/src/dovecot/src/lib/unicode-data.h

Source
#ifndef UNICODE_DATA_H
#define UNICODE_DATA_H

#include "unicode-data-tables.h"

static inline bool
unicode_general_category_is_group(enum unicode_general_category gencat)
{
  return ((gencat & 0x0f) == 0x00);
}

static inline const struct unicode_code_point_data *
unicode_code_point_get_data(uint32_t cp)
{
  unsigned int idx8 = cp >> 24;
  unsigned int blk16 = unicode_code_points_index8[idx8];
  unsigned int idx16 = (blk16 << 8) + ((cp >> 16) & 0xFF);
  unsigned int blk24 = unicode_code_points_index16[idx16];
  unsigned int idx24 = (blk24 << 8) + ((cp >> 8) & 0xFF);
  unsigned int blk32 = unicode_code_points_index24[idx24];
  unsigned int idx32 = (blk32 << 8) + (cp & 0xFF);
  unsigned int idxcp = unicode_code_points_index32[idx32];

  return &unicode_code_points[idxcp];
}

static inline bool
unicode_code_point_data_is_assigned(
  const struct unicode_code_point_data *cp_data)
{
  switch (cp_data->general_category) {
  case UNICODE_GENERAL_CATEGORY_INVALID:
  case UNICODE_GENERAL_CATEGORY_CN:
    return FALSE;
  default:
    break;
  }
  return TRUE;
}

static inline size_t
unicode_code_point_data_get_first_decomposition(
  const struct unicode_code_point_data *cp_data,
  uint8_t *type_r, const uint32_t **decomp_r)
{
  uint32_t offset;

  if (type_r != NULL)
    *type_r = cp_data->decomposition_type;
  offset = cp_data->decomposition_first_offset;
  *decomp_r = &unicode_decompositions[offset];
  return cp_data->decomposition_first_length;
}

static inline size_t
unicode_code_point_data_get_full_decomposition(
  const struct unicode_code_point_data *cp_data, bool canonical,
  const uint32_t **decomp_r)
{
  uint32_t offset;

  if (canonical) {
    offset = cp_data->decomposition_full_offset;
    *decomp_r = &unicode_decompositions[offset];
    return cp_data->decomposition_full_length;
  }
  offset = cp_data->decomposition_full_k_offset;
  *decomp_r = &unicode_decompositions[offset];
  return cp_data->decomposition_full_k_length;
}

static inline uint32_t
unicode_code_point_data_find_composition(
  const struct unicode_code_point_data *cp_data, uint32_t second)
{
  const uint32_t *compositions =
    &unicode_compositions[cp_data->composition_offset];
  size_t left_idx, right_idx;

  left_idx = 0; right_idx = cp_data->composition_count;
  while (left_idx < right_idx) {
    unsigned int idx = (left_idx + right_idx) / 2;

    if (second > compositions[idx])
      left_idx = idx + 1;
    else if (second < compositions[idx])
      right_idx = idx;
    else {
      return unicode_composition_primaries[
        cp_data->composition_offset + idx];
    }
  }

  return 0x0000;
}

static inline bool unicode_code_point_is_assigned(uint32_t cp)
{
  const struct unicode_code_point_data *cp_data =
    unicode_code_point_get_data(cp);

  return unicode_code_point_data_is_assigned(cp_data);
}

static inline size_t
unicode_code_point_get_full_decomposition(uint32_t cp, bool canonical,
            const uint32_t **decomp_r)
{
  const struct unicode_code_point_data *cp_data =
    unicode_code_point_get_data(cp);

  return unicode_code_point_data_get_full_decomposition(
    cp_data, canonical, decomp_r);
}

static inline size_t
unicode_code_point_data_get_uppercase_mapping(
  const struct unicode_code_point_data *cp_data,
  const uint32_t **map_r)
{
  uint32_t offset;

  offset = cp_data->uppercase_mapping_offset;
  *map_r = &unicode_case_mappings[offset];
  return cp_data->uppercase_mapping_length;
}

static inline size_t
unicode_code_point_data_get_lowercase_mapping(
  const struct unicode_code_point_data *cp_data,
  const uint32_t **map_r)
{
  uint32_t offset;

  offset = cp_data->lowercase_mapping_offset;
  *map_r = &unicode_case_mappings[offset];
  return cp_data->lowercase_mapping_length;
}

static inline size_t
unicode_code_point_data_get_casefold_mapping(
  const struct unicode_code_point_data *cp_data,
  const uint32_t **map_r)
{
  uint32_t offset;

  offset = cp_data->casefold_mapping_offset;
  *map_r = &unicode_case_mappings[offset];
  return cp_data->casefold_mapping_length;
}

uint8_t unicode_general_category_from_string(const char *str);

