1 /*************************************************************************
3 * Copyright (C) 2016 and later: Unicode, Inc. and others.
4 * License & terms of use: http://www.unicode.org/copyright.html#License
6 *************************************************************************
7 *************************************************************************
9 * Copyright (C) 2002-2006 IBM, Inc. All Rights Reserved.
11 *************************************************************************/
14 * This program demos string collation
17 const char gHelpString[] =
18 "usage: coll [options*] -source source_string -target target_string\n"
19 "-help Display this message.\n"
20 "-locale name ICU locale to use. Default is en_US\n"
21 "-rules rule Collation rules file (overrides locale)\n"
22 "-french French accent ordering\n"
23 "-norm Normalizing mode on\n"
24 "-shifted Shifted mode\n"
25 "-lower Lower case first\n"
26 "-upper Upper case first\n"
27 "-case Enable separate case level\n"
28 "-level n Sort level, 1 to 5, for Primary, Secndary, Tertiary, Quaternary, Identical\n"
29 "-source string Source string for comparison\n"
30 "-target string Target string for comparison\n"
31 "Example coll -rules \\u0026b\\u003ca -source a -target b\n"
32 "The format \\uXXXX is supported for the rules and comparison strings\n"
39 #include <unicode/utypes.h>
40 #include <unicode/ucol.h>
41 #include <unicode/ustring.h>
44 * Command line option variables
45 * These global variables are set according to the options specified
46 * on the command line by the user.
48 char * opt_locale = "en_US";
50 UBool opt_help = FALSE;
51 UBool opt_norm = FALSE;
52 UBool opt_french = FALSE;
53 UBool opt_shifted = FALSE;
54 UBool opt_lower = FALSE;
55 UBool opt_upper = FALSE;
56 UBool opt_case = FALSE;
58 char * opt_source = "abc";
59 char * opt_target = "abd";
60 UCollator * collator = 0;
63 * Definitions for the command line options
67 enum {FLAG, NUM, STRING} type;
72 {"-locale", OptSpec::STRING, &opt_locale},
73 {"-rules", OptSpec::STRING, &opt_rules},
74 {"-source", OptSpec::STRING, &opt_source},
75 {"-target", OptSpec::STRING, &opt_target},
76 {"-norm", OptSpec::FLAG, &opt_norm},
77 {"-french", OptSpec::FLAG, &opt_french},
78 {"-shifted", OptSpec::FLAG, &opt_shifted},
79 {"-lower", OptSpec::FLAG, &opt_lower},
80 {"-upper", OptSpec::FLAG, &opt_upper},
81 {"-case", OptSpec::FLAG, &opt_case},
82 {"-level", OptSpec::NUM, &opt_level},
83 {"-help", OptSpec::FLAG, &opt_help},
84 {"-?", OptSpec::FLAG, &opt_help},
89 * processOptions() Function to read the command line options.
91 UBool processOptions(int argc, const char **argv, OptSpec opts[])
93 for (int argNum = 1; argNum < argc; argNum ++) {
94 const char *pArgName = argv[argNum];
96 for (pOpt = opts; pOpt->name != 0; pOpt ++) {
97 if (strcmp(pOpt->name, pArgName) == 0) {
100 *(UBool *)(pOpt->pVar) = TRUE;
102 case OptSpec::STRING:
104 if (argNum >= argc) {
105 fprintf(stderr, "value expected for \"%s\" option.\n",
109 *(const char **)(pOpt->pVar) = argv[argNum];
113 if (argNum >= argc) {
114 fprintf(stderr, "value expected for \"%s\" option.\n",
119 int i = strtol(argv[argNum], &endp, 0);
120 if (endp == argv[argNum]) {
122 "integer value expected for \"%s\" option.\n",
126 *(int *)(pOpt->pVar) = i;
133 fprintf(stderr, "Unrecognized option \"%s\"\n", pArgName);
141 * ICU string comparison
147 u_unescape(opt_source, source, 100);
148 u_unescape(opt_target, target, 100);
149 UCollationResult result = ucol_strcoll(collator, source, -1, target, -1);
150 if (result == UCOL_LESS) {
153 else if (result == UCOL_GREATER) {
162 UBool processCollator()
164 // Set up an ICU collator
165 UErrorCode status = U_ZERO_ERROR;
168 if (opt_rules != 0) {
169 u_unescape(opt_rules, rules, 100);
170 collator = ucol_openRules(rules, -1, UCOL_OFF, UCOL_TERTIARY,
174 collator = ucol_open(opt_locale, &status);
176 if (U_FAILURE(status)) {
177 fprintf(stderr, "Collator creation failed.: %d\n", status);
180 if (status == U_USING_DEFAULT_WARNING) {
181 fprintf(stderr, "Warning, U_USING_DEFAULT_WARNING for %s\n",
184 if (status == U_USING_FALLBACK_WARNING) {
185 fprintf(stderr, "Warning, U_USING_FALLBACK_ERROR for %s\n",
189 ucol_setAttribute(collator, UCOL_NORMALIZATION_MODE, UCOL_ON, &status);
192 ucol_setAttribute(collator, UCOL_FRENCH_COLLATION, UCOL_ON, &status);
195 ucol_setAttribute(collator, UCOL_CASE_FIRST, UCOL_LOWER_FIRST,
199 ucol_setAttribute(collator, UCOL_CASE_FIRST, UCOL_UPPER_FIRST,
203 ucol_setAttribute(collator, UCOL_CASE_LEVEL, UCOL_ON, &status);
206 ucol_setAttribute(collator, UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED,
209 if (opt_level != 0) {
212 ucol_setAttribute(collator, UCOL_STRENGTH, UCOL_PRIMARY, &status);
215 ucol_setAttribute(collator, UCOL_STRENGTH, UCOL_SECONDARY,
219 ucol_setAttribute(collator, UCOL_STRENGTH, UCOL_TERTIARY, &status);
222 ucol_setAttribute(collator, UCOL_STRENGTH, UCOL_QUATERNARY,
226 ucol_setAttribute(collator, UCOL_STRENGTH, UCOL_IDENTICAL,
230 fprintf(stderr, "-level param must be between 1 and 5\n");
234 if (U_FAILURE(status)) {
235 fprintf(stderr, "Collator attribute setting failed.: %d\n", status);
242 * Main -- process command line, read in and pre-process the test file,
243 * call other functions to do the actual tests.
245 int main(int argc, const char** argv)
247 if (processOptions(argc, argv, opts) != TRUE || opt_help) {
252 if (processCollator() != TRUE) {
253 fprintf(stderr, "Error creating collator for comparison\n");
257 fprintf(stdout, "Comparing source=%s and target=%s\n", opt_source,
259 int result = strcmp();
261 fprintf(stdout, "source is equals to target\n");
263 else if (result < 0) {
264 fprintf(stdout, "source is less than target\n");
267 fprintf(stdout, "source is greater than target\n");
270 ucol_close(collator);