source: trunk/ICeCoffEE/ICeCoffEE/ICeCoffEEParser.m@ 375

//
//  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
        
        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");
    
    NS_DURING
        if ([[NSCharacterSet characterSetWithCharactersInString: @";,."] characterIsMember:
            [string characterAtIndex: selEnd - 1]]) {
            selEnd--;
        }
        NSCharacterSet *alphanumericCharacterSet = [NSCharacterSet alphanumericCharacterSet];
        while (![alphanumericCharacterSet characterIsMember: [string characterAtIndex: selStart]]) {
            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;
    NS_HANDLER
        free(urlData);
        [localException raise];
    NS_ENDHANDLER
    
    free(urlData);
}

NSRange ICCF_URLEnclosingRange(NSString *s, NSRange range) {
    NSCharacterSet *urlLeftDelimiters = nil, *urlRightDelimiters = nil;
    NSRange delimiterRange;
    unsigned extraLen;
    
    ICCF_CheckRange(range);
    
    ICCF_Delimiters(&urlLeftDelimiters, &urlRightDelimiters);
    
expandFront:
    // XXX instead of 0, make this stop at the max URL length to prevent protracted searches
    
    // XXX here's how this is supposed to work:
    // (http://web.sabi.net/) and <http://web.sabi.net/> should work if they are the entire document, even if clicking at the end/beginning of the document, not barfing with "no URL" (correct, as now) or selecting the final >, or ) (what would happen if we remove this "add 1" accommodation).  But how about "http://web.sabi.net/(foo)"?  That should work too, as long as it's not preceded by a (.
    // Should probably backport to ICeCoffEETerminal, now I finally understand the method to this madness.
    // 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");
        range.length += range.location - delimiterRange.location - 1;
        range.location = delimiterRange.location + 1;
        
        // in url/(parens)stuff, handle clicking inside or after (parens).
        /*if ([s characterAtIndex: delimiterRange.location] == '(' &&
            range.location > 2) { // prevent wrapping, ordinarily not necessary
            if ([s rangeOfString: @")" options: NSLiteralSearch range: range].location != NSNotFound ||
                [s rangeOfCharacterFromSet: [NSCharacterSet characterSetWithCharactersInString: @"/."]
                                   options: NSLiteralSearch range: range].location == NSNotFound) {
                range.location -= 2;
                range.length += 2;
                ICLog(@"expanding past (, now |%@|", [s substringWithRange: range]);
                goto expandFront;
            }
        } */       
    }
    
    ICCF_CheckRange(range);
    
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;
    } else {
        NSCAssert(delimiterRange.length == 1, @"Internal error: delimiter matched range is not of length 1");
        range.length += delimiterRange.location - range.location - range.length;
        
        // grow URL past closing paren/brace if we've seen an open paren/brace
        unichar closing = [s characterAtIndex: delimiterRange.location];
        NSString *opening;
        if (closing == ')') opening = @"(";
        else if (closing == '}') opening = @"{";
        else goto checkRange;
        if ([s rangeOfString: opening options: NSLiteralSearch range: range].location == NSNotFound)
            goto checkRange;
        
        if (extraLen == 0) {
            range.length += 1;
            ICLog(@"expanding past %c, now |%@|", closing, [s substringWithRange: range]);
        } else {
            range.length += 2;
            ICLog(@"expanding past %c, now |%@|", closing, [s substringWithRange: range]);
            goto expandBack;
        }
    }
    
checkRange:
    ICCF_CheckRange(range);
    
    ICCF_ParseURL([s substringWithRange: range], &range);
    
    return range;
}