parser.c File Reference

#include <assert.h>
#include <string.h>
#include <parserutils/charset/mibenum.h>
#include <parserutils/input/inputstream.h>
#include <hubbub/parser.h>
#include "charset/detect.h"
#include "tokeniser/tokeniser.h"
#include "treebuilder/treebuilder.h"
#include "utils/parserutilserror.h"

Include dependency graph for parser.c:

Classes
struct	hubbub_parser

Functions
hubbub_error	hubbub_parser_create (const char *enc, bool fix_enc, hubbub_parser **parser)
hubbub_error	hubbub_parser_destroy (hubbub_parser *parser)
hubbub_error	hubbub_parser_setopt (hubbub_parser *parser, hubbub_parser_opttype type, hubbub_parser_optparams *params)
hubbub_error	hubbub_parser_insert_chunk (hubbub_parser *parser, const uint8_t *data, size_t len)
hubbub_error	hubbub_parser_parse_chunk (hubbub_parser *parser, const uint8_t *data, size_t len)
hubbub_error	hubbub_parser_completed (hubbub_parser *parser)
const char *	hubbub_parser_read_charset (hubbub_parser *parser, hubbub_charset_source *source)

Function Documentation

hubbub_parser_completed()

hubbub_error hubbub_parser_completed

(

hubbub_parser *

parser

)

Inform the parser that the last chunk of data has been parsed

Parameters

parser

Parser to inform

Returns

HUBBUB_OK on success, appropriate error otherwise

hubbub_parser_create()

hubbub_error hubbub_parser_create	(	const char *	enc,
		bool	fix_enc,
		hubbub_parser **	parser
	)

Create a hubbub parser

Parameters

enc	Source document encoding, or NULL to autodetect
fix_enc	Permit fixing up of encoding if it's frequently misused
parser	Pointer to location to receive parser instance

Returns

HUBBUB_OK on success, HUBBUB_BADPARM on bad parameters, HUBBUB_NOMEM on memory exhaustion, HUBBUB_BADENCODING if enc is unsupported

hubbub_parser_destroy()

hubbub_error hubbub_parser_destroy

(

hubbub_parser *

parser

)

Destroy a hubbub parser

Parameters

parser

Parser instance to destroy

Returns

HUBBUB_OK on success, appropriate error otherwise

hubbub_parser_insert_chunk()

hubbub_error hubbub_parser_insert_chunk	(	hubbub_parser *	parser,
		const uint8_t *	data,
		size_t	len
	)

Insert a chunk of data into a hubbub parser input stream

Inserts the given data into the input stream ready for parsing but does not cause any additional processing of the input. This is useful to allow hubbub callbacks to add computed data to the input.

Parameters

parser	Parser instance to use
data	Data to parse (encoded in UTF-8)
len	Length, in bytes, of data

Returns

HUBBUB_OK on success, appropriate error otherwise

hubbub_parser_parse_chunk()

hubbub_error hubbub_parser_parse_chunk	(	hubbub_parser *	parser,
		const uint8_t *	data,
		size_t	len
	)

Pass a chunk of data to a hubbub parser for parsing

Parameters

parser	Parser instance to use
data	Data to parse (encoded in the input charset)
len	Length, in bytes, of data

Returns

HUBBUB_OK on success, appropriate error otherwise

hubbub_parser_read_charset()

const char * hubbub_parser_read_charset	(	hubbub_parser *	parser,
		hubbub_charset_source *	source
	)

Read the document charset

Parameters

parser	Parser instance to query
source	Pointer to location to receive charset source

Returns

Pointer to charset name (constant; do not free), or NULL if unknown

hubbub_parser_setopt()

hubbub_error hubbub_parser_setopt	(	hubbub_parser *	parser,
		hubbub_parser_opttype	type,
		hubbub_parser_optparams *	params
	)

Configure a hubbub parser

Parameters

parser	Parser instance to configure
type	Option to set
params	Option-specific parameters

Returns

HUBBUB_OK on success, appropriate error otherwise