Other/Adium Spotlight Importer/GetMetadataForFile.m

   1 #include <CoreFoundation/CoreFoundation.h>
   2 #include <CoreServices/CoreServices.h>
   3 #import "GetMetadataForHTMLLog.h"
   4 #import <AIUtilities/NSCalendarDate+ISO8601Parsing.h>
   5
   6 /*
   7  Relevant keys from MDItem.h we use or may want to use:
   8
   9  @constant kMDItemContentCreationDate
  10  This is the date that the contents of the file were created,
  11  has an application specific semantic.
  12  Type is a CFDate.
  13
  14  @constant kMDItemTextContent
  15  Contains the text content of the document. Type is a CFString.
  16
  17  @constant kMDItemDisplayName
  18  This is the localized version of the LaunchServices call
  19  LSCopyDisplayNameForURL()/LSCopyDisplayNameForRef().
  20
  21  @const  kMDItemInstantMessageAddresses
  22  Instant message addresses for this item. Type is an Array of CFStrings.
  23  */
  24
  25 /* -----------------------------------------------------------------------------
  26 Get metadata attributes from file
  27
  28 This function's job is to extract useful information your file format supports
  29 and return it as a dictionary
  30 ----------------------------------------------------------------------------- */
  31
  32 Boolean GetMetadataForXMLLog(NSMutableDictionary *attributes, NSString *pathToFile);
  33 NSString *GetTextContentForXMLLog(NSString *pathToFile);
  34
  35 Boolean GetMetadataForFile(void* thisInterface,
  36                                                    CFMutableDictionaryRef attributes,
  37                                                    CFStringRef contentTypeUTI,
  38                                                    CFStringRef pathToFile)
  39 {
  40     /* Pull any available metadata from the file at the specified path */
  41     /* Return the attribute keys and attribute values in the dict */
  42     /* Return TRUE if successful, FALSE if there was no data provided */
  43
  44         Boolean                         success = FALSE;
  45         NSAutoreleasePool       *pool;
  46         pool = [[NSAutoreleasePool alloc] init];
  47
  48         if (CFStringCompare(contentTypeUTI, (CFStringRef)@"com.adiumx.htmllog", kCFCompareBackwards) == kCFCompareEqualTo) {
  49                 success = GetMetadataForHTMLLog((NSMutableDictionary *)attributes, (NSString *)pathToFile);
  50         } else if (CFStringCompare(contentTypeUTI, (CFStringRef)@"com.adiumx.xmllog", kCFCompareBackwards) == kCFCompareEqualTo) {
  51                 success = GetMetadataForXMLLog((NSMutableDictionary *)attributes, (NSString *)pathToFile);
  52         } else {
  53                 NSLog(@"We were passed %@, of type %@, which is an unknown type",pathToFile,contentTypeUTI);
  54         }
  55
  56         [pool release];
  57
  58     return success;
  59 }
  60
  61 /*!
  62  * @brief Copy the text content for a file
  63  *
  64  * This is the text which would be the kMDItemTextContent for the file in Spotlight.
  65  *
  66  * @param contentTypeUTI The UTI type. If NULL, the extension of pathToFile will be used
  67  * @param pathToFile The full path to the file
  68  *
  69  * @result The kMDItemTextContent. Follows the Copy rule.
