source: trunk/ICeCoffEE/ICeCoffEE/ICeCoffEEParser.m@ 474

//
//  ICeCoffEEParser.m
//  ICeCoffEE
//
//  Created by Nicholas Riley on 6/21/07.
//  Copyright 2007 Nicholas Riley. All rights reserved.
//

#import "ICeCoffEEParser.h"
#import "ICeCoffEE.h"

void ICCF_Delimiters(NSCharacterSet **leftPtr, NSCharacterSet **rightPtr) {
    static NSCharacterSet *urlLeftDelimiters = nil, *urlRightDelimiters = nil;
    
    if (urlLeftDelimiters == nil || urlRightDelimiters == nil) {
        NSMutableCharacterSet *set = [[NSCharacterSet whitespaceAndNewlineCharacterSet] mutableCopy];
        NSMutableCharacterSet *tmpSet;
        [urlLeftDelimiters release];
        [urlRightDelimiters release];
        
        [set autorelease];
        [set formUnionWithCharacterSet: [[NSCharacterSet characterSetWithRange: NSMakeRange(0x21, 0x5e)] invertedSet]]; // nonprintable and non-ASCII characters
        [set formUnionWithCharacterSet: [NSCharacterSet punctuationCharacterSet]];
        // XXX obsoleted by RFC 3986 now... use §2.1, 2.2, 2.3
        [set removeCharactersInString: @";/?:@&=+$,-_.!~*'(){}[]%#"]; // RFC 2396 §2.2, 2.3, 2.4, plus % and # from "delims" set and {}, []
        
        tmpSet = [[set mutableCopy] autorelease];
        [tmpSet formUnionWithCharacterSet: [NSCharacterSet characterSetWithCharactersInString: @"<(["]];
        urlLeftDelimiters = [tmpSet copy]; // make immutable again - for efficiency
        
        tmpSet = [[set mutableCopy] autorelease];
        [tmpSet formUnionWithCharacterSet: [NSCharacterSet characterSetWithCharactersInString: @">)]"]];
        urlRightDelimiters = [tmpSet copy]; // make immutable again - for efficiency
    }
    
    *leftPtr = urlLeftDelimiters; *rightPtr = urlRightDelimiters;
}

static ICInstance ICCF_icInst = NULL;

void ICCF_StartIC() {
    OSStatus err;
    
    if (ICCF_icInst != NULL) {
        ICLog(@"ICCF_StartIC: Internet Config is already running!");
        ICCF_StopIC();
    }
    err = ICStart(&ICCF_icInst, kICCFCreator);
    NSCAssert1(err == noErr, ICCF_LocalizedString(@"Unable to start Internet Config (error %d)"), err);
}

void ICCF_StopIC() {
    if (ICCF_icInst == NULL) {
        ICLog(@"ICCF_StopIC: Internet Config is not running!");
    } else {
        ICStop(ICCF_icInst);
        ICCF_icInst = NULL;
    }
}

ICInstance ICCF_GetInst() {
    NSCAssert(ICCF_icInst != NULL, @"Internal error: Called ICCF_GetInst without ICCF_StartIC");
    return ICCF_icInst;
}

// input/output 'range' is the range of source document which contains 'string'
void ICCF_ParseURL(NSString *string, NSRange *range) {
    OSStatus err;
    Handle h;
    long selStart = 0, selEnd = range->length; // local offsets within 'string'
    char *urlData = NULL;
    
    NSCAssert(selEnd == [string length], @"Internal error: URL string is wrong length");
    
    @try {
        if ([[NSCharacterSet characterSetWithCharactersInString: @";,.!"] characterIsMember:
            [string characterAtIndex: selEnd - 1]]) {
            selEnd--;
        }
        NSCharacterSet *alphanumericCharacterSet = [NSCharacterSet alphanumericCharacterSet];
        unichar opening, closing;
        while (![alphanumericCharacterSet characterIsMember:
                 (opening = [string characterAtIndex: selStart])]) {
            closing = [string characterAtIndex: selEnd - 1];
            if ((opening == '(' && closing == ')') ||
                (opening == '{' && closing == '}') ||
                (opening == '[' && closing == ']')) {
                selEnd--;
            }
            selStart++;
            NSCAssert(selStart < selEnd, @"No URL is selected");
        }
        
        string = [string substringWithRange: NSMakeRange(selStart, selEnd - selStart)];
        
