tutorial_packet_parsing.html

---
title: Tutorials
layout: default
section: tutorials
redirect_to: https://pcapplusplus.github.io/docs/tutorials/packet-parsing
---
          <h1><a href="tutorials.html">Tutorials</a></h1>
          <h2>Part 4: Packet parsing</h2>
          <p>This part explains how to parse packets and use the different
            layers (protocol parsers)</p>
          <p>Table of contents:</p>
          <ul>
            <li><a href="tutorial_intro.html">Part 1: Introduction and basics</a></li>
            <li><a href="tutorial_pcap_files.html">Part 2: Reading and writing pcap files</a></li>
            <li><a href="tutorial_live_capture.html">Part 3: Capturing and sending packets</a></li>
            <li>Part 4: Packet parsing
              <ol>
                <li><a href="#intro">Introduction</a></li>
                <li><a href="#packet_parsing">Packet parsing basics</a> </li>
                <li><a href="#parsing_eth">Parsing Ethernet</a></li>
                <li><a href="#parsing_ipv4">Parsing IPv4</a></li>
                <li><a href="#parsing_tcp">Parsing TCP</a></li>
                <li><a href="#parsing_http">Parsing HTTP</a></li>
                <li><a href="#run_example">Running the example</a></li>
              </ol>
            </li>
            <li><a href="tutorial_packet_craft_n_edit.html">Part 5: Packet crafting and editing</a></li>
            <li><a href="tutorial_dpdk.html">Part 6: Working with DPDK</a></li>
          </ul>
          <h2 id="intro">Introduction</h2>
          <p>Packet parsing, editing and crafting are a major part of
            PcapPlusPlus and is the essence of the Packet++ library. There is a
            long list of <a href="index.html#supported-protocols">protocols
              currently supported</a>, each of them is represented by a <code class="language-cpp">Layer</code>
            class which (in most cases) supports both parsing of the protocol,
            editing and creation of new layers from scratch.</p>
          <p>This tutorial will go through the packet parsing fundamentals and
            the next tutorial will focus on packet crafting and editing. The
            tutorial demonstrate parsing on a few popular protocols:</p>
          <ul>
            <li>Ethernet</li>
            <li>IPv4</li>
            <li>TCP</li>
            <li>HTTP</li>
          </ul>
          <p>For further information about these protocols and the other
            protocols supported in PcapPlusPlus please go to the <a href="Documentation/index.html">API
              documentation</a></p>
          <h2 id="packet_parsing">Packet parsing basics</h2>
          <p>In this tutorial we'll read a packet from a pcap file, let
            PcapPlusPlus parse it, and see how we can retrieve data from each
            layer. Let's start by writing a <code class="language-cpp">main()</code> method and add
            the includes that we need:</p>
<pre><code class="language-cpp">#if !defined(WIN32) && !defined(WINx64)
#include <in.h> // this is for using ntohs() and htons() on non-Windows OS's
#endif
#include "stdlib.h"
#include "Packet.h"
#include "EthLayer.h"
#include "IPv4Layer.h"
#include "TcpLayer.h"
#include "HttpLayer.h"
#include "PcapFileDevice.h"

int main(int argc, char* argv[])
{
	// We'll write our code here
}</code>
</pre>
          <p>As you can see we added an include to <code class="language-cpp">Packet.h</code>
            which contains the basic parsed packet structures, to <code class="language-cpp">PcapFileDevice.h</code>
            which contains the API for reading from pcap files and to all of the
            layers which we want to retrieve information from. In addition we included
            <code class="language-cpp">in.h</code> for using <code class="language-cpp">htons()</code> and
            <code class="language-cpp">ntohs()</code> which we'll use later. This include is relevant for non-Windows
            operating systems only.</p>
          <p>Now let's read the packet from the pcap file. This pcap file
            contains only 1 packet, so we'll open the reader, read the packet
            and close the reader:</p>
<pre><code class="language-cpp">// use the IFileReaderDevice interface to automatically identify file type (pcap/pcap-ng)
// and create an interface instance that both readers implement
