Convert spelled out number to number

Question

Is there a way, in Objective-C/Cocoa, to convert spelled out words to an NSNumber or equivalent in multiple languages?

For example:

convert three to 3 or convert ocho to 8 (Spanish).

Also slightly different, but 3 1/2 to 3.5

I could write my own code to do this, but I was hoping there was a built-in way to do this. Getting the translations of every number in several languages is something I'd like to avoid.

Consider searching the documentation for "natural language" and "NSNumberFormatter" — Joshua Nozzi, Mar 28 '13 at 19:57
@JoshuaNozzi From what I've found, `NSNumberFormatter` can only do the reverse of this. — edc1591, Mar 28 '13 at 20:03
You can create an NSDictionnary that would have the value 1 for the key "one", 2 for "two" and so on. But you have to write each number one after another - not quite practical. — Moray, Mar 28 '13 at 20:04
Huh? I may be misunderstanding you but I thought you wanted "three" to be converted to an NSNumber instance with the value of "3". But I was wrong anyway: this only works for **NSDateFormatter**. Hence, "tomorrow" would return an NSDate object with "tomorrow's" date relative to [NSDate date]. — Joshua Nozzi, Mar 28 '13 at 20:07
The above said, what you're asking for is REALLY complicated for even one language, much less multiple. Consider lots of nasty-looking RegEx. — Joshua Nozzi, Mar 28 '13 at 20:09
@JoshuaNozzi No, you were right. That is what I want to do. So I guess localizing every single word in the range I need would be the only way? — edc1591, Mar 28 '13 at 20:12
You will likely find [ParseKit](http://parsekit.com/) helpful in implementing this. — jscs, Mar 28 '13 at 20:19
It'd be a more direct way given the current tool set, but I don't know about "the only". — Joshua Nozzi, Mar 28 '13 at 22:34

Rob · Accepted Answer · 2013-03-28T21:28:21.357

13

NSNumberFormatter can convert from text to numbers:

NSNumberFormatter *formatter = [[NSNumberFormatter alloc] init];
formatter.numberStyle = NSNumberFormatterSpellOutStyle;

NSLog(@"%@", [formatter numberFromString:@"thirty-four"]);
NSLog(@"%@", [formatter numberFromString:@"three point five"]);

formatter.locale = [[NSLocale alloc]initWithLocaleIdentifier:[NSLocale localeIdentifierFromComponents:@{NSLocaleLanguageCode: @"es"}]];

NSLog(@"%@", [formatter numberFromString:@"ocho"]);

There are serious limitations as to what it can handle (it doesn't auto detect languages, if you deviate from the expected format (e.g. "thirty four" instead of "thirty-four"), fractions, etc.), but for the narrow domain, it appears to do the job.

edited Mar 28 '13 at 21:28

answered Mar 28 '13 at 21:08

Rob

415,655
72
787
1,044

1

Awesome! I had no idea `NSNumberFormatter` could do this. Thanks! – edc1591 Mar 29 '13 at 00:49
this solution works like magic. However, could you explain why i get (null) for [formatter numberFromString:@"three point eleven"] even though [formatter numberFromString:@"three point one"] works out fine? – static0886 Feb 02 '16 at 11:00
btw i get 3.11 for [formatter numberFromString:@"three point one one"] but I get 101.3 for [formatter numberFromString:@"one one point three"] – static0886 Feb 02 '16 at 11:19
Note the string needs to be lowercase – Adrian Harris Crowne Jan 30 '18 at 19:40

score 2 · Answer 2 · answered Mar 28 '13 at 20:54

NSLinguisticTagger will flag numbers for you in multiple languages.

NSArray * texts = @[@"It's 3 degrees outside", @"Ocho tacos", @"What is 3 1/2?", @"ocho"];
for (NSString * text in texts)
{
    NSLinguisticTaggerOptions options = NSLinguisticTaggerOmitWhitespace | NSLinguisticTaggerJoinNames;
    NSArray * tagSchemes = [NSLinguisticTagger availableTagSchemesForLanguage:@"en"];
    tagSchemes = [tagSchemes arrayByAddingObjectsFromArray:[NSLinguisticTagger availableTagSchemesForLanguage:@"es"]];

    NSLinguisticTagger * tagger = [[NSLinguisticTagger alloc] initWithTagSchemes:tagSchemes
                                                                                 options:options];
    [tagger setString:text];

    [tagger enumerateTagsInRange:NSMakeRange(0, [text length])
                          scheme:NSLinguisticTagSchemeNameTypeOrLexicalClass
                         options:options
                      usingBlock:^(NSString *tag, NSRange tokenRange, NSRange sentenceRange, BOOL *stop)
        {
            NSString *token = [text substringWithRange:tokenRange];
            NSLog(@"%@: %@", token, tag);
        }];
}

This does leave you with the task of identifying how and when to do things like fraction resolution.

Convert spelled out number to number

2 Answers2

Linked

Related