webnsurlextras.mm

linux下开源浏览器WebKit的源码,市面上的很多商用浏览器都是移植自WebKit

    static int URLComponentTypeBufferLength = 2048;
    
    UInt8 staticAllBytesBuffer[URLComponentTypeBufferLength];
    UInt8 *allBytesBuffer = staticAllBytesBuffer;
    
    CFIndex bytesFilled = CFURLGetBytes((CFURLRef)self, allBytesBuffer, URLComponentTypeBufferLength);
    if (bytesFilled == -1) {
        CFIndex bytesToAllocate = CFURLGetBytes((CFURLRef)self, NULL, 0);
        allBytesBuffer = static_cast<UInt8 *>(malloc(bytesToAllocate));
        bytesFilled = CFURLGetBytes((CFURLRef)self, allBytesBuffer, bytesToAllocate);
    }
    
    CFRange range;
    if (componentType != completeURL) {
        range = CFURLGetByteRangeForComponent((CFURLRef)self, componentType, NULL);
        if (range.location == kCFNotFound) {
            return nil;
        }
    }
    else {
        range.location = 0;
        range.length = bytesFilled;
    }
    
    NSData *componentData = [NSData dataWithBytes:allBytesBuffer + range.location length:range.length]; 
    
    const unsigned char *bytes = static_cast<const unsigned char *>([componentData bytes]);
    NSMutableData *resultData = [NSMutableData data];
    // NOTE: add leading '?' to query strings non-zero length query strings.
    // NOTE: retain question-mark only query strings.
    if (componentType == kCFURLComponentQuery) {
        if (range.length > 0 || [self _web_hasQuestionMarkOnlyQueryString]) {
            [resultData appendBytes:"?" length:1];    
        }
    }
    int i;
    for (i = 0; i < range.length; i++) {
        unsigned char c = bytes[i];
        if (c <= 0x20 || c >= 0x7f) {
            char escaped[3];
            escaped[0] = '%';
            escaped[1] = hexDigit(c >> 4);
            escaped[2] = hexDigit(c & 0xf);
            [resultData appendBytes:escaped length:3];    
        }
        else {
            char b[1];
            b[0] = c;
            [resultData appendBytes:b length:1];    
        }               
    }
    
    if (staticAllBytesBuffer != allBytesBuffer) {
        free(allBytesBuffer);
    }
    
    return resultData;
}

-(NSData *)_web_schemeData
{
    return [self _web_dataForURLComponentType:kCFURLComponentScheme];
}

-(NSData *)_web_hostData
{
    NSData *result = [self _web_dataForURLComponentType:kCFURLComponentHost];
    NSData *scheme = [self _web_schemeData];
    // Take off localhost for file
    if ([scheme _web_isCaseInsensitiveEqualToCString:"file"]) {
        return ([result _web_isCaseInsensitiveEqualToCString:"localhost"]) ? nil : result;
    }
    return result;
}

- (NSString *)_web_hostString
{
    NSData *data = [self _web_hostData];
    if (!data) {
        data = [NSData data];
    }
    return [[[NSString alloc] initWithData:[self _web_hostData] encoding:NSUTF8StringEncoding] autorelease];
}

- (NSString *)_webkit_suggestedFilenameWithMIMEType:(NSString *)MIMEType
{
    return suggestedFilenameWithMIMEType(self, MIMEType);
}

@end

@implementation NSString (WebNSURLExtras)

- (BOOL)_web_isUserVisibleURL
{
    BOOL valid = YES;
    // get buffer

    char static_buffer[1024];
    const char *p;
    BOOL success = CFStringGetCString((CFStringRef)self, static_buffer, 1023, kCFStringEncodingUTF8);
    if (success) {
        p = static_buffer;
    } else {
        p = [self UTF8String];
    }

    int length = strlen(p);

    // check for characters <= 0x20 or >=0x7f, %-escape sequences of %7f, and xn--, these
    // are the things that will lead _web_userVisibleString to actually change things.
    int i;
    for (i = 0; i < length; i++) {
        unsigned char c = p[i];
        // escape control characters, space, and delete
        if (c <= 0x20 || c == 0x7f) {
            valid = NO;
            break;
        } else if (c == '%' && (i + 1 < length && isHexDigit(p[i + 1])) && i + 2 < length && isHexDigit(p[i + 2])) {
            unsigned char u = (hexDigitValue(p[i + 1]) << 4) | hexDigitValue(p[i + 2]);
            if (u > 0x7f) {
                valid = NO;
                break;
            }
            i += 2;
        } else {
            // Check for "xn--" in an efficient, non-case-sensitive, way.
            if (c == '-' && i >= 3 && (p[i - 3] | 0x20) == 'x' && (p[i - 2] | 0x20) == 'n' && p[i - 1] == '-') {
                valid = NO;
                break;
            }
        }
    }

    return valid;
}


- (BOOL)_webkit_isJavaScriptURL
{
    return [self _webkit_hasCaseInsensitivePrefix:@"javascript:"];
}

