added additional metadata.

Signed-off-by: steven <[email protected]>

kam1n0/kam1n0-apps/src/main/java/ca/mcgill/sis/dmas/kam1n0/app/adata/FunctionDataUnit.java

@@ -37,11 +37,12 @@ public class FunctionDataUnit implements Serializable {
 	public String functionName;
 	public String functionId;
 	public String startAddress;
-	public long blockSize;
+	public int blockSize;
 	public SrcFunction srcFunc;
 
 	public ArrayList<BlockDataUnit> nodes = new ArrayList<>();
 	public ArrayList<Link> links = new ArrayList<>();
+	public int codeSize;
 
 	public static class Link implements Serializable {
 		private static final long serialVersionUID = -3012737921524201819L;
@@ -152,7 +153,8 @@ public FunctionDataUnit(Function function, AsmLineNormalizer normalizer, boolean
 		this.functionId = Long.toString(function.functionId);
 		this.functionName = function.functionName;
 		this.startAddress = Long.toString(function.startingAddress);
-		this.blockSize = function.numBlocks;
+		this.blockSize = (int) function.numBlocks;
+		this.codeSize = (int) function.codeSize;
 	}
 
 }

kam1n0/kam1n0-apps/src/main/java/ca/mcgill/sis/dmas/kam1n0/app/clone/adata/FunctionCloneEntryForWeb.java

@@ -32,6 +32,7 @@ public class FunctionCloneEntryForWeb implements Serializable {
 	public String binaryId;
 	public String binaryName = StringResources.STR_EMPTY;
 	public int numBbs = 0;
+	public int codeSize = 0;
 	public double similarity = -1;
 	public FunctionDataUnit actualFunc = null;
 
@@ -57,6 +58,7 @@ public FunctionCloneEntryForWeb(FunctionCloneEntry entry) {
 		this.binaryName = entry.binaryName;
 		this.similarity = entry.similarity;
 		this.numBbs = (int) entry.numBbs;
+		this.codeSize = (int) entry.codeSize;
 
 		clonedParts = entry.clonedParts.stream() //
 				.map(set -> set//
@@ -77,6 +79,7 @@ public FunctionCloneEntryForWeb(FunctionDataUnit func, double similarity, int nu
 		this.binaryName = func.binaryName;
 
 		this.numBbs = numbbs;
+		this.codeSize = func.codeSize;
 
 	}
 

kam1n0/kam1n0-clone/src/main/java/ca/mcgill/sis/dmas/kam1n0/problem/clone/FunctionCloneEntry.java

@@ -38,6 +38,7 @@ public class FunctionCloneEntry implements Serializable, Comparable<FunctionClon
 	public long binaryId;
 	public String binaryName = StringResources.STR_EMPTY;
 	public double similarity = Double.MAX_VALUE;
+	public long codeSize;
 	public long startingEA;
 	public long numBbs = 0;
 
@@ -56,15 +57,17 @@ public FunctionCloneEntry(Function function, double similarity) {
 		this.functionName = function.functionName;
 		this.similarity = similarity;
 		this.numBbs = function.numBlocks;
+		this.codeSize = function.codeSize;
 	}
 
-	public FunctionCloneEntry(Block aBlk, double similarity, int bbcount) {
+	public FunctionCloneEntry(Block aBlk, double similarity) {
 		this.binaryId = aBlk.binaryId;
 		this.binaryName = aBlk.binaryName;
 		this.functionId = aBlk.functionId;
 		this.functionName = aBlk.functionName;
 		this.similarity = similarity;
-		this.numBbs = bbcount;
+		this.numBbs = aBlk.peerSize;
+		this.codeSize = aBlk.funcCodeSize;
 	}
 
 	@Override

kam1n0/kam1n0-clone/src/main/java/ca/mcgill/sis/dmas/kam1n0/problem/clone/detector/kam/index/ALSH.java

@@ -208,7 +208,7 @@ public <E extends VecObject<T, K>> Tuple2<List<Tuple2<Long, E>>, JavaRDD<VecEntr
 			List<? extends E> objs, int topK, Function<List<T>, List<T>> filter) {
 		List<Tuple2<Long, E>> hid_tbid_l = this.index_bucket.collectHids(rid, objs, this::hash);
 		HashSet<Long> hids = hid_tbid_l.stream().map(tp -> tp._1).collect(Collectors.toCollection(HashSet::new));
-		logger.info("hids {}", hids.size());
+		// logger.info("hids {}", hids.size());
 		// hid->info
 		JavaRDD<VecEntry<T, K>> hid_info = this.index_deduplication.getVecEntryInfoAsRDD(rid, hids, false, filter);// .cache();
 		return new Tuple2<>(hid_tbid_l, hid_info);

kam1n0/kam1n0-clone/src/main/java/ca/mcgill/sis/dmas/kam1n0/problem/clone/detector/kam/utils/SubgraphBlocksImpl3.java

