Merge remote-tracking branch 'origin/topic/matthias/opaque'

* origin/topic/matthias/opaque:
  Add new unit test for opaque serialization.
  Migrate entropy testing to opaque.
  C++ify RandTest.*
  Fix a hard-to-spot bug.
  Use more descriptive error message.
  Fix the fix :-/.
  Fix initialization of hash values.
  Be clearer about delegation.
  Implement serialization of opaque types.
  Update hash BiF documentation.
  Migrate free SHA* functions to SHA*Val::digest().
  Add missing type name that caused failing tests.
  Update base scripts and unit tests.
  Simplify hash function BiFs.
  Add support for opaque hash values.
  Adapt BiF & Bro parser to handle opaque types.
  More lexer/parser work.
  Implement equivalence relation for opaque types.
  Support basic serialization of opaque.
  Add opaque type to lexer, parser, and BroType.

Closes #925

Conflicts:
	aux/broccoli

CHANGES

@@ -1,4 +1,23 @@
 
+2.1-263 | 2012-12-20 16:22:09 -0800
+
+  * Bro's language now has a new set of types "opaque of X". (Matthias
+    Vallentin)
+
+    Opaque values can be passed around like other values but they can
+    only be manipulated with BiF functions, not with other operators.
+    Currently, the following opaque types are supported:
+
+        - opaque of md5
+        - opaque of sha1
+        - opaque of sha256
+        - opaquey of entropy.
+
+    They go along with the corrsponding BiF functions md5_*, sha1_*,
+    sha256_*, and entropy_*, respectively. Note that these functions
+    have changed their signatures to work with opaques types rather
+    than global state as it was before.
+    
 2.1-240 | 2012-12-20 15:21:07 -0800
 
   * Improve error for invalid use of types as values. Addresses #923.

NEWS

@@ -31,6 +31,22 @@ New Functionality
   "return" or "break" statement will fall through to subsequent cases.
   A default case label is allowed.
 
+- Bro's language now has a new set of types "opaque of X". Opaque
+  values can be passed around like other values but they can only be
+  manipulated with BiF functions, not with other operators. Currently,
+  the following opaque types are supported:
+
+        - opaque of md5
+        - opaque of sha1
+        - opaque of sha256
+        - opaquey of entropy.
+
+  They go along with the corrsponding BiF functions md5_*, sha1_*,
+  sha256_*, and entropy_*, respectively. Note that these functions
+  have changed their signatures to work with opaques types rather
+  than global state as it was before.
+
+
 Changed Functionality
 ~~~~~~~~~~~~~~~~~~~~~
 
@@ -44,6 +60,9 @@ Changed Functionality
           make_connection_persistent(), generate_idmef(),
           split_complete()
 
+        - md5_*, sha1_*, sha256_*, and entropy_* have all changed
+        their signatures to work with opaque types (see above).
+
 - Removed a now unused argument from "do_split" helper function.
 
 - "this" is no longer a reserved keyword.

VERSION

@@ -1 +1 @@
-2.1-240
+2.1-263

scripts/base/protocols/http/file-hash.bro

@@ -22,7 +22,7 @@ export {
 		
 		## Indicates if an MD5 sum is being calculated for the current 
 		## request/response pair.
-		calculating_md5: bool     &default=F;
+		md5_handle: opaque of md5   &optional;
 	};
 	
 	## Generate MD5 sums for these filetypes.
@@ -41,13 +41,12 @@ event http_entity_data(c: connection, is_orig: bool, length: count, data: string
 		if ( c$http$calc_md5 || 
 		     (c$http?$mime_type && generate_md5 in c$http$mime_type) )
 			{
-			c$http$calculating_md5 = T;
+			c$http$md5_handle = md5_hash_init();
-			md5_hash_init(c$id);
 			}
 		}
 	
-	if ( c$http$calculating_md5 )
+	if ( c$http?$md5_handle )
-		md5_hash_update(c$id, data);
+		md5_hash_update(c$http$md5_handle, data);
 	}
 	
 ## In the event of a content gap during a file transfer, detect the state for
@@ -55,11 +54,11 @@ event http_entity_data(c: connection, is_orig: bool, length: count, data: string
 ## incorrect anyway.
 event content_gap(c: connection, is_orig: bool, seq: count, length: count) &priority=5
 	{
-	if ( is_orig || ! c?$http || ! c$http$calculating_md5 ) return;
+	if ( is_orig || ! c?$http || ! c$http?$md5_handle ) return;
 	
 	set_state(c, F, is_orig);
-	c$http$calculating_md5 = F;
+	md5_hash_finish(c$http$md5_handle); # Ignore return value.
-	md5_hash_finish(c$id);
+	delete c$http$md5_handle;
 	}
 
 ## When the file finishes downloading, finish the hash and generate a notice.
@@ -67,11 +66,11 @@ event http_message_done(c: connection, is_orig: bool, stat: http_message_stat) &
 	{
 	if ( is_orig || ! c?$http ) return;
 	
-	if ( c$http$calculating_md5 )
+	if ( c$http?$md5_handle )
 		{
 		local url = build_url_http(c$http);
-		c$http$calculating_md5 = F;
+		c$http$md5 = md5_hash_finish(c$http$md5_handle);
-		c$http$md5 = md5_hash_finish(c$id);
+		delete c$http$md5_handle;
 		
 		NOTICE([$note=MD5, $msg=fmt("%s %s %s", c$id$orig_h, c$http$md5, url),
 		        $sub=c$http$md5, $conn=c, $URL=url]);
@@ -82,11 +81,12 @@ event connection_state_remove(c: connection) &priority=-5
 	{
 	if ( c?$http_state && 
 	     c$http_state$current_response in c$http_state$pending &&
-	     c$http_state$pending[c$http_state$current_response]$calculating_md5 )
+	     c$http_state$pending[c$http_state$current_response]?$md5_handle )
 		{
 		# The MD5 sum isn't going to be saved anywhere since the entire 
 		# body wouldn't have been seen anyway and we'd just be giving an
 		# incorrect MD5 sum.
-		md5_hash_finish(c$id);
+		md5_hash_finish(c$http$md5_handle);
+		delete c$http$md5_handle;
 		}
 	}

scripts/base/protocols/smtp/entities.bro

@@ -36,7 +36,7 @@ export {
 		calc_md5:         bool            &default=F;
 		## This boolean value indicates if an MD5 sum is being calculated 
 		## for the current file transfer.
-		calculating_md5:  bool    &default=F;
+		md5_handle:       opaque of md5   &optional;
 		