pcpp::IFileReaderDevice* reader = pcpp::IFileReaderDevice::getReader("1_http_packet.pcap");

// verify that a reader interface was indeed created
if (reader == NULL)
{
	printf("Cannot determine reader for file type\n");
	exit(1);
}

// open the reader for reading
if (!reader-&gt;open())
{
	printf("Cannot open input.pcap for reading\n");
	exit(1);
}

// read the first (and only) packet from the file
pcpp::RawPacket rawPacket;
if (!reader-&gt;getNextPacket(rawPacket))
{
	printf("Couldn't read the first packet in the file\n");
	return 1;
}

// close the file reader, we don't need it anymore
reader-&gt;close();</code>
</pre>
          <p>The next step is to let PcapPlusPlus parse the packet. We do this
            by creating an instance of the <code class="language-cpp">Packet</code>
            class and giving it in the constructor a pointer to the <code class="language-cpp">RawPacket</code>
            instance we have:</p>
<pre><code class="language-cpp">// parse the raw packet into a parsed packet
pcpp::Packet parsedPacket(&amp;rawPacket);</code>
</pre>
          <p>Before we dive into the protocols, let's remember how the <code class="language-cpp">Packet</code>
            class is <a href="tutorial_intro.html#packets_layers">built</a>: it
            contains a link list of <code class="language-cpp">Layer</code> instances,
            each layer points to the next layer in the packet. In our example:
            Ethernet layer will be the first one, it will point to IPv4 layer
            which will point to TCP layer and finally we'll have HTTP request
            layer. The <code class="language-cpp">Packet</code> class exposes this link
            list so we can iterate over the layers and retrieve basic
            information like the protocols they represent, sizes, etc. Let's see
            the code:</p>
<pre><code class="language-cpp">// first let's go over the layers one by one and find out its type, its total length, its header length and its payload length
for (pcpp::Layer* curLayer = parsedPacket.getFirstLayer(); curLayer != NULL; curLayer = curLayer-&gt;getNextLayer())
{
	printf("Layer type: %s; Total data: %d [bytes]; Layer data: %d [bytes]; Layer payload: %d [bytes]\n",
			getProtocolTypeAsString(curLayer-&gt;getProtocol()).c_str(), // get layer type
			(int)curLayer-&gt;getDataLen(),                              // get total length of the layer
			(int)curLayer-&gt;getHeaderLen(),                            // get the header length of the layer
			(int)curLayer-&gt;getLayerPayloadSize());                    // get the payload length of the layer (equals total length minus header length)
}</code>
</pre>
          <p>As you can see, we're using the <code class="language-cpp">getFirstLayer()</code>
            and <code class="language-cpp">getNextLayer()</code> APIs to iterate over
            the layers. In each layer we have the following information:</p>
          <ul>
            <li><code class="language-cpp">getProtocol()</code> - get an enum of the
              protocol the layer represents</li>
            <li><code class="language-cpp">getHeaderLen()</code> - get the size of the
              layer's header, meaning the size of the layer data</li>
            <li><code class="language-cpp">getLayerPayloadSize()</code> - get the size of
              the layer's payload, meaning the size of all layers that follows
              this layer</li>
            <li><code class="language-cpp">getDataLen()</code> - get the total size of
              the layer: header + payload</li>
          </ul>
          <p>For printing the protocols I used a simple function that takes a <code

              class="language-cpp">ProtocolType</code> enum and returns a string:</p>
<pre><code class="language-cpp">std::string getProtocolTypeAsString(pcpp::ProtocolType protocolType)
{
	switch (protocolType)
	{
	case pcpp::Ethernet:
		return "Ethernet";
	case pcpp::IPv4:
		return "IPv4";
	case pcpp::TCP:
		return "TCP";
	case pcpp::HTTPRequest:
	case pcpp::HTTPResponse:
		return "HTTP";
	default:
		return "Unknown";
	}
}</code>
</pre>
          <p>Let's see the output so far:</p>
<pre><code class="language-shell">Layer type: Ethernet; Total data: 443 [bytes]; Layer data: 14 [bytes]; Layer payload: 429 [bytes]
Layer type: IPv4; Total data: 429 [bytes]; Layer data: 20 [bytes]; Layer payload: 409 [bytes]
