Add MD5/SHA1/SHA256 file analysis hashing actions.

scripts/base/frameworks/file-analysis/main.bro

@@ -37,6 +37,12 @@ export {
 		extract_filename: string &optional;
 	};
 
+	type ActionResults: record {
+		md5: string &optional;
+		sha1: string &optional;
+		sha256: string &optional;
+	};
+
 	## Contains all metadata related to the analysis of a given file, some
 	## of which is logged.
 	type Info: record {
@@ -81,6 +87,9 @@ export {
 
 		## The corresponding arguments supplied to each element of *actions*.
 		action_args: vector of ActionArgs &default=vector();
+
+		## Some actions may directly yield results in this record.
+		action_results: ActionResults;
 	} &redef;
 
 	## TODO: document

src/CMakeLists.txt

@@ -454,6 +454,7 @@ set(bro_SRCS
     file_analysis/FileID.h
     file_analysis/Action.h
     file_analysis/Extract.cc
+    file_analysis/Hash.cc
 
     nb_dns.c
     digest.h

src/file_analysis.bif

@@ -8,6 +8,7 @@ module FileAnalysis;
 
 type Info: record;
 type ActionArgs: record;
+type ActionResults: record;
 
 ## An enumeration of possibly-interesting "events" that can occur over
 ## the course of analyzing files.  The :bro:see:`FileAnalysis::policy`
@@ -57,6 +58,9 @@ enum Trigger %{
 
 enum Action %{
 	ACTION_EXTRACT,
+	ACTION_MD5,
+	ACTION_SHA1,
+	ACTION_SHA256,
 %}
 
 function FileAnalysis::postpone_timeout%(file_id: string%): bool

src/file_analysis/Action.h

@@ -2,9 +2,12 @@
 #define FILE_ANALYSIS_ACTION_H
 
 #include "Val.h"
+#include "NetVar.h"
 
 namespace file_analysis {
 
+typedef BifEnum::FileAnalysis::Action ActionTag;
+
 class Info;
 
 /**
@@ -17,29 +20,44 @@ public:
 
 	/**
 	 * Subclasses may override this to receive file data non-sequentially.
+	 * @return true if the action is still in a valid state to continue
+	 *         receiving data/events or false if it's essentially "done".
 	 */
-	virtual void DeliverChunk(const u_char* data, uint64 len, uint64 offset) {}
+	virtual bool DeliverChunk(const u_char* data, uint64 len, uint64 offset)
+		{ return true; }
 
 	/**
 	 * Subclasses may override this to receive file sequentially.
+	 * @return true if the action is still in a valid state to continue
+	 *         receiving data/events or false if it's essentially "done".
 	 */
-	virtual void DeliverStream(const u_char* data, uint64 len) {}
+	virtual bool DeliverStream(const u_char* data, uint64 len)
+		{ return true; }
 
 	/**
 	 * Subclasses may override this to specifically handle the end of a file.
+	 * @return true if the action is still in a valid state to continue
+	 *         receiving data/events or false if it's essentially "done".
 	 */
-	virtual void EndOfFile() {}
+	virtual bool EndOfFile()
+		{ return true; }
 
 	/**
 	 * Subclasses may override this to handle missing data in a file stream.
+	 * @return true if the action is still in a valid state to continue
+	 *         receiving data/events or false if it's essentially "done".
 	 */
-	virtual void Undelivered(uint64 offset, uint64 len) {}
+	virtual bool Undelivered(uint64 offset, uint64 len)
+		{ return true; }
+
+	ActionTag Tag() const { return tag; }
 
 protected:
 
-	Action(Info* arg_info) {}
+	Action(Info* arg_info, ActionTag arg_tag) : info(arg_info), tag(arg_tag) {}
 
 	Info* info;
+	ActionTag tag;
 };
 
 typedef Action* (*ActionInstantiator)(const RecordVal* args, Info* info);

src/file_analysis/Extract.cc

@@ -6,7 +6,8 @@
 using namespace file_analysis;
 