 		## Optionally write the file to disk.  Must be set prior to first 
 		## data chunk being seen in an event.
@@ -126,18 +126,16 @@ event mime_segment_data(c: connection, length: count, data: string) &priority=-5
 
 	if ( c$smtp$current_entity$content_len == 0 )
 		{
-		if ( generate_md5 in c$smtp$current_entity$mime_type && ! never_calc_md5 )
+		local entity = c$smtp$current_entity;
-			c$smtp$current_entity$calc_md5 = T;
+		if ( generate_md5 in entity$mime_type && ! never_calc_md5 )
+			entity$calc_md5 = T;
 
-		if ( c$smtp$current_entity$calc_md5 )
+		if ( entity$calc_md5 )
-			{
+			entity$md5_handle = md5_hash_init();
-			c$smtp$current_entity$calculating_md5 = T;
-			md5_hash_init(c$id);
-			}
 		}
 
-	if ( c$smtp$current_entity$calculating_md5 )
+	if ( c$smtp$current_entity?$md5_handle )
-		md5_hash_update(c$id, data);
+		md5_hash_update(entity$md5_handle, data);
 }
 
 ## In the event of a content gap during the MIME transfer, detect the state for
@@ -147,10 +145,11 @@ event content_gap(c: connection, is_orig: bool, seq: count, length: count) &prio
 	{
 	if ( is_orig || ! c?$smtp || ! c$smtp?$current_entity ) return;
 
-	if ( c$smtp$current_entity$calculating_md5 )
+	local entity = c$smtp$current_entity;
+	if ( entity?$md5_handle )
 		{
-		c$smtp$current_entity$calculating_md5 = F;
+		md5_hash_finish(entity$md5_handle);
-		md5_hash_finish(c$id);
+		delete entity$md5_handle;
 		}
 	}
 
@@ -161,12 +160,14 @@ event mime_end_entity(c: connection) &priority=-3
 	if ( ! c?$smtp || ! c$smtp?$current_entity )
 		return;
 
-	if ( c$smtp$current_entity$calculating_md5 )
+	local entity = c$smtp$current_entity;
+	if ( entity?$md5_handle )
 		{
-		c$smtp$current_entity$md5 = md5_hash_finish(c$id);
+		entity$md5 = md5_hash_finish(entity$md5_handle);
+		delete entity$md5_handle;
 
 		NOTICE([$note=MD5, $msg=fmt("Calculated a hash for a MIME entity from %s", c$id$orig_h),
-				$sub=c$smtp$current_entity$md5, $conn=c]);
+				$sub=entity$md5, $conn=c]);
 		}
 	}
 

src/CMakeLists.txt

@@ -361,6 +361,7 @@ set(bro_SRCS
     NetVar.cc
     NetbiosSSN.cc
     Obj.cc
+    OpaqueVal.cc
     OSFinger.cc
     PacketFilter.cc
     PacketSort.cc

src/OpaqueVal.cc

@@ -0,0 +1,501 @@
+#include "OpaqueVal.h"
+#include "Reporter.h"
+#include "Serializer.h"
+
+bool HashVal::IsValid() const
+	{
+	return valid;
+	}
+
+bool HashVal::Init()
+	{
+	if ( valid )
+		return false;
+
+	valid = DoInit();
+	return valid;
+	}
+
+StringVal* HashVal::Get()
+	{
+	if ( ! valid )
+		return new StringVal("");
+
+	StringVal* result = DoGet();
+	valid = false;
+	return result;
+	}
+
+bool HashVal::Feed(const void* data, size_t size)
+	{
+	if ( valid )
+		return DoFeed(data, size);
+
+	reporter->InternalError("invalid opaque hash value");
+	return false;
+	}
+
+bool HashVal::DoInit()
+	{
+	assert(! "missing implementation of DoInit()");
+	return false;
+	}
+
+bool HashVal::DoFeed(const void*, size_t)
+	{
+	assert(! "missing implementation of DoFeed()");
+	return false;
+	}
+
+StringVal* HashVal::DoGet()
+	{
+	assert(! "missing implementation of DoGet()");
+	return new StringVal("");
+	}
+
+HashVal::HashVal(OpaqueType* t) : OpaqueVal(t)
+	{
+	valid = false;
+	}
+
+IMPLEMENT_SERIAL(HashVal, SER_HASH_VAL);
+
+bool HashVal::DoSerialize(SerialInfo* info) const
+	{
+	DO_SERIALIZE(SER_HASH_VAL, OpaqueVal);
+	return SERIALIZE(valid);
+	}
+
+bool HashVal::DoUnserialize(UnserialInfo* info)
+	{
+	DO_UNSERIALIZE(OpaqueVal);
+	return UNSERIALIZE(&valid);
+	}
+
+void MD5Val::digest(val_list& vlist, u_char result[MD5_DIGEST_LENGTH])
+	{
+	MD5_CTX h;
+	md5_init(&h);
+
+	loop_over_list(vlist, i)
+		{
+		Val* v = vlist[i];
+		if ( v->Type()->Tag() == TYPE_STRING )
+			{
+			const BroString* str = v->AsString();
+			md5_update(&h, str->Bytes(), str->Len());
+			}
+		else
+			{
+			ODesc d(DESC_BINARY);
+			v->Describe(&d);
+			md5_update(&h, (const u_char *) d.Bytes(), d.Len());
+			}
+		}
+
+	md5_final(&h, result);
+	}
+
+void MD5Val::hmac(val_list& vlist,
+                  u_char key[MD5_DIGEST_LENGTH],
+                  u_char result[MD5_DIGEST_LENGTH])
+	{
+	digest(vlist, result);
+	for ( int i = 0; i < MD5_DIGEST_LENGTH; ++i )
+		result[i] ^= key[i];
+
+	MD5(result, MD5_DIGEST_LENGTH, result);
+	}
+
+bool MD5Val::DoInit()
+	{
+	assert(! IsValid());
+	md5_init(&ctx);
+	return true;
+	}
+
+bool MD5Val::DoFeed(const void* data, size_t size)
+	{
+	if ( ! IsValid() )
+		return false;
+
+	md5_update(&ctx, data, size);
+	return true;
+	}
+
+StringVal* MD5Val::DoGet()
+	{
+	if ( ! IsValid() )
+		return new StringVal("");
+
+	u_char digest[MD5_DIGEST_LENGTH];
+	md5_final(&ctx, digest);
+	return new StringVal(md5_digest_print(digest));
+	}
+
+IMPLEMENT_SERIAL(MD5Val, SER_MD5_VAL);
+
+bool MD5Val::DoSerialize(SerialInfo* info) const
+	{
+	DO_SERIALIZE(SER_MD5_VAL, HashVal);
+
+	if ( ! IsValid() )
+		return true;
+
+	if ( ! (SERIALIZE(ctx.A) &&
+		SERIALIZE(ctx.B) &&
+		SERIALIZE(ctx.C) &&
+		SERIALIZE(ctx.D) &&
+		SERIALIZE(ctx.Nl) &&
+		SERIALIZE(ctx.Nh)) )
+		return false;
+
+	for ( int i = 0; i < MD5_LBLOCK; ++i )
+		{
+		if ( ! SERIALIZE(ctx.data[i]) )
+			return false;
+		}
+
+	if ( ! SERIALIZE(ctx.num) )
+		return false;
+
+	return true;
+	}
+
+bool MD5Val::DoUnserialize(UnserialInfo* info)
+	{
+	DO_UNSERIALIZE(HashVal);
+
+	if ( ! IsValid() )
+		return true;
+
+	if ( ! (UNSERIALIZE(&ctx.A) &&
+		UNSERIALIZE(&ctx.B) &&
+		UNSERIALIZE(&ctx.C) &&
+		UNSERIALIZE(&ctx.D) &&
+		UNSERIALIZE(&ctx.Nl) &&
+		UNSERIALIZE(&ctx.Nh)) )
+		return false;
+
+	for ( int i = 0; i < MD5_LBLOCK; ++i )
+		{
+		if ( ! UNSERIALIZE(&ctx.data[i]) )
+			return false;
+		}
+
+	if ( ! UNSERIALIZE(&ctx.num) )
+		return false;
+
+	return true;
+	}
+
+void SHA1Val::digest(val_list& vlist, u_char result[SHA_DIGEST_LENGTH])
+	{
+	SHA_CTX h;
+	sha1_init(&h);
+
+	loop_over_list(vlist, i)
+		{
+		Val* v = vlist[i];
+		if ( v->Type()->Tag() == TYPE_STRING )
+			{
+			const BroString* str = v->AsString();
+			sha1_update(&h, str->Bytes(), str->Len());
+			}
+		else
+			{
+			ODesc d(DESC_BINARY);
+			v->Describe(&d);
+			sha1_update(&h, (const u_char *) d.Bytes(), d.Len());
+			}
+		}
+
+	sha1_final(&h, result);
+	}
+
+bool SHA1Val::DoInit()
+	{
+	assert(! IsValid());