Layer type: TCP; Total data: 409 [bytes]; Layer data: 32 [bytes]; Layer payload: 377 [bytes]
Layer type: HTTP; Total data: 377 [bytes]; Layer data: 377 [bytes]; Layer payload: 0 [bytes]</code>
</pre>
          <p> </p>
          <h2 id="parsing_eth">Parsing Ethernet</h2>
          <p>Now let's see what information we can get from the first layer in
            this packet: <code class="language-cpp">EthLayer</code>. First let's get a
            pointer to this layer. We can use the methods we used before and
            cast the <code class="language-cpp">Layer*</code> to <code class="language-cpp">EthLayer*</code>
            but the <code class="language-cpp">Packet</code> class offers a more
            convenient way to do that:</p>
<pre><code class="language-cpp">// now let's get the Ethernet layer
pcpp::EthLayer* ethernetLayer = parsedPacket.getLayerOfType&lt;pcpp::EthLayer&gt;();
if (ethernetLayer == NULL)
{
	printf("Something went wrong, couldn't find Ethernet layer\n");
	exit(1);
}</code>
</pre>
          <p>As you can see we used the templated method <code class="language-cpp">getLayerOfType&lt;pcpp::EthLayer&gt;()</code>
            which returns a pointer to <code class="language-cpp">EthLayer</code> if
            exists in the packet or NULL otherwise. Now we are ready to start
            getting some information. The Ethernet layer is quite simple so
            there's not much information we can get. We can basically get the
            source and destination MAC addresses and the Ether Type of the next
            layer:</p>
<pre><code class="language-cpp">// print the source and dest MAC addresses and the Ether type
printf("\nSource MAC address: %s\n", ethernetLayer-&gt;getSourceMac().toString().c_str());
printf("Destination MAC address: %s\n", ethernetLayer-&gt;getDestMac().toString().c_str());
printf("Ether type = 0x%X\n", ntohs(ethernetLayer-&gt;getEthHeader()-&gt;etherType));</code>
</pre>
          <p>For getting the source and destination MAC addresses <code class="language-cpp">EthLayer</code>
            exposes methods which return an instance of type <code class="language-cpp">MacAddress</code>
            which encapsulates MAC addresses and provides helper function such
            as print the MAC address as a nice string (like we have in our code
            example). For getting the Ether Type we call <code class="language-cpp">getEthHeader()</code>
            which casts the raw packet bytes into a struct: <code class="language-cpp">ether_header*</code>
            and we can read the Ether Type from this struct. Since packet raw
            data is stored in network order, we need to convert the Ether Type
            value from network to host order using <code class="language-cpp">ntohs()</code></p>
          <p>The output is the following:</p>
<pre><code class="language-shell">Source MAC address: 00:50:43:01:4d:d4
Destination MAC address: 00:90:7f:3e:02:d0
Ether type = 0x800</code>
</pre>
          <p> </p>
          <h2 id="parsing_ipv4">Parsing IPv4</h2>
          <p>Now let's get the IPv4 layer, we'll do it in the same way as before
            using the template <code class="language-cpp">getLayerOfType&lt;pcpp::IPv4Layer&gt;()</code>
            method:</p>
<pre><code class="language-cpp">// let's get the IPv4 layer
pcpp::IPv4Layer* ipLayer = parsedPacket.getLayerOfType&lt;pcpp::IPv4Layer&gt;();
if (ipLayer == NULL)
{
	printf("Something went wrong, couldn't find IPv4 layer\n");
	exit(1);
}</code>
</pre>
          <p>Let's get some information from the <code class="language-cpp">IPv4Layer</code>:</p>
<pre><code class="language-cpp">// print source and dest IP addresses, IP ID and TTL
printf("\nSource IP address: %s\n", ipLayer-&gt;getSrcIpAddress().toString().c_str());
printf("Destination IP address: %s\n", ipLayer-&gt;getDstIpAddress().toString().c_str());
printf("IP ID: 0x%X\n", ntohs(ipLayer-&gt;getIPv4Header()-&gt;ipId));
printf("TTL: %d\n", ipLayer-&gt;getIPv4Header()-&gt;timeToLive);</code>
</pre>
          <p>As you can see this layer exposes 2 methods for reading the source
            and destination IP addresses in an easy-to-use wrapper class called
            <code class="language-cpp">IPv4Address</code>. This class provides various
            capabilities, one of them is printing the IP address as a string.