 Extract::Extract(Info* arg_info, const string& arg_filename)
-    : Action(arg_info), filename(arg_filename)
+    : Action(arg_info, BifEnum::FileAnalysis::ACTION_EXTRACT),
+      filename(arg_filename)
 	{
 	fd = open(filename.c_str(), O_WRONLY | O_CREAT | O_TRUNC, 0666);
 
@@ -36,11 +37,12 @@ Action* Extract::Instantiate(const RecordVal* args, Info* info)
 	return new Extract(info, v->AsString()->CheckString());
 	}
 
-void Extract::DeliverChunk(const u_char* data, uint64 len, uint64 offset)
+bool Extract::DeliverChunk(const u_char* data, uint64 len, uint64 offset)
 	{
 	Action::DeliverChunk(data, len, offset);
 
-	if ( ! fd ) return;
+	if ( ! fd ) return false;
 
 	safe_pwrite(fd, data, len, offset);
+	return true;
 	}

src/file_analysis/Extract.h

@@ -5,20 +5,21 @@
 
 #include "Val.h"
 #include "Info.h"
+#include "Action.h"
 
 namespace file_analysis {
 
 /**
  * An action to simply extract files to disk.
  */
-class Extract : Action {
+class Extract : public Action {
 public:
 
 	static Action* Instantiate(const RecordVal* args, Info* info);
 
-	~Extract();
+	virtual ~Extract();
 
-	virtual void DeliverChunk(const u_char* data, uint64 len, uint64 offset);
+	virtual bool DeliverChunk(const u_char* data, uint64 len, uint64 offset);
 
 protected:
 

src/file_analysis/Hash.cc

@@ -0,0 +1,54 @@
+#include <string>
+
+#include "Hash.h"
+#include "util.h"
+
+using namespace file_analysis;
+
+Hash::Hash(Info* arg_info, ActionTag tag, HashVal* hv)
+	: Action(arg_info, tag), hash(hv)
+	{
+	hash->Init();
+	}
+
+Hash::~Hash()
+	{
+	// maybe it's all there...
+	Finalize();
+	delete hash;
+	}
+
+bool Hash::DeliverStream(const u_char* data, uint64 len)
+	{
+	Action::DeliverStream(data, len);
+
+	if ( ! hash->IsValid() ) return false;
+
+	hash->Feed(data, len);
+	return true;
+	}
+
+bool Hash::EndOfFile()
+	{
+	Action::EndOfFile();
+	Finalize();
+	return false;
+	}
+
+bool Hash::Undelivered(uint64 offset, uint64 len)
+	{
+	return false;
+	}
+
+void Hash::Finalize()
+	{
+	if ( ! hash->IsValid() ) return;
+
+	StringVal* sv = hash->Get();
+	int i = GetResultFieldOffset();
+
+	if ( i < 0 )
+		reporter->InternalError("Hash Action result field not found");
+
+	info->Results()->Assign(i, sv);
+	}