+	sha1_init(&ctx);
+	return true;
+	}
+
+bool SHA1Val::DoFeed(const void* data, size_t size)
+	{
+	if ( ! IsValid() )
+		return false;
+
+	sha1_update(&ctx, data, size);
+	return true;
+	}
+
+StringVal* SHA1Val::DoGet()
+	{
+	if ( ! IsValid() )
+		return new StringVal("");
+
+	u_char digest[SHA_DIGEST_LENGTH];
+	sha1_final(&ctx, digest);
+	return new StringVal(sha1_digest_print(digest));
+	}
+
+IMPLEMENT_SERIAL(SHA1Val, SER_SHA1_VAL);
+
+bool SHA1Val::DoSerialize(SerialInfo* info) const
+	{
+	DO_SERIALIZE(SER_SHA1_VAL, HashVal);
+
+	if ( ! IsValid() )
+		return true;
+
+	if ( ! (SERIALIZE(ctx.h0) &&
+		SERIALIZE(ctx.h1) &&
+		SERIALIZE(ctx.h2) &&
+		SERIALIZE(ctx.h3) &&
+		SERIALIZE(ctx.h4) &&
+		SERIALIZE(ctx.Nl) &&
+		SERIALIZE(ctx.Nh)) )
+		return false;
+
+	for ( int i = 0; i < SHA_LBLOCK; ++i )
+		{
+		if ( ! SERIALIZE(ctx.data[i]) )
+			return false;
+		}
+
+	if ( ! SERIALIZE(ctx.num) )
+		return false;
+
+	return true;
+	}
+
+bool SHA1Val::DoUnserialize(UnserialInfo* info)
+	{
+	DO_UNSERIALIZE(HashVal);
+
+	if ( ! IsValid() )
+		return true;
+
+	if ( ! (UNSERIALIZE(&ctx.h0) &&
+		UNSERIALIZE(&ctx.h1) &&
+		UNSERIALIZE(&ctx.h2) &&
+		UNSERIALIZE(&ctx.h3) &&
+		UNSERIALIZE(&ctx.h4) &&
+		UNSERIALIZE(&ctx.Nl) &&
+		UNSERIALIZE(&ctx.Nh)) )
+		return false;
+
+	for ( int i = 0; i < SHA_LBLOCK; ++i )
+		{
+		if ( ! UNSERIALIZE(&ctx.data[i]) )
+			return false;
+		}
+
+	if ( ! UNSERIALIZE(&ctx.num) )
+		return false;
+
+	return true;
+	}
+
+void SHA256Val::digest(val_list& vlist, u_char result[SHA256_DIGEST_LENGTH])
+	{
+	SHA256_CTX h;
+	sha256_init(&h);
+
+	loop_over_list(vlist, i)
+		{
+		Val* v = vlist[i];
+		if ( v->Type()->Tag() == TYPE_STRING )
+			{
+			const BroString* str = v->AsString();
+			sha256_update(&h, str->Bytes(), str->Len());
+			}
+		else
+			{
+			ODesc d(DESC_BINARY);
+			v->Describe(&d);
+			sha256_update(&h, (const u_char *) d.Bytes(), d.Len());
+			}
+		}
+
+	sha256_final(&h, result);
+	}
+
+bool SHA256Val::DoInit()
+	{
+	assert( ! IsValid() );
+	sha256_init(&ctx);
+	return true;
+	}
+
+bool SHA256Val::DoFeed(const void* data, size_t size)
+	{
+	if ( ! IsValid() )
+		return false;
+
+	sha256_update(&ctx, data, size);
+	return true;
+	}
+
+StringVal* SHA256Val::DoGet()
+	{
+	if ( ! IsValid() )
+		return new StringVal("");
+
+	u_char digest[SHA256_DIGEST_LENGTH];
+	sha256_final(&ctx, digest);
+	return new StringVal(sha256_digest_print(digest));
+	}
+
+IMPLEMENT_SERIAL(SHA256Val, SER_SHA256_VAL);
+
+bool SHA256Val::DoSerialize(SerialInfo* info) const
+	{
+	DO_SERIALIZE(SER_SHA256_VAL, HashVal);
+
+	if ( ! IsValid() )
+		return true;
+
+	for ( int i = 0; i < 8; ++i )
+		{
+		if ( ! SERIALIZE(ctx.h[i]) )
+			return false;
+		}
+
+	if ( ! (SERIALIZE(ctx.Nl) &&
+		SERIALIZE(ctx.Nh)) )
+		return false;
+
+	for ( int i = 0; i < SHA_LBLOCK; ++i )
+		{
+		if ( ! SERIALIZE(ctx.data[i]) )
+			return false;
+		}
+
+	if ( ! (SERIALIZE(ctx.num) &&
+		SERIALIZE(ctx.md_len)) )
+	     return false;
+
+	return true;
+	}
+
+bool SHA256Val::DoUnserialize(UnserialInfo* info)
+	{
+	DO_UNSERIALIZE(HashVal);
+
+	if ( ! IsValid() )
+		return true;
+
+	for ( int i = 0; i < 8; ++i )
+		{
+		if ( ! UNSERIALIZE(&ctx.h[i]) )
+			return false;
+		}
+
+	if ( ! (UNSERIALIZE(&ctx.Nl) &&
+		UNSERIALIZE(&ctx.Nh)) )
+	     return false;
+
+	for ( int i = 0; i < SHA_LBLOCK; ++i )
+		{
+		if ( ! UNSERIALIZE(&ctx.data[i]) )
+			return false;
+		}
+
+
+	if ( ! (UNSERIALIZE(&ctx.num) &&
+		UNSERIALIZE(&ctx.md_len)) )
+		return false;
+
+	return true;
+	}
+
+
+bool EntropyVal::Feed(const void* data, size_t size)
+	{
+	state.add(data, size);
+	return true;
+	}
+
+bool EntropyVal::Get(double *r_ent, double *r_chisq, double *r_mean,
+                     double *r_montepicalc, double *r_scc)
+	{
+	state.end(r_ent, r_chisq, r_mean, r_montepicalc, r_scc);
+	return true;
+	}
+
+IMPLEMENT_SERIAL(EntropyVal, SER_ENTROPY_VAL);
+
+bool EntropyVal::DoSerialize(SerialInfo* info) const
+	{
+	DO_SERIALIZE(SER_ENTROPY_VAL, OpaqueVal);
+
+	for ( int i = 0; i < 256; ++i )
+		{
+		if ( ! SERIALIZE(state.ccount[i]) )
+			return false;
+		}
+
+	if ( ! (SERIALIZE(state.totalc) &&
+		SERIALIZE(state.mp) &&
+		SERIALIZE(state.sccfirst)) )
+		return false;
+
+	for ( int i = 0; i < RT_MONTEN; ++i )
+		{
+		if ( ! SERIALIZE(state.monte[i]) )
+			return false;
+		}
+
+	if ( ! (SERIALIZE(state.inmont) &&
+		SERIALIZE(state.mcount) &&
+		SERIALIZE(state.cexp) &&
+		SERIALIZE(state.montex) &&
+		SERIALIZE(state.montey) &&
+		SERIALIZE(state.montepi) &&
+		SERIALIZE(state.sccu0) &&
+		SERIALIZE(state.scclast) &&
+		SERIALIZE(state.scct1) &&
+		SERIALIZE(state.scct2) &&
+		SERIALIZE(state.scct3)) )
+		return false;
+
+	return true;
+	}
+
+bool EntropyVal::DoUnserialize(UnserialInfo* info)
+	{
+	DO_UNSERIALIZE(OpaqueVal);
+
+	for ( int i = 0; i < 256; ++i )
+		{
+		if ( ! UNSERIALIZE(&state.ccount[i]) )
+			return false;
+		}
+
+	if ( ! (UNSERIALIZE(&state.totalc) &&
+		UNSERIALIZE(&state.mp) &&
+		UNSERIALIZE(&state.sccfirst)) )
+		return false;
+
+	for ( int i = 0; i < RT_MONTEN; ++i )
+		{
+		if ( ! UNSERIALIZE(&state.monte[i]) )
+			return false;
+		}
+
+	if ( ! (UNSERIALIZE(&state.inmont) &&
+		UNSERIALIZE(&state.mcount) &&
+		UNSERIALIZE(&state.cexp) &&
+		UNSERIALIZE(&state.montex) &&
+		UNSERIALIZE(&state.montey) &&
+		UNSERIALIZE(&state.montepi) &&
+		UNSERIALIZE(&state.sccu0) &&
+		UNSERIALIZE(&state.scclast) &&
+		UNSERIALIZE(&state.scct1) &&
+		UNSERIALIZE(&state.scct2) &&
+		UNSERIALIZE(&state.scct3)) )
+		return false;
+
+	return true;
+	}

src/OpaqueVal.h

@@ -0,0 +1,110 @@
+// See the file "COPYING" in the main distribution directory for copyright.
+
+#ifndef OPAQUEVAL_H
+#define OPAQUEVAL_H
+
+#include "RandTest.h"
+#include "Val.h"
+#include "digest.h"
+
+class HashVal : public OpaqueVal {
+public:
+	virtual bool IsValid() const;
+	virtual bool Init();
+	virtual bool Feed(const void* data, size_t size);
+	virtual StringVal* Get();
+
+protected:
+	HashVal() { };
+	HashVal(OpaqueType* t);
+	virtual bool DoInit();
+	virtual bool DoFeed(const void* data, size_t size);
+	virtual StringVal* DoGet();
+
+	DECLARE_SERIAL(HashVal);
+
+private:
+	// This flag exists because Get() can only be called once.
