BLI_string_utf8.h File Reference

#include "BLI_compiler_attrs.h"
#include "BLI_sys_types.h"

Go to the source code of this file.

Macros
#define	BLI_UTF8_MAX   6
#define	BLI_UTF8_WIDTH_MAX   2 /* columns */
#define	BLI_UTF8_ERR   ((unsigned int)-1)
String Copy/Format Macros
Avoid repeating destination with sizeof(..). Note ARRAY_SIZE allows pointers on some platforms.
#define	STRNCPY_UTF8(dst, src)
#define	STRNCPY_UTF8_RLEN(dst, src)
#define	STRNLEN_UTF8(str)

Functions
char *	BLI_strncpy_utf8 (char *__restrict dst, const char *__restrict src, size_t dst_maxncpy) ATTR_NONNULL(1
char size_t	BLI_strncpy_utf8_rlen (char *__restrict dst, const char *__restrict src, size_t dst_maxncpy) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1
char size_t size_t	BLI_strncpy_utf8_rlen_unterminated (char *__restrict dst, const char *__restrict src, size_t dst_maxncpy)
ptrdiff_t	BLI_str_utf8_invalid_byte (const char *str, size_t str_len) ATTR_NONNULL(1)
int	BLI_str_utf8_invalid_strip (char *str, size_t str_len) ATTR_NONNULL(1)
int	BLI_str_utf8_invalid_substitute (char *str, size_t str_len, const char substitute) ATTR_NONNULL(1)
const char *	BLI_str_utf8_invalid_substitute_as_needed (const char *str, size_t str_len, const char substitute, char *buf, const size_t buf_maxncpy) ATTR_NONNULL(1
const char int	BLI_str_utf8_size_or_error (const char *p) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
int	BLI_str_utf8_size_safe (const char *p) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
unsigned int	BLI_str_utf8_as_unicode_or_error (const char *p) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
unsigned int	BLI_str_utf8_as_unicode_safe (const char *p) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
unsigned int	BLI_str_utf8_as_unicode_step_safe (const char *__restrict p, size_t p_len, size_t *__restrict index) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1
unsigned int unsigned int	BLI_str_utf8_as_unicode_step_or_error (const char *__restrict p, size_t p_len, size_t *__restrict index) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1
unsigned int unsigned int size_t	BLI_str_utf8_from_unicode_len (unsigned int c) ATTR_WARN_UNUSED_RESULT
size_t	BLI_str_utf8_from_unicode (unsigned int c, char *dst, size_t dst_maxncpy) ATTR_NONNULL(2)
size_t	BLI_str_utf8_as_utf32 (char32_t *__restrict dst_w, const char *__restrict src_c, size_t dst_w_maxncpy) ATTR_NONNULL(1
size_t size_t	BLI_str_utf32_as_utf8 (char *__restrict dst, const char32_t *__restrict src, size_t dst_maxncpy) ATTR_NONNULL(1
size_t size_t size_t	BLI_str_utf32_as_utf8_len_ex (const char32_t *src, size_t src_maxlen) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
size_t	BLI_str_utf32_as_utf8_len (const char32_t *src) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
const char *	BLI_str_find_prev_char_utf8 (const char *p, const char *str_start) ATTR_WARN_UNUSED_RESULT ATTR_RETURNS_NONNULL ATTR_NONNULL(1
const char const char *	BLI_str_find_next_char_utf8 (const char *p, const char *str_end) ATTR_WARN_UNUSED_RESULT ATTR_RETURNS_NONNULL ATTR_NONNULL(1
const char const char size_t	BLI_wstrlen_utf8 (const wchar_t *src) ATTR_NONNULL(1) ATTR_WARN_UNUSED_RESULT
size_t	BLI_strlen_utf8_ex (const char *strc, size_t *r_len_bytes) ATTR_NONNULL(1
size_t	BLI_strlen_utf8 (const char *strc) ATTR_NONNULL(1) ATTR_WARN_UNUSED_RESULT
size_t	BLI_strnlen_utf8_ex (const char *strc, size_t strc_maxlen, size_t *r_len_bytes) ATTR_NONNULL(1
size_t size_t	BLI_strnlen_utf8 (const char *strc, size_t strc_maxlen) ATTR_NONNULL(1) ATTR_WARN_UNUSED_RESULT
size_t	BLI_strncpy_wchar_as_utf8 (char *__restrict dst, const wchar_t *__restrict src, size_t dst_maxncpy) ATTR_NONNULL(1
size_t size_t	BLI_strncpy_wchar_from_utf8 (wchar_t *__restrict dst_w, const char *__restrict src_c, size_t dst_w_maxncpy) ATTR_NONNULL(1
size_t size_t int	BLI_wcwidth_or_error (char32_t ucs) ATTR_WARN_UNUSED_RESULT
int	BLI_wcwidth_safe (char32_t ucs) ATTR_WARN_UNUSED_RESULT
int	BLI_wcswidth_or_error (const char32_t *pwcs, size_t n) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
char32_t	BLI_str_utf32_char_to_upper (char32_t wc)
char32_t	BLI_str_utf32_char_to_lower (char32_t wc)
bool	BLI_str_utf32_char_is_breaking_space (char32_t codepoint)
bool	BLI_str_utf32_char_is_optional_break_after (char32_t codepoint, char32_t codepoint_prev)
bool	BLI_str_utf32_char_is_optional_break_before (char32_t codepoint, char32_t codepoint_prev)
int	BLI_str_utf8_char_width_or_error (const char *p) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
int	BLI_str_utf8_char_width_safe (const char *p) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
size_t	BLI_str_partition_utf8 (const char *str, const unsigned int delim[], const char **r_sep, const char **r_suf) ATTR_NONNULL(1
size_t size_t	BLI_str_rpartition_utf8 (const char *str, const unsigned int delim[], const char **r_sep, const char **r_suf) ATTR_NONNULL(1
size_t size_t size_t	BLI_str_partition_ex_utf8 (const char *str, const char *end, const unsigned int delim[], const char **r_sep, const char **r_suf, bool from_right) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1
size_t size_t size_t bool	BLI_str_utf8_truncate_at_size (char *str, const size_t str_size)
int	BLI_str_utf8_offset_to_index (const char *str, size_t str_len, int offset_target) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
int	BLI_str_utf8_offset_from_index (const char *str, size_t str_len, int index_target) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
int	BLI_str_utf8_offset_to_column (const char *str, size_t str_len, int offset_target) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
int	BLI_str_utf8_offset_from_column (const char *str, size_t str_len, int column_target) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
int	BLI_str_utf8_offset_to_column_with_tabs (const char *str, size_t str_len, int offset_target, int tab_width) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)
int	BLI_str_utf8_offset_from_column_with_tabs (const char *str, size_t str_len, int column_target, int tab_width) ATTR_WARN_UNUSED_RESULT ATTR_NONNULL(1)