src/file_analysis/Hash.h

@@ -0,0 +1,88 @@
+#ifndef FILE_ANALYSIS_HASH_H
+#define FILE_ANALYSIS_HASH_H
+
+#include <string>
+
+#include "Val.h"
+#include "OpaqueVal.h"
+#include "Info.h"
+#include "Action.h"
+
+namespace file_analysis {
+
+/**
+ * An action to produce a hash of file contents.
+ */
+class Hash : public Action {
+public:
+
+	virtual ~Hash();
+
+	virtual bool DeliverStream(const u_char* data, uint64 len);
+
+	virtual bool EndOfFile();
+
+	virtual bool Undelivered(uint64 offset, uint64 len);
+
+protected:
+
+	Hash(Info* arg_info, ActionTag arg_tag, HashVal* hv);
+
+	void Finalize();
+
+	virtual int GetResultFieldOffset() const = 0;
+
+	HashVal* hash;
+};
+
+class MD5 : public Hash {
+public:
+
+	static Action* Instantiate(const RecordVal* args, Info* info)
+		{ return new MD5(info); }
+
+protected:
+
+	MD5(Info* arg_info)
+	 : Hash(arg_info, BifEnum::FileAnalysis::ACTION_MD5, new MD5Val()) {}
+
+	virtual int GetResultFieldOffset() const
+		{ return BifType::Record::FileAnalysis::ActionResults->
+		  FieldOffset("md5"); }
+};
+
+class SHA1 : public Hash {
+public:
+
+	static Action* Instantiate(const RecordVal* args, Info* info)
+		{ return new SHA1(info); }
+
+protected:
+
+	SHA1(Info* arg_info)
+	 : Hash(arg_info, BifEnum::FileAnalysis::ACTION_SHA1, new SHA1Val()) {}
+
+	virtual int GetResultFieldOffset() const
+		{ return BifType::Record::FileAnalysis::ActionResults->
+		  FieldOffset("sha1"); }
+};
+
+class SHA256 : public Hash {
+public:
+
+	static Action* Instantiate(const RecordVal* args, Info* info)
+		{ return new SHA256(info); }
+
+protected:
+
+	SHA256(Info* arg_info)
+	 : Hash(arg_info, BifEnum::FileAnalysis::ACTION_SHA256, new SHA256Val()) {}
+
+	virtual int GetResultFieldOffset() const
+		{ return BifType::Record::FileAnalysis::ActionResults->
+		  FieldOffset("sha256"); }
+};
+
+} // namespace file_analysis
+
+#endif

src/file_analysis/Info.cc

@@ -8,12 +8,16 @@
 
 #include "Action.h"
 #include "Extract.h"
+#include "Hash.h"
 
 using namespace file_analysis;
 
 // keep in order w/ declared enum values in file_analysis.bif
 static ActionInstantiator action_factory[] = {
     Extract::Instantiate,
+    MD5::Instantiate,
+    SHA1::Instantiate,
+    SHA256::Instantiate,
 };
 
 static TableVal* empty_conn_id_set()
@@ -53,6 +57,7 @@ int Info::overflow_bytes_idx = -1;
 int Info::timeout_interval_idx = -1;
 int Info::actions_idx = -1;
 int Info::action_args_idx = -1;
+int Info::action_results_idx = -1;
 
 void Info::InitFieldIndices()
 	{
@@ -69,6 +74,7 @@ void Info::InitFieldIndices()
 	timeout_interval_idx = Idx("timeout_interval");
 	actions_idx = Idx("actions");
 	action_args_idx = Idx("action_args");
+	action_results_idx = Idx("action_results");
 	}
 
 Info::Info(const string& unique, Connection* conn, const string& protocol)
@@ -149,6 +155,11 @@ double Info::TimeoutInterval() const
 	return LookupFieldDefaultInterval(timeout_interval_idx);
 	}
 
+RecordVal* Info::Results() const
+	{
+	return val->Lookup(action_results_idx)->AsRecordVal();
+	}
+
 void Info::IncrementByteCount(uint64 size, int field_idx)
 	{
 	uint64 old = LookupFieldDefaultCount(field_idx);
@@ -174,22 +185,25 @@ void Info::ScheduleInactivityTimer() const
 	timer_mgr->Add(new InfoTimer(network_time, file_id, TimeoutInterval()));
 	}
 
-bool Info::AddAction(EnumVal* act, RecordVal* args)
+bool Info::AddAction(ActionTag act, RecordVal* args)
 	{
-	if ( actions.find(act->AsEnum()) != actions.end() ) return false;
+	if ( actions.find(act) != actions.end() ) return false;
 
-	Action* a = action_factory[act->AsEnum()](args, this);
+	ActionTag tag = static_cast<ActionTag>(act);
+
+	Action* a = action_factory[act](args, this);
 
 	if ( ! a ) return false;
 
