Many resources are needed to download a project. Please understand that we have to compensate our server costs. Thank you in advance. Project price only 1 $
You can buy this project and download/modify it how often you want.
/*
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS HEADER.
*
* Copyright 1997-2008 Sun Microsystems, Inc. All rights reserved.
*
* The contents of this file are subject to the terms of either the GNU
* General Public License Version 2 only ("GPL") or the Common Development
* and Distribution License("CDDL") (collectively, the "License"). You
* may not use this file except in compliance with the License. You can obtain
* a copy of the License at https://glassfish.dev.java.net/public/CDDL+GPL.html
* or glassfish/bootstrap/legal/LICENSE.txt. See the License for the specific
* language governing permissions and limitations under the License.
*
* When distributing the software, include this License Header Notice in each
* file and include the License file at glassfish/bootstrap/legal/LICENSE.txt.
* Sun designates this particular file as subject to the "Classpath" exception
* as provided by Sun in the GPL Version 2 section of the License file that
* accompanied this code. If applicable, add the following below the License
* Header, with the fields enclosed by brackets [] replaced by your own
* identifying information: "Portions Copyrighted [year]
* [name of copyright owner]"
*
* Contributor(s):
*
* If you wish your version of this file to be governed by only the CDDL or
* only the GPL Version 2, indicate your decision by adding "[Contributor]
* elects to include this software in this distribution under the [CDDL or GPL
* Version 2] license." If you don't indicate a single choice of license, a
* recipient has the option to distribute your version of this file under
* either the CDDL, the GPL Version 2 or to extend the choice of license to
* its licensees as provided above. However, if you add GPL Version 2 code
* and therefore, elected the GPL Version 2 license, then the option applies
* only if the new code is made subject to such option by the copyright
* holder.
*
*
* This file incorporates work covered by the following copyright and
* permission notice:
*
* Copyright 2004 The Apache Software Foundation
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package com.sun.grizzly.util.http;
import java.io.PrintWriter;
import java.io.StringWriter;
import java.util.StringTokenizer;
import com.sun.grizzly.util.buf.ByteChunk;
import com.sun.grizzly.util.buf.MessageBytes;
import java.util.logging.Level;
import java.util.logging.Logger;
/**
* A collection of cookies - reusable and tuned for server side performance.
* Based on RFC2965 ( and 2109 )
*
* This class is not synchronized.
*
* @author Costin Manolache
* @author kevin seguin
*/
public final class Cookies { // extends MultiMap {
private static Logger logger = Logger.getLogger("Grizzly");
// expected average number of cookies per request
public static final int INITIAL_SIZE=4;
ServerCookie scookies[]=new ServerCookie[INITIAL_SIZE];
int cookieCount=0;
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.
*
* @param headers Cookies are lazy-evaluated and will extract the
* information from the provided headers.
*/
public Cookies(MimeHeaders headers) {
this.headers=headers;
}
/**
* Construct a new uninitialized cookie collection.
* Use {@link #setHeaders} to initialize.
*/
// [seguin] added so that an empty Cookies object could be
// created, have headers set, then recycled.
public Cookies() {
}
/**
* Set the headers from which cookies will be pulled.
* This has the side effect of recycling the object.
*
* @param headers Cookies are lazy-evaluated and will extract the
* information from the provided headers.
*/
// [seguin] added so that an empty Cookies object could be
// created, have headers set, then recycled.
public void setHeaders(MimeHeaders headers) {
recycle();
this.headers=headers;
}
/**
* Recycle.
*/
public void recycle() {
for( int i=0; i< cookieCount; i++ ) {
if( scookies[i]!=null )
scookies[i].recycle();
}
cookieCount=0;
unprocessed=true;
}
/**
* EXPENSIVE!!! only for debugging.
*/
public String toString() {
StringWriter sw = new StringWriter();
PrintWriter pw = new PrintWriter(sw);
pw.println("=== Cookies ===");
int count = getCookieCount();
for (int i = 0; i < count; ++i) {
pw.println(getCookie(i).toString());
}
return sw.toString();
}
// -------------------- Indexed access --------------------
public ServerCookie getCookie( int idx ) {
if( unprocessed ) {
getCookieCount(); // will also update the cookies
}
return scookies[idx];
}
public int getCookieCount() {
if( unprocessed ) {
processCookies(headers);
}
return cookieCount;
}
// -------------------- Adding cookies --------------------
/** Register a new, unitialized cookie. Cookies are recycled, and
* most of the time an existing ServerCookie object is returned.