#endif

Line	Count	Source
1		#ifndef UNICODE_DATA_H
2		#define UNICODE_DATA_H
3
4		#include "unicode-data-tables.h"
5
6		static inline bool
7		unicode_general_category_is_group(enum unicode_general_category gencat)
8	0	{
9	0	return ((gencat & 0x0f) == 0x00);
10	0	} Unexecuted instantiation: unichar.c:unicode_general_category_is_group Unexecuted instantiation: unicode-break.c:unicode_general_category_is_group Unexecuted instantiation: unicode-transform.c:unicode_general_category_is_group
11
12		static inline const struct unicode_code_point_data *
13		unicode_code_point_get_data(uint32_t cp)
14	0	{
15	0	unsigned int idx8 = cp >> 24;
16	0	unsigned int blk16 = unicode_code_points_index8[idx8];
17	0	unsigned int idx16 = (blk16 << 8) + ((cp >> 16) & 0xFF);
18	0	unsigned int blk24 = unicode_code_points_index16[idx16];
19	0	unsigned int idx24 = (blk24 << 8) + ((cp >> 8) & 0xFF);
20	0	unsigned int blk32 = unicode_code_points_index24[idx24];
21	0	unsigned int idx32 = (blk32 << 8) + (cp & 0xFF);
22	0	unsigned int idxcp = unicode_code_points_index32[idx32];
23
24	0	return &unicode_code_points[idxcp];
25	0	} Unexecuted instantiation: unichar.c:unicode_code_point_get_data Unexecuted instantiation: unicode-break.c:unicode_code_point_get_data Unexecuted instantiation: unicode-transform.c:unicode_code_point_get_data
26
27		static inline bool
28		unicode_code_point_data_is_assigned(
29		const struct unicode_code_point_data *cp_data)
30	0	{
31	0	switch (cp_data->general_category) {
32	0	case UNICODE_GENERAL_CATEGORY_INVALID:
33	0	case UNICODE_GENERAL_CATEGORY_CN:
34	0	return FALSE;
35	0	default:
36	0	break;
37	0	}
38	0	return TRUE;
39	0	} Unexecuted instantiation: unichar.c:unicode_code_point_data_is_assigned Unexecuted instantiation: unicode-break.c:unicode_code_point_data_is_assigned Unexecuted instantiation: unicode-transform.c:unicode_code_point_data_is_assigned
40
41		static inline size_t
42		unicode_code_point_data_get_first_decomposition(
43		const struct unicode_code_point_data *cp_data,
44		uint8_t type_r, const uint32_t *decomp_r)
45	0	{
46	0	uint32_t offset;
47	0
48	0	if (type_r != NULL)
49	0	*type_r = cp_data->decomposition_type;
50	0	offset = cp_data->decomposition_first_offset;
51	0	*decomp_r = &unicode_decompositions[offset];
52	0	return cp_data->decomposition_first_length;
53	0	} Unexecuted instantiation: unichar.c:unicode_code_point_data_get_first_decomposition Unexecuted instantiation: unicode-break.c:unicode_code_point_data_get_first_decomposition Unexecuted instantiation: unicode-transform.c:unicode_code_point_data_get_first_decomposition
54
55		static inline size_t
56		unicode_code_point_data_get_full_decomposition(
57		const struct unicode_code_point_data *cp_data, bool canonical,
58		const uint32_t **decomp_r)
59	0	{
60	0	uint32_t offset;
61
62	0	if (canonical) {
63	0	offset = cp_data->decomposition_full_offset;
64	0	*decomp_r = &unicode_decompositions[offset];
65	0	return cp_data->decomposition_full_length;
66	0	}
67	0	offset = cp_data->decomposition_full_k_offset;
68	0	*decomp_r = &unicode_decompositions[offset];
69	0	return cp_data->decomposition_full_k_length;
70	0	} Unexecuted instantiation: unichar.c:unicode_code_point_data_get_full_decomposition Unexecuted instantiation: unicode-break.c:unicode_code_point_data_get_full_decomposition Unexecuted instantiation: unicode-transform.c:unicode_code_point_data_get_full_decomposition
71
72		static inline uint32_t
73		unicode_code_point_data_find_composition(
74		const struct unicode_code_point_data *cp_data, uint32_t second)
75	0	{
76	0	const uint32_t *compositions =
77	0	&unicode_compositions[cp_data->composition_offset];
78	0	size_t left_idx, right_idx;
79
