1
0
Fork 0
mirror of https://github.com/KingDuckZ/dindexer.git synced 2025-07-03 14:14:11 +00:00

Limit insertion to 100 records per query, but this is still suboptimal.

This commit is contained in:
King_DuckZ 2015-11-11 12:35:52 +00:00
parent 4236b2ece8
commit 444d2a2fa3

View file

@ -23,6 +23,7 @@
namespace din { namespace din {
namespace { namespace {
const std::size_t g_batch_size = 100;
} //unnamed namespace } //unnamed namespace
void write_to_db (const std::vector<FileRecordData>& parData) { void write_to_db (const std::vector<FileRecordData>& parData) {
@ -30,15 +31,18 @@ namespace din {
return; return;
} }
std::ostringstream query;
query << "BEGIN;\n";
query << "INSERT INTO \"Files\" (path, hash, level, group_id, is_directory, is_symlink, size) VALUES ";
pq::Connection conn("michele", "password", "dindexer", "100.200.100.200", 5432); pq::Connection conn("michele", "password", "dindexer", "100.200.100.200", 5432);
conn.connect(); conn.connect();
conn.query_void("BEGIN;");
//TODO: use COPY instead of INSERT INTO
for (std::size_t z = 0; z < parData.size(); z += g_batch_size) {
std::ostringstream query;
query << "INSERT INTO \"Files\" (path, hash, level, group_id, is_directory, is_symlink, size) VALUES ";
const char* comma = ""; const char* comma = "";
for (const auto& itm : parData) { for (auto i = z; i < std::min(z + g_batch_size, parData.size()); ++i) {
const auto& itm = parData[i];
query << comma; query << comma;
query << '(' << conn.escape_literal(itm.path) << ",'" << itm.hash << "'," query << '(' << conn.escape_literal(itm.path) << ",'" << itm.hash << "',"
<< itm.level << ',' << itm.level << ','
@ -48,8 +52,10 @@ namespace din {
comma = ","; comma = ",";
} }
query << ';'; query << ';';
query << "\nCOMMIT;"; //query << "\nCOMMIT;";
conn.query_void(query.str()); conn.query_void(query.str());
} }
conn.query_void("COMMIT;");
}
} //namespace din } //namespace din