+	bool valid;
+};
+
+class MD5Val : public HashVal {
+public:
+	static void digest(val_list& vlist, u_char result[MD5_DIGEST_LENGTH]);
+
+	static void hmac(val_list& vlist,
+			 u_char key[MD5_DIGEST_LENGTH],
+			 u_char result[MD5_DIGEST_LENGTH]);
+
+	MD5Val() : HashVal(new OpaqueType("md5"))	{ }
+
+protected:
+	friend class Val;
+
+	virtual bool DoInit() /* override */;
+	virtual bool DoFeed(const void* data, size_t size) /* override */;
+	virtual StringVal* DoGet() /* override */;
+
+	DECLARE_SERIAL(MD5Val);
+
+private:
+	MD5_CTX ctx;
+};
+
+class SHA1Val : public HashVal {
+public:
+	static void digest(val_list& vlist, u_char result[SHA_DIGEST_LENGTH]);
+
+	SHA1Val() : HashVal(new OpaqueType("sha1")) { }
+
+protected:
+	friend class Val;
+
+	virtual bool DoInit() /* override */;
+	virtual bool DoFeed(const void* data, size_t size) /* override */;
+	virtual StringVal* DoGet() /* override */;
+
+	DECLARE_SERIAL(SHA1Val);
+
+private:
+	SHA_CTX ctx;
+};
+
+class SHA256Val : public HashVal {
+public:
+	static void digest(val_list& vlist, u_char result[SHA256_DIGEST_LENGTH]);
+
+	SHA256Val() : HashVal(new OpaqueType("sha256"))	{ }
+
+protected:
+	friend class Val;
+
+	virtual bool DoInit() /* override */;
+	virtual bool DoFeed(const void* data, size_t size) /* override */;
+	virtual StringVal* DoGet() /* override */;
+
+	DECLARE_SERIAL(SHA256Val);
+
+private:
+	SHA256_CTX ctx;
+};
+
+class EntropyVal : public OpaqueVal {
+public:
+	EntropyVal() : OpaqueVal(new OpaqueType("entropy"))	{ }
+
+	bool Feed(const void* data, size_t size);
+	bool Get(double *r_ent, double *r_chisq, double *r_mean,
+		 double *r_montepicalc, double *r_scc);
+
+protected:
+	friend class Val;
+	EntropyVal(OpaqueType* t);
+
+	DECLARE_SERIAL(EntropyVal);
+
+private:
+	RandTest state;
+};
+
+#endif

src/RandTest.cc

@@ -12,7 +12,18 @@
 	Modified for Bro by Seth Hall - July 2010
 */
 
-#include <RandTest.h>
+#include <math.h>
+#include "RandTest.h"
+
+#define log2of10 3.32192809488736234787
+/*  RT_LOG2  --  Calculate log to the base 2  */
+static double rt_log2(double x)
+{
+    return log2of10 * log10(x);
+}
+
+// RT_INCIRC = pow(pow(256.0, (double) (RT_MONTEN / 2)) - 1, 2.0);
+#define RT_INCIRC 281474943156225.0
 
 RandTest::RandTest()
 	{
@@ -28,9 +39,9 @@ RandTest::RandTest()
 		}
 	}
 
-void RandTest::add(void *buf, int bufl)
+void RandTest::add(const void *buf, int bufl)
 	{
-	unsigned char *bp = (unsigned char*)buf;
+	const unsigned char *bp = static_cast<const unsigned char*>(buf);
 	int oc;
 
 	while (bufl-- > 0)
@@ -78,8 +89,8 @@ void RandTest::add(void *buf, int bufl)
 		}
 	}
 
-void RandTest::end(double *r_ent, double *r_chisq,
+void RandTest::end(double* r_ent, double* r_chisq,
-                   double *r_mean, double *r_montepicalc, double *r_scc)
+                   double* r_mean, double* r_montepicalc, double* r_scc)
 	{
 	int i;
 	double ent, chisq, scc, datasum;

src/RandTest.h

@@ -1,34 +1,33 @@
-#include <math.h>
+#ifndef RANDTEST_H
+#define RANDTEST_H
 
-#define log2of10 3.32192809488736234787
+#include "util.h"
-/*  RT_LOG2  --  Calculate log to the base 2  */
-static double rt_log2(double x)
-{
-    return log2of10 * log10(x);
-}
 
 #define RT_MONTEN 6  /* Bytes used as Monte Carlo
                         co-ordinates. This should be no more
                         bits than the mantissa of your "double"
                         floating point type. */
+class EntropyVal;
 
-// RT_INCIRC = pow(pow(256.0, (double) (RT_MONTEN / 2)) - 1, 2.0);
-#define RT_INCIRC 281474943156225.0
 
 class RandTest {
 	public:
 		RandTest();
-		void add(void *buf, int bufl);
+		void add(const void* buf, int bufl);
-		void end(double *r_ent, double *r_chisq, double *r_mean,
+		void end(double* r_ent, double* r_chisq, double* r_mean,
-		         double *r_montepicalc, double *r_scc);
+		         double* r_montepicalc, double* r_scc);
 
 	private:
-		long ccount[256];  /* Bins to count occurrences of values */
+	  friend class EntropyVal;
-		long totalc;       /* Total bytes counted */
+
+		int64 ccount[256];  /* Bins to count occurrences of values */
+		int64 totalc;       /* Total bytes counted */
 		int mp;
 		int sccfirst;
 		unsigned int monte[RT_MONTEN];
-		long inmont, mcount;
+		int64 inmont, mcount;
 		double cexp, montex, montey, montepi,
 		       sccu0, scclast, scct1, scct2, scct3;
-	};
+};
+
+#endif

src/SerialTypes.h

@@ -98,6 +98,12 @@ SERIAL_VAL(RECORD_VAL, 10)
 SERIAL_VAL(ENUM_VAL, 11)
 SERIAL_VAL(VECTOR_VAL, 12)
 SERIAL_VAL(MUTABLE_VAL, 13)
+SERIAL_VAL(OPAQUE_VAL, 14)
+SERIAL_VAL(HASH_VAL, 15)
+SERIAL_VAL(MD5_VAL, 16)
+SERIAL_VAL(SHA1_VAL, 17)
+SERIAL_VAL(SHA256_VAL, 18)
+SERIAL_VAL(ENTROPY_VAL, 19)
 
 #define SERIAL_EXPR(name, val) SERIAL_CONST(name, val, EXPR)
 SERIAL_EXPR(EXPR, 1)
@@ -178,6 +184,7 @@ SERIAL_TYPE(SUBNET_TYPE, 8)
 SERIAL_TYPE(FILE_TYPE, 9)
 SERIAL_TYPE(ENUM_TYPE, 10)
 SERIAL_TYPE(VECTOR_TYPE, 11)
+SERIAL_TYPE(OPAQUE_TYPE, 12)
 
 SERIAL_CONST2(ATTRIBUTES)
 SERIAL_CONST2(EVENT_HANDLER)

src/Type.cc

@@ -30,6 +30,7 @@ const char* type_name(TypeTag t)
 		"table", "union", "record", "types",
 		"func",
 		"file",
+		"opaque",
 		"vector",
 		"type",
 		"error",
@@ -96,6 +97,7 @@ BroType::BroType(TypeTag t, bool arg_base_type)
 	case TYPE_LIST:
 	case TYPE_FUNC:
 	case TYPE_FILE:
+	case TYPE_OPAQUE:
 	case TYPE_VECTOR:
 	case TYPE_TYPE:
 		internal_tag = TYPE_INTERNAL_OTHER;
@@ -1262,6 +1264,41 @@ bool FileType::DoUnserialize(UnserialInfo* info)
 	return yield != 0;
 	}
 
+OpaqueType::OpaqueType(const string& arg_name) : BroType(TYPE_OPAQUE)
+	{
+	name = arg_name;
+	}
+
+void OpaqueType::Describe(ODesc* d) const
+	{
+	if ( d->IsReadable() )
+		d->AddSP("opaque of");
+	else
+		d->Add(int(Tag()));
+
+	d->Add(name.c_str());
+	}
+
+IMPLEMENT_SERIAL(OpaqueType, SER_OPAQUE_TYPE);
+
+bool OpaqueType::DoSerialize(SerialInfo* info) const
+	{
+	DO_SERIALIZE(SER_OPAQUE_TYPE, BroType);
+	return SERIALIZE(name);
+	}
+
+bool OpaqueType::DoUnserialize(UnserialInfo* info)
+	{
+	DO_UNSERIALIZE(BroType);
+
+	char const* n;
+	if ( ! UNSERIALIZE_STR(&n, 0) )
+		return false;
+
+	name = n;
+	return true;
+	}
+
 EnumType::EnumType(const string& arg_name)
 : BroType(TYPE_ENUM)
 	{
@@ -1716,6 +1753,13 @@ int same_type(const BroType* t1, const BroType* t2, int is_init)
 	case TYPE_FILE:
 		return same_type(t1->YieldType(), t2->YieldType(), is_init);
 
+	case TYPE_OPAQUE:
+		{
+		const OpaqueType* ot1 = (const OpaqueType*) t1;
+		const OpaqueType* ot2 = (const OpaqueType*) t2;
+		return ot1->Name() == ot2->Name() ? 1 : 0;
+		}
+
 	case TYPE_TYPE:
 		return same_type(t1, t2, is_init);
 
@@ -1805,6 +1849,7 @@ int is_assignable(BroType* t)
 
 	case TYPE_VECTOR:
 	case TYPE_FILE:
+	case TYPE_OPAQUE:
 	case TYPE_TABLE:
 	case TYPE_TYPE:
 		return 1;

src/Type.h

@@ -29,6 +29,7 @@ typedef enum {
 	TYPE_LIST,
 	TYPE_FUNC,
 	TYPE_FILE,
+	TYPE_OPAQUE,
 	TYPE_VECTOR,
 	TYPE_TYPE,
 	TYPE_ERROR
@@ -499,6 +500,23 @@ protected:
 	BroType* yield;
 };
 
+class OpaqueType : public BroType {
+public:
+	OpaqueType(const string& name);
+	virtual ~OpaqueType() { };
+
+	const string& Name() const { return name; }
+
+	void Describe(ODesc* d) const;
+
+protected:
+	OpaqueType() { }
+
+	DECLARE_SERIAL(OpaqueType)
+
+	string name;
+};
+
 class EnumType : public BroType {
 public:
 	EnumType(const string& arg_name);

src/Val.cc

@@ -3114,6 +3114,27 @@ void VectorVal::ValDescribe(ODesc* d) const
 	d->Add("]");
 	}
 
+OpaqueVal::OpaqueVal(OpaqueType* t) : Val(t)
+	{
+	}
+
+OpaqueVal::~OpaqueVal()
+	{
+	}
+
+IMPLEMENT_SERIAL(OpaqueVal, SER_OPAQUE_VAL);
+
+bool OpaqueVal::DoSerialize(SerialInfo* info) const
+	{
+	DO_SERIALIZE(SER_OPAQUE_VAL, Val);
+	return true;
+	}
+
+bool OpaqueVal::DoUnserialize(UnserialInfo* info)
+	{
+	DO_UNSERIALIZE(Val);
+	return true;
+	}
 
 Val* check_and_promote(Val* v, const BroType* t, int is_init)
 	{

src/Val.h

@@ -1013,6 +1013,20 @@ protected:
 	VectorType* vector_type;
 };
 
+// Base class for values with types that are managed completely internally,
+// with no further script-level operators provided (other than bif
+// functions). See OpaqueVal.h for derived classes.
