source/samples/coll/coll.cpp

   1 /*************************************************************************
   2  *
   3  *   Copyright (C) 2016 and later: Unicode, Inc. and others.
   4  *   License & terms of use: http://www.unicode.org/copyright.html#License
   5  *
   6  *************************************************************************
   7  *************************************************************************
   8  * COPYRIGHT:
   9  * Copyright (C) 2002-2006 IBM, Inc.   All Rights Reserved.
  10  *
  11  *************************************************************************/
  12
  13 /**
  14  * This program demos string collation
  15  */
  16
  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"
  33         ;
  34
  35 #include <stdio.h>
  36 #include <string.h>
  37 #include <stdlib.h>
  38
  39 #include <unicode/utypes.h>
  40 #include <unicode/ucol.h>
  41 #include <unicode/ustring.h>
  42
  43 /**
  44  * Command line option variables
  45  *    These global variables are set according to the options specified
  46  *    on the command line by the user.
  47  */
  48 char * opt_locale     = "en_US";
  49 char * opt_rules      = 0;
  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;
  57 int    opt_level      = 0;
  58 char * opt_source     = "abc";
  59 char * opt_target     = "abd";
  60 UCollator * collator  = 0;
  61
  62 /**
  63  * Definitions for the command line options
  64  */
  65 struct OptSpec {
  66     const char *name;
  67     enum {FLAG, NUM, STRING} type;
  68     void *pVar;
  69 };
  70
  71 OptSpec opts[] = {
  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},
  85     {0, OptSpec::FLAG, 0}
  86 };
  87
  88 /**
  89  * processOptions()  Function to read the command line options.
  90  */
  91 UBool processOptions(int argc, const char **argv, OptSpec opts[])
  92 {
  93     for (int argNum = 1; argNum < argc; argNum ++) {
  94         const char *pArgName = argv[argNum];
  95         OptSpec *pOpt;
  96         for (pOpt = opts;  pOpt->name != 0; pOpt ++) {
  97             if (strcmp(pOpt->name, pArgName) == 0) {
  98                 switch (pOpt->type) {
  99                 case OptSpec::FLAG:
 100                     *(UBool *)(pOpt->pVar) = TRUE;
 101                     break;
 102                 case OptSpec::STRING:
 103                     argNum ++;
 104                     if (argNum >= argc) {
 105                         fprintf(stderr, "value expected for \"%s\" option.\n",
 106                                                             pOpt->name);
 107                         return FALSE;
 108                     }
 109                     *(const char **)(pOpt->pVar) = argv[argNum];
 110                     break;
 111                 case OptSpec::NUM:
 112                     argNum ++;
 113                     if (argNum >= argc) {
 114                         fprintf(stderr, "value expected for \"%s\" option.\n",
 115                                                             pOpt->name);
 116                         return FALSE;
 117                     }
 118                     char *endp;
 119                     int i = strtol(argv[argNum], &endp, 0);
 120                     if (endp == argv[argNum]) {
 121                         fprintf(stderr,
 122                                                             "integer value expected for \"%s\" option.\n",
 123                                                                 pOpt->name);
 124                         return FALSE;
 125                     }
 126                     *(int *)(pOpt->pVar) = i;
 127                 }
 128                 break;
 129             }
 130         }
 131         if (pOpt->name == 0)
 132         {
 133             fprintf(stderr, "Unrecognized option \"%s\"\n", pArgName);
 134             return FALSE;
 135         }
 136     }
 137         return TRUE;
 138 }
 139
 140 /**
 141  * ICU string comparison
 142  */
 143 int strcmp()
 144 {
 145         UChar source[100];
 146         UChar target[100];
 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) {
 151                 return -1;
 152         }
 153     else if (result == UCOL_GREATER) {
 154                 return 1;
 155         }
