Other/Adium Spotlight Importer/GetMetadataForHTMLLog-Additions.m

   1 //
   2 //  GetMetadataForHTMLLog-Additions.m
   3 //  AdiumSpotlightImporter
   4 //
   5 //  Created by Evan Schoenberg on 5/25/06.
   6 //
   7
   8 #import "GetMetadataForHTMLLog-Additions.h"
   9
  10 /*
  11  * @brief These additions are all from AIUtilities
  12  *
  13  * The spotlight importer should include this file to get these specific additions.
  14  * If the GetMetadataForHTMLLog class is used in a situation in which AIUtilities is linked in already, it is
  15  * not necessary to include this implementation file.
  16  */
  17 @implementation NSScanner (AdiumSpotlightImporterAdditions)
  18
  19 - (BOOL)scanUnsignedInt:(unsigned int *)unsignedIntValue
  20 {
  21         //skip characters if necessary
  22         NSCharacterSet *skipSet = [self charactersToBeSkipped];
  23         [self setCharactersToBeSkipped:nil];
  24         [self scanCharactersFromSet:skipSet intoString:NULL];
  25         [self setCharactersToBeSkipped:skipSet];
  26
  27         NSString *string = [self string];
  28         NSRange range = NSMakeRange([self scanLocation], 0);
  29         register unsigned length = [string length] - range.location; //register because it is used in the loop below.
  30         range.length = length;
  31
  32         unichar *buf = malloc(length * sizeof(unichar));
  33         [string getCharacters:buf range:range];
  34
  35         register unsigned i = 0;
  36
  37         if (length && (buf[i] == '+')) {
  38                 ++i;
  39         }
  40         if (i >= length) return NO;
  41         if ((buf[i] < '0') || (buf[i] > '9')) return NO;
  42
  43         unsigned total = 0;
  44         while (i < length) {
  45                 if ((buf[i] >= '0') && (buf[i] <= '9')) {
  46                         total *= 10;
  47                         total += buf[i] - '0';
  48                         ++i;
  49                 } else {
  50                         break;
  51                 }
  52         }
  53         [self setScanLocation:i];
  54         *unsignedIntValue = total;
  55         return YES;
  56 }
  57
  58 @end
  59
  60 //From AIUtilities
  61 @implementation NSString (AdiumSpotlightImporterAdditions)
  62
  63 BOOL AIGetSurrogates(UTF32Char in, UTF16Char *outHigh, UTF16Char *outLow)
  64 {
  65         if (in < 0x10000) {
  66                 if (outHigh) *outHigh = 0;
  67                 if (outLow)  *outLow  = in;
  68                 return NO;
  69         } else {
  70                 enum {
  71                         UTF32LowShiftToUTF16High = 10,
  72                         UTF32HighShiftToUTF16High,
  73                         UTF16HighMask = 31,  //0b0000 0111 1100 0000
  74                         UTF16LowMask  = 63,  //0b0000 0000 0011 1111
  75                         UTF32LowMask = 1023, //0b0000 0011 1111 1111
  76                         UTF16HighAdditiveMask = 55296, //0b1101 1000 0000 0000
  77                         UTF16LowAdditiveMask  = 56320, //0b1101 1100 0000 0000
  78                 };
  79
  80                 if (outHigh) {
  81                         *outHigh = \
  82                         ((in >> UTF32HighShiftToUTF16High) & UTF16HighMask) \
  83                         | ((in >> UTF32LowShiftToUTF16High) & UTF16LowMask) \
  84                         | UTF16HighAdditiveMask;
  85                 }
  86
  87                 if (outLow) {
  88                         *outLow = (in & UTF32LowMask) | UTF16LowAdditiveMask;
  89                 }
  90
  91                 return YES;
  92         }
  93 }
  94
  95 - (NSString *)stringByUnescapingFromHTML
  96 {
  97         if ([self length] == 0) return [[self copy] autorelease]; //avoids various RangeExceptions.
  98
  99         static NSString *ampersand = @"&", *semicolon = @";";
 100
 101         NSString *segment = nil, *entity = nil;
 102         NSScanner *scanner = [NSScanner scannerWithString:self];
 103         [scanner setCaseSensitive:YES];
 104         unsigned myLength = [self length];
 105         NSMutableString *result = [NSMutableString string];
 106
 107         do {
 108                 if ([scanner scanUpToString:ampersand intoString:&segment] || [self characterAtIndex:[scanner scanLocation]] == '&') {
 109                         if (segment) {
 110                                 [result appendString:segment];
 111                                 segment = nil;
 112                         }
 113                         if (![scanner isAtEnd]) {
 114                                 [scanner setScanLocation:[scanner scanLocation]+1];
 115                         }
 116                 }
 117                 if ([scanner scanUpToString:semicolon intoString:&entity]) {
 118                         unsigned number;
 119                         if ([entity characterAtIndex:0] == '#') {
 120                                 NSScanner       *numScanner;
 121                                 unichar         secondCharacter;
 122                                 BOOL            appendIt = NO;
 123
 124                                 numScanner = [NSScanner scannerWithString:entity];
 125                                 [numScanner setCaseSensitive:YES];
 126                                 secondCharacter = [entity characterAtIndex:1];
 127
 128                                 if (secondCharacter == 'x' || secondCharacter == 'X') {
 129                                         //hexadecimal: "#x..." or "#X..."
