UTF-8 manipulation functions (implementation). More...

#include <assert.h>
#include <errno.h>
#include <stdlib.h>
#include <string.h>
#include <strings.h>
#include <iconv.h>
#include <parserutils/charset/utf8.h>
#include "utils/config.h"
#include "utils/log.h"
#include "utils/utf8.h"
#include "netsurf/inttypes.h"
#include "netsurf/utf8.h"
#include "desktop/gui_internal.h"

Include dependency graph for utf8.c:

Functions
uint32_t	utf8_to_ucs4 (const char *s_in, size_t l)
	Convert a UTF-8 multibyte sequence into a single UCS4 character. More...

size_t	utf8_from_ucs4 (uint32_t c, char *s)
	Convert a single UCS4 character into a UTF-8 multibyte sequence. More...

size_t	utf8_length (const char *s)
	Calculate the length (in characters) of a NULL-terminated UTF-8 string. More...

size_t	utf8_bounded_length (const char *s, size_t l)
	Calculated the length (in characters) of a bounded UTF-8 string. More...

size_t	utf8_bounded_byte_length (const char *s, size_t l, size_t c)
	Calculate the length (in bytes) of a bounded UTF-8 string. More...

size_t	utf8_char_byte_length (const char *s)
	Calculate the length (in bytes) of a UTF-8 character. More...

size_t	utf8_prev (const char *s, size_t o)
	Find previous legal UTF-8 char in string. More...

size_t	utf8_next (const char *s, size_t l, size_t o)
	Find next legal UTF-8 char in string. More...

static void	utf8_clear_cd_cache (void)

static nserror	get_cached_cd (const char enc_from, const char enc_to, iconv_t *cd_out)
	obtain a cached conversion descriptor More...

nserror	utf8_finalise (void)
	Finalise the UTF-8 library. More...

static nserror	utf8_convert (const char string, size_t slen, const char from, const char to, char result_out, size_t result_len_out)
	Convert a string from one encoding to another. More...

nserror	utf8_to_enc (const char string, const char encname, size_t len, char **result)
	Convert a UTF8 string into the named encoding. More...

nserror	utf8_from_enc (const char string, const char encname, size_t len, char *result, size_t result_len)
	Convert a string in the named encoding into a UTF-8 string. More...

static nserror	utf8_convert_html_chunk (iconv_t cd, const char chunk, size_t inlen, char out, size_t outlen)
	convert a chunk of html data More...

nserror	utf8_to_html (const char string, const char encname, size_t len, char **result_out)
	Convert a UTF-8 encoded string into a string of the given encoding, applying HTML escape sequences where necessary. More...

bool	utf8_save_text (const char utf8_text, const char path)
	Save the given utf8 text to a file, converting to local encoding. More...

Variables
struct {
char from [32]
	Encoding name to convert from. More...

char to [32]
	Encoding name to convert to. More...

iconv_t cd
	Iconv conversion descriptor. More...

}	last_cd

Detailed Description

UTF-8 manipulation functions (implementation).

Definition in file utf8.c.

Function Documentation

◆ get_cached_cd()

static nserror get_cached_cd	(	const char *	enc_from,
		const char *	enc_to,
		iconv_t *	cd_out
	)

static

obtain a cached conversion descriptor

either return the cached conversion descriptor or create one if required

Definition at line 161 of file utf8.c.

References cd, last_cd, NSERROR_BAD_ENCODING, NSERROR_NOMEM, and NSERROR_OK.

Referenced by utf8_convert(), and utf8_to_html().

Here is the caller graph for this function:

◆ utf8_bounded_byte_length()

size_t utf8_bounded_byte_length	(	const char *	s,
		size_t	l,
		size_t	c
	)

Calculate the length (in bytes) of a bounded UTF-8 string.

Parameters

s	The string
l	Maximum length of input (in bytes)
c	Maximum number of characters to measure

Returns: Length of string, in bytes

Definition at line 93 of file utf8.c.

