mathpad/MathPad/MPExpressionTokenizer.m

//
//  MPExpressionLexer.m
//  MathPad
//
//  Created by Kim Wittenburg on 19.09.14.
//  Copyright (c) 2014 Kim Wittenburg. All rights reserved.
//

#import "MPExpressionTokenizer.h"

#import "MPExpression.h"
#import "MPExpressionElement.h"

#import "MPToken.h"

#import "NSRegularExpression+MPParsingAdditions.h"


#define MPRangeExists(range) (range.location != NSNotFound)



@implementation MPExpressionTokenizer

+ (NSArray *)tokenizeExpression:(MPExpression *)expression
{
    NSMutableArray *tokens = [[NSMutableArray alloc] init];
    NSUInteger symbolIndex = 0;
    for (NSUInteger index = 0; index < [expression countItemsInReferenceFrame:MPElementReferenceFrame]; index++) {
        id <MPExpressionElement> element = [expression itemAtIndex:index referenceFrame:MPElementReferenceFrame];
        if ([element isFunction]) {
            [tokens addObject:element];
        } else {
            [tokens addObjectsFromArray:[self tokenizeElement:(NSString *)element
                                           elementSymbolIndex:symbolIndex]];
        }
        symbolIndex += element.length;
    }
    return tokens;
}



+ (NSArray *)tokenizeElement:(NSString *)element
          elementSymbolIndex:(NSUInteger)symbolIndex
{
    NSUInteger lexLocation = 0;
    
    NSString *decimalSeparator = [NSRegularExpression escapedPatternForString:[[NSLocale currentLocale] objectForKey:NSLocaleDecimalSeparator]];
    NSString *regexStringFormat = @"\\A(?:"
                            @"([\\*∙⋅])|"
                            @"([+-](?:\\s*[+-])*)|"
                            @"((?:\\d+(?:%@\\d+)?)|(?:%@\\d+))|"
                            @"(sin)|"
                            @"(cos)|"
                            @"(tan)|"
                            @"(asin)|"
                            @"(acos)|"
                            @"(atan)|"
                            @"([A-Za-z])|"
                            @"(!)|"
                            @"(=)|"
                            @"(\\s+)"
                            @")";
    NSString *regexString = [NSString stringWithFormat:regexStringFormat, decimalSeparator, decimalSeparator];
    NSRegularExpression *regex = [NSRegularExpression regularExpressionWithPattern:regexString
                                                                           options:0
                                                                             error:NULL];
    NSMutableArray *tokens = [[NSMutableArray alloc] init];
    while (lexLocation < element.length) {
        NSTextCheckingResult *match = [regex firstMatchInString:element fromIndex:lexLocation];
        
        NSRange range = NSMakeRange(lexLocation, 1);
        MPTokenType tokenType = MPUnidentifiedToken;
        if (match) {
            NSRange multiplicationSymbolRange = [match rangeAtIndex:1];
            NSRange operatorRange = [match rangeAtIndex:2];
            NSRange numberRange = [match rangeAtIndex:3];
            NSRange sinRange = [match rangeAtIndex:4];
            NSRange cosRange = [match rangeAtIndex:5];
            NSRange tanRange = [match rangeAtIndex:6];
            NSRange asinRange = [match rangeAtIndex:7];
            NSRange acosRange = [match rangeAtIndex:8];
            NSRange atanRange = [match rangeAtIndex:9];
            NSRange variableRange = [match rangeAtIndex:10];
            NSRange factorialRange = [match rangeAtIndex:11];
            NSRange equalsRange = [match rangeAtIndex:12];
            NSRange whitespaceRange = [match rangeAtIndex:13];
        
            if (MPRangeExists(multiplicationSymbolRange)) {
                range = multiplicationSymbolRange;
                tokenType = MPMultiplicationSymbolToken;
            } else if (MPRangeExists(operatorRange)) {
                range = operatorRange;
                tokenType = MPOperatorListToken;
            } else if (MPRangeExists(numberRange)) {
                range = numberRange;
                tokenType = MPNumberToken;
            } else if (MPRangeExists(sinRange)) {
                range = sinRange;
                tokenType = MPSinToken;
            } else if (MPRangeExists(cosRange)) {
                range = cosRange;
                tokenType = MPCosToken;
            } else if (MPRangeExists(tanRange)) {
                range = tanRange;
                tokenType = MPTanToken;
            } else if (MPRangeExists(asinRange)) {
                range = asinRange;
                tokenType = MPASinToken;
            } else if (MPRangeExists(acosRange)) {
                range = acosRange;
                tokenType = MPACosToken;
            } else if (MPRangeExists(atanRange)) {
                range = atanRange;
                tokenType = MPATanToken;
            } else if (MPRangeExists(variableRange)) {
                range = variableRange;
                tokenType = MPVariableToken;
            } else if (MPRangeExists(factorialRange)) {
                range = factorialRange;
                tokenType = MPFactorialToken;
            } else if (MPRangeExists(equalsRange)) {
                range = equalsRange;
                tokenType = MPEqualsToken;
            } else if (MPRangeExists(whitespaceRange)) {
                range = whitespaceRange;
                tokenType = MPWhitespaceToken;
            } else {
                // Should not get here
                range = NSMakeRange(lexLocation, 1);
                tokenType = MPUnidentifiedToken;
            }
        }
        
        lexLocation = NSMaxRange(range);
        NSString *tokenStringValue = [element substringWithRange:range];
        range.location += symbolIndex;
        [tokens addObject:[[MPToken alloc] initWithTokenType:tokenType
                                                       range:range
                                                 stringValue:tokenStringValue]];
    }
    
    return tokens;
}

@end