<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE preface PUBLIC "-//OASIS//DTD DocBook XML V4.4//EN"
                 "http://www.oasis-open.org/docbook/xml/4.4/docbookx.dtd">
<!--
   ====================================================================
   Licensed to the Apache Software Foundation (ASF) under one
   or more contributor license agreements.  See the NOTICE file
   distributed with this work for additional information
   regarding copyright ownership.  The ASF licenses this file
   to you 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.
   ====================================================================

-->
<chapter id="advanced">
    <title>Advanced topics</title>
    <section>
        <title>HTTP message parsing and formatting framework</title>
        <para>
        HTTP message processing framework is designed to be expressive and flexible while remaining
        memory efficient and fast. HttpCore HTTP message processing code achieves near zero
        intermediate garbage and near zero-copy buffering for its parsing and formatting
        operations. The same HTTP message parsing and formatting API and implementations are used
        by both the blocking and non-blocking transport implementations, which helps ensure a
        consistent behavior of HTTP services regardless of the I/O model.
        </para>
        <section>
            <title>HTTP line parsing and formatting</title>
            <para>
            HttpCore utilizes a number of low level components for all its line parsing and
            formatting methods.
            </para>
            <para>
            <classname>CharArrayBuffer</classname> represents a sequence of characters, usually a
            single line in an HTTP message stream such as a request line, a status line or a
            header. Internally <classname>CharArrayBuffer</classname> is backed by an array of
            chars, which can be expanded to accommodate more input if needed. <classname>
            CharArrayBuffer</classname> also provides a number of utility methods for manipulating
            content of the buffer, storing more data and retrieving subsets of data.
            </para>
            <programlisting><![CDATA[
CharArrayBuffer buf = new CharArrayBuffer(64);
buf.append("header:  data ");
int i = buf.indexOf(':');
String s = buf.substringTrimmed(i + 1, buf.length());
System.out.println(s);
System.out.println(s.length());
]]></programlisting>
            <para>stdout &gt;</para>
            <programlisting><![CDATA[
data
4
]]></programlisting>
            <para>
            <classname>ParserCursor</classname> represents a context of a parsing operation: the
            bounds limiting the scope of the parsing operation and the current position the parsing
            operation is expected to start at.
            </para>
            <programlisting><![CDATA[
CharArrayBuffer buf = new CharArrayBuffer(64);
buf.append("header:  data ");
int i = buf.indexOf(':');
ParserCursor cursor = new ParserCursor(0, buf.length());
cursor.updatePos(i + 1);
System.out.println(cursor);
]]></programlisting>
            <para>stdout &gt;</para>
            <programlisting><![CDATA[
[0>7>14]
]]></programlisting>
            <para>
            <interfacename>LineParser</interfacename> is the interface for parsing lines in the
            head section of an HTTP message. There are individual methods for parsing a request
            line, a status line, or a header line. The lines to parse are passed in-memory, the
            parser does not depend on any specific I/O mechanism.
            </para>
            <programlisting><![CDATA[
CharArrayBuffer buf = new CharArrayBuffer(64);
buf.append("HTTP/1.1 200");
ParserCursor cursor = new ParserCursor(0, buf.length());