References utf8_next().

Referenced by textarea_insert_text(), and textarea_set_caret().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ utf8_bounded_length()

size_t utf8_bounded_length	(	const char *	s,
		size_t	l
	)

Calculated the length (in characters) of a bounded UTF-8 string.

Parameters

s	The string
l	Maximum length of input (in bytes)

Returns: Length of string, in characters

Definition at line 80 of file utf8.c.

Referenced by nsfont_width(), textarea_insert_text(), textarea_replace_text_internal(), and utf8_length().

Here is the caller graph for this function:

◆ utf8_char_byte_length()

size_t utf8_char_byte_length ( const char * s )

Calculate the length (in bytes) of a UTF-8 character.

Parameters

s	Pointer to start of character

Returns: Length of character, in bytes

Definition at line 104 of file utf8.c.

Referenced by ami_key_to_nskey().

Here is the caller graph for this function:

◆ utf8_clear_cd_cache()

static void utf8_clear_cd_cache ( void )

inlinestatic

Definition at line 148 of file utf8.c.

References last_cd.

Referenced by utf8_convert(), utf8_finalise(), and utf8_to_html().

Here is the caller graph for this function:

◆ utf8_convert()

static nserror utf8_convert	(	const char *	string,
		size_t	slen,
		const char *	from,
		const char *	to,
		char **	result_out,
		size_t *	result_len_out
	)

static

Convert a string from one encoding to another.

Parameters

string	The NULL-terminated string to convert
slen	Length of input string to consider (in bytes), or 0
from	The encoding name to convert from
to	The encoding name to convert to
result_out	Pointer to location in which to store result.
result_len_out	Pointer to location in which to store result length.

Returns: NSERROR_OK for no error, NSERROR_NOMEM on allocation error, NSERROR_BAD_ENCODING for a bad character encoding

Todo:: handle the various cases properly There are 3 possible error cases: a) Insufficiently large output buffer b) Invalid input byte sequence c) Incomplete input sequence

Definition at line 222 of file utf8.c.

References cd, from, get_cached_cd(), last_cd, NSERROR_NOMEM, NSERROR_OK, result, strndup(), to, and utf8_clear_cd_cache().

Referenced by utf8_from_enc(), and utf8_to_enc().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ utf8_convert_html_chunk()

static nserror utf8_convert_html_chunk	(	iconv_t	cd,
		const char *	chunk,
		size_t	inlen,
		char **	out,
		size_t *	outlen
	)

static

convert a chunk of html data

Definition at line 331 of file utf8.c.

References cd, NSERROR_NOMEM, NSERROR_OK, utf8_next(), and utf8_to_ucs4().

Referenced by utf8_to_html().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ utf8_finalise()

nserror utf8_finalise ( void )

Finalise the UTF-8 library.

Definition at line 197 of file utf8.c.

References last_cd, NSERROR_OK, and utf8_clear_cd_cache().

Referenced by netsurf_exit().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ utf8_from_enc()

nserror utf8_from_enc	(	const char *	string,
		const char *	encname,
		size_t	len,
		char **	result,
		size_t *	result_len
	)

Convert a string in the named encoding into a UTF-8 string.

Parameters

string	The NULL-terminated string to convert
encname	The encoding name (suitable for passing to iconv)
len	Length of input string to consider (in bytes), or 0
result	Pointer to location to store result (allocated on heap)
result_len	The length of the data placed in result.

Returns: standard nserror value

Definition at line 321 of file utf8.c.

References result, and utf8_convert().

Referenced by ami_clipboard_cat_collection(), nsgtk_viewsource(), and utf8_from_local_encoding().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ utf8_from_ucs4()

size_t utf8_from_ucs4	(	uint32_t	c,
		char *	s
	)

Convert a single UCS4 character into a UTF-8 multibyte sequence.

Encoding of UCS values outside the UTF-16 plane has been removed from RFC3629. This function conforms to RFC2279, however.