Variables
size_t	ATTR_WARN_UNUSED_RESULT

Macro Definition Documentation

BLI_UTF8_ERR

#define BLI_UTF8_ERR   ((unsigned int)-1)

Definition at line 286 of file BLI_string_utf8.h.

Referenced by blf_glyph_from_utf8_and_step(), BLI_str_partition_ex_utf8(), BLI_str_utf8_as_unicode_or_error(), BLI_str_utf8_as_unicode_safe(), BLI_str_utf8_as_unicode_step_or_error(), BLI_str_utf8_as_unicode_step_safe(), BLI_str_utf8_as_utf32(), BLI_str_utf8_char_width_or_error(), and BLI_str_utf8_char_width_safe().

BLI_UTF8_MAX

#define BLI_UTF8_MAX   6

Size in bytes.

Definition at line 284 of file BLI_string_utf8.h.

Referenced by console_insert_invoke(), draw_suggestion_list(), find_family_object(), text_insert_invoke(), txt_add_char_intern(), and txt_replace_char().

BLI_UTF8_WIDTH_MAX

#define BLI_UTF8_WIDTH_MAX   2 /* columns */

Definition at line 285 of file BLI_string_utf8.h.

Referenced by textview_wrap_offsets().

STRNCPY_UTF8

#define STRNCPY_UTF8	(		dst,
			src )

Value:

BLI_strncpy_utf8(dst, src, ARRAY_SIZE(dst))

Definition at line 294 of file BLI_string_utf8.h.

Referenced by action_groups_add_new(), ANIM_armature_bonecoll_name_set(), ANIM_bonecoll_new(), arg_handle_engine_set(), asset_metadata_tag_add(), BKE_gpencil_layer_addnew(), BKE_id_new_name_validate(), BKE_image_add_renderslot(), BKE_keyblock_add(), BKE_keyingset_add(), BKE_object_defgroup_set_name(), BKE_preferences_asset_library_name_set(), BKE_preferences_extension_repo_name_set(), BKE_shaderfx_new(), BKE_view_layer_add(), BKE_view_layer_add_aov(), BKE_view_layer_add_lightgroup(), BKE_view_layer_rename(), BKE_view_layer_rename_lightgroup(), blo_do_versions_userdef(), boid_new_rule(), blender::ed::transform::calculatePropRatio(), blender::animrig::Channelbag::channel_group_create(), blender::animrig::convert_to_layered_action(), CustomData_set_layer_unique_name(), do_versions_nodetree_file_output_layers_2_64_5(), driver_add_new_variable(), dynamicPaintSurface_setUniqueName(), ED_info_draw_stats(), ED_node_shader_default(), blender::seq::ensure_unique_name(), blender::init_data(), blender::ed::asset::index::init_indexer_entry_from_value(), blender::animrig::Action::layer_add(), blender::bke::greasepencil::convert::legacy_object_modifier_common(), blender::ed::object::modifier_add(), modifier_allocate_and_init(), ntreeCompositOutputFileAddSocket(), ntreeCompositOutputFileSetLayer(), ntreeCompositOutputFileSetPath(), blender::ed::object::object_speaker_add_exec(), preferences_extension_repo_add_exec(), proj_paint_add_slot(), blender::ed::object::shaderfx_add(), blender::animrig::Action::slot_add(), blender::animrig::Action::slot_identifier_define(), blender::animrig::Action::slot_identifier_propagate(), blender::animrig::tests::TEST_F(), blender::animrig::tests::TEST_F(), blender::animrig::tests::TEST_F(), blender::animrig::tests::TEST_F(), ui_but_anim_expression_create(), ui_but_anim_expression_set(), blender::ed::space_node::ui_node_menu_column(), view_layer_add(), WM_dropboxmap_find(), WM_keyconfig_new(), and wm_keymap_new().

