inputstream.c File Reference

#include <assert.h>
#include <stdlib.h>
#include <string.h>
#include <parserutils/charset/mibenum.h>
#include <parserutils/charset/utf8.h>
#include <parserutils/input/inputstream.h>
#include "input/filter.h"
#include "utils/utils.h"

Include dependency graph for inputstream.c:

Classes
struct	parserutils_inputstream_private

Macros
#define	IS_ASCII(x)   (((x) & 0x80) == 0)
#define	UTF32_BOM_LEN   (4)
#define	UTF16_BOM_LEN   (2)
#define	UTF8_BOM_LEN   (3)

Typedefs
typedef struct parserutils_inputstream_private	parserutils_inputstream_private

Functions
parserutils_error	parserutils_inputstream_create (const char *enc, uint32_t encsrc, parserutils_charset_detect_func csdetect, parserutils_inputstream **stream)
parserutils_error	parserutils_inputstream_destroy (parserutils_inputstream *stream)
parserutils_error	parserutils_inputstream_append (parserutils_inputstream *stream, const uint8_t *data, size_t len)
parserutils_error	parserutils_inputstream_insert (parserutils_inputstream *stream, const uint8_t *data, size_t len)
parserutils_error	parserutils_inputstream_peek_slow (parserutils_inputstream *stream, size_t offset, const uint8_t **ptr, size_t *length)
const char *	parserutils_inputstream_read_charset (parserutils_inputstream *stream, uint32_t *source)
parserutils_error	parserutils_inputstream_change_charset (parserutils_inputstream *stream, const char *enc, uint32_t source)

Macro Definition Documentation

IS_ASCII

#define IS_ASCII

(

x

)

(((x) & 0x80) == 0)

UTF16_BOM_LEN

#define UTF16_BOM_LEN   (2)

UTF32_BOM_LEN

#define UTF32_BOM_LEN   (4)

UTF8_BOM_LEN

#define UTF8_BOM_LEN   (3)

Typedef Documentation

parserutils_inputstream_private

typedef struct parserutils_inputstream_private parserutils_inputstream_private

Private input stream definition

Function Documentation

parserutils_inputstream_append()

parserutils_error parserutils_inputstream_append	(	parserutils_inputstream *	stream,
		const uint8_t *	data,
		size_t	len
	)

Append data to an input stream

Parameters

stream	Input stream to append data to
data	Data to append (in document charset), or NULL to flag EOF
len	Length, in bytes, of data

Returns

PARSERUTILS_OK on success, appropriate error otherwise

parserutils_inputstream_change_charset()

parserutils_error parserutils_inputstream_change_charset	(	parserutils_inputstream *	stream,
		const char *	enc,
		uint32_t	source
	)

Change the source charset of the input stream

Parameters

stream	Input stream to modify
enc	Charset name
source	Charset source identifier

Returns

PARSERUTILS_OK on success, PARSERUTILS_BADPARM on invalid parameters, PARSERUTILS_INVALID if called after data has been read from stream, PARSERUTILS_BADENCODING if the encoding is unsupported, PARSERUTILS_NOMEM on memory exhaustion.

parserutils_inputstream_create()

parserutils_error parserutils_inputstream_create	(	const char *	enc,
		uint32_t	encsrc,
		parserutils_charset_detect_func	csdetect,
		parserutils_inputstream **	stream
	)

Create an input stream

Parameters

enc	Document charset, or NULL to autodetect
encsrc	Value for encoding source, if specified, or 0
csdetect	Charset detection function, or NULL
stream	Pointer to location to receive stream instance

Returns

PARSERUTILS_OK on success, PARSERUTILS_BADPARM on bad parameters, PARSERUTILS_NOMEM on memory exhaustion, PARSERUTILS_BADENCODING on unsupported encoding

The value 0 is defined as being the lowest priority encoding source (i.e. the default fallback encoding). Beyond this, no further interpretation is made upon the encoding source.

parserutils_inputstream_destroy()

parserutils_error parserutils_inputstream_destroy

(

parserutils_inputstream *

stream

)

Destroy an input stream

Parameters

stream

Input stream to destroy

Returns

PARSERUTILS_OK on success, appropriate error otherwise

parserutils_inputstream_insert()

parserutils_error parserutils_inputstream_insert	(	parserutils_inputstream *	stream,
		const uint8_t *	data,
		size_t	len
	)

Insert data into stream at current location

Parameters

stream	Input stream to insert into
data	Data to insert (UTF-8 encoded)
len	Length, in bytes, of data

Returns

PARSERUTILS_OK on success, appropriate error otherwise

parserutils_inputstream_peek_slow()

parserutils_error parserutils_inputstream_peek_slow	(	parserutils_inputstream *	stream,
		size_t	offset,
		const uint8_t **	ptr,
		size_t *	length
	)

Look at the character in the stream that starts at offset bytes from the cursor (slow version)

Parameters

stream	Stream to look in
offset	Byte offset of start of character
ptr	Pointer to location to receive pointer to character data
length	Pointer to location to receive character length (in bytes)

Returns

PARSERUTILS_OK on success, _NEEDDATA on reaching the end of available input, _EOF on reaching the end of all input, _BADENCODING if the input cannot be decoded, _NOMEM on memory exhaustion, _BADPARM if bad parameters are passed.

Once the character pointed to by the result of this call has been advanced past (i.e. parserutils_inputstream_advance has caused the stream cursor to pass over the character), then no guarantee is made as to the validity of the data pointed to. Thus, any attempt to dereference the pointer after advancing past the data it points to is a bug.

parserutils_inputstream_read_charset()

const char * parserutils_inputstream_read_charset	(	parserutils_inputstream *	stream,
		uint32_t *	source
	)

Read the source charset of the input stream

Parameters

stream	Input stream to query
source	Pointer to location to receive charset source identifier

Returns

Pointer to charset name (constant; do not free)