 156         return 0;
 157 }
 158
 159 /**
 160  * Creates a collator
 161  */
 162 UBool processCollator()
 163 {
 164         // Set up an ICU collator
 165     UErrorCode status = U_ZERO_ERROR;
 166         UChar rules[100];
 167
 168     if (opt_rules != 0) {
 169                 u_unescape(opt_rules, rules, 100);
 170         collator = ucol_openRules(rules, -1, UCOL_OFF, UCOL_TERTIARY,
 171                                           NULL, &status);
 172     }
 173     else {
 174         collator = ucol_open(opt_locale, &status);
 175     }
 176         if (U_FAILURE(status)) {
 177         fprintf(stderr, "Collator creation failed.: %d\n", status);
 178         return FALSE;
 179     }
 180     if (status == U_USING_DEFAULT_WARNING) {
 181         fprintf(stderr, "Warning, U_USING_DEFAULT_WARNING for %s\n",
 182                             opt_locale);
 183     }
 184     if (status == U_USING_FALLBACK_WARNING) {
 185         fprintf(stderr, "Warning, U_USING_FALLBACK_ERROR for %s\n",
 186                             opt_locale);
 187     }
 188     if (opt_norm) {
 189         ucol_setAttribute(collator, UCOL_NORMALIZATION_MODE, UCOL_ON, &status);
 190     }
 191     if (opt_french) {
 192         ucol_setAttribute(collator, UCOL_FRENCH_COLLATION, UCOL_ON, &status);
 193     }
 194     if (opt_lower) {
 195         ucol_setAttribute(collator, UCOL_CASE_FIRST, UCOL_LOWER_FIRST,
 196                                       &status);
 197     }
 198     if (opt_upper) {
 199         ucol_setAttribute(collator, UCOL_CASE_FIRST, UCOL_UPPER_FIRST,
 200                                       &status);
 201     }
 202     if (opt_case) {
 203         ucol_setAttribute(collator, UCOL_CASE_LEVEL, UCOL_ON, &status);
 204     }
 205     if (opt_shifted) {
 206         ucol_setAttribute(collator, UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED,
 207                                       &status);
 208     }
 209     if (opt_level != 0) {
 210         switch (opt_level) {
 211         case 1:
 212             ucol_setAttribute(collator, UCOL_STRENGTH, UCOL_PRIMARY, &status);
 213             break;
 214         case 2:
 215             ucol_setAttribute(collator, UCOL_STRENGTH, UCOL_SECONDARY,
 216                                               &status);
 217             break;
 218         case 3:
 219             ucol_setAttribute(collator, UCOL_STRENGTH, UCOL_TERTIARY, &status);
 220             break;
 221         case 4:
 222             ucol_setAttribute(collator, UCOL_STRENGTH, UCOL_QUATERNARY,
 223                                               &status);
 224             break;
 225         case 5:
 226             ucol_setAttribute(collator, UCOL_STRENGTH, UCOL_IDENTICAL,
 227                                               &status);
 228             break;
 229         default:
 230             fprintf(stderr, "-level param must be between 1 and 5\n");
 231             return FALSE;
 232         }
 233     }
 234     if (U_FAILURE(status)) {
 235         fprintf(stderr, "Collator attribute setting failed.: %d\n", status);
 236         return FALSE;
 237     }
 238         return TRUE;
 239 }
 240
 241 /**
 242  * Main   --  process command line, read in and pre-process the test file,
 243  *            call other functions to do the actual tests.
 244  */
 245 int main(int argc, const char** argv)
 246 {
 247     if (processOptions(argc, argv, opts) != TRUE || opt_help) {
 248         printf(gHelpString);
 249         return -1;
 250     }
 251
 252     if (processCollator() != TRUE) {
 253                 fprintf(stderr, "Error creating collator for comparison\n");
 254                 return -1;
 255         }
 256
 257         fprintf(stdout, "Comparing source=%s and target=%s\n", opt_source,
 258                         opt_target);
 259         int result = strcmp();
 260         if (result == 0) {
 261                 fprintf(stdout, "source is equals to target\n");
 262         }
 263         else if (result < 0) {
 264                 fprintf(stdout, "source is less than target\n");
 265         }
 266         else {
 267                 fprintf(stdout, "source is greater than target\n");
 268         }
 269
 270         ucol_close(collator);
 271         return 0;
 272 }