            Next, we use the <code class="language-cpp">getIPv4Header()</code> method
            which casts the raw packet bytes to a struct called <code class="language-cpp">iphdr*</code>
            and we can retrieve the rest of the data from there. Since the
            packet data is in network order, we need to use <code class="language-cpp">ntohs()</code>
            when getting data larger than 1 byte (like when reading the IP ID).</p>
          <p>Here is the output:</p>
<pre><code class="language-shell">Source IP address: 172.16.133.132
Destination IP address: 98.139.161.29
IP ID: 0x36E4
TTL: 64</code>
</pre>
          <p> </p>
          <h2 id="parsing_tcp">Parsing TCP</h2>
          <p>Let's get the TCP layer:</p>
<pre><code class="language-cpp">// let's get the TCP layer
pcpp::TcpLayer* tcpLayer = parsedPacket.getLayerOfType&lt;pcpp::TcpLayer&gt;();
if (tcpLayer == NULL)
{
	printf("Something went wrong, couldn't find TCP layer\n");
	exit(1);
}</code>
</pre>
          <p>Now let's get the TCP data:</p>
<pre><code class="language-cpp">// printf TCP source and dest ports, window size, and the TCP flags that are set in this layer
printf("\nSource TCP port: %d\n", (int)ntohs(tcpLayer-&gt;getTcpHeader()-&gt;portSrc));
printf("Destination TCP port: %d\n", (int)ntohs(tcpLayer-&gt;getTcpHeader()-&gt;portDst));
printf("Window size: %d\n", (int)ntohs(tcpLayer-&gt;getTcpHeader()-&gt;windowSize));
printf("TCP flags: %s\n", printTcpFlags(tcpLayer).c_str());</code>
</pre>
          <p>Here also, like the layers we saw before, TCP layer exposes a
            method <code class="language-cpp">getTcpHeader()</code> which casts the raw
            packet bytes to a struct <code class="language-cpp">tpchdr*</code> which
            contains all of the TCP fields. That way we can get the source and
            destination ports, the windows size and any other TCP field. Notice
            the need of using <code class="language-cpp">ntohs()</code> to convert the
            data from network to host byte order because the raw packet arrives
            in network order. I also wrote a small function that gathers all of
            the TCP flags on the packet and prints them nicely:</p>
<pre><code class="language-cpp">std::string printTcpFlags(pcpp::TcpLayer* tcpLayer)
{
	std::string result = "";
	if (tcpLayer-&gt;getTcpHeader()-&gt;synFlag == 1)
		result += "SYN ";
	if (tcpLayer-&gt;getTcpHeader()-&gt;ackFlag == 1)
		result += "ACK ";
	if (tcpLayer-&gt;getTcpHeader()-&gt;pshFlag == 1)
		result += "PSH ";
	if (tcpLayer-&gt;getTcpHeader()-&gt;cwrFlag == 1)
		result += "CWR ";
	if (tcpLayer-&gt;getTcpHeader()-&gt;urgFlag == 1)
		result += "URG ";
	if (tcpLayer-&gt;getTcpHeader()-&gt;eceFlag == 1)
		result += "ECE ";
	if (tcpLayer-&gt;getTcpHeader()-&gt;rstFlag == 1)
		result += "RST ";
	if (tcpLayer-&gt;getTcpHeader()-&gt;finFlag == 1)
		result += "FIN ";

	return result;
}</code>
</pre>
          <p>Another cool feature <code class="language-cpp">TcpLayer</code> provides is
            retrieving information about the TCP options (if exist). We can
            iterate the TCP options using the methods <code class="language-cpp">getFirstTcpOption()</code>
            and <code class="language-cpp">getNextTcpOption(tcpOption)</code> and
            extract all the information on the TCP option such as type, length
            and value. In our example let's iterate over them and print their
            type:</p>
<pre><code class="language-cpp">// go over all TCP options in this layer and print its type
printf("TCP options: ");
for (pcpp::TcpOption tcpOption = tcpLayer-&gt;getFirstTcpOption(); tcpOption.isNotNull(); tcpOption = tcpLayer-&gt;getNextTcpOption(tcpOption))
