hash: report number of processed bytes

src/Hash.h

@@ -12,6 +12,7 @@ struct Hash : protected MurMurHash3 {
     struct digest {
         enum { DIGEST_WORDS = algorithm::DIGEST_WORDS };
         uint32_t value[algorithm::DIGEST_WORDS];
+        uint32_t length;
 
         bool operator==(const digest &other) const {
             for (unsigned i = 0; i < DIGEST_WORDS; i++) {
@@ -55,7 +56,7 @@ struct Hash : protected MurMurHash3 {
     const digest getDigest() const {
         Hash copy = *this;
         digest ret;
-        copy.finalize(ret.value);
+        ret.length = copy.finalize(ret.value);
         return ret;
     }
 
@@ -99,6 +100,7 @@ struct Hash : protected MurMurHash3 {
     }
 
     Hash& operator<<(const digest & x) {
+        m_byteCount += x.length;
         return  processBytes(x.value, sizeof(x.value));
     }
 

src/MurMurHash3.h

@@ -152,11 +152,12 @@ public:
         return *this;
     }
 
-    void finalize(uint32_t *digest) {
+    uint32_t finalize(uint32_t *digest) {
         //----------
         // tail
         uint64_t k1 = 0;
         uint64_t k2 = 0;
+        // We use m_byteCount NOT here. On Purpose.
         switch(m_blockByteIndex & 15)
         {
         case 15: k2 ^= ((uint64_t)m_block[14]) << 48;
@@ -194,9 +195,11 @@ public:
 
         ((uint64_t*)digest)[0] = h1;
         ((uint64_t*)digest)[1] = h2;
+
+        return m_byteCount;
     }
 
-private:
+protected:
     void processBlock() {
         uint64_t k1 = getblock64((uint64_t*)m_block, 0);
         uint64_t k2 = getblock64((uint64_t*)m_block, 1);

src/SHA1.h

@@ -70,7 +70,7 @@ public:
         return *this;
     }
 
-    void finalize(uint32_t *digest) {
+    uint32_t finalize(uint32_t *digest) {
         size_t bitCount = m_byteCount * 8;
         processByte(0x80);
         if (m_blockByteIndex > 56) {
@@ -95,6 +95,8 @@ public:
         processByte( static_cast<unsigned char>((bitCount)     & 0xFF));
 
         memcpy(digest, m_digest, sizeof(m_digest));
+
+        return m_byteCount;
     }
 
 private:

src/clang-hash.cc

@@ -20,13 +20,16 @@ public:
         Visitor.hashDecl(Context.getTranslationUnitDecl());
 
         // Context.getTranslationUnitDecl()->dump();
+        unsigned processed_bytes;
+        std::string hash = Visitor.GetHash(&processed_bytes);
+
         if (toplevel_hash_stream) {
-            std::string hash = Visitor.GetHash();
             toplevel_hash_stream->write(hash.c_str(), hash.length());
-
             delete toplevel_hash_stream;
         }
-        llvm::errs() << "top-level-hash: " << Visitor.GetHash() << "\n";
+        llvm::errs() << "top-level-hash: " << hash << "\n";
+        llvm::errs() << "processed bytes: " << processed_bytes << "\n";
+
     }
 private:
     raw_ostream *toplevel_hash_stream;

src/hash-visitor.cc

@@ -10,8 +10,12 @@ using namespace std;
 
 typedef TranslationUnitHashVisitor HashVisitor;
 
-string HashVisitor::GetHash() {
-    return toplevel_hash.getDigest().getHexDigest();
+string HashVisitor::GetHash(unsigned *processed_bytes) {
+    auto hash = toplevel_hash.getDigest();
+    if (processed_bytes) {
+        *processed_bytes = hash.length;
+    }
+    return hash.getHexDigest();
 }
 
 /// Declarations

src/hash-visitor.h

@@ -107,7 +107,7 @@ public:
 	bool VisitAdjustedType(const AdjustedType *T);
 	bool VisitElaboratedType(const ElaboratedType *T);
 
-	std::string GetHash();
+	std::string GetHash(unsigned *processed_bytes = nullptr);
 
 	//C Exprs (no clang-builtins, ...)
 	/*bool VisitExpr(const Expr *Node);*/