Be more memory-efficient when parsing long strings containing escaped…

… characters. Suggested by Bob McCune.
OldFrank · Aug 27, 2011 · aaa6b06 · aaa6b06
1 parent f723881
commit aaa6b06
Show file tree

Hide file tree

Showing 4 changed files with 118 additions and 105 deletions.
diff --git a/Classes/SBJsonStreamParser.m b/Classes/SBJsonStreamParser.m
@@ -264,90 +264,97 @@ - (void) handleTokenNotExpectedHere: (sbjson_token_t) tok  {
 }
 
 - (SBJsonStreamParserStatus)parse:(NSData *)data_ {
-	[tokeniser appendData:data_];
-
-	for (;;) {
-
-        if ([state isError])
-            return SBJsonStreamParserError;
-
-        NSObject *token;
-		sbjson_token_t tok = [tokeniser getToken:&token];
-		switch (tok) {
-			case sbjson_token_eof:
-                return [state parserShouldReturn:self];
-				break;
-
-			case sbjson_token_error:
-				self.state = [SBJsonStreamParserStateError sharedInstance];
-				self.error = tokeniser.error;
-				return SBJsonStreamParserError;
-				break;
-
-			default:
-
-				if (![state parser:self shouldAcceptToken:tok]) {
-                    [self handleTokenNotExpectedHere: tok];
-					return SBJsonStreamParserError;
-				}
-
-				switch (tok) {
-					case sbjson_token_object_start:
-						[self handleObjectStart];
-						break;
-
-					case sbjson_token_object_end:
-                        [self handleObjectEnd: tok];
-						break;
-
-					case sbjson_token_array_start:
-						[self handleArrayStart];
-						break;
-
-					case sbjson_token_array_end:
-                        [self handleArrayEnd: tok];
-						break;
-
-					case sbjson_token_separator:
-					case sbjson_token_keyval_separator:
-						[state parser:self shouldTransitionTo:tok];
-						break;
-
-					case sbjson_token_true:
-                        [self parserFoundObject:kTrue];
-						[state parser:self shouldTransitionTo:tok];
-						break;
-
-					case sbjson_token_false:
-                        [self parserFoundObject:kFalse];
-						[state parser:self shouldTransitionTo:tok];
-						break;
-
-					case sbjson_token_null:
-                        [self parserFoundObject:kNull];
-						[state parser:self shouldTransitionTo:tok];
-						break;
-
-					case sbjson_token_number:
-                        [self parserFoundObject:token];
-						[state parser:self shouldTransitionTo:tok];
-						break;
-
-					case sbjson_token_string:
-                        if ([state needKey])
-                            [keyStack addObject:token];
-                        else
+    NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
+    @try {
+        [tokeniser appendData:data_];
+
+        for (;;) {
+
+            if ([state isError])
+                return SBJsonStreamParserError;
+
+            NSObject *token;
+            sbjson_token_t tok = [tokeniser getToken:&token];
+            switch (tok) {
+                case sbjson_token_eof:
+                    return [state parserShouldReturn:self];
+                    break;
+
+                case sbjson_token_error:
+                    self.state = [SBJsonStreamParserStateError sharedInstance];
+                    self.error = tokeniser.error;
+                    return SBJsonStreamParserError;
+                    break;
+
+                default:
+
+                    if (![state parser:self shouldAcceptToken:tok]) {
+                        [self handleTokenNotExpectedHere: tok];
+                        return SBJsonStreamParserError;
+                    }
+
+                    switch (tok) {
+                        case sbjson_token_object_start:
+                            [self handleObjectStart];
+                            break;
+
+                        case sbjson_token_object_end:
+                            [self handleObjectEnd: tok];
+                            break;
+
+                        case sbjson_token_array_start:
+                            [self handleArrayStart];
+                            break;
+
+                        case sbjson_token_array_end:
+                            [self handleArrayEnd: tok];
+                            break;
+
+                        case sbjson_token_separator:
+                        case sbjson_token_keyval_separator:
+                            [state parser:self shouldTransitionTo:tok];
+                            break;
+
+                        case sbjson_token_true:
+                            [self parserFoundObject:kTrue];
+                            [state parser:self shouldTransitionTo:tok];
+                            break;
+
+                        case sbjson_token_false:
+                            [self parserFoundObject:kFalse];
+                            [state parser:self shouldTransitionTo:tok];
+                            break;
+
+                        case sbjson_token_null:
+                            [self parserFoundObject:kNull];
+                            [state parser:self shouldTransitionTo:tok];
+                            break;
+
+                        case sbjson_token_number:
                             [self parserFoundObject:token];
-						[state parser:self shouldTransitionTo:tok];
-						break;
-
-					default:
-						break;
-				}
-				break;
-		}
-	}
-	return SBJsonStreamParserComplete;
+                            [state parser:self shouldTransitionTo:tok];
+                            break;
+
+                        case sbjson_token_string:
+                            if ([state needKey])
+                                [keyStack addObject:token];
+                            else
+                                [self parserFoundObject:token];
+                            [state parser:self shouldTransitionTo:tok];
+                            break;
+
+                        default:
+                            break;
+                    }
+                    break;
+            }
+        }
+        return SBJsonStreamParserComplete;
+
+    }
+    @finally {
+        [pool drain];
+    }
 }
 