@@ -303,12 +303,8 @@ public static FunctionCloneEntry mergeSingles2(int funcLength, Iterable<Tuple3<B
 
 		}
 
-		FunctionCloneEntry entry = new FunctionCloneEntry();
 		Block block = subgraphs.get(0).stream().findAny().get().src;
-		entry.functionId = block.functionId;
-		entry.functionName = block.functionName;
-		entry.binaryId = block.binaryId;
-		entry.binaryName = block.binaryName;
+		FunctionCloneEntry entry = new FunctionCloneEntry(block, 0);
 
 		subgraphs.sort((g1, g2) -> Double.compare(g1.score, g2.score));
 		ArrayList<Subgraph2> picks = new ArrayList<>();
@@ -349,7 +345,8 @@ public static FunctionCloneEntry mergeSingles2(int funcLength, Iterable<Tuple3<B
 				(k, v) -> v == null ? lk.score * lk.tar.codesSize : Math.max(lk.score * lk.tar.codesSize, v)));
 		// entry.similarity = picks.stream().mapToDouble(g -> g.score).sum() * 1.0 /
 		// (Math.abs(funcLength));
-		entry.similarity = hashMap.values().stream().mapToDouble(v -> v).sum() / Math.abs(funcLength);
+		double sum = hashMap.values().stream().mapToDouble(v -> v).sum() * 2;
+		entry.similarity = sum / (Math.abs(funcLength) + Math.abs(entry.codeSize));
 		return entry;
 	}
 

kam1n0/kam1n0-commons/src/main/java/ca/mcgill/sis/dmas/kam1n0/framework/disassembly/BinarySurrogate.java

@@ -125,6 +125,7 @@ public Function toFunction(FunctionSurrogate func) {
 		ofunc.srcName = func.srcName;
 		ofunc.blocks = new ArrayList<>();
 		ofunc.architecture = architecture;
+		ofunc.codeSize = func.blocks.stream().mapToLong(blk -> blk.asmLines().size()).sum();
 		func.blocks.forEach(blk -> ofunc.blockIds.add(blk.id));
 		func.blocks.forEach(blk -> {
 			Block oblk = new Block();
@@ -135,6 +136,7 @@ public Function toFunction(FunctionSurrogate func) {
 			oblk.callingBlocks = blk.call;
 			oblk.codes = new ArrayList<>(blk.asmLines());
 			oblk.codesSize = oblk.codes.size();
+			oblk.funcCodeSize = ofunc.codeSize;
 			oblk.functionId = func.id;
 			oblk.functionName = func.name;
 			oblk.peerSize = func.blocks.size();
@@ -144,13 +146,12 @@ public Function toFunction(FunctionSurrogate func) {
 			oblk.architecture = architecture;
 			ofunc.blocks.add(oblk);
 		});
+
 		ofunc.numBlocks = ofunc.blocks.size();
-		ofunc.comments = func.comments.stream()
-				.map(cmm -> {
-					String content = cmm.comment.replaceAll("(\r\n|\n)", "\n\n");
-					return new Comment(func.id, content, cmm.type, new Date().getTime(), "user_ida", cmm.offset);
-				})
-				.collect(Collectors.toList());
+		ofunc.comments = func.comments.stream().map(cmm -> {
+			String content = cmm.comment.replaceAll("(\r\n|\n)", "\n\n");
+			return new Comment(func.id, content, cmm.type, new Date().getTime(), "user_ida", cmm.offset);
+		}).collect(Collectors.toList());
 		return ofunc;
 	}
 

kam1n0/kam1n0-commons/src/main/java/ca/mcgill/sis/dmas/kam1n0/framework/storage/Block.java

@@ -63,6 +63,8 @@ public class Block implements AsmFragment, Serializable {
 	public String blockName = StringResources.STR_EMPTY;
 
 	public long codesSize = -1;
+
+	public long funcCodeSize = -1;
 
 	@AsBytes
 	public List<List<String>> codes;

kam1n0/kam1n0-commons/src/main/java/ca/mcgill/sis/dmas/kam1n0/framework/storage/Function.java

@@ -55,6 +55,8 @@ public class Function implements Serializable, Iterable<Block> {
 	public String binaryName = StringResources.STR_EMPTY;
 
 	public long numBlocks;
+
+	public long codeSize;
 
 	@AsBytes
 	public Set<Long> blockIds;