STRNCPY_UTF8_RLEN

#define STRNCPY_UTF8_RLEN	(		dst,
			src )

Value:

BLI_strncpy_utf8_rlen(dst, src, ARRAY_SIZE(dst))

Definition at line 295 of file BLI_string_utf8.h.

Referenced by bone_autoside_name().

STRNLEN_UTF8

#define STRNLEN_UTF8

(

str

)

Value:

BLI_strnlen_utf8(str, ARRAY_SIZE(str))

Definition at line 297 of file BLI_string_utf8.h.

Function Documentation

BLI_str_find_next_char_utf8()

const char const char * BLI_str_find_next_char_utf8	(	const char *	p,
		const char *	str_end )

Parameters

p	a pointer to a position within a UTF8 encoded string
str_end	a pointer to the byte following the end of the string.

Finds the start of the next UTF8 character in the string after p

p does not have to be at the beginning of a UTF8 character. No check is made to see if the character found is actually valid other than it starts with an appropriate byte.

Returns

a pointer to the found character or a pointer to the null terminating character '\0'.

References ATTR_RETURNS_NONNULL, and ATTR_WARN_UNUSED_RESULT.

Referenced by BLI_str_cursor_step_next_utf8(), ui_text_clip_give_next_off(), and ui_text_position_from_hidden().

BLI_str_find_prev_char_utf8()

const char * BLI_str_find_prev_char_utf8	(	const char *	p,
		const char *	str_start )

BLI_str_find_prev_char_utf8:

Parameters

p	pointer to some position within str
str_start	pointer to the beginning of a UTF8 encoded string

Given a position p with a UTF8 encoded string str, find the start of the previous UTF8 character starting before. p Returns str_start if no UTF8 characters are present in str_start before p.

p does not have to be at the beginning of a UTF8 character. No check is made to see if the character found is actually valid other than it starts with an appropriate byte.

Returns

A pointer to the found character.

References ATTR_RETURNS_NONNULL, and ATTR_WARN_UNUSED_RESULT.

Referenced by blf_font_width_to_rstrlen(), BLI_str_cursor_step_prev_utf8(), draw_brackets(), blender::nodes::geo_eval_log::StringLog::StringLog(), text_delete_exec(), txt_wrap_move_eol(), ui_text_clip_cursor(), ui_text_clip_give_prev_off(), ui_text_clip_right_label(), and unit_find_str().

BLI_str_partition_ex_utf8()

size_t size_t size_t BLI_str_partition_ex_utf8	(	const char *	str,
		const char *	end,
		const unsigned int	delim[],
		const char **	r_sep,
		const char **	r_suf,
		bool	from_right )

References ATTR_NONNULL, ATTR_WARN_UNUSED_RESULT, BLI_str_partition_ex_utf8(), BLI_str_utf8_offset_from_column(), BLI_str_utf8_offset_from_column_with_tabs(), BLI_str_utf8_offset_from_index(), BLI_str_utf8_offset_to_column(), BLI_str_utf8_offset_to_column_with_tabs(), BLI_str_utf8_offset_to_index(), BLI_str_utf8_truncate_at_size(), and str.

Referenced by BLI_str_partition_ex_utf8(), blender::ed::space_node::node_select_grouped_name(), and TEST().

BLI_str_partition_utf8()

size_t BLI_str_partition_utf8	(	const char *	str,
		const unsigned int	delim[],
		const char **	r_sep,
		const char **	r_suf )

References ATTR_NONNULL, BLI_str_partition_utf8(), and str.

Referenced by BLI_str_partition_utf8(), and TEST().

BLI_str_rpartition_utf8()

size_t size_t BLI_str_rpartition_utf8	(	const char *	str,
		const unsigned int	delim[],
		const char **	r_sep,
		const char **	r_suf )

References ATTR_NONNULL, BLI_str_rpartition_utf8(), and str.

Referenced by BLI_str_rpartition_utf8(), and TEST().

BLI_str_utf32_as_utf8()

size_t size_t BLI_str_utf32_as_utf8	(	char *__restrict	dst,
		const char32_t *__restrict	src,
		size_t	dst_maxncpy )

References ATTR_WARN_UNUSED_RESULT.

Referenced by copy_selection(), blender::nodes::node_geo_string_to_curves_cc::create_curve_instances(), ED_curve_editfont_load(), font_select_to_buffer(), insert_text_invoke(), and paste_text_exec().

BLI_str_utf32_as_utf8_len()

size_t BLI_str_utf32_as_utf8_len

(

const char32_t *

src

)

Returns

The UTF32 len in UTF8.