+class OpaqueVal : public Val {
+public:
+	OpaqueVal(OpaqueType* t);
+	virtual ~OpaqueVal();
+
+protected:
+	friend class Val;
+	OpaqueVal() { }
+
+	DECLARE_SERIAL(OpaqueVal);
+};
 
 // Checks the given value for consistency with the given type.  If an
 // exact match, returns it.  If promotable, returns the promoted version,

src/bro.bif

@@ -530,82 +530,7 @@ function piped_exec%(program: string, to_write: string%): bool
 	%}
 
 %%{
-static void hash_md5_val(val_list& vlist, unsigned char digest[16])
+#include "OpaqueVal.h"
-	{
-	MD5_CTX h;
-
-	md5_init(&h);
-	loop_over_list(vlist, i)
-		{
-		Val* v = vlist[i];
-		if ( v->Type()->Tag() == TYPE_STRING )
-			{
-			const BroString* str = v->AsString();
-			md5_update(&h, str->Bytes(), str->Len());
-			}
-		else
-			{
-			ODesc d(DESC_BINARY);
-			v->Describe(&d);
-			md5_update(&h, (const u_char *) d.Bytes(), d.Len());
-			}
-		}
-	md5_final(&h, digest);
-	}
-
-static void hmac_md5_val(val_list& vlist, unsigned char digest[16])
-	{
-	hash_md5_val(vlist, digest);
-	for ( int i = 0; i < 16; ++i )
-		digest[i] = digest[i] ^ shared_hmac_md5_key[i];
-	MD5(digest, 16, digest);
-	}
-
-static void hash_sha1_val(val_list& vlist, unsigned char digest[20])
-	{
-	SHA_CTX h;
-
-	sha1_init(&h);
-	loop_over_list(vlist, i)
-		{
-		Val* v = vlist[i];
-		if ( v->Type()->Tag() == TYPE_STRING )
-			{
-			const BroString* str = v->AsString();
-			sha1_update(&h, str->Bytes(), str->Len());
-			}
-		else
-			{
-			ODesc d(DESC_BINARY);
-			v->Describe(&d);
-			sha1_update(&h, (const u_char *) d.Bytes(), d.Len());
-			}
-		}
-	sha1_final(&h, digest);
-	}
-
-static void hash_sha256_val(val_list& vlist, unsigned char digest[32])
-	{
-	SHA256_CTX h;
-
-	sha256_init(&h);
-	loop_over_list(vlist, i)
-		{
-		Val* v = vlist[i];
-		if ( v->Type()->Tag() == TYPE_STRING )
-			{
-			const BroString* str = v->AsString();
-			sha256_update(&h, str->Bytes(), str->Len());
-			}
-		else
-			{
-			ODesc d(DESC_BINARY);
-			v->Describe(&d);
-			sha256_update(&h, (const u_char *) d.Bytes(), d.Len());
-			}
-		}
-	sha256_final(&h, digest);
-	}
 %%}
 
 ## Computes the MD5 hash value of the provided list of arguments.
@@ -623,8 +548,8 @@ static void hash_sha256_val(val_list& vlist, unsigned char digest[32])
 ##      friends.
 function md5_hash%(...%): string
 	%{
-	unsigned char digest[16];
+	unsigned char digest[MD5_DIGEST_LENGTH];
-	hash_md5_val(@ARG@, digest);
+	MD5Val::digest(@ARG@, digest);
 	return new StringVal(md5_digest_print(digest));
 	%}
 
@@ -643,8 +568,8 @@ function md5_hash%(...%): string
 ##      friends.
 function sha1_hash%(...%): string
 	%{
-	unsigned char digest[20];
+	unsigned char digest[SHA_DIGEST_LENGTH];
-	hash_sha1_val(@ARG@, digest);
+	SHA1Val::digest(@ARG@, digest);
 	return new StringVal(sha1_digest_print(digest));
 	%}
 
@@ -663,8 +588,8 @@ function sha1_hash%(...%): string
 ##      friends.
 function sha256_hash%(...%): string
 	%{
-	unsigned char digest[32];
+	unsigned char digest[SHA256_DIGEST_LENGTH];
-	hash_sha256_val(@ARG@, digest);
+	SHA256Val::digest(@ARG@, digest);
 	return new StringVal(sha256_digest_print(digest));
 	%}
 
@@ -679,288 +604,183 @@ function sha256_hash%(...%): string
 ##    sha256_hash sha256_hash_init sha256_hash_update sha256_hash_finish
 function md5_hmac%(...%): string
 	%{
-	unsigned char digest[16];
+	unsigned char hmac[MD5_DIGEST_LENGTH];
-	hmac_md5_val(@ARG@, digest);
+	MD5Val::hmac(@ARG@, shared_hmac_md5_key, hmac);
-	return new StringVal(md5_digest_print(digest));
+	return new StringVal(md5_digest_print(hmac));
 	%}
 
-%%{
+## Constructs an MD5 handle to enable incremental hash computation. You can
-static map<BroString, MD5_CTX> md5_states;
+## feed data to the returned opaque value with :bro:id:`md5_hash_update` and
-static map<BroString, SHA_CTX> sha1_states;
+## eventually need to call :bro:id:`md5_hash_finish` to finish the computation
-static map<BroString, SHA256_CTX> sha256_states;
+## and get the hash digest as result.
-
-BroString* convert_index_to_string(Val* index)
-	{
-	ODesc d;
-	index->Describe(&d);
-	BroString* s = new BroString(1, d.TakeBytes(), d.Len());
-	s->SetUseFreeToDelete(1);
-	return s;
-	}
-%%}
-
-## Initializes MD5 state to enable incremental hash computation. After
-## initializing the MD5 state with this function, you can feed data to
-## :bro:id:`md5_hash_update` and finally need to call :bro:id:`md5_hash_finish`
-## to finish the computation and get the final hash value.
 ##
 ## For example, when computing incremental MD5 values of transferred files in
-## multiple concurrent HTTP connections, one would call ``md5_hash_init(c$id)``
+## multiple concurrent HTTP connections, one keeps an optional handle in the
-## once before invoking ``md5_hash_update(c$id, some_more_data)`` in the
+## HTTP session record. Then, one would call
+## ``c$http$md5_handle = md5_hash_init()`` once before invoking
+## ``md5_hash_update(c$http$md5_handle, some_more_data)`` in the
 ## :bro:id:`http_entity_data` event handler. When all data has arrived, a call
 ## to :bro:id:`md5_hash_finish` returns the final hash value.
 ## 
-## index: The unique identifier to associate with this hash computation.
+## Returns: The opaque handle associated with this hash computation.
 ##
 ## .. bro:see:: md5_hmac md5_hash md5_hash_update md5_hash_finish
 ##    sha1_hash sha1_hash_init sha1_hash_update sha1_hash_finish
 ##    sha256_hash sha256_hash_init sha256_hash_update sha256_hash_finish
-function md5_hash_init%(index: any%): bool
+function md5_hash_init%(%): opaque of md5
 	%{
-	BroString* s = convert_index_to_string(index);
+	HashVal* digest = new MD5Val();
-	int status = 0;
+	digest->Init();
-
+	return digest;
-	if ( md5_states.count(*s) < 1 )
-		{
-		MD5_CTX h;
-		md5_init(&h);
-		md5_states[*s] = h;
-		status = 1;
-		}
-
-	delete s;
-	return new Val(status, TYPE_BOOL);
 	%}
 
-## Initializes SHA1 state to enable incremental hash computation. After
+## Constructs an SHA1 handle to enable incremental hash computation. You can
-## initializing the SHA1 state with this function, you can feed data to
+## feed data to the returned opaque value with :bro:id:`sha1_hash_update` and
-## :bro:id:`sha1_hash_update` and finally need to call
+## finally need to call :bro:id:`sha1_hash_finish` to finish the computation
-## :bro:id:`sha1_hash_finish` to finish the computation and get the final hash
+## and get the hash digest as result.
-## value.
 ##
 ## For example, when computing incremental SHA1 values of transferred files in
-## multiple concurrent HTTP connections, one would call ``sha1_hash_init(c$id)``
+## multiple concurrent HTTP connections, one keeps an optional handle in the
-## once before invoking ``sha1_hash_update(c$id, some_more_data)`` in the
+## HTTP session record. Then, one would call
+## ``c$http$sha1_handle = sha1_hash_init()`` ## once before invoking
+## ``sha1_hash_update(c$http$sha1_handle, some_more_data)`` in the
 ## :bro:id:`http_entity_data` event handler. When all data has arrived, a call
 ## to :bro:id:`sha1_hash_finish` returns the final hash value.
 ##
-## index: The unique identifier to associate with this hash computation.
+## Returns: The opaque handle associated with this hash computation.