80	0	left_idx = 0; right_idx = cp_data->composition_count;
81	0	while (left_idx < right_idx) {
82	0	unsigned int idx = (left_idx + right_idx) / 2;
83
84	0	if (second > compositions[idx])
85	0	left_idx = idx + 1;
86	0	else if (second < compositions[idx])
87	0	right_idx = idx;
88	0	else {
89	0	return unicode_composition_primaries[
90	0	cp_data->composition_offset + idx];
91	0	}
92	0	}
93
94	0	return 0x0000;
95	0	} Unexecuted instantiation: unichar.c:unicode_code_point_data_find_composition Unexecuted instantiation: unicode-break.c:unicode_code_point_data_find_composition Unexecuted instantiation: unicode-transform.c:unicode_code_point_data_find_composition
96
97		static inline bool unicode_code_point_is_assigned(uint32_t cp)
98	0	{
99	0	const struct unicode_code_point_data *cp_data =
100	0	unicode_code_point_get_data(cp);
101	0
102	0	return unicode_code_point_data_is_assigned(cp_data);
103	0	} Unexecuted instantiation: unichar.c:unicode_code_point_is_assigned Unexecuted instantiation: unicode-break.c:unicode_code_point_is_assigned Unexecuted instantiation: unicode-transform.c:unicode_code_point_is_assigned
104
105		static inline size_t
106		unicode_code_point_get_full_decomposition(uint32_t cp, bool canonical,
107		const uint32_t **decomp_r)
108	0	{
109	0	const struct unicode_code_point_data *cp_data =
110	0	unicode_code_point_get_data(cp);
111
112	0	return unicode_code_point_data_get_full_decomposition(
113	0	cp_data, canonical, decomp_r);
114	0	} Unexecuted instantiation: unichar.c:unicode_code_point_get_full_decomposition Unexecuted instantiation: unicode-break.c:unicode_code_point_get_full_decomposition Unexecuted instantiation: unicode-transform.c:unicode_code_point_get_full_decomposition
115
116		static inline size_t
117		unicode_code_point_data_get_uppercase_mapping(
118		const struct unicode_code_point_data *cp_data,
119		const uint32_t **map_r)
120	0	{
121	0	uint32_t offset;
122
123	0	offset = cp_data->uppercase_mapping_offset;
124	0	*map_r = &unicode_case_mappings[offset];
125	0	return cp_data->uppercase_mapping_length;
126	0	} Unexecuted instantiation: unichar.c:unicode_code_point_data_get_uppercase_mapping Unexecuted instantiation: unicode-break.c:unicode_code_point_data_get_uppercase_mapping Unexecuted instantiation: unicode-transform.c:unicode_code_point_data_get_uppercase_mapping
127
128		static inline size_t
129		unicode_code_point_data_get_lowercase_mapping(
130		const struct unicode_code_point_data *cp_data,
131		const uint32_t **map_r)
132	0	{
133	0	uint32_t offset;
134
135	0	offset = cp_data->lowercase_mapping_offset;
136	0	*map_r = &unicode_case_mappings[offset];
137	0	return cp_data->lowercase_mapping_length;
138	0	} Unexecuted instantiation: unichar.c:unicode_code_point_data_get_lowercase_mapping Unexecuted instantiation: unicode-break.c:unicode_code_point_data_get_lowercase_mapping Unexecuted instantiation: unicode-transform.c:unicode_code_point_data_get_lowercase_mapping
139
140		static inline size_t
141		unicode_code_point_data_get_casefold_mapping(
142		const struct unicode_code_point_data *cp_data,
143		const uint32_t **map_r)
144	0	{
145	0	uint32_t offset;
146
147	0	offset = cp_data->casefold_mapping_offset;
148	0	*map_r = &unicode_case_mappings[offset];
149	0	return cp_data->casefold_mapping_length;
150	0	} Unexecuted instantiation: unichar.c:unicode_code_point_data_get_casefold_mapping Unexecuted instantiation: unicode-break.c:unicode_code_point_data_get_casefold_mapping Unexecuted instantiation: unicode-transform.c:unicode_code_point_data_get_casefold_mapping
151
152		uint8_t unicode_general_category_from_string(const char *str);
153
154		#endif

Coverage Report

Created: 2026-05-16 06:51