182 of file tokenizer.h.

lxb_html_tokenizer_begin()

LXB_API 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()

lxb_inline void * lxb_html_tokenizer_callback_token_done_ctx

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 216 of file tokenizer.h.

lxb_html_tokenizer_callback_token_done_ctx_noi()

LXB_API 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()

lxb_inline void lxb_html_tokenizer_callback_token_done_set	(	lxb_html_tokenizer_t *	tkz,
		lxb_html_tokenizer_token_f	call_func,
		void *	ctx )

Definition at line 207 of file tokenizer.h.

lxb_html_tokenizer_callback_token_done_set_noi()

LXB_API 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_change_incoming()

LXB_API const lxb_char_t * lxb_html_tokenizer_change_incoming	(	lxb_html_tokenizer_t *	tkz,
		const lxb_char_t *	pos )

lxb_html_tokenizer_chunk()

LXB_API 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()

LXB_API void lxb_html_tokenizer_clean

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 199 of file tokenizer.c.

lxb_html_tokenizer_create()

LXB_API lxb_html_tokenizer_t * lxb_html_tokenizer_create

(

void

)

Definition at line 39 of file tokenizer.c.

lxb_html_tokenizer_current_namespace()

LXB_API 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_API 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_API 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_API 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_API lxb_status_t lxb_html_tokenizer_init

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 45 of file tokenizer.c.

lxb_html_tokenizer_mraw()

lxb_inline lexbor_mraw_t * lxb_html_tokenizer_mraw

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 248 of file tokenizer.h.

lxb_html_tokenizer_mraw_noi()

LXB_API 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_API 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()

LXB_API 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()

lxb_inline void lxb_html_tokenizer_state_set	(	lxb_html_tokenizer_t *	tkz,
		lxb_html_tokenizer_state_f	state )

Definition at line 222 of file tokenizer.h.

lxb_html_tokenizer_state_set_noi()

LXB_API 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()

lxb_inline void lxb_html_tokenizer_status_set	(	lxb_html_tokenizer_t *	tkz,
		lxb_status_t	status )

Definition at line 164 of file tokenizer.h.

lxb_html_tokenizer_status_set_noi()

LXB_API 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()

lxb_inline lexbor_hash_t * lxb_html_tokenizer_tags

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 176 of file tokenizer.h.

lxb_html_tokenizer_tags_destroy()

LXB_API 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_API 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()

LXB_API lexbor_hash_t * lxb_html_tokenizer_tags_noi

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 518 of file tokenizer.c.

lxb_html_tokenizer_tags_set()

lxb_inline void lxb_html_tokenizer_tags_set	(	lxb_html_tokenizer_t *	tkz,
		lexbor_hash_t *	tags )

Definition at line 170 of file tokenizer.h.

lxb_html_tokenizer_temp_append()

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

Definition at line 289 of file tokenizer.h.

lxb_html_tokenizer_temp_append_data()

lxb_inline lxb_status_t lxb_html_tokenizer_temp_append_data	(	lxb_html_tokenizer_t *	tkz,
		const lxb_char_t *	data )

Definition at line 272 of file tokenizer.h.

lxb_html_tokenizer_temp_realloc()

lxb_inline lxb_status_t lxb_html_tokenizer_temp_realloc	(	lxb_html_tokenizer_t *	tkz,
		size_t	size )

Definition at line 254 of file tokenizer.h.

lxb_html_tokenizer_tmp_tag_id_set()

lxb_inline void lxb_html_tokenizer_tmp_tag_id_set	(	lxb_html_tokenizer_t *	tkz,
		lxb_tag_id_t	tag_id )

Definition at line 229 of file tokenizer.h.

lxb_html_tokenizer_tmp_tag_id_set_noi()

LXB_API 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()

lxb_inline lxb_html_tree_t * lxb_html_tokenizer_tree

(

lxb_html_tokenizer_t *

tkz

)

Definition at line 236 of file tokenizer.h.

lxb_html_tokenizer_tree_noi()

LXB_API 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()

lxb_inline void lxb_html_tokenizer_tree_set	(	lxb_html_tokenizer_t *	tkz,
		lxb_html_tree_t *	tree )

Definition at line 242 of file tokenizer.h.

lxb_html_tokenizer_tree_set_noi()

LXB_API 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_API 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

extern

Definition at line 30 of file tokenizer.c.