Bodged together test of QSSense.m

test-qssense.m

#import <Cocoa/Cocoa.h>

//
// QSense.m
// QSqSense
//
// Created by Alcor on 11/22/04.
// Copyright 2004 Blacktree. All rights reserved.
//

#import "QSSense.h"

#define MIN_ABBR_OPTIMIZE 0
#define IGNORED_SCORE 0.9
#define SKIPPED_SCORE 0.15



CGFloat QSScoreForAbbreviationWithRanges(CFStringRef str, CFStringRef abbr, id mask, CFRange strRange, CFRange abbrRange);

CGFloat QSScoreForAbbreviation(CFStringRef str, CFStringRef abbr, id mask) {
    return QSScoreForAbbreviationOrTransliteration(str, abbr, mask);
}

CGFloat QSScoreForAbbreviationOrTransliteration(CFStringRef str, CFStringRef abbr, id mask) {
    CGFloat score = QSScoreForAbbreviationWithRanges(str, abbr, mask, CFRangeMake(0, CFStringGetLength(str)), CFRangeMake(0, CFStringGetLength(abbr)));
    if (score == 0) {
        CFMutableStringRef mutableString = CFStringCreateMutableCopy(kCFAllocatorDefault, 0, str);
        CFStringTransform(mutableString, nil, kCFStringTransformToLatin, false);
        
        if (CFStringCompare(str, mutableString, 0) != kCFCompareEqualTo) {
            // only do this if the two strings are not equal (otherwise it's a wasted compute)
            score = QSScoreForAbbreviationWithRanges(mutableString, abbr, nil, CFRangeMake(0, CFStringGetLength(mutableString)), CFRangeMake(0, CFStringGetLength(abbr)));
        }
        if (mutableString) {
            CFRelease(mutableString);
        }
        return score;
    }
    return score;
}

CGFloat QSScoreForAbbreviationWithRanges(CFStringRef str, CFStringRef abbr, id mask, CFRange strRange, CFRange abbrRange) {
    printf("call: %s\n", CFStringGetCStringPtr(abbr, kCFStringEncodingMacRoman));
    
    if (!abbrRange.length)
        return IGNORED_SCORE; //deduct some points for all remaining letters
    
    if (abbrRange.length > strRange.length)
        return 0.0;
    
    static CFCharacterSetRef wordSeparator = NULL;
    static CFCharacterSetRef uppercase = NULL;
    static dispatch_once_t onceToken;
    dispatch_once(&onceToken, ^{
        wordSeparator = CFCharacterSetCreateMutableCopy(NULL, CFCharacterSetGetPredefined(kCFCharacterSetWhitespace));
        CFCharacterSetAddCharactersInString((CFMutableCharacterSetRef)wordSeparator, (CFStringRef)@".");
        
        uppercase = CFCharacterSetGetPredefined(kCFCharacterSetUppercaseLetter);
    });
    
    // Create an inline buffer version of str.  Will be used in loop below
    // for faster lookups.
    CFStringInlineBuffer inlineBuffer;
    CFStringInitInlineBuffer(str, &inlineBuffer, strRange);
// this will correctly fill the buffer with the entire string
//    CFStringInitInlineBuffer(str, &inlineBuffer, CFRangeMake(0, CFStringGetLength(str)));
    CFLocaleRef userLoc = CFLocaleCopyCurrent();
    
    CGFloat score = 0.0, remainingScore = 0.0;
    CFIndex i, j;
    CFRange matchedRange, remainingStrRange, adjustedStrRange = strRange;
    
    // Search for steadily smaller portions of the abbreviation
    for (i = abbrRange.length; i > 0; i--) {
        CFStringRef curAbbr = CFStringCreateWithSubstring (NULL, abbr, CFRangeMake(abbrRange.location, i) );
        // terminality
        // axeen
        