Definition at line 1156 of file string_utf8.cc.

References BLI_str_utf8_from_unicode_len(), and len.

Referenced by BKE_vfont_clipboard_set(), and ED_curve_editfont_load().

BLI_str_utf32_as_utf8_len_ex()

size_t size_t size_t BLI_str_utf32_as_utf8_len_ex	(	const char32_t *	src,
		size_t	src_maxlen )

Returns

The UTF32 len in UTF8 with a clamped length.

Definition at line 1144 of file string_utf8.cc.

References BLI_str_utf8_from_unicode_len(), and len.

Referenced by font_select_to_buffer().

BLI_str_utf32_char_is_breaking_space()

bool BLI_str_utf32_char_is_breaking_space

(

char32_t

codepoint

)

Definition at line 826 of file string_utf8.cc.

References ELEM.

Referenced by blf_font_wrap_apply(), BLI_str_utf32_char_is_optional_break_after(), BLI_str_utf32_char_is_optional_break_before(), and BLI_strncpy_wchar_from_utf8().

BLI_str_utf32_char_is_optional_break_after()

bool BLI_str_utf32_char_is_optional_break_after	(	char32_t	codepoint,
		char32_t	codepoint_prev )

Definition at line 848 of file string_utf8.cc.

References BLI_str_utf32_char_is_breaking_space(), and ELEM.

Referenced by blf_font_wrap_apply(), and BLI_strncpy_wchar_from_utf8().

BLI_str_utf32_char_is_optional_break_before()

bool BLI_str_utf32_char_is_optional_break_before	(	char32_t	codepoint,
		char32_t	codepoint_prev )

Definition at line 889 of file string_utf8.cc.

References BLI_str_utf32_char_is_breaking_space(), and ELEM.

Referenced by blf_font_wrap_apply(), and BLI_strncpy_wchar_from_utf8().

BLI_str_utf32_char_to_lower()

char32_t BLI_str_utf32_char_to_lower

(

char32_t

wc

)

Return the lowercase of a 32-bit character or the character when no case change is needed.

Note

A 1:1 mapping doesn't account for multiple characters as part of conversion in some cases.

Definition at line 713 of file string_utf8.cc.

References ARRAY_SIZE, max, min, and U.

Referenced by BLI_strncpy_wchar_from_utf8(), and set_case().

BLI_str_utf32_char_to_upper()

char32_t BLI_str_utf32_char_to_upper

(

char32_t

wc

)

Return the uppercase of a 32-bit character or the character when no case change is needed.

Note

A 1:1 mapping doesn't account for multiple characters as part of conversion in some cases.

Definition at line 602 of file string_utf8.cc.

References ARRAY_SIZE, max, min, and U.

Referenced by BLI_strncpy_wchar_from_utf8(), and set_case().

BLI_str_utf8_as_unicode_or_error()

unsigned int BLI_str_utf8_as_unicode_or_error

(

const char *

p

)

Parameters

p	a pointer to Unicode character encoded as UTF8

Converts a sequence of bytes encoded as UTF8 to a Unicode character. If p does not point to a valid UTF8 encoded character, results are undefined. If you are not sure that the bytes are complete valid Unicode characters, you should use g_utf8_get_char_validated() instead.

Return value: the resulting character

Definition at line 966 of file string_utf8.cc.

References BLI_UTF8_ERR, len, mask(), UNLIKELY, utf8_char_compute_skip_or_error_with_mask(), and utf8_char_decode().

Referenced by BLI_str_partition_ex_utf8(), BLI_str_utf8_as_unicode_safe(), BLI_str_utf8_char_width_or_error(), BLI_str_utf8_char_width_safe(), insert_text_invoke(), key_event_glyph_or_text(), text_autocomplete_build(), and text_insert_invoke().

BLI_str_utf8_as_unicode_safe()

unsigned int BLI_str_utf8_as_unicode_safe

(

const char *

p

)

Definition at line 980 of file string_utf8.cc.

References BLI_str_utf8_as_unicode_or_error(), BLI_UTF8_ERR, result, and UNLIKELY.

Referenced by blf_font_glyph_advance(), blf_font_wrap_apply(), blender::string_search::extract_normalized_words(), and blender::string_search::get_fuzzy_match_errors().

BLI_str_utf8_as_unicode_step_or_error()

unsigned int unsigned int BLI_str_utf8_as_unicode_step_or_error	(	const char *__restrict	p,
		size_t	p_len,
		size_t *__restrict	index )

UTF8 decoding that steps over the index (unless an error is encountered).

Parameters

p	The text to step over.
p_len	The length of p.
index	Index of p to step over.

Returns

the code-point or BLI_UTF8_ERR if there is a decoding error.

Note

The behavior for clipped text (where p_len limits decoding trailing bytes) must have the same behavior is encountering a nil byte, so functions that only use the first part of a string has matching behavior to functions that null terminate the text.

References ATTR_WARN_UNUSED_RESULT.

Referenced by cursor_delim_type_utf8().