LineParser parser = new BasicLineParser();
ProtocolVersion ver = parser.parseProtocolVersion(buf, cursor);
System.out.println(ver);
System.out.println(buf.substringTrimmed(
    cursor.getPos(),
    cursor.getUpperBound()));
]]></programlisting>
            <para>stdout &gt;</para>
            <programlisting><![CDATA[
HTTP/1.1
200
]]></programlisting>
            <programlisting><![CDATA[
CharArrayBuffer buf = new CharArrayBuffer(64);
buf.append("HTTP/1.1 200 OK");
ParserCursor cursor = new ParserCursor(0, buf.length());
LineParser parser = new BasicLineParser();
StatusLine sl = parser.parseStatusLine(buf, cursor);
System.out.println(sl.getReasonPhrase());
]]></programlisting>
            <para>stdout &gt;</para>
            <programlisting><![CDATA[
OK
]]></programlisting>
            <para>
            <interfacename>LineFormatter</interfacename> for formatting elements of the head
            section of an HTTP message. This is the complement to <interfacename>LineParser
            </interfacename>. There are individual methods for formatting a request line, a status
            line, or a header line.
            </para>
            <para>
            Please note the formatting does not include the trailing line break sequence
            <literal>CR-LF</literal>.
            </para>
            <programlisting><![CDATA[
CharArrayBuffer buf = new CharArrayBuffer(64);
LineFormatter formatter = new BasicLineFormatter();
formatter.formatRequestLine(buf,
    new BasicRequestLine("GET", "/", HttpVersion.HTTP_1_1));
System.out.println(buf.toString());
formatter.formatHeader(buf,
    new BasicHeader("Content-Type", "text/plain"));
System.out.println(buf.toString());
]]></programlisting>
            <para>stdout &gt;</para>
            <programlisting><![CDATA[
GET / HTTP/1.1
Content-Type: text/plain
]]></programlisting>
            <para>
            <interfacename>HeaderValueParser</interfacename> is the interface for parsing header
            values into elements.
            </para>
            <programlisting><![CDATA[
CharArrayBuffer buf = new CharArrayBuffer(64);
HeaderValueParser parser = new BasicHeaderValueParser();
buf.append("name1=value1; param1=p1, " +
    "name2 = \"value2\", name3  = value3");
ParserCursor cursor = new ParserCursor(0, buf.length());
System.out.println(parser.parseHeaderElement(buf, cursor));
System.out.println(parser.parseHeaderElement(buf, cursor));
System.out.println(parser.parseHeaderElement(buf, cursor));
]]></programlisting>
            <para>stdout &gt;</para>
            <programlisting><![CDATA[
name1=value1; param1=p1
name2=value2
name3=value3
]]></programlisting>
            <para>
            <interfacename>HeaderValueFormatter</interfacename> is the interface for formatting
            elements of a header value. This is the complement to <interfacename>HeaderValueParser
            </interfacename>.
            </para>
            <programlisting><![CDATA[
CharArrayBuffer buf = new CharArrayBuffer(64);
HeaderValueFormatter formatter = new BasicHeaderValueFormatter();
HeaderElement[] hes = new HeaderElement[] {
        new BasicHeaderElement("name1", "value1",
                new NameValuePair[] {
                    new BasicNameValuePair("param1", "p1")} ),
        new BasicHeaderElement("name2", "value2"),
        new BasicHeaderElement("name3", "value3"),
};
formatter.formatElements(buf, hes, true);
System.out.println(buf.toString());
]]></programlisting>
            <para>stdout &gt;</para>
            <programlisting><![CDATA[
name1="value1"; param1="p1", name2="value2", name3="value3"
]]></programlisting>
        </section>
        <section>
            <title>HTTP message streams and session I/O buffers</title>
            <para>
            HttpCore provides a number of utility classes for the blocking and non-blocking I/O
            models that facilitate the processing of HTTP message streams, simplify handling of
            <literal>CR-LF</literal> delimited lines in HTTP messages and manage intermediate data
            buffering.
            </para>
            <para>
            HTTP connection implementations usually rely on session input/output buffers for
            reading and writing data from and to an HTTP message stream. Session input/output
            buffer implementations are I/O model specific and are optimized either for blocking or
            non-blocking operations.
            </para>
            <para>
            Blocking HTTP connections use socket bound session buffers to transfer data. Session
            buffer interfaces are similar to <classname>java.io.InputStream</classname> /
            <classname>java.io.OutputStream</classname> classes, but they also provide methods for
            reading and writing <literal>CR-LF</literal> delimited lines.
            </para>
            <programlisting><![CDATA[
Socket socket1;
Socket socket2;
HttpParams params = new BasicHttpParams();
SessionInputBuffer inbuffer = new SocketInputBuffer(
    socket1, 4096, params);
SessionOutputBuffer outbuffer = new SocketOutputBuffer(
    socket2, 4096, params);