        ICLog(@"Parsing URL |%@|", string);
        
        NSCAssert([string canBeConvertedToEncoding: NSASCIIStringEncoding], @"No URL is selected");
        
        urlData = (char *)malloc( (range->length + 1) * sizeof(char));
        NSCAssert(urlData != NULL, @"Internal error: can't allocate memory for URL string");
        
        // XXX getCString: is deprecated in 10.4, but this is safe and shouldn't assert because we've already verified the string can be converted to ASCII, which should be a subset of any possible system encoding.  The replacement (getCString:maxLength:encoding:) is not available until 10.4, so we leave this until we dump Internet Config and gain IDN friendliness.
        [string getCString: urlData];
        
        h = NewHandle(0);
        NSCAssert(h != NULL, @"Internal error: can't allocate URL handle");
        
        err = ICParseURL(ICCF_GetInst(), "\pmailto", urlData, range->length, &selStart, &selEnd, h);
        DisposeHandle(h);
        
        ICCF_OSErrCAssert(err, @"ICParseURL");
        
        range->length = selEnd - selStart;
        range->location += selStart;
    } @finally {
        free(urlData);
    }
}

static BOOL ICCF_StringIncludesCharacter(NSString *s, unichar character, NSRange range) {
    NSRange result = [s rangeOfCharacterFromSet: [NSCharacterSet characterSetWithCharactersInString:
                                         [NSString stringWithCharacters: &character length: 1]]
                                        options: NSLiteralSearch range: range];
    return (result.location != NSNotFound);
}

static BOOL ICCF_IsLikelyURI(NSString *s, NSRange range) {
    return ([s rangeOfCharacterFromSet: [NSCharacterSet characterSetWithCharactersInString: @":/.@"]
                               options: NSLiteralSearch range: range].location != NSNotFound);
}

static BOOL ICCF_IsLikelyIPv6Address(NSString *s, NSRange range) {
    return ([s rangeOfCharacterFromSet:
             [[NSCharacterSet characterSetWithCharactersInString: @"ABCDEFabcdef0123456789:"] invertedSet]
                               options: NSLiteralSearch range: range].location == NSNotFound);
}

NSRange ICCF_URLEnclosingRange(NSString *s, NSRange range) {
    NSCharacterSet *urlLeftDelimiters = nil, *urlRightDelimiters = nil;
    NSRange delimiterRange;
    unsigned extraLen;
    BOOL multiLine = NO;
    
    ICCF_CheckRange(range);
    
    ICCF_Delimiters(&urlLeftDelimiters, &urlRightDelimiters);
    
    // right delimiter selected?  Yes, this can break with ...)URL(....  Oh well.
    if (range.location > 0 && [urlRightDelimiters characterIsMember: [s characterAtIndex: range.location]]) {
        --range.location;
        ++range.length;
        ICLog(@"expanding past initial %c, now |%@|", [s characterAtIndex: range.location + 1],
              [s substringWithRange: range]);
    }
        
expandFront:
    // XXX instead of 0, make this stop at the max URL length to prevent protracted searches
    // XXX backport to ICeCoffEETerminal
    // add 1 to range to trap delimiters that are on the edge of the selection (i.e., <...)
    delimiterRange = [s rangeOfCharacterFromSet: urlLeftDelimiters
                                        options: NSLiteralSearch | NSBackwardsSearch
                                          range: NSMakeRange(0, range.location + (range.location != [s length]))];
    if (delimiterRange.location == NSNotFound) {
        // extend to beginning of string
        range.length += range.location;
        range.location = 0;
    } else {
        NSCAssert(delimiterRange.length == 1, @"Internal error: delimiter matched range is not of length 1");
        if ([s characterAtIndex: delimiterRange.location] == '<') { // XXX move to expandBoth to handle clicking in middle
            multiLine = YES;
            urlRightDelimiters = [NSCharacterSet characterSetWithCharactersInString: @">"];
        }
        range.length += range.location - delimiterRange.location - 1;
        range.location = delimiterRange.location + 1;
    }
    

