Just hash pointers and keep objects around, instead of hashing SHA-1s

arrowstore.hpp

@@ -7,7 +7,9 @@
 using namespace std;
 using namespace std::tr1;
 
-typedef unordered_multimap<string, string> arrow_map_t;
+class GitObject;
+
+typedef unordered_multimap<GitObject *, GitObject *> arrow_map_t;
 
 class ArrowStore
 {
@@ -18,7 +20,7 @@ class ArrowStore
   ArrowStore( void ) {}
   ~ArrowStore() {}
 
-  void add( string src, string dest )
+  void add( GitObject *src, GitObject *dest )
   {
     arrow_map.insert( arrow_map_t::value_type( dest, src ) );
   }

deltadb.cpp

@@ -76,7 +76,7 @@ int DeltaDB::recursive_traverse( GitObject *obj, GitObject *parent, GitObject *b
   obj->inflate_object();
   obj->apply_delta( parent );
 
-  base->parse( obj, arrows );
+  base->parse( obj, arrows, this );
 
   int size = obj->get_delta_decoded_size();
 

gitflip.cpp

@@ -17,9 +17,8 @@ int main( int argc, char *argv[] )
   char *pack_filename = argv[ 1 ];
   char *idx_filename = argv[ 2 ];
 
-  ArrowStore *arrows = new ArrowStore();
-
   try {
+    ArrowStore *arrows = new ArrowStore();
     Pack *pack = new Pack( pack_filename, idx_filename );
     DeltaDB *deltas = new DeltaDB( pack, arrows );
 

objects.hpp

@@ -129,31 +129,31 @@ class GitObject
     return delta_decoded_data + index;
   }
 
-  virtual void parse( GitObject *obj, ArrowStore *arrows ) = 0;
+  virtual void parse( GitObject *obj, ArrowStore *arrows, const DeltaDB *db ) = 0;
 };
 
 class Commit : public GitObject
 {
 public:
-  void parse( GitObject *obj, ArrowStore *arrows );
+  void parse( GitObject *obj, ArrowStore *arrows, const DeltaDB *db );
 };
 
 class Tree : public GitObject
 {
 public:
-  void parse( GitObject *obj, ArrowStore *arrows );
+  void parse( GitObject *obj, ArrowStore *arrows, const DeltaDB *db );
 };
 
 class Blob : public GitObject
 {
 public:
-  void parse( GitObject *obj, ArrowStore *arrows );
+  void parse( GitObject *obj, ArrowStore *arrows, const DeltaDB *db );
 };
 
 class Tag : public GitObject
 {
 public:
-  void parse( GitObject *obj, ArrowStore *arrows );
+  void parse( GitObject *obj, ArrowStore *arrows, const DeltaDB *db );
 };
 
 class Delta : public GitObject
@@ -170,7 +170,7 @@ class Delta : public GitObject
   Delta() : reference_object( NULL ) {}
   GitObject* get_reference( void ) const { return reference_object; }
   void apply_delta( GitObject *parent );
-  void parse( GitObject *obj, ArrowStore *arrows );
+  void parse( GitObject *obj, ArrowStore *arrows, const DeltaDB *db );
 };
 
 class Ofs_Delta : public Delta

objparse.cpp

@@ -2,19 +2,20 @@
 
 #include "objects.hpp"
 #include "arrowstore.hpp"
+#include "deltadb.hpp"
 
 /* These are member functions of the base object in a delta chain.
    They parse either the base object or a derived, decoded delta,
    based on its type. (A delta doesn't know its own type.)
 */
 
-void Commit::parse( GitObject *obj, ArrowStore *arrows )
+void Commit::parse( GitObject *obj, ArrowStore *arrows, const DeltaDB *db )
 {
   /* confirm "tree" */
   unixassert( strncmp( (char *)obj->get_buf( 0, 5 ), "tree ", 5 ) );
   sha1 tree;
   tree.read( obj->get_buf( 5, 40 ) );
-  arrows->add( obj->get_hash()->str(), tree.str() );
+  arrows->add( obj, db->lookup_hash( tree ) );
 
   /* get as many parents as are present */
   /* we assume every commit has an author and commiter,
@@ -24,12 +25,12 @@ void Commit::parse( GitObject *obj, ArrowStore *arrows )
   while ( 0 == strncmp( (char *)obj->get_buf( i, 7 ), "parent ", 7 ) ) {
     sha1 parent;
     parent.read( obj->get_buf( i + 7, 40 ) );
-    arrows->add( obj->get_hash()->str(), parent.str() );
+    arrows->add( obj, db->lookup_hash( parent ) );
     i += 48;
   }
 }
 
-void Tree::parse( GitObject *obj, ArrowStore *arrows )
+void Tree::parse( GitObject *obj, ArrowStore *arrows, const DeltaDB *db )
 {
   /* null bytes signal end of filename, and immediately precede SHA-1 */
   /* note tree uses binary SHA-1, unlike ASCII form */
@@ -48,31 +49,31 @@ void Tree::parse( GitObject *obj, ArrowStore *arrows )
 	printf( "\n" );
       }
       sha1 entry( (char *)(obj->get_buf( i + 1, 20 ) ) );
-      arrows->add( obj->get_hash()->str(), entry.str() );
+      arrows->add( obj, db->lookup_hash( entry ) );
       i += 21;
     } else {
       i++;
     }
   }
 }
 
-void Tag::parse( GitObject *obj, ArrowStore *arrows )
+void Tag::parse( GitObject *obj, ArrowStore *arrows, const DeltaDB *db )
 {
   /* confirm beginning */
   unixassert( strncmp( (char *)obj->get_buf( 0, 7 ), "object ", 7 ) );
 
   sha1 reference;
   reference.read( obj->get_buf( 7, 40 ) );
-  arrows->add( obj->get_hash()->str(), reference.str() );
+  arrows->add( obj, db->lookup_hash( reference ) );
 }
 
-void Blob::parse( GitObject *obj, ArrowStore *arrows )
+void Blob::parse( GitObject *obj, ArrowStore *arrows, const DeltaDB *db )
 {
   /* We're not interested in these. */
   throw InternalError();
 }
 
-void Delta::parse( GitObject *obj, ArrowStore *arrows )
+void Delta::parse( GitObject *obj, ArrowStore *arrows, const DeltaDB *db )
 {
   /* Delta should never be the base of a delta chain. */
   throw InternalError();

templates.cpp

@@ -9,4 +9,4 @@ template class unordered_map<string, GitObject *>;
 template class unordered_map<off_t, GitObject *>;
 template class unordered_multimap<GitObject *, GitObject *>;
 template class unordered_map<GitObject *, char>;
-template class unordered_multimap<string, string>;
+