 

diff --git a/Classes/SBJsonTokeniser.m b/Classes/SBJsonTokeniser.m
@@ -145,30 +145,36 @@ - (sbjson_token_t)getStringToken:(NSObject**)token {
         unichar ch;
         {
             NSMutableString *string = nil;
-            if (![_stream getSimpleString:&string])
-                return sbjson_token_eof;
+            @try {
+                if (![_stream getRetainedStringFragment:&string])
+                    return sbjson_token_eof;
 
-            if (!string) {
-                self.error = @"Broken Unicode encoding";
-                return sbjson_token_error;
-            }
-
-
-            if (![_stream getUnichar:&ch])
-                return sbjson_token_eof;
-
-            if (acc) {
-                [acc appendString:string];
+                if (!string) {
+                    self.error = @"Broken Unicode encoding";
+                    return sbjson_token_error;
+                }
 
-            } else if (ch == '"') {
-                *token = string;
-                [_stream skip];
-                return sbjson_token_string;
+                if (![_stream getUnichar:&ch]) {
+                    return sbjson_token_eof;
+                }
+
+                if (acc) {
+                    [acc appendString:string];
 
-            } else {
-                acc = [[string mutableCopy] autorelease];
+                } else if (ch == '"') {
+                    *token = [[string copy] autorelease];
+                    [_stream skip];
+                    return sbjson_token_string;
+
+                } else {
+                    acc = [[string mutableCopy] autorelease];
+                }
+            }
+            @finally {
+                [string release];
             }
         }
+
 
         switch (ch) {
             case 0 ... 0x1F:

diff --git a/Classes/SBJsonUTF8Stream.h b/Classes/SBJsonUTF8Stream.h
@@ -51,7 +51,7 @@
 
 - (BOOL)getUnichar:(unichar*)ch;
 - (BOOL)getNextUnichar:(unichar*)ch;
-- (BOOL)getSimpleString:(NSString**)string;
+- (BOOL)getRetainedStringFragment:(NSString**)string;
 
 - (NSString*)stringWithRange:(NSRange)range;
 

diff --git a/Classes/SBJsonUTF8Stream.m b/Classes/SBJsonUTF8Stream.m
@@ -83,14 +83,14 @@ - (BOOL)getNextUnichar:(unichar*)ch {
     return NO;
 }
 
-- (BOOL)getSimpleString:(NSString **)string {
+- (BOOL)getRetainedStringFragment:(NSString **)string {
     NSUInteger start = _index;
     while (_index < _length) {
         switch (_bytes[_index]) {
             case '"':
             case '\\':
             case 0 ... 0x1f:
-                *string = [[[NSString alloc] initWithBytes:(_bytes + start) length:(_index - start) encoding:NSUTF8StringEncoding] autorelease];
+                *string = [[NSString alloc] initWithBytes:(_bytes + start) length:(_index - start) encoding:NSUTF8StringEncoding];
                 return YES;
                 break;
             default: