CODEWITHSUNDEEP
iosobjective-cnsstringuisearchdisplaycontrollernsrange
user1445205
user1445205

Reputation:

Search for any characters in an NSString

I have the following code to search an NSString:

for (NSDictionary *obj in data) {
        NSString *objQuestion = [obj objectForKey:@"Question"];
        NSRange dataRange = [objQuestion rangeOfString:searchText options:NSCaseInsensitiveSearch];
        if (dataRange.location != NSNotFound) {
            [filteredData addObject:obj];
        }
    }

This works fine, but there is a problem. If objQuestion is: "Green Yellow Red" and I search for "Yellow Green Red", the object will not show up as my search is not in the correct order.

How would I change my code so that no matter what order I search the words in, the object will show?

Upvotes: 0

Views: 753

Answers (3)

Monolo
Monolo

Reputation: 18253

You might want to consider that the search input string is not always as clean as you expect, and could contain punctuation, brackets, etc.

You'd also want to be lax with accents.

I like to use regular expressions for this sort of problem, and since you are looking for a solution that allows arbitrary ordering of the search terms, we'd need to re-work the search string. We can use regular expressions for that, too - so the pattern is constructed by a regex substitution, just out of principle. You may want to document it thoroughly.

So here is a code snippet that will do these things:

// Use the Posix locale as the lowest common denominator of locales to
// remove accents.
NSLocale *enLoc = [[NSLocale alloc] initWithLocaleIdentifier: @"en_US_POSIX"];

// Mixed bag of genres, but for testing purposes we get all the accents we need
NSString *orgString = @"Beyoncé Motörhead Händel"; 

// Clean string by removing accents and upper case letters in Posix encoding
NSString *string = [orgString stringByFoldingWithOptions: NSCaseInsensitiveSearch | NSDiacriticInsensitiveSearch
                                                  locale: enLoc ];

// What the user has typed in, with misplaced umlaut and all
NSString *orgSearchString = @"handel, mötorhead, beyonce";

// Clean the search string, too
NSString *searchString = [orgSearchString stringByFoldingWithOptions: NSCaseInsensitiveSearch | NSDiacriticInsensitiveSearch | NSWidthInsensitiveSearch
                                                              locale: enLoc ];

// Turn the search string into a regex pattern.
// Create a pattern that looks like: "(?=.*handel)(?=.*motorhead)(?=.*beyonce)"
// This pattern uses positive lookahead to create an AND logic that will
// accept arbitrary ordering of the words in the pattern.
// The \b expression matches a word boundary, so gets rid of punctuation, etc.

// We use a regex to create the regex pattern.
NSString *regexifyPattern = @"(?w)(\\W*)(\\b.+?\\b)(\\W*)";

NSString *pattern = [searchString stringByReplacingOccurrencesOfString: regexifyPattern
                                                            withString: @"(?=.*$2)"
                                                               options: NSRegularExpressionSearch
                                                                 range: NSMakeRange(0, searchString.length) ];

NSError *error;
NSRegularExpression *anyOrderRegEx = [NSRegularExpression regularExpressionWithPattern: pattern
                                                                               options: 0
                                                                                 error: &error];

if ( !anyOrderRegEx ) {

    // Regex patterns are tricky, programmatically constructed ones even more.
    // So we check if it went well and do something intelligent if it didn't

    // ...
}

// Match the constructed pattern with the string
NSUInteger numberOfMatches = [anyOrderRegEx numberOfMatchesInString: string
                                                    options: 0
                                                      range: NSMakeRange(0, string.length)];


BOOL found = (numberOfMatches > 0);

The use of the Posix locale identifier is discussed in this tech note from Apple.

In theory there is an edge case here if the user enters characters with a special meaning for regexes, but since the first regex removes non-word characters it should be solved that way. A bit of an un-planned positive side effect, so could be worth verifying.

Should you not be interested in a regex-based solution, the code folding may still be useful for "normal" NSString-based searching.

Upvotes: 0

Tala
Tala

Reputation: 8928

You should be breaking your search text into words and search each word.

NSArray *wordArray= [searchText componentsSeparatedByString: @" "];
for (NSDictionary *obj in data) {
    NSString *objQuestion = [obj objectForKey:@"Question"];        
    BOOL present = NO;
    for (NSString *s in wordArray) {
        if (s) {                
            NSRange dataRange = [objQuestion rangeOfString:s options:NSCaseInsensitiveSearch];
            if (dataRange.location != NSNotFound) {
               present = YES;
            }
        } 
    }
    if (present) {
        [filteredData addObject:obj];
    }
}

Upvotes: 2

MZimmerman6
MZimmerman6

Reputation: 8593

So you want to basically do a keyword search? I would recommend doing a regular expression search. where the words can be in any order.

Something like this.

(your|test|data)? *(your|test|data)? *(your|test|data)?

Which you can use in a NSRegularExpressoin

NSRegularExpression *regex = [NSRegularExpression regularExpressionWithPattern:@"(your|test|data)? *(your|test|data)? *(your|test|data)?" options:NSRegularExpressionCaseInsensitive error:&error];
int numMatches = [regex numberOfMatchesInString:searchString options:0 range:NSMakeRange(0, [searchString length])];];

This will match any ordering in an efficient manner.

Not sure if regex is okay for Obj C, because I do not have a mac in front of me right now, but it should be okay.

Upvotes: 0

Related Questions