 ##
 ## .. bro:see:: md5_hmac md5_hash md5_hash_init md5_hash_update md5_hash_finish
 ##    sha1_hash sha1_hash_update sha1_hash_finish
 ##    sha256_hash sha256_hash_init sha256_hash_update sha256_hash_finish
-function sha1_hash_init%(index: any%): bool
+function sha1_hash_init%(%): opaque of sha1
 	%{
-	BroString* s = convert_index_to_string(index);
+	HashVal* digest = new SHA1Val();
-	int status = 0;
+	digest->Init();
-
+	return digest;
-	if ( sha1_states.count(*s) < 1 )
-		{
-		SHA_CTX h;
-		sha1_init(&h);
-		sha1_states[*s] = h;
-		status = 1;
-		}
-
-	delete s;
-	return new Val(status, TYPE_BOOL);
 	%}
 
-## Initializes SHA256 state to enable incremental hash computation. After
+## Constructs an SHA256 handle to enable incremental hash computation. You can
-## initializing the SHA256 state with this function, you can feed data to
+## feed data to the returned opaque value with :bro:id:`sha256_hash_update` and
-## :bro:id:`sha256_hash_update` and finally need to call
+## finally need to call :bro:id:`sha256_hash_finish` to finish the computation
-## :bro:id:`sha256_hash_finish` to finish the computation and get the final hash
+## and get the hash digest as result.
-## value.
 ##
 ## For example, when computing incremental SHA256 values of transferred files in
-## multiple concurrent HTTP connections, one would call
+## multiple concurrent HTTP connections, one keeps an optional handle in the
-## ``sha256_hash_init(c$id)`` once before invoking
+## HTTP session record. Then, one would call
-## ``sha256_hash_update(c$id, some_more_data)`` in the
+## ``c$http$sha256_handle = sha256_hash_init()`` ## once before invoking
+## ``sha256_hash_update(c$http$sha256_handle, some_more_data)`` in the
 ## :bro:id:`http_entity_data` event handler. When all data has arrived, a call
 ## to :bro:id:`sha256_hash_finish` returns the final hash value.
 ##
-## index: The unique identifier to associate with this hash computation.
+## Returns: The opaque handle associated with this hash computation.
 ##
 ## .. bro:see:: md5_hmac md5_hash md5_hash_init md5_hash_update md5_hash_finish
 ##    sha1_hash sha1_hash_init sha1_hash_update sha1_hash_finish
 ##    sha256_hash sha256_hash_update sha256_hash_finish
-function sha256_hash_init%(index: any%): bool
+function sha256_hash_init%(%): opaque of sha256
 	%{
-	BroString* s = convert_index_to_string(index);
+	HashVal* digest = new SHA256Val();
-	int status = 0;
+	digest->Init();
-
+	return digest;
-	if ( sha256_states.count(*s) < 1 )
-		{
-		SHA256_CTX h;
-		sha256_init(&h);
-		sha256_states[*s] = h;
-		status = 1;
-		}
-
-	delete s;
-	return new Val(status, TYPE_BOOL);
 	%}
 
-## Update the MD5 value associated with a given index. It is required to
+## Updates the MD5 value associated with a given index. It is required to
 ## call :bro:id:`md5_hash_init` once before calling this
 ## function.
 ##
-## index: The unique identifier to associate with this hash computation.
+## handle: The opaque handle associated with this hash computation.
 ##
 ## data: The data to add to the hash computation.
 ##
+## Returns: True on success.
+##
 ## .. bro:see:: md5_hmac md5_hash md5_hash_init md5_hash_finish
 ##    sha1_hash sha1_hash_init sha1_hash_update sha1_hash_finish
 ##    sha256_hash sha256_hash_init sha256_hash_update sha256_hash_finish
-function md5_hash_update%(index: any, data: string%): bool
+function md5_hash_update%(handle: opaque of md5, data: string%): bool
 	%{
-	BroString* s = convert_index_to_string(index);
+	bool rc = static_cast<HashVal*>(handle)->Feed(data->Bytes(), data->Len());
-	int status = 0;
+	return new Val(rc, TYPE_BOOL);
-
-	if ( md5_states.count(*s) > 0 )
-		{
-		md5_update(&md5_states[*s], data->Bytes(), data->Len());
-		status = 1;
-		}
-
-	delete s;
-	return new Val(status, TYPE_BOOL);
 	%}
 
-## Update the SHA1 value associated with a given index. It is required to
+## Updates the SHA1 value associated with a given index. It is required to
 ## call :bro:id:`sha1_hash_init` once before calling this
 ## function.
 ##
-## index: The unique identifier to associate with this hash computation.
+## handle: The opaque handle associated with this hash computation.
 ##
 ## data: The data to add to the hash computation.
 ##
+## Returns: True on success.
+##
 ## .. bro:see:: md5_hmac md5_hash md5_hash_init md5_hash_update md5_hash_finish
 ##    sha1_hash sha1_hash_init sha1_hash_finish
 ##    sha256_hash sha256_hash_init sha256_hash_update sha256_hash_finish
-function sha1_hash_update%(index: any, data: string%): bool
+function sha1_hash_update%(handle: opaque of sha1, data: string%): bool
 	%{
-	BroString* s = convert_index_to_string(index);
+	bool rc = static_cast<HashVal*>(handle)->Feed(data->Bytes(), data->Len());
-	int status = 0;
+	return new Val(rc, TYPE_BOOL);
-
-	if ( sha1_states.count(*s) > 0 )
-		{
-		sha1_update(&sha1_states[*s], data->Bytes(), data->Len());
-		status = 1;
-		}
-
-	delete s;
-	return new Val(status, TYPE_BOOL);
 	%}
 
-## Update the SHA256 value associated with a given index. It is required to
+## Updates the SHA256 value associated with a given index. It is required to
 ## call :bro:id:`sha256_hash_init` once before calling this
 ## function.
 ##
-## index: The unique identifier to associate with this hash computation.
+## handle: The opaque handle associated with this hash computation.
 ##
 ## data: The data to add to the hash computation.
 ##
+## Returns: True on success.
+##
 ## .. bro:see:: md5_hmac md5_hash md5_hash_init md5_hash_update md5_hash_finish
 ##    sha1_hash sha1_hash_init sha1_hash_update sha1_hash_finish
 ##    sha256_hash sha256_hash_init sha256_hash_finish
-function sha256_hash_update%(index: any, data: string%): bool
+function sha256_hash_update%(handle: opaque of sha256, data: string%): bool
 	%{
-	BroString* s = convert_index_to_string(index);
+	bool rc = static_cast<HashVal*>(handle)->Feed(data->Bytes(), data->Len());
-	int status = 0;
+	return new Val(rc, TYPE_BOOL);
-
-	if ( sha256_states.count(*s) > 0 )
-		{
-		sha256_update(&sha256_states[*s], data->Bytes(), data->Len());
-		status = 1;
-		}
-
-	delete s;
-	return new Val(status, TYPE_BOOL);
 	%}
 
 ## Returns the final MD5 digest of an incremental hash computation.
 ##
-## index: The unique identifier of this hash computation.
+## handle: The opaque handle associated with this hash computation.
 ##
-## Returns: The hash value associated with the computation at *index*.
+## Returns: The hash value associated with the computation of *handle*.
 ##
 ## .. bro:see:: md5_hmac md5_hash md5_hash_init md5_hash_update
 ##    sha1_hash sha1_hash_init sha1_hash_update sha1_hash_finish
 ##    sha256_hash sha256_hash_init sha256_hash_update sha256_hash_finish
-function md5_hash_finish%(index: any%): string
+function md5_hash_finish%(handle: opaque of md5%): string
 	%{
-	BroString* s = convert_index_to_string(index);
+	return static_cast<HashVal*>(handle)->Get();
-	StringVal* printable_digest;
-
-	if ( md5_states.count(*s) > 0 )
-		{
-		unsigned char digest[16];
-		md5_final(&md5_states[*s], digest);
-		md5_states.erase(*s);
-		printable_digest = new StringVal(md5_digest_print(digest));
-		}
-	else
-		printable_digest = new StringVal("");
-
-	delete s;
-	return printable_digest;
 	%}
 
 ## Returns the final SHA1 digest of an incremental hash computation.
 ##
-## index: The unique identifier of this hash computation.
+## handle: The opaque handle associated with this hash computation.
 ##
-## Returns: The hash value associated with the computation at *index*.
+## Returns: The hash value associated with the computation of *handle*.
 ##
 ## .. bro:see:: md5_hmac md5_hash md5_hash_init md5_hash_update md5_hash_finish
 ##    sha1_hash sha1_hash_init sha1_hash_update
 ##    sha256_hash sha256_hash_init sha256_hash_update sha256_hash_finish
-function sha1_hash_finish%(index: any%): string
+function sha1_hash_finish%(handle: opaque of sha1%): string
 	%{
-	BroString* s = convert_index_to_string(index);
+	return static_cast<HashVal*>(handle)->Get();
-	StringVal* printable_digest;
-
-	if ( sha1_states.count(*s) > 0 )
-		{
-		unsigned char digest[20];
-		sha1_final(&sha1_states[*s], digest);
-		sha1_states.erase(*s);
-		printable_digest = new StringVal(sha1_digest_print(digest));
-		}
-	else
-		printable_digest = new StringVal("");
-
-	delete s;
-	return printable_digest;
 	%}
 
 ## Returns the final SHA256 digest of an incremental hash computation.
 ##
-## index: The unique identifier of this hash computation.