BLI_str_utf8_as_unicode_step_safe()

unsigned int BLI_str_utf8_as_unicode_step_safe	(	const char *__restrict	p,
		size_t	p_len,
		size_t *__restrict	index )

UTF8 decoding that steps over the index. When an error is encountered fall back to LATIN1, stepping over a single byte.

Parameters

p	The text to step over.
p_len	The length of p.
index	Index of p to step over.

Returns

the code-point (p + *index) if there is a decoding error.

References ATTR_WARN_UNUSED_RESULT.

Referenced by blf_glyph_from_utf8_and_step(), blender::string_search::damerau_levenshtein_distance(), blender::string_search::extract_normalized_words(), blender::string_search::match_word_initials(), text_autocomplete_build(), text_insert_exec(), txt_insert_buf(), txt_replace_char(), and utf8_as_char32().

BLI_str_utf8_as_utf32()

size_t BLI_str_utf8_as_utf32	(	char32_t *__restrict	dst_w,
		const char *__restrict	src_c,
		size_t	dst_w_maxncpy )

Referenced by ED_curve_editfont_make(), font_paste_utf8(), blender::nodes::node_geo_string_to_curves_cc::get_text_layout(), insert_text_exec(), and vfont_to_curve().

BLI_str_utf8_char_width_or_error()

int BLI_str_utf8_char_width_or_error

(

const char *

p

)

Warning

can return -1 on bad chars.

Definition at line 572 of file string_utf8.cc.

References BLI_str_utf8_as_unicode_or_error(), BLI_UTF8_ERR, and BLI_wcwidth_or_error().

Referenced by blf_str_offset_from_cursor_position(), BLI_str_cursor_step_next_utf8(), BLI_str_cursor_step_prev_utf8(), and BLI_strncpy_wchar_from_utf8().

BLI_str_utf8_char_width_safe()

int BLI_str_utf8_char_width_safe

(

const char *

p

)

Definition at line 582 of file string_utf8.cc.

References BLI_str_utf8_as_unicode_or_error(), BLI_UTF8_ERR, and BLI_wcwidth_safe().

Referenced by BLI_strncpy_wchar_from_utf8(), console_cursor_wrap_offset(), flatten_column_to_offset(), flatten_width(), space_text_cursor_set_to_pos_wrapped(), space_text_draw(), space_text_draw_wrapped(), space_text_get_char_pos(), space_text_get_cursor_rel(), space_text_get_visible_lines(), space_text_wrap_offset_in_line(), textview_wrap_offsets(), txt_utf8_forward_columns(), txt_wrap_move_bol(), txt_wrap_move_eol(), and while().

BLI_str_utf8_from_unicode()

size_t BLI_str_utf8_from_unicode	(	unsigned int	c,
		char *	dst,
		size_t	dst_maxncpy )

BLI_str_utf8_from_unicode:

Parameters

c	a Unicode character code
dst	output buffer, must have at least dst_maxncpy bytes of space. If the length required by c exceeds dst_maxncpy, the bytes available bytes will be zeroed and dst_maxncpy returned.

Converts a single character to UTF8.

Returns

number of bytes written.

Definition at line 1067 of file string_utf8.cc.

References BLI_string_debug_size, i, len, UNLIKELY, and UTF8_VARS_FROM_CHAR32.

Referenced by BLI_str_utf32_as_utf8(), BLI_strncpy_wchar_as_utf8(), find_family_object(), blender::io::usd::make_safe_name(), txt_add_char_intern(), txt_extended_ascii_as_utf8(), txt_replace_char(), and wm_event_add_ghostevent().

BLI_str_utf8_from_unicode_len()

unsigned int unsigned int size_t BLI_str_utf8_from_unicode_len

(

unsigned int

c

)

Definition at line 1055 of file string_utf8.cc.

References len, and UTF8_VARS_FROM_CHAR32.

Referenced by blf_font_wrap_apply(), BLI_str_utf32_as_utf8_len(), BLI_str_utf32_as_utf8_len_ex(), BLI_wstrlen_utf8(), and text_insert_invoke().

BLI_str_utf8_invalid_byte()

ptrdiff_t BLI_str_utf8_invalid_byte	(	const char *	str,
		size_t	str_len )

Find first UTF8 invalid byte in given str, of str_len bytes.

Returns

the offset of the first invalid byte.

Definition at line 150 of file string_utf8.cc.

References ELEM, str, and utf8_char_compute_skip().

Referenced by blender::string_search::StringSearchBase::add_impl(), blf_str_is_utf8_valid_lazy_init(), BLI_str_utf8_invalid_strip(), BLI_str_utf8_invalid_substitute(), BLI_str_utf8_invalid_substitute_as_needed(), blender::StringRefBase::copy_utf8_truncated(), blender::nodes::node_geo_import_text::node_geo_exec(), TEST(), txt_extended_ascii_as_utf8(), and UI_text_clip_middle_ex().

BLI_str_utf8_invalid_strip()

int BLI_str_utf8_invalid_strip	(	char *	str,
		size_t	str_len )

Remove any invalid UTF8 byte (taking into account multi-bytes sequences).

