[jbossweb-commits] JBossWeb SVN: r284 - trunk/java/org/apache/tomcat/util/http.

Author: remy.maucherat(a)jboss.com Date: 2007-09-25 18:31:25 -0400 (Tue, 25 Sep 2007) New Revision: 284 Modified: trunk/java/org/apache/tomcat/util/http/Cookies.java Log: - Cookie parser update. - Remove debugger style logging. Modified: trunk/java/org/apache/tomcat/util/http/Cookies.java =================================================================== --- trunk/java/org/apache/tomcat/util/http/Cookies.java 2007-09-25 10:28:47 UTC (rev 283) +++ trunk/java/org/apache/tomcat/util/http/Cookies.java 2007-09-25 22:31:25 UTC (rev 284) @@ -35,9 +35,6 @@ */ public final class Cookies { // extends MultiMap { - private static org.jboss.logging.Logger log= - org.jboss.logging.Logger.getLogger(Cookies.class ); - // expected average number of cookies per request public static final int INITIAL_SIZE=4; ServerCookie scookies[]=new ServerCookie[INITIAL_SIZE]; @@ -45,7 +42,28 @@ boolean unprocessed=true; MimeHeaders headers; - + + /* + List of Separator Characters (see isSeparator()) + Excluding the '/' char violates the RFC, but + it looks like a lot of people put '/' + in unquoted values: '/': ; //47 + '\t':9 ' ':32 '\"':34 '\'':39 '(':40 ')':41 ',':44 ':':58 ';':59 '<':60 + '=':61 '>':62 '?':63 '@':64 '[':91 '\\':92 ']':93 '{':123 '}':125 + */ + public static final char SEPARATORS[] = { '\t', ' ', '\"', '\'', '(', ')', ',', + ':', ';', '<', '=', '>', '?', '@', '[', '\\', ']', '{', '}' }; + + protected static final boolean separators[] = new boolean[128]; + static { + for (int i = 0; i < 128; i++) { + separators[i] = false; + } + for (int i = 0; i < SEPARATORS.length; i++) { + separators[SEPARATORS[i]] = true; + } + } + /** * Construct a new cookie collection, that will extract * the information from headers. @@ -169,194 +187,17 @@ // Uncomment to test the new parsing code if( cookieValue.getType() == MessageBytes.T_BYTES ) { - if( dbg>0 ) log( "Parsing b[]: " + cookieValue.toString()); ByteChunk bc=cookieValue.getByteChunk(); processCookieHeader( bc.getBytes(), bc.getOffset(), bc.getLength()); } else { - if( dbg>0 ) log( "Parsing S: " + cookieValue.toString()); processCookieHeader( cookieValue.toString() ); } pos++;// search from the next position } } - /** Process a byte[] header - allowing fast processing of the - * raw data - */ - void processCookieHeader( byte bytes[], int off, int len ) - { - if( len<=0 || bytes==null ) return; - int end=off+len; - int pos=off; - - int version=0; //sticky - ServerCookie sc=null; - - - while( pos<end ) { - byte cc; - // [ skip_spaces name skip_spaces "=" skip_spaces value EXTRA ; ] * - if( dbg>0 ) log( "Start: " + pos + " " + end ); - - pos=skipSpaces(bytes, pos, end); - if( pos>=end ) - return; // only spaces - int startName=pos; - if( dbg>0 ) log( "SN: " + pos ); - - // Version should be the first token - boolean isSpecial=false; - if(bytes[pos]=='$') { pos++; isSpecial=true; } - - pos= findDelim1( bytes, startName, end); // " =;," - int endName=pos; - // current = "=" or " " or DELIM - pos= skipSpaces( bytes, endName, end ); - if( dbg>0 ) log( "DELIM: " + endName + " " + (char)bytes[pos]); - - if(pos >= end ) { - // it's a name-only cookie ( valid in RFC2109 ) - if( ! isSpecial ) { - sc=addCookie(); - sc.getName().setBytes( bytes, startName, - endName-startName ); - sc.getValue().setString(""); - sc.setVersion( version ); - if( dbg>0 ) log( "Name only, end: " + startName + " " + - endName); - } - return; - } - - cc=bytes[pos]; - pos++; - if( cc==';' || cc==',' || pos>=end ) { - if( ! isSpecial && startName!