+## handle: The opaque handle associated with this hash computation.
 ##
-## Returns: The hash value associated with the computation at *index*.
+## Returns: The hash value associated with the computation of *handle*.
 ##
 ## .. bro:see:: md5_hmac md5_hash md5_hash_init md5_hash_update md5_hash_finish
 ##    sha1_hash sha1_hash_init sha1_hash_update sha1_hash_finish
 ##    sha256_hash sha256_hash_init sha256_hash_update
-function sha256_hash_finish%(index: any%): string
+function sha256_hash_finish%(handle: opaque of sha256%): string
 	%{
-	BroString* s = convert_index_to_string(index);
+	return static_cast<HashVal*>(handle)->Get();
-	StringVal* printable_digest;
-
-	if ( sha256_states.count(*s) > 0 )
-		{
-		unsigned char digest[32];
-		sha256_final(&sha256_states[*s], digest);
-		sha256_states.erase(*s);
-		printable_digest = new StringVal(sha256_digest_print(digest));
-		}
-	else
-		printable_digest = new StringVal("");
-
-	delete s;
-	return printable_digest;
 	%}
 
 ## Generates a random number.
@@ -1058,11 +878,6 @@ function identify_data%(data: string, return_mime: bool%): string
 	return new StringVal(descr);
 	%}
 
-%%{
-#include <RandTest.h>
-static map<BroString, RandTest*> entropy_states;
-%%}
-
 ## Performs an entropy test on the given data.
 ## See http://www.fourmilab.ch/random.
 ##
@@ -1107,13 +922,11 @@ function find_entropy%(data: string%): entropy_test_result
 	%{
 	double montepi, scc, ent, mean, chisq;
 	montepi = scc = ent = mean = chisq = 0.0;
+	EntropyVal e;
+	e.Feed(data->Bytes(), data->Len());
+	e.Get(&ent, &chisq, &mean, &montepi, &scc);
+
 	RecordVal* ent_result = new RecordVal(entropy_test_result);
-	RandTest *rt = new RandTest();
-
-	rt->add((char*) data->Bytes(), data->Len());
-	rt->end(&ent, &chisq, &mean, &montepi, &scc);
-	delete rt;
-
 	ent_result->Assign(0, new Val(ent,     TYPE_DOUBLE));
 	ent_result->Assign(1, new Val(chisq,   TYPE_DOUBLE));
 	ent_result->Assign(2, new Val(mean,    TYPE_DOUBLE));
@@ -1124,85 +937,52 @@ function find_entropy%(data: string%): entropy_test_result
 
 ## Initializes data structures for incremental entropy calculation.
 ##
-## index: An arbitrary unique value per distinct computation.
+## Returns: An opaque handle to be used in subsequent operations.
-##
-## Returns: True on success.
 ##
 ## .. bro:see:: find_entropy entropy_test_add entropy_test_finish
-function entropy_test_init%(index: any%): bool
+function entropy_test_init%(%): opaque of entropy
 	%{
-	BroString* s = convert_index_to_string(index);
+	return new EntropyVal();
-	int status = 0;
-
-	if ( entropy_states.count(*s) < 1 )
-		{
-		entropy_states[*s] = new RandTest();
-		status = 1;
-		}
-
-	delete s;
-	return new Val(status, TYPE_BOOL);
 	%}
 
-## Adds data to an incremental entropy calculation. Before using this function,
+## Adds data to an incremental entropy calculation.
-## one needs to invoke :bro:id:`entropy_test_init`.
+##
+## handle: The opaque handle representing the entropy calculation state.
 ##
 ## data: The data to add to the entropy calculation.
 ##
-## index: An arbitrary unique value that identifies a particular entropy
-##        computation.
-##
 ## Returns: True on success.
 ##
 ## .. bro:see:: find_entropy entropy_test_add entropy_test_finish
-function entropy_test_add%(index: any, data: string%): bool
+function entropy_test_add%(handle: opaque of entropy, data: string%): bool
 	%{
-	BroString* s = convert_index_to_string(index);
+	bool status = static_cast<EntropyVal*>(handle)->Feed(data->Bytes(),
-	int status = 0;
+	                                                     data->Len());
-
-	if ( entropy_states.count(*s) > 0 )
-		{
-		entropy_states[*s]->add((char*) data->Bytes(), data->Len());
-		status = 1;
-		}
-
-	delete s;
 	return new Val(status, TYPE_BOOL);
 	%}
 
 ## Finishes an incremental entropy calculation. Before using this function,
-## one needs to initialize the computation with :bro:id:`entropy_test_init` and
+## one needs to obtain an opaque handle with :bro:id:`entropy_test_init` and
 ## add data to it via :bro:id:`entropy_test_add`.
 ##
-## index: An arbitrary unique value that identifies a particular entropy
+## handle: The opaque handle representing the entropy calculation state.
-##        computation.
 ##
 ## Returns: The result of the entropy test. See :bro:id:`find_entropy` for a
 ##          description of the individual components.
 ##
 ## .. bro:see:: find_entropy entropy_test_init entropy_test_add
-function entropy_test_finish%(index: any%): entropy_test_result
+function entropy_test_finish%(handle: opaque of entropy%): entropy_test_result
 	%{
-	BroString* s = convert_index_to_string(index);
 	double montepi, scc, ent, mean, chisq;
 	montepi = scc = ent = mean = chisq = 0.0;
+	static_cast<EntropyVal*>(handle)->Get(&ent, &chisq, &mean, &montepi, &scc);
+
 	RecordVal* ent_result = new RecordVal(entropy_test_result);
-
-	if ( entropy_states.count(*s) > 0 )
-		{
-		RandTest *rt = entropy_states[*s];
-		rt->end(&ent, &chisq, &mean, &montepi, &scc);
-		entropy_states.erase(*s);
-		delete rt;
-		}
-
 	ent_result->Assign(0, new Val(ent,     TYPE_DOUBLE));
 	ent_result->Assign(1, new Val(chisq,   TYPE_DOUBLE));
 	ent_result->Assign(2, new Val(mean,    TYPE_DOUBLE));
 	ent_result->Assign(3, new Val(montepi, TYPE_DOUBLE));
 	ent_result->Assign(4, new Val(scc,     TYPE_DOUBLE));
-
-	delete s;
 	return ent_result;
 	%}
 

src/builtin-func.l

@@ -74,6 +74,8 @@ HEX	[0-9a-fA-F]+
 "set"		return check_c_mode(TOK_SET);
 "table"		return check_c_mode(TOK_TABLE);
 "vector"	return check_c_mode(TOK_VECTOR);
+"of"            return check_c_mode(TOK_OF);
+"opaque"        return check_c_mode(TOK_OPAQUE);
 "module"        return check_c_mode(TOK_MODULE);
 
 "@ARG@"		return TOK_ARG;

src/builtin-func.y

@@ -269,15 +269,15 @@ void print_event_c_body(FILE *fp)
 
 %token TOK_LPP TOK_RPP TOK_LPB TOK_RPB TOK_LPPB TOK_RPPB TOK_VAR_ARG
 %token TOK_BOOL
-%token TOK_FUNCTION TOK_EVENT TOK_CONST TOK_ENUM
+%token TOK_FUNCTION TOK_EVENT TOK_CONST TOK_ENUM TOK_OF
-%token TOK_TYPE TOK_RECORD TOK_SET TOK_VECTOR TOK_TABLE TOK_MODULE
+%token TOK_TYPE TOK_RECORD TOK_SET TOK_VECTOR TOK_OPAQUE TOK_TABLE TOK_MODULE
 %token TOK_ARGS TOK_ARG TOK_ARGC
 %token TOK_ID TOK_ATTR TOK_CSTR TOK_LF TOK_WS TOK_COMMENT
 %token TOK_ATOM TOK_INT TOK_C_TOKEN
 
 %left ',' ':'
 
-%type <str> TOK_C_TOKEN TOK_ID TOK_CSTR TOK_WS TOK_COMMENT TOK_ATTR TOK_INT opt_ws
+%type <str> TOK_C_TOKEN TOK_ID TOK_CSTR TOK_WS TOK_COMMENT TOK_ATTR TOK_INT opt_ws type
 %type <val> TOK_ATOM TOK_BOOL
 
 %union	{
@@ -584,7 +584,17 @@ args_1:		args_1 ',' opt_ws arg opt_ws
 			{ /* empty */ }
 	;
 
-arg:		TOK_ID opt_ws ':' opt_ws TOK_ID
+// TODO: Migrate all other compound types to this rule. Once the BiF language
+// can parse all regular Bro types, we can throw out the unnecessary
+// boilerplate typedefs for addr_set, string_set, etc.