CharArrayBuffer linebuf = new CharArrayBuffer(1024);
inbuffer.readLine(linebuf);
outbuffer.writeLine(linebuf);
]]></programlisting>
            <para>
            Non-blocking HTTP connections use session buffers optimized for reading and writing
            data from and to non-blocking NIO channels. NIO session input/output sessions help deal
            with <literal>CR-LF</literal> delimited lines in a non-blocking I/O mode.
            </para>
            <programlisting><![CDATA[
ReadableByteChannel channel1;
WritableByteChannel channel2;

HttpParams params = new BasicHttpParams();
SessionInputBuffer inbuffer = new SessionInputBufferImpl(
    4096, 1024, params);
SessionOutputBuffer outbuffer = new SessionOutputBufferImpl(
    4096, 1024, params);

CharArrayBuffer linebuf = new CharArrayBuffer(1024);
boolean endOfStream = false;
int bytesRead = inbuffer.fill(channel1);
if (bytesRead == -1) {
    endOfStream = true;
}
if (inbuffer.readLine(linebuf, endOfStream)) {
    outbuffer.writeLine(linebuf);
}
if (outbuffer.hasData()) {
    outbuffer.flush(channel2);
}
]]></programlisting>
        </section>
        <section>
            <title>HTTP message parsers and formatters</title>
            <para>
            HttpCore also provides coarse-grained facade type interfaces for parsing and
            formatting of HTTP messages. Default implementations of those interfaces build upon the
            functionality provided by <interfacename>SessionInputBuffer</interfacename> /
            <interfacename>SessionOutputBuffer</interfacename> and <interfacename>HttpLineParser
            </interfacename>/ <interfacename>HttpLineFormatter</interfacename> implementations.
            </para>
            <para>
            Example of HTTP request parsing / writing for blocking HTTP connections:
            </para>
            <programlisting><![CDATA[
SessionInputBuffer inbuffer;
SessionOutputBuffer outbuffer;

HttpParams params = new BasicHttpParams();

HttpMessageParser requestParser = new HttpRequestParser(
        inbuffer,
        new BasicLineParser(),
        new DefaultHttpRequestFactory(),
        params);

HttpRequest request = (HttpRequest) requestParser.parse();

HttpMessageWriter requestWriter = new HttpRequestWriter(
        outbuffer,
        new BasicLineFormatter(),
        params);

requestWriter.write(request);
]]></programlisting>
            <para>
            Example of HTTP response parsing / writing for blocking HTTP connections:
            </para>
            <programlisting><![CDATA[
SessionInputBuffer inbuffer;
SessionOutputBuffer outbuffer;

HttpParams params = new BasicHttpParams();

HttpMessageParser responseParser = new HttpResponseParser(
        inbuffer,
        new BasicLineParser(),
        new DefaultHttpResponseFactory(),
        params);

HttpResponse response = (HttpResponse) responseParser.parse();

HttpMessageWriter responseWriter = new HttpResponseWriter(
        outbuffer,
        new BasicLineFormatter(),
        params);

responseWriter.write(response);
]]></programlisting>
            <para>
            Example of HTTP request parsing / writing for non-blocking HTTP connections:
            </para>
            <programlisting><![CDATA[
SessionInputBuffer inbuffer;
SessionOutputBuffer outbuffer;

HttpParams params = new BasicHttpParams();

NHttpMessageParser requestParser = new DefaultHttpRequestParser(
        inbuffer,
        new BasicLineParser(),
        new DefaultHttpRequestFactory(),
        params);

HttpRequest request = (HttpRequest) requestParser.parse();

NHttpMessageWriter requestWriter = new DefaultHttpRequestWriter(
        outbuffer,
        new BasicLineFormatter(),
        params);

requestWriter.write(request);
]]></programlisting>
            <para>
            Example of HTTP response parsing / writing for non-blocking HTTP connections:
            </para>
            <programlisting><![CDATA[
SessionInputBuffer inbuffer;
SessionOutputBuffer outbuffer;