= endName ) { - sc=addCookie(); - sc.getName().setBytes( bytes, startName, - endName-startName ); - sc.getValue().setString(""); - sc.setVersion( version ); - if( dbg>0 ) log( "Name only: " + startName + " " + endName); - } - continue; - } - - // we should have "=" ( tested all other alternatives ) - int startValue=skipSpaces( bytes, pos, end); - int endValue=startValue; - - cc=bytes[pos]; - if( cc=='"' ) { - endValue=findDelim3( bytes, startValue+1, end, cc ); - if (endValue == -1) { - endValue=findDelim2( bytes, startValue+1, end ); - } else startValue++; - pos=endValue+1; // to skip to next cookie - } else { - endValue=findDelim2( bytes, startValue, end ); - pos=endValue+1; - } - - // if not $Version, etc - if( ! isSpecial ) { - sc=addCookie(); - sc.getName().setBytes( bytes, startName, endName-startName ); - sc.getValue().setBytes( bytes, startValue, endValue-startValue); - sc.setVersion( version ); - if( dbg>0 ) { - log( "New: " + sc.getName() + "X=X" + sc.getValue()); - } - continue; - } - - // special - Path, Version, Domain, Port - if( dbg>0 ) log( "Special: " + startName + " " + endName); - // XXX TODO - if( equals( "$Version", bytes, startName, endName ) ) { - if(dbg>0 ) log( "Found version " ); - if( bytes[startValue]=='1' && endValue==startValue+1 ) { - version=1; - if(dbg>0 ) log( "Found version=1" ); - } - continue; - } - if( sc==null ) { - // Path, etc without a previous cookie - continue; - } - if( equals( "$Path", bytes, startName, endName ) ) { - sc.getPath().setBytes( bytes, - startValue, - endValue-startValue ); - } - if( equals( "$Domain", bytes, startName, endName ) ) { - sc.getDomain().setBytes( bytes, - startValue, - endValue-startValue ); - } - if( equals( "$Port", bytes, startName, endName ) ) { - // sc.getPort().setBytes( bytes, - // startValue, - // endValue-startValue ); - } - } - } - - // -------------------- Utils -------------------- - public static int skipSpaces( byte bytes[], int off, int end ) { - while( off < end ) { - byte b=bytes[off]; - if( b!= ' ' ) return off; - off ++; - } - return off; - } - - public static int findDelim1( byte bytes[], int off, int end ) - { - while( off < end ) { - byte b=bytes[off]; - if( b==' ' || b=='=' || b==';' || b==',' ) - return off; - off++; - } - return off; - } - - public static int findDelim2( byte bytes[], int off, int end ) - { - while( off < end ) { - byte b=bytes[off]; - if( b==';' || b==',' ) - return off; - off++; - } - return off; - } - - /* - * search for cc but skip \cc as required by rfc2616 - * (according to rfc2616 cc should be ") - */ - public static int findDelim3( byte bytes[], int off, int end, byte cc ) - { - while( off < end ) { - byte b=bytes[off]; - if ( b== '\\' ) { - off++; - off++; - continue; - } - if( b==cc ) - return off; - off++; - } - return -1; - } - // XXX will be refactored soon! public static boolean equals( String s, byte b[], int start, int end) { int blen = end-start; @@ -378,7 +219,7 @@ private void processCookieHeader( String cookieString ) { - if( dbg>0 ) log( "Parsing cookie header " + cookieString ); + // normal cookie, with a string value. // This is the original code, un-optimized - it shouldn't // happen in normal case @@ -402,7 +243,7 @@ cookie.getName().setString(name); cookie.getValue().setString(value); - if( dbg > 0 ) log( "Add cookie " + name + "=" + value); + } else { // we have a bad cookie.... just let it go } @@ -433,49 +274,291 @@ } - // log - static final int dbg=0; - public void log(String s ) { - if (log.isDebugEnabled()) - log.debug("Cookies: " + s); + /** + * Returns true if the byte is a separator character as + * defined in RFC2619. Since this is called often, this + * function should be organized with the most probable + * outcomes first. + * JVK + */ + public static final boolean isSeparator(final byte c) { + if (c > 0 && c < 126) + return separators[c]; + else + return false; } + + /** + * Returns true if the byte is a whitespace character as + * defined in RFC2619 + * JVK + */ + public static final boolean isWhiteSpace(final byte