  70  */
  71 CFStringRef CopyTextContentForFile(CFStringRef contentTypeUTI,
  72                                                                    CFStringRef pathToFile)
  73 {
  74         NSAutoreleasePool       *pool;
  75         CFStringRef                     textContent;
  76         pool = [[NSAutoreleasePool alloc] init];
  77
  78         //Deteremine the UTI type if we weren't passed one
  79         if (contentTypeUTI == NULL) {
  80                 if (CFStringCompare((CFStringRef)[(NSString *)pathToFile pathExtension],
  81                                                         CFSTR("chatLog"),
  82                                                         (kCFCompareBackwards | kCFCompareCaseInsensitive)) == kCFCompareEqualTo) {
  83                         contentTypeUTI = CFSTR("com.adiumx.xmllog");
  84                 } else if (CFStringCompare((CFStringRef)[(NSString *)pathToFile pathExtension],
  85                                                         CFSTR("AdiumXMLLog"),
  86                                                         (kCFCompareBackwards | kCFCompareCaseInsensitive)) == kCFCompareEqualTo) {
  87                         contentTypeUTI = CFSTR("com.adiumx.xmllog");
  88                 } else {
  89                         //Treat all other log extensions as HTML logs (plaintext will come out fine this way, too)
  90                         contentTypeUTI = CFSTR("com.adiumx.htmllog");
  91                 }
  92         }
  93
  94         if (CFStringCompare(contentTypeUTI, CFSTR("com.adiumx.htmllog"), kCFCompareBackwards) == kCFCompareEqualTo) {
  95                 textContent = (CFStringRef)GetTextContentForHTMLLog((NSString *)pathToFile);
  96         } else if (CFStringCompare(contentTypeUTI, (CFStringRef)@"com.adiumx.xmllog", kCFCompareBackwards) == kCFCompareEqualTo) {
  97                 textContent = (CFStringRef)GetTextContentForXMLLog((NSString *)pathToFile);
  98
  99         } else {
 100                 textContent = nil;
 101                 NSLog(@"We were passed %@, of type %@, which is an unknown type",pathToFile,contentTypeUTI);
 102         }
 103
 104         if (textContent) CFRetain(textContent);
 105         [pool release];
 106
 107         return textContent;
 108 }
 109
 110 /*!
 111  * @brief get metadata for an XML file
 112  *
 113  * This function gets the metadata contained within a universal chat log format file
 114  * @param attributes The dictionary in which to store the metadata
 115  * @param pathToFile The path to the file to index
 116  *
 117  * @result true if successful, false if not
 118  */
 119 Boolean GetMetadataForXMLLog(NSMutableDictionary *attributes, NSString *pathToFile)
 120 {
 121         Boolean ret = YES;
 122         NSXMLDocument *xmlDoc;
 123         NSError *err=nil;
 124         NSURL *furl = [NSURL fileURLWithPath:(NSString *)pathToFile];
 125         xmlDoc = [[NSXMLDocument alloc] initWithContentsOfURL:furl
 126                                                                                                                                                 options:NSXMLNodePreserveCDATA
 127                                                                                                                                                   error:&err];
 128
 129         if (xmlDoc)
 130         {
 131                 NSArray *authorsArray = [xmlDoc nodesForXPath:@"//message/@sender"
 132                                                                                                 error:&err];
 133                 NSSet *duplicatesRemover = [NSSet setWithArray: authorsArray];
 134                 authorsArray = [duplicatesRemover allObjects];
 135
 136                 [(NSMutableDictionary *)attributes setObject:authorsArray
 137                                                                                           forKey:(NSString *)kMDItemAuthors];
 138
 139                 NSArray *contentArray = [xmlDoc nodesForXPath:@"//message//text()"
 140                                                                                                 error:&err];
 141                 NSString *contentString = [contentArray componentsJoinedByString:@" "];
 142
 143                 [attributes setObject:contentString
 144                                            forKey:(NSString *)kMDItemTextContent];
 145
 146                 NSString *serviceString = [[[xmlDoc rootElement] attributeForName:@"service"] objectValue];
 147                 if(serviceString != nil)
 148                         [attributes setObject:serviceString
 149                                                    forKey:@"com_adiumX_service"];
 150
 151                 NSArray                 *children = [[xmlDoc rootElement] children];
 152                 NSCalendarDate  *startDate = nil, *endDate = nil;
 153
 154                 if ([children count]) {
 155                         NSString                *dateStr;
 156
 157                         dateStr = [[(NSXMLElement *)[children objectAtIndex:0] attributeForName:@"time"] objectValue];
 158                         startDate = (dateStr ? [NSCalendarDate calendarDateWithString:dateStr] : nil);
 159                         if (startDate)
 160                                 [(NSMutableDictionary *)attributes setObject:startDate