- (BOOL)_webkit_isFileURL
{
    return [self rangeOfString:@"file:" options:(NSCaseInsensitiveSearch | NSAnchoredSearch)].location != NSNotFound;
}

- (NSString *)_webkit_stringByReplacingValidPercentEscapes
{
    return decodeURLEscapeSequences(self);
}

- (NSString *)_webkit_scriptIfJavaScriptURL
{
    if (![self _webkit_isJavaScriptURL]) {
        return nil;
    }
    return [[self substringFromIndex:11] _webkit_stringByReplacingValidPercentEscapes];
}

- (BOOL)_webkit_isFTPDirectoryURL
{
    int length = [self length];
    if (length < 5) {  // 5 is length of "ftp:/"
        return NO;
    }
    unichar lastChar = [self characterAtIndex:length - 1];
    return lastChar == '/' && [self _webkit_hasCaseInsensitivePrefix:@"ftp:"];
}


static BOOL readIDNScriptWhiteListFile(NSString *filename)
{
    if (!filename) {
        return NO;
    }
    FILE *file = fopen([filename fileSystemRepresentation], "r");
    if (file == NULL) {
        return NO;
    }

    // Read a word at a time.
    // Allow comments, starting with # character to the end of the line.
    while (1) {
        // Skip a comment if present.
        int result = fscanf(file, " #%*[^\n\r]%*[\n\r]");
        if (result == EOF) {
            break;
        }

        // Read a script name if present.
        char word[33];
        result = fscanf(file, " %32[^# \t\n\r]%*[^# \t\n\r] ", word);
        if (result == EOF) {
            break;
        }
        if (result == 1) {
            // Got a word, map to script code and put it into the array.
            int32_t script = u_getPropertyValueEnum(UCHAR_SCRIPT, word);
            if (script >= 0 && script < USCRIPT_CODE_LIMIT) {
                size_t index = script / 32;
                uint32_t mask = 1 << (script % 32);
                IDNScriptWhiteList[index] |= mask;
            }
        }
    }
    fclose(file);
    return YES;
}

static void readIDNScriptWhiteList(void)
{
    // Read white list from library.
    NSArray *dirs = NSSearchPathForDirectoriesInDomains(NSLibraryDirectory, NSAllDomainsMask, YES);
    int i, numDirs = [dirs count];
    for (i = 0; i < numDirs; i++) {
        NSString *dir = [dirs objectAtIndex:i];
        if (readIDNScriptWhiteListFile([dir stringByAppendingPathComponent:@"IDNScriptWhiteList.txt"])) {
            return;
        }
    }

    // Fall back on white list inside bundle.
    NSBundle *bundle = [NSBundle bundleWithIdentifier:@"com.apple.WebKit"];
    readIDNScriptWhiteListFile([bundle pathForResource:@"IDNScriptWhiteList" ofType:@"txt"]);
}

static BOOL allCharactersInIDNScriptWhiteList(const UChar *buffer, int32_t length)
{
    pthread_once(&IDNScriptWhiteListFileRead, readIDNScriptWhiteList);

    int32_t i = 0;
    while (i < length) {
        UChar32 c;
        U16_NEXT(buffer, i, length, c)
        UErrorCode error = U_ZERO_ERROR;
        UScriptCode script = uscript_getScript(c, &error);
        if (error != U_ZERO_ERROR) {
            LOG_ERROR("got ICU error while trying to look at scripts: %d", error);
            return NO;
        }
        if (script < 0) {
            LOG_ERROR("got negative number for script code from ICU: %d", script);
            return NO;
        }
        if (script >= USCRIPT_CODE_LIMIT) {
            return NO;
        }
        size_t index = script / 32;
        uint32_t mask = 1 << (script % 32);
        if (!(IDNScriptWhiteList[index] & mask)) {
            return NO;
        }

        if (isLookalikeCharacter(c))
            return NO;
    }
    return YES;
}

// Return value of nil means no mapping is necessary.
// If makeString is NO, then return value is either nil or self to indicate mapping is necessary.
// If makeString is YES, then return value is either nil or the mapped string.
- (NSString *)_web_mapHostNameWithRange:(NSRange)range encode:(BOOL)encode makeString:(BOOL)makeString
{
    if (range.length > HOST_NAME_BUFFER_LENGTH) {
        return nil;
    }

    if ([self length] == 0)
        return nil;
    