-	DBG_LOG(DBG_FILE_ANALYSIS, "Add action %d for file id %s", act->AsEnum(),
+	DBG_LOG(DBG_FILE_ANALYSIS, "Add action %d for file id %s", act,
 	        file_id.c_str());
-	actions[act->AsEnum()] = a;
+	actions[act] = a;
 
 	VectorVal* av = val->LookupWithDefault(actions_idx)->AsVectorVal();
 	VectorVal* aav = val->LookupWithDefault(action_args_idx)->AsVectorVal();
 
-	av->Assign(av->Size(), act->Ref(), 0);
+	EnumVal* ev = new EnumVal(act, BifType::Enum::FileAnalysis::Action);
+	av->Assign(av->Size(), ev, 0);
 	aav->Assign(aav->Size(), args->Ref(), 0);
 
 	Unref(av);
@@ -198,13 +212,18 @@ bool Info::AddAction(EnumVal* act, RecordVal* args)
 	return true;
 	}
 
-bool Info::RemoveAction(EnumVal* act)
+bool Info::RemoveAction(ActionTag act)
 	{
-	ActionMap::iterator it = actions.find(act->AsEnum());
+	ActionMap::iterator it = actions.find(act);
 
 	if ( it == actions.end() ) return false;
 
-	DBG_LOG(DBG_FILE_ANALYSIS, "Remove action %d for file id %s", act->AsEnum(),
+	return RemoveAction(it);
+	}
+
+bool Info::RemoveAction(const ActionMap::iterator& it)
+	{
+	DBG_LOG(DBG_FILE_ANALYSIS, "Remove action %d for file id %s", it->first,
 	        file_id.c_str());
 	delete it->second;
 	actions.erase(it);
@@ -213,12 +232,18 @@ bool Info::RemoveAction(EnumVal* act)
 
 void Info::DataIn(const u_char* data, uint64 len, uint64 offset)
 	{
-	ActionMap::const_iterator it;
-	for ( it = actions.begin(); it != actions.end(); ++it )
-		it->second->DeliverChunk(data, len, offset);
+	ActionMap::iterator it = actions.begin();
+	while ( it != actions.end() )
+		if ( ! it->second->DeliverChunk(data, len, offset) )
+			RemoveAction(it++);
+		else
+			++it;
 
 	// TODO: check reassembly requirement based on buffer size in record
-	if ( ! need_reassembly ) return;
+	if ( need_reassembly )
+		{
+		// TODO
+		}
 
 	// TODO: reassembly stuff, possibly having to deliver chunks if buffer full
 	//       and incrememt overflow bytes
@@ -228,13 +253,22 @@ void Info::DataIn(const u_char* data, uint64 len, uint64 offset)
 
 void Info::DataIn(const u_char* data, uint64 len)
 	{
-	ActionMap::const_iterator it;
-	for ( it = actions.begin(); it != actions.end(); ++it )
+	ActionMap::iterator it = actions.begin();
+	while ( it != actions.end() )
 		{
-		it->second->DeliverStream(data, len);
+		if ( ! it->second->DeliverStream(data, len) )
+			{
+			RemoveAction(it++);
+			continue;
+			}
+
 		uint64 offset = LookupFieldDefaultCount(seen_bytes_idx) +
 		                LookupFieldDefaultCount(missing_bytes_idx);
-		it->second->DeliverChunk(data, len, offset);
+
+		if ( ! it->second->DeliverChunk(data, len, offset) )
+			RemoveAction(it++);
+		else
+			++it;
 		}
 