Parameters

c	The character to process (0 <= c <= 0x7FFFFFFF)
s	Pointer to 6 byte long output buffer

Returns: Length of multibyte sequence

Definition at line 56 of file utf8.c.

Referenced by fire_dom_keyboard_event(), ro_textarea_key_press(), and textarea_keypress().

Here is the caller graph for this function:

◆ utf8_length()

size_t utf8_length ( const char * s )

Calculate the length (in characters) of a NULL-terminated UTF-8 string.

Parameters

s	The string

Returns: Length of string

Definition at line 74 of file utf8.c.

References utf8_bounded_length().

Referenced by ro_textarea_insert_text(), ro_textarea_key_press(), ro_textarea_replace_text(), ro_textarea_set_caret(), textarea_replace_text_internal(), and textarea_set_text().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ utf8_next()

size_t utf8_next	(	const char *	s,
		size_t	l,
		size_t	o
	)

Find next legal UTF-8 char in string.

Parameters

s	The string
l	Maximum offset in string
o	Offset in the string to start at

Returns: Offset of first byte of next legal character

Definition at line 129 of file utf8.c.

Referenced by ami_font_bm_convert_local_to_utf8_offset(), amiga_nsfont_position_in_string(), amiga_nsfont_split(), fb_font_position(), fb_font_split(), fb_font_width(), framebuffer_plot_text(), nsgtk_cw_input_method_commit(), nsgtk_window_input_method_commit(), ro_gui_window_import_text(), ro_textarea_get_caret(), ro_textarea_insert_text(), ro_textarea_replace_text(), ro_textarea_set_caret(), ro_textarea_set_caret_xy(), textarea_char_to_byte_offset(), textarea_keypress(), textplain_coord_from_offset(), textplain_offset_from_coords(), textplain_redraw(), utf8_bounded_byte_length(), utf8_convert_html_chunk(), utf8_to_html(), and utf8_to_local_encoding().

Here is the caller graph for this function:

◆ utf8_prev()

size_t utf8_prev	(	const char *	s,
		size_t	o
	)

Find previous legal UTF-8 char in string.

Parameters

s	The string
o	Offset in the string to start at

Returns: Offset of first byte of previous legal character

Definition at line 117 of file utf8.c.

Referenced by textarea_keypress().

Here is the caller graph for this function:

◆ utf8_save_text()

bool utf8_save_text	(	const char *	utf8_text,
		const char *	path
	)

Save the given utf8 text to a file, converting to local encoding.

Parameters

utf8_text	text to save to file
path	pathname to save to

Returns: true iff the save succeeded

Definition at line 467 of file utf8.c.

References guit, NSERROR_OK, NSLOG, path(), netsurf_table::utf8, and gui_utf8_table::utf8_to_local.

Referenced by ro_gui_save_content().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ utf8_to_enc()

nserror utf8_to_enc	(	const char *	string,
		const char *	encname,
		size_t	len,
		char **	result
	)

Convert a UTF8 string into the named encoding.

Parameters

string	The NULL-terminated string to convert
encname	The encoding name (suitable for passing to iconv)
len	Length of input string to consider (in bytes), or 0
result	Pointer to location to store result (allocated on heap)

Returns: standard nserror value

Definition at line 314 of file utf8.c.

References result, and utf8_convert().

Referenced by ami_font_unicode_width(), amiga_nsfont_position_in_string(), amiga_nsfont_split(), amiga_nsfont_text(), form_encode_item(), utf8_to_font_encoding(), utf8_to_local(), and utf8_to_local_encoding().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ utf8_to_html()

nserror utf8_to_html	(	const char *	string,
		const char *	encname,
		size_t	len,
		char **	result
	)

Convert a UTF-8 encoded string into a string of the given encoding, applying HTML escape sequences where necessary.

Parameters

string	String to convert (NUL-terminated)
encname	Name of encoding to convert to
len	Length, in bytes, of the input string, or 0
result	Pointer to location to receive result