expose node hashes; fixes bblfsh#78

Signed-off-by: Denys Smirnov <[email protected]>

src/libuast.hpp

@@ -58,6 +58,15 @@ namespace uast {
         virtual T node() = 0;
     };
 
+    // NodeHash is a hash of a node subtree.
+    struct NodeHash {
+        // The data are a cryptographic-quality fingerprint of the tree structure.
+        // The specific algorithm is not specified and may change across library versions.
+        // Clients may compare hash values for equality to match equivalent nodes, but cannot
+        // recompute hash values directly, as the input depends on details of serialization.
+        uint8_t data[UAST_HASH_SIZE];
+    };
+
     // Context is a common interface implemented by all UAST contexts.
     template<class T> class Context {
     public:
@@ -75,6 +84,7 @@ namespace uast {
 
         virtual Iterator<T>* Filter(T root, std::string query) = 0;
         virtual Iterator<T>* Iterate(T root, TreeOrder order) = 0;
+        virtual NodeHash Hash(T root, HashFlags flags) = 0;
     };
 
     // NodeCreator is an interface that creates new UAST nodes.
@@ -351,6 +361,11 @@ namespace uast {
             CheckError();
             return new RawIterator(it);
         }
+        NodeHash Hash(NodeHandle root, HashFlags flags) {
+            NodeHash h;
+            UastHash(ctx, root, (void*)&h, flags);
+            return h;
+        }
     };
 
     // PtrIterator is an iterator that casts NodeHandle directly to pointer type T.
@@ -414,6 +429,9 @@ namespace uast {
             auto it = new PtrIterator<T>(raw, true);
             return it;
         }
+        NodeHash Hash(T root, HashFlags flags) {
+            return ctx->Hash(ToHandle(root), flags);
+        }
         void CheckError(){
             ctx->CheckError();
         }

src/uast.h

@@ -90,7 +90,9 @@ typedef struct UastIterator {
 
 typedef enum { UAST_BINARY = 0, UAST_YAML = 1 } UastFormat;
 
+// HashFlags is a bit-field with different flags which controls node hashing.
 typedef enum {
+    HASH_ALL = 0x0,
     HASH_NO_POS = 0x1,
 } HashFlags;
 

tests/main.cc

@@ -28,6 +28,7 @@ int main() {
 
   // add the tests to the suite
   ADD_TEST(suite, "test of RoleNameForId()", TestRoleNameForId);
+  ADD_TEST(suite, "test node hash", TestNodeHash);
   ADD_TEST(suite, "test of UastFilter() pointers", TestUastFilterPointers);
   ADD_TEST(suite, "test iteration (preorder)", TestUastIteratorPreOrder);
   ADD_TEST(suite, "test of UastFilter() counting", TestUastFilterCount);

tests/nodes_test.h

@@ -642,6 +642,36 @@ void TestNodeFindError() {
   UastFree(ctx);
 }
 
+void TestNodeHash() {
+  Uast *ctx = NewUastMock();
+  Node* module = newObject("Module");
+  Node* child = newObject("Child");
+  module->SetChild("field", child);
+
+  unsigned char hash[UAST_HASH_SIZE];
+  // This value must be updated if the algorithm for structural hashing is changed,
+  // or if the tree structure for the test module changes in a way that modifies the
+  // structural hash. To update it, copy the actual value from the test failure and
+  // update this array.
+  unsigned char exp[UAST_HASH_SIZE] = {
+    0xe6, 0xd2, 0x53, 0xe1, 0x26, 0x0a, 0xaa, 0xa9,
+    0x37, 0xcc, 0xfc, 0x42, 0x0f, 0x52, 0x65, 0x48,
+    0x1d, 0x59, 0x18, 0xce, 0x01, 0xad, 0xda, 0xa2,
+    0x82, 0x4c, 0x74, 0x77, 0xae, 0xa1, 0x26, 0xb5};
+  UastHash(ctx, NodeHandle(module), (void*)hash, HASH_ALL);
+  bool ok = memcmp(hash, exp, UAST_HASH_SIZE) == 0;
+  if (!ok) {
+    printf("unexpected hash value:\n");
+    for (int i = 0; i < UAST_HASH_SIZE; i++) {
+      printf("0x%02x, ", (unsigned char)hash[i]);
+      if (i%8 == 7) printf("\n");
+    }
+  }
+  CU_ASSERT_FATAL(ok);
+
+  UastFree(ctx);
+}
+
 void TestEmptyResult() {
   Uast *ctx = NewUastMock();
   Node* module = newObject("Module");