Outputs where all amounts are known spent can now be pruned

Only for pre rct for obvious reasons.

Note: DO NOT use a known spent list which includes outputs
which are not known spent. If the list includes any output
that's just strongly thought to be spent, but not provably
so, you risk finding yourself unable to sync past the point
where that output is spent.

I estimate only 200 MB saved on current mainnet though,
unless the new blackballing rule unearths a good amount of
large-amount-set extra spent outs.

3 files changed, 75 insertions, 4 deletions

diff --git a/src/blockchain_db/blockchain_db.h b/src/blockchain_db/blockchain_db.h
index 53e33898a..a8f4eafe0 100644
--- a/src/blockchain_db/blockchain_db.h
+++ b/src/blockchain_db/blockchain_db.h
@@ -1406,6 +1406,13 @@ public:
   virtual cryptonote::blobdata get_txpool_tx_blob(const crypto::hash& txid) const = 0;
 
   /**
+   * @brief prune output data for the given amount
+   *
+   * @param amount the amount for which to prune data
+   */
+  virtual void prune_outputs(uint64_t amount) = 0;
+
+  /**
    * @brief runs a function over all txpool transactions
    *
    * The subclass should run the passed function for each txpool tx it has
diff --git a/src/blockchain_db/lmdb/db_lmdb.cpp b/src/blockchain_db/lmdb/db_lmdb.cpp
index d260caa75..142e86b5d 100644
--- a/src/blockchain_db/lmdb/db_lmdb.cpp
+++ b/src/blockchain_db/lmdb/db_lmdb.cpp
@@ -1077,6 +1077,60 @@ void BlockchainLMDB::remove_output(const uint64_t amount, const uint64_t& out_in
     throw0(DB_ERROR(lmdb_error(std::string("Error deleting amount for output index ").append(boost::lexical_cast<std::string>(out_index).append(": ")).c_str(), result).c_str()));
 }
 
+void BlockchainLMDB::prune_outputs(uint64_t amount)
+{
+  LOG_PRINT_L3("BlockchainLMDB::" << __func__);
+  check_open();
+  mdb_txn_cursors *m_cursors = &m_wcursors;
+  CURSOR(output_amounts);
+  CURSOR(output_txs);
+
+  MINFO("Pruning outputs for amount " << amount);
+
+  MDB_val v;
+  MDB_val_set(k, amount);
+  int result = mdb_cursor_get(m_cur_output_amounts, &k, &v, MDB_SET);
+  if (result == MDB_NOTFOUND)
+    return;
+  if (result)
+    throw0(DB_ERROR(lmdb_error("Error looking up outputs: ", result).c_str()));
+
+  // gather output ids
+  mdb_size_t num_elems;
+  mdb_cursor_count(m_cur_output_amounts, &num_elems);
+  MINFO(num_elems << " outputs found");
+  std::vector<uint64_t> output_ids;
+  output_ids.reserve(num_elems);
+  while (1)
+  {
+    const pre_rct_outkey *okp = (const pre_rct_outkey *)v.mv_data;
+    output_ids.push_back(okp->output_id);
+    MDEBUG("output id " << okp->output_id);
+    result = mdb_cursor_get(m_cur_output_amounts, &k, &v, MDB_NEXT_DUP);
+    if (result == MDB_NOTFOUND)
+      break;
+    if (result)
+      throw0(DB_ERROR(lmdb_error("Error counting outputs: ", result).c_str()));
+  }
+  if (output_ids.size() != num_elems)
+    throw0(DB_ERROR("Unexpected number of outputs"));
+
+  result = mdb_cursor_del(m_cur_output_amounts, MDB_NODUPDATA);
+  if (result)
+    throw0(DB_ERROR(lmdb_error("Error deleting outputs: ", result).c_str()));
+
+  for (uint64_t output_id: output_ids)
+  {
+    MDB_val_set(v, output_id);
+    result = mdb_cursor_get(m_cur_output_txs, (MDB_val *)&zerokval, &v, MDB_GET_BOTH);
+    if (result)
+      throw0(DB_ERROR(lmdb_error("Error looking up output: ", result).c_str()));
+    result = mdb_cursor_del(m_cur_output_txs, 0);
+    if (result)
+      throw0(DB_ERROR(lmdb_error("Error deleting output: ", result).c_str()));
+  }
+}
+
 void BlockchainLMDB::add_spent_key(const crypto::key_image& k_image)
 {
   LOG_PRINT_L3("BlockchainLMDB::" << __func__);
@@ -2231,11 +2285,19 @@ uint64_t BlockchainLMDB::num_outputs() const
   TXN_PREFIX_RDONLY();
   int result;
 
-  // get current height
-  MDB_stat db_stats;
-  if ((result = mdb_stat(m_txn, m_output_txs, &db_stats)))
+  RCURSOR(output_txs)
+
+  uint64_t num = 0;
+  MDB_val k, v;
+  result = mdb_cursor_get(m_cur_output_txs, &k, &v, MDB_LAST);
+  if (result == MDB_NOTFOUND)
+    num = 0;
+  else if (result == 0)
+    num = 1 + ((const outtx*)v.mv_data)->output_id;
+  else
     throw0(DB_ERROR(lmdb_error("Failed to query m_output_txs: ", result).c_str()));
-  return db_stats.ms_entries;
+
+  return num;
 }
 
 bool BlockchainLMDB::tx_exists(const crypto::hash& h) const
diff --git a/src/blockchain_db/lmdb/db_lmdb.h b/src/blockchain_db/lmdb/db_lmdb.h
index 26159ab4d..6db241240 100644
--- a/src/blockchain_db/lmdb/db_lmdb.h
+++ b/src/blockchain_db/lmdb/db_lmdb.h
@@ -345,6 +345,8 @@ private:
 
   void remove_output(const uint64_t amount, const uint64_t& out_index);
 
+  virtual void prune_outputs(uint64_t amount);
+
   virtual void add_spent_key(const crypto::key_image& k_image);
 
   virtual void remove_spent_key(const crypto::key_image& k_image);