1 3448adb0 2022-11-02 op /* See LICENSE file for copyright and license details. */
  2 3448adb0 2022-11-02 op #include <stdbool.h>
  3 3448adb0 2022-11-02 op #include <stdint.h>
  4 3448adb0 2022-11-02 op 
  5 3448adb0 2022-11-02 op #include "../gen/sentence-test.h"
  6 3448adb0 2022-11-02 op #include "../grapheme.h"
  7 3448adb0 2022-11-02 op #include "util.h"
  8 3448adb0 2022-11-02 op 
  9 3448adb0 2022-11-02 op static const struct unit_test_next_break next_sentence_break[] = {
 10 3448adb0 2022-11-02 op 	{
 11 3448adb0 2022-11-02 op 		.description = "NULL input",
 12 3448adb0 2022-11-02 op 		.input = {
 13 3448adb0 2022-11-02 op 			.src    = NULL,
 14 3448adb0 2022-11-02 op 			.srclen = 0,
 15 3448adb0 2022-11-02 op 		},
 16 3448adb0 2022-11-02 op 		.output = { 0 },
 17 3448adb0 2022-11-02 op 	},
 18 3448adb0 2022-11-02 op 	{
 19 3448adb0 2022-11-02 op 		.description = "empty input",
 20 3448adb0 2022-11-02 op 		.input = {
 21 3448adb0 2022-11-02 op 			.src    = (uint_least32_t *)(uint_least32_t[]){ 0x0 },
 22 3448adb0 2022-11-02 op 			.srclen = 0,
 23 3448adb0 2022-11-02 op 		},
 24 3448adb0 2022-11-02 op 		.output = { 0 },
 25 3448adb0 2022-11-02 op 	},
 26 3448adb0 2022-11-02 op 	{
 27 3448adb0 2022-11-02 op 		.description = "empty input, null-terminated",
 28 3448adb0 2022-11-02 op 		.input = {
 29 3448adb0 2022-11-02 op 			.src    = (uint_least32_t *)(uint_least32_t[]){ 0x0 },
 30 3448adb0 2022-11-02 op 			.srclen = SIZE_MAX,
 31 3448adb0 2022-11-02 op 		},
 32 3448adb0 2022-11-02 op 		.output = { 0 },
 33 3448adb0 2022-11-02 op 	},
 34 3448adb0 2022-11-02 op 	{
 35 3448adb0 2022-11-02 op 		.description = "one sentence",
 36 3448adb0 2022-11-02 op 		.input = {
 37 3448adb0 2022-11-02 op 			.src    = (uint_least32_t *)(uint_least32_t[]){ 0x1F1E9, 0x1F1EA, 0x2E, 0x20, 0x2A },
 38 3448adb0 2022-11-02 op 			.srclen = 5,
 39 3448adb0 2022-11-02 op 		},
 40 3448adb0 2022-11-02 op 		.output = { 4 },
 41 3448adb0 2022-11-02 op 	},
 42 3448adb0 2022-11-02 op 	{
 43 3448adb0 2022-11-02 op 		.description = "one sentence, null-terminated",
 44 3448adb0 2022-11-02 op 		.input = {
 45 3448adb0 2022-11-02 op 			.src    = (uint_least32_t *)(uint_least32_t[]){ 0x1F1E9, 0x1F1EA, 0x2E, 0x20, 0x2A, 0x0 },
 46 3448adb0 2022-11-02 op 			.srclen = SIZE_MAX,
 47 3448adb0 2022-11-02 op 		},
 48 3448adb0 2022-11-02 op 		.output = { 4 },
 49 3448adb0 2022-11-02 op 	},
 50 3448adb0 2022-11-02 op };
 51 3448adb0 2022-11-02 op 
 52 3448adb0 2022-11-02 op static const struct unit_test_next_break_utf8 next_sentence_break_utf8[] = {
 53 3448adb0 2022-11-02 op 	{
 54 3448adb0 2022-11-02 op 		.description = "NULL input",
 55 3448adb0 2022-11-02 op 		.input = {
 56 3448adb0 2022-11-02 op 			.src    = NULL,
 57 3448adb0 2022-11-02 op 			.srclen = 0,
 58 3448adb0 2022-11-02 op 		},
 59 3448adb0 2022-11-02 op 		.output = { 0 },
 60 3448adb0 2022-11-02 op 	},
 61 3448adb0 2022-11-02 op 	{
 62 3448adb0 2022-11-02 op 		.description = "empty input",
 63 3448adb0 2022-11-02 op 		.input = { "", 0 },
 64 3448adb0 2022-11-02 op 		.output = { 0 },
 65 3448adb0 2022-11-02 op 	},
 66 3448adb0 2022-11-02 op 	{
 67 3448adb0 2022-11-02 op 		.description = "empty input, NUL-terminated",
 68 3448adb0 2022-11-02 op 		.input = { "", SIZE_MAX },
 69 3448adb0 2022-11-02 op 		.output = { 0 },
 70 3448adb0 2022-11-02 op 	},
 71 3448adb0 2022-11-02 op 	{
 72 3448adb0 2022-11-02 op 		.description = "one sentence",