 	IncrementByteCount(len, seen_bytes_idx);
@@ -242,16 +276,22 @@ void Info::DataIn(const u_char* data, uint64 len)
 
 void Info::EndOfFile()
 	{
-	ActionMap::const_iterator it;
-	for ( it = actions.begin(); it != actions.end(); ++it )
-		it->second->EndOfFile();
+	ActionMap::iterator it = actions.begin();
+	while ( it != actions.end() )
+		if ( ! it->second->EndOfFile() )
+			RemoveAction(it++);
+		else
+			++it;
 	}
 
 void Info::Gap(uint64 offset, uint64 len)
 	{
-	ActionMap::const_iterator it;
-	for ( it = actions.begin(); it != actions.end(); ++it )
-		it->second->Undelivered(offset, len);
+	ActionMap::iterator it = actions.begin();
+	while ( it != actions.end() )
+		if ( ! it->second->Undelivered(offset, len) )
+			RemoveAction(it++);
+		else
+			++it;
 
 	IncrementByteCount(len, missing_bytes_idx);
 	}

src/file_analysis/Info.h

@@ -29,6 +29,11 @@ public:
 	 */
 	FileID GetFileID() const { return file_id; }
 
+	/**
+	 * @return record val of the "action_results" field from #val record.
+	 */
+	RecordVal* Results() const;
+
 	/**
 	 * @return the string which uniquely identifies the file.
 	 */
@@ -67,13 +72,13 @@ public:
 	 * Attaches an action.  Only one action per type can be attached at a time.
 	 * @return true if the action was attached, else false.
 	 */
-	bool AddAction(EnumVal* act, RecordVal* args);
+	bool AddAction(ActionTag act, RecordVal* args);
 
 	/**
 	 * Removes an action.
 	 * @return true if the action was removed, else false.
 	 */
-	bool RemoveAction(EnumVal* act);
+	bool RemoveAction(ActionTag act);
 
 	/**
 	 * Pass in non-sequential data and deliver to attached actions/analyzers.
@@ -99,6 +104,8 @@ protected:
 
 	friend class Manager;
 
+	typedef map<ActionTag, Action*> ActionMap;
+
 	/**
 	 * Constructor; only file_analysis::Manager should be creating these.
 	 */
@@ -128,16 +135,19 @@ protected:
 	 */
 	double LookupFieldDefaultInterval(int idx) const;
 
+	/**
+	 * Removes an action.
+	 * @return true if the action was removed, else false.
+	 */
+	bool RemoveAction(const ActionMap::iterator& it);
+
 	FileID file_id;            /**< A pretty hash that likely identifies file*/
 	string unique;             /**< A string that uniquely identifies file */
 	RecordVal* val;            /**< \c FileAnalysis::Info from script layer. */
 	double last_activity_time; /**< Time of last activity. */
 	bool postpone_timeout;     /**< Whether postponing timeout is requested. */
 	bool need_reassembly;      /**< Whether file stream reassembly is needed. */
-
-	typedef map<int, Action*> ActionMap;
-
-	ActionMap actions;
+	ActionMap actions;         /**< Actions/analysis to perform on file. */
 
 	/**
 	 * @return the field offset in #val record corresponding to \a field_name.
@@ -161,6 +171,7 @@ protected:
 	static int timeout_interval_idx;
 	static int actions_idx;
 	static int action_args_idx;
+	static int action_results_idx;
 };
 
 } // namespace file_analysis

src/file_analysis/Manager.cc

@@ -3,6 +3,7 @@
 
 #include "Manager.h"
 #include "Info.h"
+#include "Action.h"
 
 using namespace file_analysis;
 
@@ -106,7 +107,7 @@ bool Manager::AddAction(const FileID& file_id, EnumVal* act,
 
 	if ( ! info ) return false;
 
-	return info->AddAction(act, args);
+	return info->AddAction(static_cast<ActionTag>(act->AsEnum()), args);
 	}
 
 bool Manager::RemoveAction(const FileID& file_id, EnumVal* act) const
@@ -115,7 +116,7 @@ bool Manager::RemoveAction(const FileID& file_id, EnumVal* act) const
 
 	if ( ! info ) return false;
 
-	return info->RemoveAction(act);
+	return info->RemoveAction(static_cast<ActionTag>(act->AsEnum()));
 	}
 
 Info* Manager::GetInfo(const string& unique, Connection* conn,