{
	printf("%s ", printTcpOptionType(tcpOption.getTcpOptionType()).c_str());
}
printf("\n");</code>
</pre>
          <p>Let's see the method that gets the TCP option type as enum and
            converts it to string. Notice this method handles only the TCP
            options we have on the specific packet we're parsing, PcapPlusPlus
            support all TCP options types:</p>
<pre><code class="language-cpp">std::string printTcpOptionType(pcpp::TcpOptionType optionType)
{
	switch (optionType)
	{
	case pcpp::PCPP_TCPOPT_NOP:
		return "NOP";
	case pcpp::PCPP_TCPOPT_TIMESTAMP:
		return "Timestamp";
	default:
		return "Other";
	}
}</code>
</pre>
          <p> </p>
          <h2 id="parsing_http">Parsing HTTP</h2>
          <p>Finally, let's see the capabilities <code class="language-cpp">HttpRequestLayer</code>
            has to offer. First let's extract the layer from the packet:</p>
<pre><code class="language-cpp">// let's get the HTTP request layer
pcpp::HttpRequestLayer* httpRequestLayer = parsedPacket.getLayerOfType&lt;pcpp::httprequestlayer&gt;();
if (httpRequestLayer == NULL)
{
	printf("Something went wrong, couldn't find HTTP request layer\n");
	exit(1);
}</code>
</pre>
          <p>Of course there is a similar class <code class="language-cpp">HttpResponseLayer</code>
            for HTTP responses.</p>
          <p>HTTP messages (both requests and responses) have 3 main parts:</p>
          <ul>
            <li>The first line (also known as request-line or status-line) which
              includes the HTTP version, HTTP method (for requests) or status
              code (for responses) and the URI (for requests)</li>
            <li>Message headers which include all header fields (e.g host,
              user-agent, cookie, content-type etc.)</li>
            <li>Message body</li>
          </ul>
          <p>The HTTP layer classes provide access to all of these parts. Let's
            start with showing how to get data from the first line:</p>
<pre><code class="language-cpp">// print HTTP method and URI. Both appear in the first line of the HTTP request
printf("\nHTTP method: %s\n", printHttpMethod(httpRequestLayer-&gt;getFirstLine()-&gt;getMethod()).c_str());
printf("HTTP URI: %s\n", httpRequestLayer-&gt;getFirstLine()-&gt;getUri().c_str());</code></pre>
          <p>As you can see the <code class="language-cpp">HttpRequestLayer </code>class
            exposes a getter (<code class="language-cpp">getFirstLine()</code>) that
            retrieves an object of type <code class="language-cpp">HttpRequestFirstLine</code>
            that contain all of the first-line data: method, URI,etc. The method
            is returned as an enum so I added a simple function <code class="language-cpp">printHttpMethod</code>
            to print it as a string:</p>
<pre><code class="language-cpp">std::string printHttpMethod(pcpp::HttpRequestLayer::HttpMethod httpMethod)
{
	switch (httpMethod)
	{
	case pcpp::HttpRequestLayer::HttpGET:
		return "GET";
	case pcpp::HttpRequestLayer::HttpPOST:
		return "POST";
	default:
		return "Other";
	}
}</code>
</pre>
          <p>Now let's see how to get header fields data:</p>
<pre><code class="language-cpp">// print values of the following HTTP field: Host, User-Agent and Cookie
printf("HTTP host: %s\n", httpRequestLayer-&gt;getFieldByName(PCPP_HTTP_HOST_FIELD)-&gt;getFieldValue().c_str());
printf("HTTP user-agent: %s\n", httpRequestLayer-&gt;getFieldByName(PCPP_HTTP_USER_AGENT_FIELD)-&gt;getFieldValue().c_str());
printf("HTTP cookie: %s\n", httpRequestLayer-&gt;getFieldByName(PCPP_HTTP_COOKIE_FIELD)-&gt;getFieldValue().c_str());</code>
</pre>