        BOOL found = CFStringFindWithOptionsAndLocale(str, curAbbr,
                                                      CFRangeMake(adjustedStrRange.location, adjustedStrRange.length - abbrRange.length + i),
                                                      kCFCompareCaseInsensitive | kCFCompareDiacriticInsensitive | kCFCompareLocalized,
                                                      userLoc, &matchedRange);

        printf("%i found: %i\n", (int) i, found);
// this prints null when the string gets down to 1 or 2 chars.  ffs.
//        printf("%i %s\n", (int) i, CFStringGetCStringPtr(curAbbr, kCFStringEncodingUTF8));
        CFShow(curAbbr);
        
        CFRelease(curAbbr);
        
        if (!found) {
            continue;
        }
        
        if (mask) {
            [mask addIndexesInRange:NSMakeRange(matchedRange.location, matchedRange.length)];
        }
        
        remainingStrRange.location = matchedRange.location + matchedRange.length;
        remainingStrRange.length = strRange.location + strRange.length - remainingStrRange.location;
        
        // Search what is left of the string with the rest of the abbreviation
        remainingScore = QSScoreForAbbreviationWithRanges(str, abbr, mask, remainingStrRange, CFRangeMake(abbrRange.location + i, abbrRange.length - i));
        
        if (remainingScore) {
            score = remainingStrRange.location-strRange.location;
            // ignore skipped characters if is first letter of a word
            if (matchedRange.location > strRange.location) {
                // if some letters were skipped
                UniChar previousChar = CFStringGetCharacterFromInlineBuffer(&inlineBuffer, matchedRange.location - 1);
                UniChar character = CFStringGetCharacterFromInlineBuffer(&inlineBuffer, matchedRange.location);
                
                printf("prev and current: %c %c (%i %i)\n", previousChar, character, (int) matchedRange.location - 1, (int) matchedRange.location);
                
                if (CFCharacterSetIsCharacterMember(wordSeparator, previousChar)) {
                    // We're on the first letter of a word
                    for (j = matchedRange.location - 2; j >= strRange.location; j--) {
                        if (CFCharacterSetIsCharacterMember(wordSeparator, CFStringGetCharacterFromInlineBuffer(&inlineBuffer, j)))
                            score--;
                        else
                            score -= SKIPPED_SCORE;
                    }
                } else if (CFCharacterSetIsCharacterMember(uppercase, character)) {
                    for (j = matchedRange.location - 1; j >= strRange.location; j--) {
                        if (CFCharacterSetIsCharacterMember(uppercase, CFStringGetCharacterFromInlineBuffer(&inlineBuffer, j)))
                            score--;
                        else
                            score -= SKIPPED_SCORE;
                    }
                } else {
                    printf("before: %.20f minus %i %f\n", score, (matchedRange.location-strRange.location)/2, (float) (matchedRange.location-strRange.location)/2);
                    score -= (matchedRange.location-strRange.location)/2;
                    printf("after: %.20f\n", score);
                }
            }
            score += remainingScore * remainingStrRange.length;
            score /= strRange.length;
            CFRelease(userLoc);
            
            printf("score: %.20f\n", score);
            return score;
        }
    }
    CFRelease(userLoc);
    return 0;
}

int main(int argc, const char * argv[]) {
    CFStringRef str  = CFSTR("Test string");
//    printf("%.20f\n", QSScoreForAbbreviation(str, CFSTR("t"), nil));
//    printf("%.20f\n", QSScoreForAbbreviation(str, CFSTR("ts"), nil));
    printf("%.20f\n", QSScoreForAbbreviation(str, CFSTR("tet"), nil));
//    printf("%.20f\n", QSScoreForAbbreviation(str, CFSTR("str"), nil));
//    printf("%.20f\n", QSScoreForAbbreviation(str, CFSTR("ng"), nil));
    printf("%.20f\n", QSScoreForAbbreviation(CFSTR("react hot loader"), CFSTR("rhl"), nil));
    return 1;
}