1 // SPDX-License-Identifier: GPL-2.0
3  * ucs.c - Universal Character Set processing
23 static int interval16_cmp(const void *key, const void *element)  in interval16_cmp()  argument
25 	u16 cp = *(u16 *)key;  in interval16_cmp()
28 	if (cp < entry->first)  in interval16_cmp()
29 		return -1;  in interval16_cmp()
30 	if (cp > entry->last)  in interval16_cmp()
35 static int interval32_cmp(const void *key, const void *element)  in interval32_cmp()  argument
37 	u32 cp = *(u32 *)key;  in interval32_cmp()
40 	if (cp < entry->first)  in interval32_cmp()
41 		return -1;  in interval32_cmp()
42 	if (cp > entry->last)  in interval32_cmp()
49 	if (cp < ranges[0].first || cp > ranges[size - 1].last)  in cp_in_range16()
58 	if (cp < ranges[0].first || cp > ranges[size - 1].last)  in cp_in_range32()
68  * ucs_is_zero_width() - Determine if a Unicode code point is zero-width.
69  * @cp: Unicode code point (UCS-4)
71  * Return: true if the character is zero-width, false otherwise
84  * ucs_is_double_width() - Determine if a Unicode code point is double-width.
85  * @cp: Unicode code point (UCS-4)
87  * Return: true if the character is double-width, false otherwise
116 static int recomposition_cmp(const void *key, const void *element)  in recomposition_cmp()  argument
118 	const struct compare_key *search_key = key;  in recomposition_cmp()
122 	if (search_key->base < entry->base)  in recomposition_cmp()
123 		return -1;  in recomposition_cmp()
124 	if (search_key->base > entry->base)  in recomposition_cmp()
128 	if (search_key->mark < entry->mark)  in recomposition_cmp()
129 		return -1;  in recomposition_cmp()
130 	if (search_key->mark > entry->mark)  in recomposition_cmp()
138  * ucs_recompose() - Attempt to recompose two Unicode characters into a single character.
139  * @base: Base Unicode code point (UCS-4)
140  * @mark: Combining mark Unicode code point (UCS-4)
142  * Return: Recomposed Unicode code point, or 0 if no recomposition is possible
151 	struct compare_key key = { base, mark };  in ucs_recompose()  local
153 		__inline_bsearch(&key, ucs_recomposition_table,  in ucs_recompose()
158 	return result ? result->recomposed : 0;  in ucs_recompose()
162  * The fallback table structures implement a 2-level lookup.
166 	u8 page;	/* Page index (high byte of code points) */
172 	u8 offset;	/* Offset within page (0-255) */
178 static int ucs_page_desc_cmp(const void *key, const void *element)  in ucs_page_desc_cmp()  argument
180 	u8 page = *(u8 *)key;  in ucs_page_desc_cmp()
183 	if (page < entry->page)  in ucs_page_desc_cmp()
184 		return -1;  in ucs_page_desc_cmp()
185 	if (page > entry->page)  in ucs_page_desc_cmp()
190 static int ucs_page_entry_cmp(const void *key, const void *element)  in ucs_page_entry_cmp()  argument
192 	u8 offset = *(u8 *)key;  in ucs_page_entry_cmp()
195 	if (offset < entry->offset)  in ucs_page_entry_cmp()
196 		return -1;  in ucs_page_entry_cmp()
197 	if (entry->fallback == UCS_PAGE_ENTRY_RANGE_MARKER) {  in ucs_page_entry_cmp()
201 		if (offset > entry->offset)  in ucs_page_entry_cmp()
208  * ucs_get_fallback() - Get a substitution for the provided Unicode character
209  * @cp: Unicode code point (UCS-4)
216  * Return: Fallback Unicode code point, or 0 if none is available
228 	 * Full-width to ASCII mapping (covering all printable ASCII 33-126)  in ucs_get_fallback()
229 	 * 0xFF01 (！) to 0xFF5E (～) -> ASCII 33 (!) to 126 (~)  in ucs_get_fallback()
233 		return cp - 0xFF01 + 33;  in ucs_get_fallback()
242 	entry = __inline_bsearch(&offset, ucs_fallback_entries + page->start,  in ucs_get_fallback()
243 				 page->count, sizeof(*ucs_fallback_entries),  in ucs_get_fallback()
248 	if (entry->fallback == UCS_PAGE_ENTRY_RANGE_MARKER)  in ucs_get_fallback()
250 	return entry->fallback;  in ucs_get_fallback()