// TODO: what if there is long time between response and reply packet? Should we add a threshold and exclude those cases?
}
-
/**
- * Given a list of TCP conversations and associated DNS mappings, groups the conversations by hostname.
- * @param tcpConversations The list of TCP conversations.
+ * Given a collection of TCP conversations and associated DNS mappings, groups the conversations by hostname.
+ * @param tcpConversations The collection of TCP conversations.
* @param ipHostnameMappings The associated DNS mappings.
* @return A map where each key is a hostname and its associated value is a list of conversations where one of the
* two communicating hosts is that hostname (i.e. its IP maps to the hostname).
*/
- public static Map<String, List<Conversation>> groupConversationsByHostname(List<Conversation> tcpConversations, DnsMap ipHostnameMappings) {
+ public static Map<String, List<Conversation>> groupConversationsByHostname(Collection<Conversation> tcpConversations, DnsMap ipHostnameMappings) {
HashMap<String, List<Conversation>> result = new HashMap<>();
for (Conversation c : tcpConversations) {
if (c.getPackets().size() == 0) {
return result;
}
+ public static Map<String, Integer> countPacketSequenceFrequencies(Collection<Conversation> conversations) {
+ Map<String, Integer> result = new HashMap<>();
+ for (Conversation conv : conversations) {
+ if (conv.getPackets().size() == 0) {
+ // Skip conversations with no payload packets.
+ continue;
+ }
+ StringBuilder sb = new StringBuilder();
+ for (PcapPacket pp : conv.getPackets()) {
+ sb.append(pp.length() + " ");
+ }
+ result.merge(sb.toString(), 1, (i1, i2) -> i1+i2);
+ }
+ return result;
+ }
+
+ /**
+ * Given a {@link Collection} of {@link Conversation}s, builds a {@link Map} from {@link String} to {@link List}
+ * of {@link Conversation}s such that each key is the <em>concatenation of the packet lengths of all payload packets
+ * (i.e., the set of packets returned by {@link Conversation#getPackets()}) separated by a delimiter</em> of any
+ * {@link Conversation} pointed to by that key. In other words, what the {@link Conversation}s {@code cs} pointed to
+ * by the key {@code s} have in common is that they all contain exactly the same number of payload packets <em>and
+ * </em> these payload packets are identical across all {@code Conversation}s in {@code convs} in terms of packet
+ * length and packet order. For example, if the key is "152 440 550", this means that every individual
+ * {@code Conversation} in the list of {@code Conversation}s pointed to by that key contain exactly three payload
+ * packet of lengths 152, 440, and 550, and these three packets are ordered the in the order prescribed by the key.
+ *
+ * @param conversations The collection of {@code Conversation}s to group by packet sequence.
+ * @return a {@link Map} from {@link String} to {@link List} of {@link Conversation}s such that each key is the
+ * <em>concatenation of the packet lengths of all payload packets (i.e., the set of packets returned by
+ * {@link Conversation#getPackets()}) separated by a delimiter</em> of any {@link Conversation} pointed to
+ * by that key.
+ */
+ public static Map<String, List<Conversation>> groupConversationsByPacketSequence(Collection<Conversation> conversations) {
+ Map<String, List<Conversation>> result = new HashMap<>();
+ for (Conversation conv : conversations) {
+ if (conv.getPackets().size() == 0) {
+ // Skip conversations with no payload packets.
+ continue;
+ }
+ StringBuilder sb = new StringBuilder();
+ for (PcapPacket pp : conv.getPackets()) {
+ if (sb.length() != 0) {
+ // only add a space if there's preceding content
+ sb.append(" ");
+ }
+ sb.append(pp.length());
+ }
+ List<Conversation> oneItemList = new ArrayList<>();
+ oneItemList.add(conv);
+ result.merge(sb.toString(), oneItemList, (oldList, newList) -> {
+ oldList.addAll(newList);
+ return oldList;
+ });
+ }
+ return result;
+ }
+
+ /**
+ * Given a {@link Conversation}, counts the frequencies of each unique packet length seen as part of the
+ * {@code Conversation}.
+ * @param c The {@code Conversation} for which unique packet length frequencies are to be determined.
+ * @return A mapping from packet length to its frequency.
+ */
+ public static Map<Integer, Integer> countPacketLengthFrequencies(Conversation c) {
+ Map<Integer, Integer> result = new HashMap<>();
+ for (PcapPacket packet : c.getPackets()) {
+ result.merge(packet.length(), 1, (i1, i2) -> i1 + i2);
+ }
+ return result;
+ }
+ /**
+ * Like {@link #countPacketLengthFrequencies(Conversation)}, but counts packet length frequencies for a collection
+ * of {@code Conversation}s, i.e., the frequency of a packet length becomes the total number of packets with that
+ * length across <em>all</em> {@code Conversation}s in {@code conversations}.
+ * @param conversations The collection of {@code Conversation}s for which packet length frequencies are to be
+ * counted.
+ * @return A mapping from packet length to its frequency.
+ */
+ public static Map<Integer, Integer> countPacketLengthFrequencies(Collection<Conversation> conversations) {
+ Map<Integer, Integer> result = new HashMap<>();
+ for (Conversation c : conversations) {
+ Map<Integer, Integer> intermediateResult = countPacketLengthFrequencies(c);
+ for (Map.Entry<Integer, Integer> entry : intermediateResult.entrySet()) {
+ result.merge(entry.getKey(), entry.getValue(), (i1, i2) -> i1 + i2);
+ }
+ }
+ return result;
+ }
+
+ public static Map<String, Integer> countPacketPairFrequencies(Collection<PcapPacketPair> pairs) {
+ Map<String, Integer> result = new HashMap<>();
+ for (PcapPacketPair ppp : pairs) {
+ result.merge(ppp.toString(), 1, (i1, i2) -> i1 + i2);
+ }
+ return result;
+ }
+
+ public static Map<String, Map<String, Integer>> countPacketPairFrequenciesByHostname(Collection<Conversation> tcpConversations, DnsMap ipHostnameMappings) {
+ Map<String, List<Conversation>> convsByHostname = groupConversationsByHostname(tcpConversations, ipHostnameMappings);
+ HashMap<String, Map<String, Integer>> result = new HashMap<>();
+ for (Map.Entry<String, List<Conversation>> entry : convsByHostname.entrySet()) {
+ // Merge all packet pairs exchanged during the course of all conversations with hostname into one list
+ List<PcapPacketPair> allPairsExchangedWithHostname = new ArrayList<>();
+ entry.getValue().forEach(conversation -> allPairsExchangedWithHostname.addAll(extractPacketPairs(conversation)));
+ // Then count the frequencies of packet pairs exchanged with the hostname, irrespective of individual
+ // conversations
+ result.put(entry.getKey(), countPacketPairFrequencies(allPairsExchangedWithHostname));
+ }
+ return result;
+ }
}