HttpParams params = new BasicHttpParams();

NHttpMessageParser responseParser = new DefaultHttpResponseParser(
        inbuffer,
        new BasicLineParser(),
        new DefaultHttpResponseFactory(),
        params);

HttpResponse response = (HttpResponse) responseParser.parse();

NHttpMessageWriter responseWriter = new DefaultHttpResponseWriter(
        outbuffer,
        new BasicLineFormatter(),
        params);

responseWriter.write(response);
]]></programlisting>
        </section>
        <section>
            <title>HTTP header parsing on demand</title>
            <para>
            The default implementations of <interfacename>HttpMessageParser</interfacename> and
            <interfacename>NHttpMessageParser</interfacename> interfaces do not parse HTTP headers
            immediately. Parsing of header value is deferred until its properties are accessed.
            Those headers that are never used by the application will not be parsed at all. The
            <classname>CharArrayBuffer</classname> backing the header can be obtained through an
            optional <interfacename>FormattedHeader</interfacename> interface.
            </para>
            <programlisting><![CDATA[
Header h1 = response.getFirstHeader("Content-Type");
if (h1 instanceof FormattedHeader) {
    CharArrayBuffer buf = ((FormattedHeader) h1).getBuffer();
    System.out.println(buf);
}
]]></programlisting>
        </section>
    </section>
    <section>
        <title>Customizing HTTP connections</title>
        <para>
        One can customize the way HTTP connections parse and format HTTP messages by extending the
        default implementations and overriding factory methods and replacing the default parser or
        formatter implementations with a custom one.
        </para>
        <para>
        For blocking HTTP connections one also can provide custom implementation of session
        input/output buffers.
        </para>
        <programlisting><![CDATA[
class MyDefaultHttpClientConnection
                        extends DefaultHttpClientConnection {

    @Override
    protected SessionInputBuffer createSessionInputBuffer(
            Socket socket,
            int buffersize,
            HttpParams params) throws IOException {
        return new MySocketInputBuffer(socket, buffersize, params);
    }

    @Override
    protected SessionOutputBuffer createSessionOutputBuffer(
            Socket socket,
            int buffersize,
            HttpParams params) throws IOException {
        return new MySocketOutputBuffer(socket, buffersize, params);
    }

    @Override
    protected HttpMessageWriter createRequestWriter(
            SessionOutputBuffer buffer,
            HttpParams params) {
        return new MyHttpRequestWriter(
            buffer, new BasicLineFormatter(), params);
    }

    @Override
    protected HttpMessageParser createResponseParser(
            SessionInputBuffer buffer,
            HttpResponseFactory responseFactory,
            HttpParams params) {
        return new MyHttpResponseParser(
        buffer, new BasicLineParser(), responseFactory, params);
    }

};
]]></programlisting>
        <para>
        For non-blocking HTTP connection implementation one can replace the default HTTP message
        parser and formatter implementations. The session input/output buffer implementations can
        be overridden at the I/O reactor level.
        </para>
        <programlisting><![CDATA[
class MyDefaultNHttpClientConnection
                        extends DefaultNHttpClientConnection {

    public MyDefaultNHttpClientConnection(
            IOSession session,
            HttpResponseFactory responseFactory,
            ByteBufferAllocator allocator,
            HttpParams params) {
        super(session, responseFactory, allocator, params);
    }

    @Override
    protected NHttpMessageWriter createRequestWriter(
            SessionOutputBuffer buffer,
            HttpParams params) {
        return new HttpRequestWriter(
            buffer, new BasicLineFormatter(), params);
    }

    @Override
    protected NHttpMessageParser createResponseParser(
            SessionInputBuffer buffer,
            HttpResponseFactory responseFactory,
            HttpParams params) {
        return new HttpResponseParser(
            buffer, new BasicLineParser(), responseFactory, params);
    }

};
]]></programlisting>
    </section>
</chapter>