expandBack:
    // XXX instead of length of string, make this stop at the max URL length to prevent protracted searches
    // add 1 to range to trap delimiters that are on the edge of the selection (i.e., ...>)
    extraLen = [s length] - range.location - range.length;
    delimiterRange = [s rangeOfCharacterFromSet: urlRightDelimiters
                                        options: NSLiteralSearch
                                          range: NSMakeRange(range.location + range.length - (range.length != 0),
                                                             extraLen + (range.length != 0))];
    if (delimiterRange.location == NSNotFound) {
        // extend to end of string
        range.length += extraLen;
        extraLen = 0;
    } else {
        NSCAssert(delimiterRange.length == 1, @"Internal error: delimiter matched range is not of length 1");
        range.length += delimiterRange.location - range.location - range.length;
        extraLen = [s length] - NSMaxRange(range);

        unichar opening, closing = [s characterAtIndex: delimiterRange.location];
        if (closing == '>' && !multiLine && ICCF_StringIncludesCharacter(s, '<', NSMakeRange(0, range.location))) {
            urlLeftDelimiters = [NSCharacterSet characterSetWithCharactersInString: @"<"];
            goto expandFront; // XXX move to expandBoth to handle clicking in middle
        }
        // grow URL past closing paren/brace/bracket if we've seen an open paren/brace/bracket
        if (closing == ')') opening = '(';
        else if (closing == '}') opening = '{';
        else if (closing == ']') opening = '[';
        else goto expandBoth;
        if (!ICCF_StringIncludesCharacter(s, opening, range))
            goto expandBoth;
        
        if (extraLen == 1) {
            range.length += 1;
            --extraLen;
            ICLog(@"expanding past %c, now |%@|", closing, [s substringWithRange: range]);
        } else {
            range.length += 2;
            ICLog(@"expanding past %c, now |%@|", closing, [s substringWithRange: range]);
            goto expandBack;
        }
    }
    
expandBoth:
    if (range.location <= 1)
        goto checkRange; // nowhere to expand
    unichar opening = [s characterAtIndex: range.location - 1], closing;
    if (opening == '(') closing = ')';
    else if (opening == '{') closing = '}';
    else if (opening == '[') closing = ']';
    else goto checkRange;

    ICLog(@"extraLen = %d", extraLen);
    // check if we're inside a partial delimited URL: not foolproof, but handles (foo), {UUID} and [IPv6]
    if (delimiterRange.location != NSNotFound && [s characterAtIndex: delimiterRange.location] == closing && 
        ((opening == '[' && ICCF_IsLikelyIPv6Address(s, range)) || !ICCF_IsLikelyURI(s, range))) {
        ICLog(@"expanding past %c...%c, was |%@|", opening, closing, [s substringWithRange: range]);
        range.location -= 2;
        if (extraLen > 1)
            range.length += 4;
        else
            range.length += 2 + extraLen;
        ICLog(@"expanding past %c...%c, now |%@|", opening, closing, [s substringWithRange: range]);
        goto expandFront;
    }

    if (ICCF_StringIncludesCharacter(s, closing, range) &&
        ((opening == '[' &&
          ICCF_IsLikelyIPv6Address(s, 
               NSMakeRange(range.location,
                           [s rangeOfString: @"]" options: NSLiteralSearch range: range].location -
                           range.location)))
         || !ICCF_IsLikelyURI(s, range))) {
        range.location -= 2;
        range.length += 2;
        ICLog(@"expanding past %c, now |%@|", opening, [s substringWithRange: range]);
        goto expandFront;
    }

checkRange:
    ICCF_CheckRange(range);
    
    ICCF_ParseURL([s substringWithRange: range], &range);
    
    return range;
}