Parameters

str	a null terminated string.
str_len	the result of strlen(str).

Returns

number of stripped bytes.

Definition at line 285 of file string_utf8.cc.

References BLI_assert, BLI_str_utf8_invalid_byte(), and str.

Referenced by BKE_id_new_name_validate(), BKE_vfontdata_from_freetypefont(), blo_do_versions_450(), blender::seq::edit_strip_name_set(), id_name_final_build(), outputNumInput(), TEST(), ui_textedit_end(), and wm_clipboard_text_get_ex().

BLI_str_utf8_invalid_substitute()

int BLI_str_utf8_invalid_substitute	(	char *	str,
		size_t	str_len,
		const char	substitute )

Substitute any invalid UTF8 byte with substitute (taking into account multi-bytes sequences). The length of the string remains unchanged.

Parameters

str	a null terminated string.
str_len	the result of strlen(str).

Returns

number of bytes replaced.

Definition at line 310 of file string_utf8.cc.

References BLI_assert, BLI_str_utf8_invalid_byte(), and str.

Referenced by BLI_str_utf8_invalid_substitute_as_needed(), and TEST().

BLI_str_utf8_invalid_substitute_as_needed()

const char * BLI_str_utf8_invalid_substitute_as_needed ( const char * str, size_t str_len, const char substitute, char * buf, const size_t buf_maxncpy )

nodiscard

A utility for BLI_str_utf8_invalid_substitute that returns str when it contains a valid UTF8 string. Otherwise it is copied into buf with invalid byte sequences substituted for substitute.

Note

This is intended for situations when the string is expected to be valid, where copying and substituting values is typically not needed.

References ATTR_WARN_UNUSED_RESULT, and str.

Referenced by WM_window_title().

BLI_str_utf8_offset_from_column()

int BLI_str_utf8_offset_from_column	(	const char *	str,
		size_t	str_len,
		int	column_target )

Definition at line 1324 of file string_utf8.cc.

References BLI_str_utf8_as_unicode_step_safe(), BLI_wcwidth_safe(), and str.

Referenced by BLI_str_partition_ex_utf8().

BLI_str_utf8_offset_from_column_with_tabs()

int BLI_str_utf8_offset_from_column_with_tabs	(	const char *	str,
		size_t	str_len,
		int	column_target,
		int	tab_width )

Definition at line 1357 of file string_utf8.cc.

References BLI_str_utf8_as_unicode_step_safe(), BLI_wcwidth_safe(), and str.

Referenced by BLI_str_partition_ex_utf8(), text_convert_whitespace_exec(), textview_draw_string(), txt_move_down(), txt_move_up(), and txt_wrap_move_bol().

BLI_str_utf8_offset_from_index()

int BLI_str_utf8_offset_from_index	(	const char *	str,
		size_t	str_len,
		int	index_target )

Return the byte offset in str from index_target.

Parameters

index_target

The unicode index, where multi-byte characters are counted once. There is no need to clamp this value, the index is logically clamped to BLI_strlen_utf8(str) or below.

Definition at line 1296 of file string_utf8.cc.

References BLI_assert, BLI_str_utf8_as_unicode_step_safe(), str, and UNUSED_VARS.

Referenced by BLI_str_partition_ex_utf8(), ED_space_text_region_location_from_cursor(), blender::nodes::node_geo_string_to_curves_cc::get_text_layout(), blender::nodes::node_fn_slice_string_cc::node_build_multi_function(), TEST(), and txt_sel_set().

BLI_str_utf8_offset_to_column()

int BLI_str_utf8_offset_to_column	(	const char *	str,
		size_t	str_len,
		int	offset_target )

Definition at line 1310 of file string_utf8.cc.

References BLI_assert, BLI_str_utf8_as_unicode_step_safe(), BLI_wcwidth_safe(), and str.

Referenced by BLI_str_partition_ex_utf8(), and space_text_wrap_offset_in_line().

BLI_str_utf8_offset_to_column_with_tabs()

int BLI_str_utf8_offset_to_column_with_tabs	(	const char *	str,
		size_t	str_len,
		int	offset_target,
		int	tab_width )

Definition at line 1339 of file string_utf8.cc.

References BLI_assert, BLI_str_utf8_as_unicode_step_safe(), BLI_wcwidth_safe(), and str.

Referenced by BLI_str_partition_ex_utf8(), text_convert_whitespace_exec(), textview_draw_sel(), txt_move_down(), and txt_move_up().

BLI_str_utf8_offset_to_index()

int BLI_str_utf8_offset_to_index	(	const char *	str,
		size_t	str_len,
		int	offset_target )

Definition at line 1279 of file string_utf8.cc.

References BLI_assert, BLI_str_utf8_as_unicode_step_safe(), str, and UNUSED_VARS.

Referenced by BLI_str_partition_ex_utf8(), draw_brackets(), and text_jump_to_file_at_point_exec().

BLI_str_utf8_size_or_error()

const char int BLI_str_utf8_size_or_error

(

const char *

p

)

Returns

The size (in bytes) of a single UTF8 char.