          <p>The HTTP request and response layers exposes a method <code class="language-cpp">getFieldByName()</code>
            to get a header field data by it's name. The class representing a
            field is called <code class="language-cpp">HttpField </code>and has some interesting
            API, but probably the most important method for parsing is <code class="language-cpp">getFieldValue()</code>
            which returns the value of this header field as string. Please
            notice that I didn't write the header field names as strings but
            rather used a macro defined in PcapPlusPlus for some of the most
            useful HTTP fields (like host, cookie, user-agent, etc.).</p>
          <p>Finally, let's see another cool method in <code class="language-cpp">HttpRequestLayer </code>which
            is <code class="language-cpp">getURL()</code> that forms and returns the full URL from
            the request (including host-name from "Host" header field + URI from
            the request-line):</p>
<pre><code class="language-cpp">// print the full URL of this request
printf("HTTP full URL: %s\n", httpRequestLayer-&gt;getUrl().c_str());</code>
</pre>
          <p>Now let's see the output:</p>
<pre><code class="language-shell">HTTP method: GET
HTTP URI: /serv?s=19190039&amp;t=1361916157&amp;f=us-p9h3
HTTP host: geo.yahoo.com
HTTP user-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_7_3) AppleWebKit/534.55.3 (KHTML, like Gecko) Version/5.1.3 Safari/534.53.10
HTTP cookie: B=fdnulql8iqc6l&amp;b=3&amp;s=ps
HTTP full URL: geo.yahoo.com/serv?s=19190039&amp;t=1361916157&amp;f=us-p9h3</code>
</pre>
          <h2 id="run_example">Running the example</h2>
          <p>All code that was covered in this tutorial can be found <a href="https://github.com/seladb/PcapPlusPlus/tree/master/Examples/Tutorials/Tutorial-PacketParsing">here</a>.
            In order to compile and run the code please first download and
            compile PcapPlusPlus code or downloaded a pre-compiled version from
            <a href="https://github.com/seladb/PcapPlusPlus/releases/latest">the
              latest PcapPlusPlus release</a>. Then follow these instruction,
            according to your platform:</p>
          <ul>
            <li>Linux and MacOS - make sure PcapPlusPlus is installed (by running <strong>sudo make install</strong> in PcapPlusPlus main directory). Then either change the <code class="language-shell">Makefile.non_windows</code>
              file name to <code class="language-shell">Makefile</code> and run <code class="language-shell">make all</code>,
              or run <code class="language-shell">make -f Makefile.non_windows all</code></li>
            <li>Windows using MinGW or MinGW-w64 - either change the <code class="language-shell">Makefile.windows</code>
              file name to <code class="language-shell">Makefile</code> and run <code class="language-shell">make all</code>,
              or run <code class="language-shell">make -f Makefile.windows all</code></li>
            <li>Windows using Visual Studio 2015 - there is a Visual Studio 2015
              solution containing all tutorials <a href="https://github.com/seladb/PcapPlusPlus/tree/master/mk/vs2015/Tutorials.sln">here</a>.
              Just open it and compile all tutorials</li>
          </ul>
          <p>In all options the compiled executable will be inside the tutorial
            directory (<code class="language-shell">[PcapPlusPlus
              Folder]/Examples/Tutorials/Tutorial-PacketParsing</code>)</p>
          <script>function topFunction() { document.body.scrollTop = 0; document.documentElement.scrollTop = 0; }</script>
          <p><a onclick="topFunction()" href="#">Go to top &uarr;</a></p>