 73 3448adb0 2022-11-02 op 		.input = { "\xF0\x9F\x87\xA9\xF0\x9F\x87\xAA is the flag of Germany.  It", 36 },
 74 3448adb0 2022-11-02 op 		.output = { 34 },
 75 3448adb0 2022-11-02 op 	},
 76 3448adb0 2022-11-02 op 	{
 77 3448adb0 2022-11-02 op 		.description = "one sentence, fragment",
 78 3448adb0 2022-11-02 op 		.input = { "\xF0\x9F\x87\xA9\xF0", 5 },
 79 3448adb0 2022-11-02 op 		.output = { 4 },
 80 3448adb0 2022-11-02 op 	},
 81 3448adb0 2022-11-02 op 	{
 82 3448adb0 2022-11-02 op 		.description = "one sentence, NUL-terminated",
 83 3448adb0 2022-11-02 op 		.input = { "\xF0\x9F\x87\xA9\xF0\x9F\x87\xAA is the flag of Germany.  It", SIZE_MAX },
 84 3448adb0 2022-11-02 op 		.output = { 34 },
 85 3448adb0 2022-11-02 op 	},
 86 3448adb0 2022-11-02 op 	{
 87 3448adb0 2022-11-02 op 		.description = "one sentence, fragment, NUL-terminated",
 88 3448adb0 2022-11-02 op 		.input = { "\xF0\x9F\x87\xA9\xF0\x9F", SIZE_MAX },
 89 3448adb0 2022-11-02 op 		.output = { 6 },
 90 3448adb0 2022-11-02 op 	},
 91 3448adb0 2022-11-02 op };
 92 3448adb0 2022-11-02 op 
 93 3448adb0 2022-11-02 op static int
 94 3448adb0 2022-11-02 op unit_test_callback_next_sentence_break(const void *t, size_t off,
 95 3448adb0 2022-11-02 op                                              const char *name,
 96 3448adb0 2022-11-02 op                                              const char *argv0)
 97 3448adb0 2022-11-02 op {
 98 3448adb0 2022-11-02 op 	return unit_test_callback_next_break(t, off,
 99 3448adb0 2022-11-02 op 	                                     grapheme_next_sentence_break,
100 3448adb0 2022-11-02 op 	                                     name, argv0);
101 3448adb0 2022-11-02 op }
102 3448adb0 2022-11-02 op 
103 3448adb0 2022-11-02 op static int
104 3448adb0 2022-11-02 op unit_test_callback_next_sentence_break_utf8(const void *t, size_t off,
105 3448adb0 2022-11-02 op                                              const char *name,
106 3448adb0 2022-11-02 op                                              const char *argv0)
107 3448adb0 2022-11-02 op {
108 3448adb0 2022-11-02 op 	return unit_test_callback_next_break_utf8(t, off,
109 3448adb0 2022-11-02 op 	                                          grapheme_next_sentence_break_utf8,
110 3448adb0 2022-11-02 op 	                                          name, argv0);
111 3448adb0 2022-11-02 op }
112 3448adb0 2022-11-02 op 
113 3448adb0 2022-11-02 op int
114 3448adb0 2022-11-02 op main(int argc, char *argv[])
115 3448adb0 2022-11-02 op {
116 3448adb0 2022-11-02 op 	(void)argc;
117 3448adb0 2022-11-02 op 
118 3448adb0 2022-11-02 op 	return run_break_tests(grapheme_next_sentence_break,
119 3448adb0 2022-11-02 op 	                       sentence_break_test,
120 3448adb0 2022-11-02 op 	                       LEN(sentence_break_test), argv[0]) +
121 3448adb0 2022-11-02 op 	       run_unit_tests(unit_test_callback_next_sentence_break,
122 3448adb0 2022-11-02 op 	                      next_sentence_break, LEN(next_sentence_break),
123 3448adb0 2022-11-02 op 	                      "grapheme_next_sentence_break", argv[0]) +
124 3448adb0 2022-11-02 op 	       run_unit_tests(unit_test_callback_next_sentence_break_utf8,
125 3448adb0 2022-11-02 op 	                      next_sentence_break_utf8, LEN(next_sentence_break_utf8),
126 3448adb0 2022-11-02 op 	                      "grapheme_next_character_break_utf8", argv[0]);
127 3448adb0 2022-11-02 op }