 161                                                                                                           forKey:(NSString *)kMDItemContentCreationDate];
 162
 163                         dateStr = [[(NSXMLElement *)[children objectAtIndex:0] attributeForName:@"time"] objectValue];
 164                         endDate = (dateStr ? [NSCalendarDate calendarDateWithString:dateStr] : nil);
 165                         if (endDate)
 166                                 [(NSMutableDictionary *)attributes setObject:[NSNumber numberWithInt:[endDate timeIntervalSinceDate:startDate]]
 167                                                                                                           forKey:(NSString *)kMDItemDurationSeconds];
 168                 }
 169
 170                 NSString *accountString = [[[xmlDoc rootElement] attributeForName:@"account"] objectValue];
 171                 if (accountString) {
 172                         [attributes setObject:accountString
 173                                                    forKey:@"com_adiumX_chatSource"];
 174                         NSMutableArray *otherAuthors = [authorsArray mutableCopy];
 175                         [otherAuthors removeObject:accountString];
 176                         //pick the first author for this.  likely a bad idea
 177                         if (startDate && [otherAuthors count]) {
 178                                 NSString *toUID = [otherAuthors objectAtIndex:0];
 179                                 [attributes setObject:[NSString stringWithFormat:@"%@ on %@",toUID,[startDate descriptionWithCalendarFormat:@"%y-%m-%d"
 180                                                                                                                                                                                                                                    timeZone:nil
 181                                                                                                                                                                                                                                          locale:nil]]
 182                                                            forKey:(NSString *)kMDItemDisplayName];
 183                         }
 184                         [otherAuthors release];
 185
 186                 }
 187                 [attributes setObject:@"Chat log"
 188                                            forKey:(NSString *)kMDItemKind];
 189
 190                 [xmlDoc release];
 191         }
 192         else
 193                 ret = NO;
 194
 195         return ret;
 196 }
 197
 198 NSString *killXMLTags(NSString *inString)
 199 {
 200     NSScanner *scan = [NSScanner scannerWithString:inString];
 201     NSString *tempString = nil;
 202     NSMutableString *ret = [NSMutableString string];
 203     NSRange rng;
 204
 205     while(![scan isAtEnd]){
 206         tempString = nil;
 207         [scan scanUpToString:@"<" intoString:&tempString];
 208         if(tempString != nil)
 209             [ret appendFormat:@"%@ ", tempString];
 210         [scan scanString:@"<" intoString:nil];
 211         [scan scanUpToString:@">" intoString:&tempString];
 212         if([tempString hasPrefix:@"br"])
 213             [ret appendString:@"\n"];
 214         [scan scanString:@">" intoString:nil];
 215     }
 216     rng.location = -1;
 217     do {
 218         NSRange searchRange = NSMakeRange(rng.location+1, [ret length]-rng.location-1);
 219         rng = [ret rangeOfString:@"&lt;" options:0 range:searchRange];
 220         if (rng.length > 0)
 221             [ret replaceCharactersInRange: rng withString: @"<"];
 222     } while (rng.length > 0);
 223     rng.location = -1;
 224     do {
 225         NSRange searchRange = NSMakeRange(rng.location+1, [ret length]-rng.location-1);
 226         rng = [ret rangeOfString:@"&gt;" options:0 range:searchRange];
 227         if (rng.length > 0)
 228             [ret replaceCharactersInRange: rng withString: @">"];
 229     } while (rng.length > 0);
 230     rng.location = -1;
 231     do {
 232         NSRange searchRange = NSMakeRange(rng.location+1, [ret length]-rng.location-1);
 233         rng = [ret rangeOfString:@"&amp;" options:0 range:searchRange];
 234         if (rng.length > 0)
 235             [ret replaceCharactersInRange: rng withString: @"&"];
 236     } while (rng.length > 0);
 237     return ret;
 238 }
 239
 240 NSString *GetTextContentForXMLLog(NSString *pathToFile)
 241 {
 242         NSError *err=nil;
 243         NSURL *furl = [NSURL fileURLWithPath:(NSString *)pathToFile];
 244         NSString *contentString;
 245         NSXMLDocument *xmlDoc;
 246         xmlDoc = [[NSXMLDocument alloc] initWithContentsOfURL:furl
 247                                                                                                   options:NSXMLNodePreserveCDATA
 248                                                                                                         error:&err];
 249         NSArray *contentArray = [xmlDoc nodesForXPath:@"//message//text()"
 250                                                                                         error:&err];
 251         contentString = [contentArray componentsJoinedByString:@" "];
 252         [xmlDoc release];
 253         return contentString;
 254 }