* The caller can set the name/value and attributes for the cookie
*/
public ServerCookie addCookie() {
if( cookieCount >= scookies.length ) {
ServerCookie scookiesTmp[]=new ServerCookie[2*cookieCount];
System.arraycopy( scookies, 0, scookiesTmp, 0, cookieCount);
scookies=scookiesTmp;
}
ServerCookie c = scookies[cookieCount];
if( c==null ) {
c= new ServerCookie();
scookies[cookieCount]=c;
}
cookieCount++;
return c;
}
// code from CookieTools
/** Add all Cookie found in the headers of a request.
*/
public void processCookies( MimeHeaders headers ) {
unprocessed=false;
if( headers==null )
return;// nothing to process
// process each "cookie" header
int pos=0;
while( pos>=0 ) {
// Cookie2: version ? not needed
pos=headers.findHeader( "Cookie", pos );
// no more cookie headers headers
if( pos<0 ) break;
MessageBytes cookieValue=headers.getValue( pos );
if( cookieValue==null || cookieValue.isNull() ) {
pos++;
continue;
}
// 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
}
}
// XXX will be refactored soon!
public static boolean equals( String s, byte b[], int start, int end) {
int blen = end-start;
if (b == null || blen != s.length()) {
return false;
}
int boff = start;
for (int i = 0; i < blen; i++) {
if (b[boff++] != s.charAt(i)) {
return false;
}
}
return true;
}
// ---------------------------------------------------------
// -------------------- DEPRECATED, OLD --------------------
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
StringTokenizer tok = new StringTokenizer(cookieString,
";", false);
while (tok.hasMoreTokens()) {
String token = tok.nextToken();
int i = token.indexOf("=");
if (i > -1) {
// XXX
// the trims here are a *hack* -- this should
// be more properly fixed to be spec compliant
String name = token.substring(0, i).trim();
String value = token.substring(i+1, token.length()).trim();
// RFC 2109 and bug
value=stripQuote( value );
ServerCookie cookie = addCookie();
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
}
}
}
/**
*
* Strips quotes from the start and end of the cookie string
* This conforms to RFC 2109
*
* @param value a String specifying the cookie
* value (possibly quoted).
*
* @see #setValue
*
*/
private static String stripQuote( String value )
{
// log("Strip quote from " + value );
if (((value.startsWith("\"")) && (value.endsWith("\""))) ||
((value.startsWith("'") && (value.endsWith("'"))))) {
try {
return value.substring(1,value.length()-1);
} catch (Exception ex) {
}
}
return value;
}
// log
static final int dbg=0;
public void log(String s ) {
if (logger.isLoggable(Level.FINE))
logger.fine("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.
*/
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.
*/
public static final boolean isWhiteSpace(final byte c) {
return (c == ' ' || c == '\t' || c == '\n' || c == '\r' || c == '\f');
}
/**
* Parses a cookie header after the initial "Cookie:"
* [WS][$]token[WS]=[WS](token|QV)[;|,]
* RFC 2965
* JVK
*/
public final void processCookieHeader(byte bytes[], int off, int len){
if( len<=0 || bytes==null ) return;
int end=off+len;
int pos=off;
int nameStart=0;
int nameEnd=0;
int valueStart=0;
int valueEnd=0;
int version = 0;
ServerCookie sc=null;
boolean isSpecial;
boolean isQuoted;
while (pos < end) {
isSpecial = false;
isQuoted = false;
// Skip whitespace and non-token characters (separators)
while (pos < end &&
(isSeparator(bytes[pos]) || isWhiteSpace(bytes[pos]))) {
pos++;
}
if (pos >= end) {
return;
}
// Detect Special cookies
if (bytes[pos] == '$') {
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
isQuoted = true;
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.
log("Invalid cookie. Value not a token or quoted value");
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 + 1)) {
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, complain
log("Unknown Special 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);
if (isQuoted) {
// We know this is a byte value so this is safe
ServerCookie.unescapeDoubleQuotes(
sc.getValue().getByteChunk());
}
} else {
// Name Only
sc.getValue().setString("");
}
continue;
}
}
}
/**
* 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;
}
}