    UChar sourceBuffer[HOST_NAME_BUFFER_LENGTH];
    UChar destinationBuffer[HOST_NAME_BUFFER_LENGTH];
    
    NSString *string = self;
    if (encode && [self rangeOfString:@"%" options:NSLiteralSearch range:range].location != NSNotFound) {
        NSString *substring = [self substringWithRange:range];
        substring = WebCFAutorelease(CFURLCreateStringByReplacingPercentEscapes(NULL, (CFStringRef)substring, CFSTR("")));
        if (substring != nil) {
            string = substring;
            range = NSMakeRange(0, [string length]);
        }
    }
    
    int length = range.length;
    [string getCharacters:sourceBuffer range:range];

    UErrorCode error = U_ZERO_ERROR;
    int32_t numCharactersConverted = (encode ? uidna_IDNToASCII : uidna_IDNToUnicode)
        (sourceBuffer, length, destinationBuffer, HOST_NAME_BUFFER_LENGTH, UIDNA_ALLOW_UNASSIGNED, NULL, &error);
    if (error != U_ZERO_ERROR) {
        return nil;
    }
    if (numCharactersConverted == length && memcmp(sourceBuffer, destinationBuffer, length * sizeof(UChar)) == 0) {
        return nil;
    }
    if (!encode && !allCharactersInIDNScriptWhiteList(destinationBuffer, numCharactersConverted)) {
        return nil;
    }
    return makeString ? (NSString *)[NSString stringWithCharacters:destinationBuffer length:numCharactersConverted] : (NSString *)self;
}

- (BOOL)_web_hostNameNeedsDecodingWithRange:(NSRange)range
{
    return [self _web_mapHostNameWithRange:range encode:NO makeString:NO] != nil;
}

- (BOOL)_web_hostNameNeedsEncodingWithRange:(NSRange)range
{
    return [self _web_mapHostNameWithRange:range encode:YES makeString:NO] != nil;
}

- (NSString *)_web_decodeHostNameWithRange:(NSRange)range
{
    return [self _web_mapHostNameWithRange:range encode:NO makeString:YES];
}

- (NSString *)_web_encodeHostNameWithRange:(NSRange)range
{
    return [self _web_mapHostNameWithRange:range encode:YES makeString:YES];
}

- (NSString *)_web_decodeHostName
{
    NSString *name = [self _web_mapHostNameWithRange:NSMakeRange(0, [self length]) encode:NO makeString:YES];
    return name == nil ? self : name;
}

- (NSString *)_web_encodeHostName
{
    NSString *name = [self _web_mapHostNameWithRange:NSMakeRange(0, [self length]) encode:YES makeString:YES];
    return name == nil ? self : name;
}

-(NSRange)_webkit_rangeOfURLScheme
{
    NSRange colon = [self rangeOfString:@":"];
    if (colon.location != NSNotFound && colon.location > 0) {
        NSRange scheme = {0, colon.location};
        static NSCharacterSet *InverseSchemeCharacterSet = nil;
        if (!InverseSchemeCharacterSet) {
            /*
             This stuff is very expensive.  10-15 msec on a 2x1.2GHz.  If not cached it swamps
             everything else when adding items to the autocomplete DB.  Makes me wonder if we
             even need to enforce the character set here.
            */
            NSString *acceptableCharacters = @"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+.-";
            InverseSchemeCharacterSet = [[[NSCharacterSet characterSetWithCharactersInString:acceptableCharacters] invertedSet] retain];
        }
        NSRange illegals = [self rangeOfCharacterFromSet:InverseSchemeCharacterSet options:0 range:scheme];
        if (illegals.location == NSNotFound)
            return scheme;
    }
    return NSMakeRange(NSNotFound, 0);
}

-(BOOL)_webkit_looksLikeAbsoluteURL
{
    // Trim whitespace because _web_URLWithString allows whitespace.
    return [[self _webkit_stringByTrimmingWhitespace] _webkit_rangeOfURLScheme].location != NSNotFound;
}

- (NSString *)_webkit_URLFragment
{
    NSRange fragmentRange;
    
    fragmentRange = [self rangeOfString:@"#" options:NSLiteralSearch];
    if (fragmentRange.location == NSNotFound)
        return nil;
    return [self substringFromIndex:fragmentRange.location + 1];
}

@end