Warning

Can return -1 on bad chars.

Definition at line 956 of file string_utf8.cc.

References utf8_char_compute_skip_or_error().

Referenced by handleNumInput(), ui_do_but_textedit(), ui_handle_menu_letter_press_search(), wm_event_add_ghostevent(), WM_event_print(), and WM_event_utf8_to_ascii().

BLI_str_utf8_size_safe()

int BLI_str_utf8_size_safe

(

const char *

p

)

Use when we want to skip errors.

Definition at line 961 of file string_utf8.cc.

References utf8_char_compute_skip().

Referenced by BLI_strlen_utf8_ex(), BLI_strnlen_utf8_ex(), blender::seq::build_character_info(), console_cursor_wrap_offset(), console_indent_or_autocomplete_exec(), console_insert_invoke(), draw_brackets(), flatten_column_to_offset(), flatten_string(), flatten_width(), blender::string_search::get_fuzzy_match_errors(), blender::ed::vse::sequencer_text_insert_invoke(), space_text_cursor_set_to_pos_wrapped(), space_text_draw(), space_text_draw_wrapped(), space_text_get_char_pos(), space_text_get_cursor_rel(), space_text_get_visible_lines(), space_text_wrap_offset_in_line(), text_convert_whitespace_exec(), text_font_draw_character_utf8(), text_format_fill(), text_insert_invoke(), textview_wrap_offsets(), txt_utf8_forward_columns(), txt_wrap_move_bol(), txt_wrap_move_eol(), txtfmt_glsl_format_line(), txtfmt_osl_format_line(), txtfmt_pov_format_line(), txtfmt_pov_ini_format_line(), txtfmt_py_format_line(), ui_text_clip_cursor(), and while().

BLI_str_utf8_truncate_at_size()

size_t size_t size_t bool BLI_str_utf8_truncate_at_size	(	char *	str,
		const size_t	str_size )

Ensure that str has a null byte in the range of [0..str_size], while not generating any invalid UTF-8 code. The resulting strlen(str) is guaranteed to be less than str_size.

Returns

true when str was truncated.

Definition at line 1256 of file string_utf8.cc.

References BLI_assert, BLI_strnlen_utf8_ex(), and str.

Referenced by BLI_str_partition_ex_utf8(), bone_autoside_name(), long_id_names_process_action_slots_identifiers(), and read_id_struct().

BLI_strlen_utf8()

size_t BLI_strlen_utf8

(

const char *

strc

)

Definition at line 492 of file string_utf8.cc.

References BLI_strlen_utf8_ex().

Referenced by blo_do_versions_260(), blender::seq::calc_text_runtime(), insert_text_exec(), key_event_glyph_or_text(), blender::nodes::node_fn_string_length_cc::node_build_multi_function(), radial_control_paint_cursor(), TEST(), txt_sel_set(), ui_but_text_password_hide(), and blender::ed::object::voxel_size_edit_draw().

BLI_strlen_utf8_ex()

size_t BLI_strlen_utf8_ex	(	const char *	strc,
		size_t *	r_len_bytes )

References BLI_strlen_utf8_ex().

Referenced by BKE_curve_init(), BLI_strlen_utf8_ex(), curve_blend_read_data(), blender::nodes::node_geo_string_to_curves_cc::get_text_layout(), and txt_add_object().

BLI_strncpy_utf8()

char * BLI_strncpy_utf8	(	char *__restrict	dst,
		const char *__restrict	src,
		size_t	dst_maxncpy )

Referenced by acf_filldrivers_name(), acf_generic_idblock_name(), acf_nla_controls_name(), acf_summary_name(), BKE_preferences_extension_remote_to_name(), blender::animrig::versioning::convert_legacy_action_assignments(), CustomData_validate_layer_name(), blender::seq::edit_strip_name_set(), fileselect_ensure_updated_file_params(), blender::animrig::generic_assign_action_slot(), node_blend_label(), node_filter_label(), blender::nodes::node_fn_boolean_math_cc::node_label(), blender::nodes::node_fn_compare_cc::node_label(), blender::nodes::node_fn_float_to_int_cc::node_label(), blender::nodes::node_fn_match_string_cc::node_label(), blender::nodes::node_geo_closure_cc::input_node::node_label(), blender::nodes::node_geo_foreach_geometry_element_cc::input_node::node_label(), blender::nodes::node_geo_repeat_cc::repeat_input_node::node_label(), blender::nodes::node_geo_simulation_cc::sim_input_node::node_label(), blender::nodes::node_geo_warning_cc::node_label(), blender::nodes::node_composite_mask_cc::node_mask_label(), node_math_label(), node_vector_math_label(), pyrna_py_to_prop(), blender::nodes::node_sh_mix_cc::sh_node_mix_label(), blender::animrig::Action::slot_add(), blender::animrig::Action::slot_add_for_id_type(), blender::animrig::Action::slot_display_name_define(), ui_but_string_get_ex(), ui_but_string_set(), blender::ed::space_node::ui_node_sock_name(), and ui_textedit_string_set().