c) { + // This switch statement is slightly slower + // for my vm than the if statement. + // Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0_07-164) + /* + switch (c) { + case ' ':; + case '\t':; + case '\n':; + case '\r':; + case '\f':; + return true; + default:; + return false; + } + */ + if (c == ' ' || c == '\t' || c == '\n' || c == '\r' || c == '\f') + return true; + else + return false; + } - /* - public static void main( String args[] ) { - test("foo=bar; a=b"); - test("foo=bar;a=b"); - test("foo=bar;a=b;"); - test("foo=bar;a=b; "); - test("foo=bar;a=b; ;"); - test("foo=;a=b; ;"); - test("foo;a=b; ;"); - // v1 - test("$Version=1; foo=bar;a=b"); - test("$Version=\"1\"; foo='bar'; $Path=/path; $Domain=\"localhost\""); - test("$Version=1;foo=bar;a=b; ; "); - test("$Version=1;foo=;a=b; ; "); - test("$Version=1;foo= ;a=b; ; "); - test("$Version=1;foo;a=b; ; "); - test("$Version=1;foo=\"bar\";a=b; ; "); - test("$Version=1;foo=\"bar\";$Path=/examples;a=b; ; "); - test("$Version=1;foo=\"bar\";$Domain=apache.org;a=b"); - test("$Version=1;foo=\"bar\";$Domain=apache.org;a=b;$Domain=yahoo.com"); - // rfc2965 - test("$Version=1;foo=\"bar\";$Domain=apache.org;$Port=8080;a=b"); + /** + * Parses a cookie header after the initial "Cookie:" + * [WS][$]token[WS]=[WS](token|QV)[;|,] + * RFC 2965 + * JVK + */ + public final void processCookieHeader(byte bytes[], int off, int len){ + if( len<=0 || bytes==null ) return; + int end=off+len; + int pos=off; + int nameStart=0; + int nameEnd=0; + int valueStart=0; + int valueEnd=0; + int version = 0; + ServerCookie sc=null; + boolean isSpecial; - // wrong - test("$Version=1;foo=\"bar\";$Domain=apache.org;$Port=8080;a=b"); + while (pos < end) { + isSpecial = false; + + // Skip whitespace and non-token characters (separators) + while (pos < end && + (isSeparator(bytes[pos]) || isWhiteSpace(bytes[pos]))) + {pos++; } + + if (pos >= end) + return; + + // Detect Special cookies + if (bytes[pos] == '$') { + isSpecial = true; + pos++; + } + + // Get the cookie name. This must be a token + valueEnd = valueStart = nameStart = pos; + pos = nameEnd = getTokenEndPosition(bytes,pos,end); + + // Skip whitespace + while (pos < end && isWhiteSpace(bytes[pos])) {pos++; }; + + + // Check for an '=' -- This could also be a name-only + // cookie at the end of the cookie header, so if we + // are past the end of the header, but we have a name + // skip to the name-only part. + if (pos < end && bytes[pos] == '=') { + + // Skip whitespace + do { + pos++; + } while (pos < end && isWhiteSpace(bytes[pos])); + + if (pos >= end) + return; + + // Determine what type of value this is, quoted value, + // token, name-only with an '=', or other (bad) + switch (bytes[pos]) { + case '"':; // Quoted Value + valueStart=pos + 1; // strip " + // getQuotedValue returns the position before + // at the last qoute. This must be dealt with + // when the bytes are copied into the cookie + valueEnd=getQuotedValueEndPosition(bytes, + valueStart, end); + // We need pos to advance + pos = valueEnd; + // Handles cases where the quoted value is + // unterminated and at the end of the header, + // e.g. [myname="value] + if (pos >= end) + return; + break; + case ';': + case ',': + // Name-only cookie with an '=' after the name token + // This may not be RFC compliant + valueStart = valueEnd = -1; + // The position is OK (On a delimiter) + break; + default:; + if (!isSeparator(bytes[pos])) { + // Token + valueStart=pos; + // getToken returns the position at the delimeter + // or other non-token character + valueEnd=getTokenEndPosition(bytes, valueStart, end); + // We need pos to advance + pos = valueEnd; + } else { + // INVALID COOKIE, advance to next delimiter + // The starting character of the cookie value was + // not valid. + while (pos < end && bytes[pos] != ';' && + bytes[pos] != ',') + {pos++; }; + pos++; + // Make sure no special avpairs can be attributed to + // the previous cookie by setting the current cookie + // to null + sc = null; + continue; + } + } + } else { + // Name only cookie + valueStart = valueEnd = -1; + pos = nameEnd; + + } + + // We should have an avpair or name-only cookie at this + // point. Perform some basic checks to make sure we are + // in a good state. + + // Skip whitespace + while (pos < end && isWhiteSpace(bytes[pos])) {pos++; }; + + + // Make sure that after the cookie we have a separator. This + // is only important if this is not the last cookie pair + while (pos < end && bytes[pos] != ';' && bytes[pos] != ',') { + pos++; + } + + pos++; + + /* + if (nameEnd <= nameStart || valueEnd < valueStart ) { + // Something is wrong, but this may be a case + // of having two ';' characters in a row. + // log("Cookie name/value does not conform to RFC 2965"); + // Advance to next delimiter (ignoring everything else) + while (pos < end && bytes[pos] != ';' && bytes[pos] != ',') + { pos++; }; + pos++; + // Make sure no special cookies can be attributed to + // the previous cookie by setting the current cookie + // to null + sc = null; + continue; + } + */ + + // All checks passed. Add the cookie, start with the + // special avpairs first + if (isSpecial) { + isSpecial = false; + // $Version must be the first avpair in the cookie header + // (sc must be null) + if (equals( "Version", bytes, nameStart, nameEnd) && + sc == null) { + // Set version + if( bytes[valueStart] =='1' && valueEnd == valueStart) { + version=1; + } else { + // unknown version (Versioning is not very strict) + } + continue; + } + + // We need an active cookie for Path/Port/etc. + if (sc == null) { + continue; + } + + // Domain is more common, so it goes first + if (equals( "Domain", bytes, nameStart, nameEnd)) { + sc.getDomain().setBytes( bytes, + valueStart, + valueEnd-valueStart); + continue; + } + + if (equals( "Path", bytes, nameStart, nameEnd)) { + sc.getPath().setBytes( bytes, + valueStart, + valueEnd-valueStart); + continue; + } + + + if (equals( "Port", bytes, nameStart, nameEnd)) { + // sc.getPort is not currently implemented. + // sc.getPort().setBytes( bytes, + // valueStart, + // valueEnd-valueStart ); + continue; + } + + // Unknown cookie + + } else { // Normal Cookie + sc = addCookie(); + sc.setVersion( version ); + sc.getName().setBytes( bytes, nameStart, + nameEnd-nameStart); + + if (valueStart != -1) { // Normal AVPair + sc.getValue().setBytes( bytes, valueStart, + valueEnd-valueStart); + } else { + // Name Only + sc.getValue().setString(""); + } + continue; + } + } } - public static void test( String s ) { - System.out.println("Processing " + s ); - Cookies cs=new Cookies(null); - cs.processCookieHeader( s.getBytes(), 0, s.length()); - for( int i=0; i< cs.getCookieCount() ; i++ ) { - System.out.println("Cookie: " + cs.getCookie( i )); + /** + * Given the starting position of a token, this gets the end of the + * token, with no separator characters in between. + * JVK + */ + public static final int getTokenEndPosition(byte bytes[], int off, int end){ + int pos = off; + while (pos < end && !isSeparator(bytes[pos])) {pos++; }; + + if (pos > end) + return end; + return pos; + } + + /** + * Given a starting position after an initial quote chracter, this gets + * the position of the end quote. This escapes anything after a '\' char + * JVK RFC 2616 + */ + public static final int getQuotedValueEndPosition(byte bytes[], int off, int end){ + int pos = off; + while (pos < end) { + if (bytes[pos] == '"') { + return pos; + } else if (bytes[pos] == '\\' && pos < (end - 1)) { + pos+=2; + } else { + pos++; + } } - + // Error, we have reached the end of the header w/o a end quote + return end; } - */ }