Linguistic tagger examples.

wordcounters.m

#import <Foundation/Foundation.h>

void taggerCounter(NSString* string, NSString* scheme)
{
  NSLog(@"=== %@", scheme);
  NSArray* schemes = [NSLinguisticTagger availableTagSchemesForLanguage:@"en"];
  NSLinguisticTagger* tagger = [[NSLinguisticTagger alloc] initWithTagSchemes:schemes
                                                                      options:0];
  [tagger setString:string];
  __block NSUInteger words = 0;
  __block NSUInteger sentences = 0;
  __block NSUInteger current_sentence = 0;
  __block NSMutableSet* tags = [[NSMutableSet alloc] init];
  [tagger enumerateTagsInRange:NSMakeRange(0, [string length])
                        scheme:scheme
                       options:0
                    usingBlock:^(NSString* tag, NSRange token, NSRange sentence, BOOL *stop) {
	if (tag) [tags addObject:tag];
	if ([tag isEqual:NSLinguisticTagWord]) ++words;
    if (!sentences || current_sentence != sentence.location) ++sentences;
    current_sentence = sentence.location;
    NSLog(@"token %d-%d sentence %d-%d %@", token.location, token.length, sentence.location, sentence.length, tag);
  }];
  NSLog(@"%d sentences, tag list: %@", sentences, tags);
}

int main(int argc, char *argv[]) {
  NSAutoreleasePool* pool = [[NSAutoreleasePool alloc] init];

  NSString* coffee = @"What I want - is a proper cup ’o coffee,"
                     @" Made in a proper copper coffee pot."
                     @" Ik kan van mijn punt."
                     @" Аз не мога да ми."
                     @" Mais je veux du café o une tasse à partir d'un pot de café bon.";

  // uncomment as desired...

  //taggerCounter(coffee, NSLinguisticTagSchemeTokenType);
  //taggerCounter(coffee, NSLinguisticTagSchemeLexicalClass);
  //taggerCounter(coffee, NSLinguisticTagSchemeNameType);
  //taggerCounter(coffee, NSLinguisticTagSchemeNameTypeOrLexicalClass);
  //taggerCounter(coffee, NSLinguisticTagSchemeLemma);
  taggerCounter(coffee, NSLinguisticTagSchemeLanguage);
  //taggerCounter(coffee, NSLinguisticTagSchemeScript);

  [pool release];
}