+type:
+                TOK_OPAQUE opt_ws TOK_OF opt_ws TOK_ID
+                        { $$ = concat("opaque of ", $5); }
+        |       TOK_ID
+                        { $$ = $1; }
+        ;
+
+arg:		TOK_ID opt_ws ':' opt_ws type
 			{ args.push_back(new BuiltinFuncArg($1, $5)); }
 	|	TOK_VAR_ARG
 			{
@@ -594,7 +604,7 @@ arg:		TOK_ID opt_ws ':' opt_ws TOK_ID
 			}
 	;
 
-return_type:	':' opt_ws TOK_ID opt_ws
+return_type:	':' opt_ws type opt_ws
 			{
 			BuiltinFuncArg* ret = new BuiltinFuncArg("", $3);
 			ret->PrintBro(fp_bro_init);

src/parse.y

@@ -11,7 +11,7 @@
 %token TOK_DOUBLE TOK_ELSE TOK_ENUM TOK_EVENT TOK_EXPORT TOK_FILE TOK_FOR
 %token TOK_FUNCTION TOK_GLOBAL TOK_HOOK TOK_ID TOK_IF TOK_INT
 %token TOK_INTERVAL TOK_LIST TOK_LOCAL TOK_MODULE
-%token TOK_NEXT TOK_OF TOK_PATTERN TOK_PATTERN_TEXT
+%token TOK_NEXT TOK_OF TOK_OPAQUE TOK_PATTERN TOK_PATTERN_TEXT
 %token TOK_PORT TOK_PRINT TOK_RECORD TOK_REDEF
 %token TOK_REMOVE_FROM TOK_RETURN TOK_SCHEDULE TOK_SET
 %token TOK_STRING TOK_SUBNET TOK_SWITCH TOK_TABLE
@@ -899,6 +899,12 @@ type:
 				$$ = new FileType(base_type(TYPE_STRING));
 				}
 
+	|	TOK_OPAQUE TOK_OF TOK_ID
+				{
+				set_location(@1, @3);
+				$$ = new OpaqueType($3);
+				}
+
 	|	resolve_id
 			{
 			if ( ! $1 || ! ($$ = $1->AsType()) )

src/scan.l

@@ -298,6 +298,7 @@ local	return TOK_LOCAL;
 module	return TOK_MODULE;
 next	return TOK_NEXT;
 of	return TOK_OF;
+opaque	return TOK_OPAQUE;
 pattern	return TOK_PATTERN;
 port	return TOK_PORT;
 print	return TOK_PRINT;

testing/btest/Baseline/istate.opaque/expected.log

@@ -0,0 +1,4 @@
+acbd18db4cc2f85cedef654fccc4a4d8
+0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33
+2c26b46b68ffc68ff99b453c1d30413413422d706483bfa0f98a5e886266e7ae
+[entropy=0.918296, chi_square=423.666667, mean=108.0, monte_carlo_pi=nan, serial_correlation=-0.5]

testing/btest/Baseline/istate.opaque/output.log

@@ -0,0 +1,4 @@
+acbd18db4cc2f85cedef654fccc4a4d8
+0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33
+2c26b46b68ffc68ff99b453c1d30413413422d706483bfa0f98a5e886266e7ae
+[entropy=0.918296, chi_square=423.666667, mean=108.0, monte_carlo_pi=nan, serial_correlation=-0.5]

testing/btest/bifs/entropy_test.bro

@@ -5,20 +5,14 @@
 event bro_init()
 	{
 	local a = "dh3Hie02uh^s#Sdf9L3frd243h$d78r2G4cM6*Q05d(7rh46f!0|4-f";
-	if ( entropy_test_init(1) != T )
+	local handle = entropy_test_init();
+	if ( ! entropy_test_add(handle, a) )
 		exit(1);
-
+	print entropy_test_finish(handle);
-	if ( entropy_test_add(1, a) != T )
-		exit(1);
-
-	print entropy_test_finish(1);
 
 	local b = "0011000aaabbbbcccc000011111000000000aaaabbbbcccc0000000";
-	if ( entropy_test_init(2) != T )
+	handle = entropy_test_init();
+	if ( ! entropy_test_add(handle, b) )
 		exit(1);
-
+	print entropy_test_finish(handle);
-	if ( entropy_test_add(2, b) != T )
-		exit(1);
-
-	print entropy_test_finish(2);
 	}

testing/btest/bifs/md5.test

@@ -4,16 +4,16 @@
 print md5_hash("one");
 print md5_hash("one", "two", "three");
 
-md5_hash_init("a");
+local a = md5_hash_init();
-md5_hash_init("b");
+local b = md5_hash_init();
 
-md5_hash_update("a", "one");
+md5_hash_update(a, "one");
-md5_hash_update("b", "one");
+md5_hash_update(b, "one");
-md5_hash_update("b", "two");
+md5_hash_update(b, "two");
-md5_hash_update("b", "three");
+md5_hash_update(b, "three");
 
-print md5_hash_finish("a");
+print md5_hash_finish(a);
-print md5_hash_finish("b");
+print md5_hash_finish(b);
 
 print md5_hmac("one");
 print md5_hmac("one", "two", "three");

testing/btest/bifs/sha1.test

@@ -4,13 +4,13 @@
 print sha1_hash("one");
 print sha1_hash("one", "two", "three");
 
-sha1_hash_init("a");
+local a = sha1_hash_init();
-sha1_hash_init("b");
+local b = sha1_hash_init();
 
-sha1_hash_update("a", "one");
+sha1_hash_update(a, "one");
-sha1_hash_update("b", "one");
+sha1_hash_update(b, "one");
-sha1_hash_update("b", "two");
+sha1_hash_update(b, "two");
-sha1_hash_update("b", "three");
+sha1_hash_update(b, "three");
 
-print sha1_hash_finish("a");
+print sha1_hash_finish(a);
-print sha1_hash_finish("b");
+print sha1_hash_finish(b);

testing/btest/bifs/sha256.test

@@ -4,13 +4,13 @@
 print sha256_hash("one");
 print sha256_hash("one", "two", "three");
 
-sha256_hash_init("a");
+local a = sha256_hash_init();
-sha256_hash_init("b");
+local b = sha256_hash_init();
 
-sha256_hash_update("a", "one");
+sha256_hash_update(a, "one");
-sha256_hash_update("b", "one");
+sha256_hash_update(b, "one");
-sha256_hash_update("b", "two");
+sha256_hash_update(b, "two");
-sha256_hash_update("b", "three");
+sha256_hash_update(b, "three");
 
-print sha256_hash_finish("a");
+print sha256_hash_finish(a);
-print sha256_hash_finish("b");
+print sha256_hash_finish(b);

testing/btest/istate/opaque.bro

@@ -0,0 +1,77 @@
+#
+# @TEST-EXEC: bro -r $TRACES/empty.trace write.bro
+# @TEST-EXEC: bro read.bro
+# @TEST-EXEC: btest-diff expected.log
+# @TEST-EXEC: btest-diff output.log
+# @TEST-EXEC: cmp output.log expected.log
+
+@TEST-START-FILE read.bro
+
+global md5_handle: opaque of md5 &persistent &synchronized;
+global sha1_handle: opaque of sha1 &persistent &synchronized;
+global sha256_handle: opaque of sha256 &persistent &synchronized;
+global entropy_handle: opaque of entropy &persistent &synchronized;
+
+event bro_done()
+  {
+  local out = open("output.log");
+
+  # Finish incremental operations started by a previous Bro.
+  if ( md5_hash_update(md5_handle, "oo") )
+    print out, md5_hash_finish(md5_handle);
+  else
+    print out, "md5_hash_update() failed";
+
+  if ( sha1_hash_update(sha1_handle, "oo") )
+    print out, sha1_hash_finish(sha1_handle);
+  else
+    print out, "sha1_hash_update() failed";
+
+  if ( sha256_hash_update(sha256_handle, "oo") )
+    print out, sha256_hash_finish(sha256_handle);
+  else
+    print out, "sha256_hash_update() failed";
+
+  if ( entropy_test_add(entropy_handle, "oo") )
+    print out, entropy_test_finish(entropy_handle);
+  else
+    print out, "entropy_test_add() failed";
+  }
+
+@TEST-END-FILE
+
+@TEST-START-FILE write.bro
+
+global md5_handle: opaque of md5 &persistent &synchronized;
+global sha1_handle: opaque of sha1 &persistent &synchronized;
+global sha256_handle: opaque of sha256 &persistent &synchronized;
+global entropy_handle: opaque of entropy &persistent &synchronized;
+
+event bro_init()
+  {
+	local out = open("expected.log");
+	print out, md5_hash("foo");
+	print out, sha1_hash("foo");
+	print out, sha256_hash("foo");
+	print out, find_entropy("foo");
+
+  # Begin incremental operations. Our goal is to feed the data string "foo" to
+  # the computation, but split into "f" and "oo" in two instances..
+  md5_handle = md5_hash_init();
+  if ( ! md5_hash_update(md5_handle, "f") )
+    print out, "md5_hash_update() failed";
+
+  sha1_handle = sha1_hash_init();
+  if ( ! sha1_hash_update(sha1_handle, "f") )
+    print out, "sha1_hash_update() failed";
+
+  sha256_handle = sha256_hash_init();
+  if ( ! sha256_hash_update(sha256_handle, "f") )
+    print out, "sha256_hash_update() failed";
+
+  entropy_handle = entropy_test_init();
+  if ( ! entropy_test_add(entropy_handle, "f") )
+    print out, "entropy_test_add() failed";
+  }
+
+@TEST-END-FILE