BLI_strncpy_utf8_rlen()

char size_t BLI_strncpy_utf8_rlen	(	char *__restrict	dst,
		const char *__restrict	src,
		size_t	dst_maxncpy )

References ATTR_WARN_UNUSED_RESULT, and str.

Referenced by BLI_uniquename_cb().

BLI_strncpy_utf8_rlen_unterminated()

char size_t size_t BLI_strncpy_utf8_rlen_unterminated	(	char *__restrict	dst,
		const char *__restrict	src,
		size_t	dst_maxncpy )

A version of BLI_strncpy_utf8_rlen that doesn't null terminate the string.

Note

Useful for C++ APIs that don't null terminate strings.

Definition at line 422 of file string_utf8.cc.

References BLI_string_debug_size, and str_utf8_copy_max_bytes_impl().

Referenced by blender::StringRefBase::copy_utf8_truncated().

BLI_strncpy_wchar_as_utf8()

size_t BLI_strncpy_wchar_as_utf8	(	char *__restrict	dst,
		const wchar_t *__restrict	src,
		size_t	dst_maxncpy )

References ATTR_NONNULL, and BLI_strncpy_wchar_as_utf8().

Referenced by BKE_appdir_font_folder_default(), BLI_current_working_dir(), and BLI_strncpy_wchar_as_utf8().

BLI_strncpy_wchar_from_utf8()

size_t size_t BLI_strncpy_wchar_from_utf8	(	wchar_t *__restrict	dst_w,
		const char *__restrict	src_c,
		size_t	dst_w_maxncpy )

References ATTR_NONNULL, ATTR_WARN_UNUSED_RESULT, BLI_str_utf32_char_is_breaking_space(), BLI_str_utf32_char_is_optional_break_after(), BLI_str_utf32_char_is_optional_break_before(), BLI_str_utf32_char_to_lower(), BLI_str_utf32_char_to_upper(), BLI_str_utf8_char_width_or_error(), BLI_str_utf8_char_width_safe(), BLI_strncpy_wchar_from_utf8(), BLI_wcswidth_or_error(), BLI_wcwidth_or_error(), and BLI_wcwidth_safe().

Referenced by BLI_strncpy_wchar_from_utf8().

BLI_strnlen_utf8()

size_t size_t BLI_strnlen_utf8	(	const char *	strc,
		size_t	strc_maxlen )

Parameters

strc	the string to measure the length.
strc_maxlen	the string length (in bytes)

Returns

the unicode length (not in bytes!)

Definition at line 526 of file string_utf8.cc.

References BLI_strnlen_utf8_ex().

Referenced by BLI_strnlen_utf8_ex(), blender::string_search::count_utf8_code_points(), space_text_draw_wrapped(), TEST(), and ui_text_position_to_hidden().

BLI_strnlen_utf8_ex()

size_t BLI_strnlen_utf8_ex	(	const char *	strc,
		size_t	strc_maxlen,
		size_t *	r_len_bytes )

References ATTR_NONNULL, ATTR_WARN_UNUSED_RESULT, BLI_strnlen_utf8(), and BLI_strnlen_utf8_ex().

Referenced by BLI_strnlen_utf8_ex(), blender::ed::vse::sequencer_text_edit_paste_exec(), blender::nodes::node_fn_find_in_string_cc::string_find(), and ui_textedit_insert_buf().

BLI_wcswidth_or_error()

int BLI_wcswidth_or_error	(	const char32_t *	pwcs,
		size_t	n )

Definition at line 567 of file string_utf8.cc.

Referenced by BLI_strncpy_wchar_from_utf8().

BLI_wcwidth_or_error()

size_t size_t int BLI_wcwidth_or_error

(

char32_t

ucs

)

Count columns that character/string occupies (based on wcwidth.co).

Definition at line 549 of file string_utf8.cc.

Referenced by blf_glyph_render(), BLI_str_cursor_step_next_utf32(), BLI_str_cursor_step_prev_utf32(), BLI_str_utf8_char_width_or_error(), BLI_strncpy_wchar_from_utf8(), and BLI_wcwidth_safe().

BLI_wcwidth_safe()

int BLI_wcwidth_safe

(

char32_t

ucs

)

Definition at line 558 of file string_utf8.cc.

References BLI_wcwidth_or_error().

Referenced by blf_font_draw_mono(), BLI_str_utf8_char_width_safe(), BLI_str_utf8_offset_from_column(), BLI_str_utf8_offset_from_column_with_tabs(), BLI_str_utf8_offset_to_column(), BLI_str_utf8_offset_to_column_with_tabs(), and BLI_strncpy_wchar_from_utf8().

BLI_wstrlen_utf8()

const char const char size_t BLI_wstrlen_utf8

(

const wchar_t *

src

)

Returns

the wchar_t length in UTF8.

Definition at line 456 of file string_utf8.cc.

References BLI_str_utf8_from_unicode_len(), and len.

Variable Documentation

ATTR_WARN_UNUSED_RESULT

size_t ATTR_WARN_UNUSED_RESULT

Definition at line 184 of file BLI_string_utf8.h.