 130                                         [numScanner setScanLocation:2];
 131                                         appendIt = [numScanner scanHexInt:&number];
 132
 133                                 } else {
 134                                         //decimal: "#..."
 135                                         [numScanner setScanLocation:1];
 136                                         appendIt = [numScanner scanUnsignedInt:&number];
 137                                 }
 138
 139                                 if (appendIt) {
 140                                         unichar chars[2] = { number, 0xffff };
 141                                         CFIndex length = 1;
 142                                         if (number > 0xffff) {
 143                                                 //split into surrogate pair
 144                                                 AIGetSurrogates(number, &chars[0], &chars[1]);
 145                                                 ++length;
 146                                         }
 147                                         CFStringAppendCharacters((CFMutableStringRef)result, chars, length);
 148                                 }
 149                         } else {
 150                                 //named entity. for now, we only support the five essential ones.
 151                                 static NSDictionary *entityNames = nil;
 152                                 if (entityNames == nil) {
 153                                         entityNames = [[NSDictionary alloc] initWithObjectsAndKeys:
 154                                                 [NSNumber numberWithUnsignedInt:'"'], @"quot",
 155                                                 [NSNumber numberWithUnsignedInt:'&'], @"amp",
 156                                                 [NSNumber numberWithUnsignedInt:'<'], @"lt",
 157                                                 [NSNumber numberWithUnsignedInt:'>'], @"gt",
 158                                                 [NSNumber numberWithUnsignedInt:' '], @"nbsp",
 159                                                 nil];
 160                                 }
 161                                 number = [[entityNames objectForKey:[entity lowercaseString]] unsignedIntValue];
 162                                 if (number) {
 163                                         [result appendFormat:@"%C", (unichar)number];
 164                                 }
 165                         }
 166                         if (![scanner isAtEnd]) {
 167                                 [scanner setScanLocation:[scanner scanLocation]+1];
 168                         }
 169                 } //if ([scanner scanUpToString:semicolon intoString:&entity])
 170         } while ([scanner scanLocation] < myLength);
 171         //      NSLog(@"unescaped %@\ninto %@", self, result);
 172         return result;
 173 }
 174
 175
 176 /*
 177  * @brief Read a string from a file, assuming it to be UTF8
 178  *
 179  * If it can not be read as UTF8, it will be read as ASCII.
 180  */
 181 + (NSString *)stringWithContentsOfUTF8File:(NSString *)path
 182 {
 183         NSString        *string;
 184
 185         if ((floor(kCFCoreFoundationVersionNumber) > kCFCoreFoundationVersionNumber10_3)) {
 186                 NSError *error = nil;
 187
 188                 string = [NSString stringWithContentsOfFile:path
 189                                                                                    encoding:NSUTF8StringEncoding
 190                                                                                           error:&error];
 191
 192                 if (error) {
 193                         BOOL    handled = NO;
 194
 195                         if ([[error domain] isEqualToString:NSCocoaErrorDomain]) {
 196                                 int             errorCode = [error code];
 197
 198                                 //XXX - I'm sure these constants are defined somewhere, but I can't find them. -eds
 199                                 if (errorCode == 260) {
 200                                         //File not found.
 201                                         string = nil;
 202                                         handled = YES;
 203
 204                                 } else if (errorCode == 261) {
 205                                         /* Reason: File could not be opened using text encoding Unicode (UTF-8).
 206                                         * Description: Text encoding Unicode (UTF-8) is not applicable.
 207                                         *
 208                                         * We couldn't read the file as UTF8.  Let the system try to determine the encoding.
 209                                         */
 210                                         NSError                         *newError = nil;
 211
 212                                         string = [NSString stringWithContentsOfFile:path
 213                                                                                                            encoding:NSASCIIStringEncoding
 214                                                                                                                   error:&newError];
 215
 216                                         //If there isn't a new error, we recovered reasonably successfully...
 217                                         if (!newError) {
 218                                                 handled = YES;
 219                                         }
 220                                 }
 221                         }
 222
 223                         if (!handled) {
 224                                 NSLog(@"Error reading %@:\n%@; %@.",path,
 225                                           [error localizedDescription], [error localizedFailureReason]);
 226                         }
 227                 }
 228
 229         } else {
 230                 NSData  *data = [NSData dataWithContentsOfFile:path];
 231
 232                 if (data) {
 233                         string = [[[NSString alloc] initWithData:data
 234                                                                                         encoding:NSUTF8StringEncoding] autorelease];
 235                         if (!string) {
 236                                 string = [[[NSString alloc] initWithData:data
 237                                                                                                 encoding:NSASCIIStringEncoding] autorelease];
 238                         }
 239
 240                         if (!string) {
 241                                 NSLog(@"Error reading %@",path);
 242                         }
 243                 } else {
 244                         //File not found
 245                         string = nil;
 246                 }
 247         }
 248
 249         return string;
 250 }
 251
 252 @end