Don't save to DB if the disk has been added already.

src/dbbackend.cpp

@@ -21,6 +21,8 @@
 #include <string>
 #include <sstream>
 #include <utility>
+#include <boost/lexical_cast.hpp>
+#include <exception>
 
 namespace din {
 	namespace {
@@ -36,6 +38,53 @@ namespace din {
 		}
 	} //unnamed namespace
 
+	bool read_from_db (FileRecordData& parItem, SetRecordDataFull& parSet, const DinDBSettings& parDB, std::string&& parHash) {
+		using boost::lexical_cast;
+
+		pq::Connection conn(std::string(parDB.username), std::string(parDB.password), std::string(parDB.dbname), std::string(parDB.address), parDB.port);
+		conn.connect();
+
+		uint32_t group_id;
+		{
+			std::ostringstream oss;
+			oss << "SELECT path,level,group_id,is_directory,is_symlink,size FROM files WHERE hash=" <<
+				conn.escaped_literal(parHash) <<
+				" LIMIT 1;";
+
+			auto resultset = conn.query(oss.str());
+			if (resultset.empty()) {
+				return false;
+			}
+
+			auto row = resultset[0];
+			parItem.path = row["path"];
+			parItem.hash = std::move(parHash);
+			parItem.level = lexical_cast<uint16_t>(row["level"]);
+			parItem.size = lexical_cast<uint64_t>(row["size"]);
+			parItem.is_directory = (row["is_directory"] == "t" ? true : false);
+			parItem.is_symlink = (row["is_symlink"] == "t" ? true : false);
+			group_id = lexical_cast<uint32_t>(row["group_id"]);
+		}
+
+		{
+			std::ostringstream oss;
+			oss << "SELECT \"desc\",\"type\",\"disk_number\" FROM sets WHERE \"id\"=" << group_id << ';';
+
+			auto resultset = conn.query(oss.str());
+			if (resultset.empty()) {
+				std::ostringstream err_msg;
+				err_msg << "Missing set: found a record with group_id=" << group_id;
+				err_msg << " but there is no such id in table \"sets\"";
+				throw std::length_error(err_msg.str());
+			}
+			auto row = resultset[0];
+			parSet.type = lexical_cast<char>(row["type"]);
+			parSet.name = row["desc"];
+			parSet.disk_number = lexical_cast<uint32_t>(row["disk_number"]);
+		}
+		return true;
+	}
+
 	void write_to_db (const DinDBSettings& parDB, const std::vector<FileRecordData>& parData, const SetRecordData& parSetData) {
 		if (parData.empty()) {
 			return;

src/dbbackend.hpp

@@ -27,12 +27,18 @@ namespace din {
 	struct DinDBSettings;
 
 	struct FileRecordData {
-		const std::string path;
-		const std::string hash;
-		const uint16_t level;
-		const uint64_t size;
-		const bool is_directory;
-		const bool is_symlink;
+		std::string path;
+		std::string hash;
+		uint16_t level;
+		uint64_t size;
+		bool is_directory;
+		bool is_symlink;
+	};
+
+	struct SetRecordDataFull {
+		std::string name;
+		uint32_t disk_number;
+		char type;
 	};
 
 	struct SetRecordData {
@@ -41,6 +47,7 @@ namespace din {
 	};
 
 	void write_to_db ( const DinDBSettings& parDB, const std::vector<FileRecordData>& parData, const SetRecordData& parSetData );
+	bool read_from_db ( FileRecordData& parItem, SetRecordDataFull& parSet, const DinDBSettings& parDB, std::string&& parHash );
 } //namespace din
 
 #endif

src/indexer.cpp

@@ -277,10 +277,21 @@ namespace din {
 #endif
 	}
 
-	void Indexer::add_to_db (const std::string& parSetName, char parType) const {
+	bool Indexer::add_to_db (const std::string& parSetName, char parType, bool parForce) const {
 #if defined(WITH_PROGRESS_FEEDBACK)
 		assert(m_local_data->done_count == m_local_data->file_count);
 #endif
+
+		if (not parForce) {
+			std::string first_hash(tiger_to_string(m_local_data->paths.front().hash, true));
+			FileRecordData itm;
+			SetRecordDataFull set;
+			const bool already_in_db = read_from_db(itm, set, m_local_data->db_settings, std::move(first_hash));
+			if (already_in_db) {
+				return false;
+			}
+		}
+
 		PathName base_path(m_local_data->paths.front().path);
 		std::vector<FileRecordData> data;
 		data.reserve(m_local_data->paths.size());
@@ -297,6 +308,7 @@ namespace din {
 
 		SetRecordData set_data {parSetName, parType};
 		write_to_db(m_local_data->db_settings, data, set_data);
+		return true;
 	}
 
 	bool Indexer::add_path (const char* parPath, int parLevel, bool parIsDir, bool parIsSymLink) {

src/indexer.hpp

@@ -54,7 +54,7 @@ namespace din {
 		std::condition_variable& step_notify ( void );
 #endif
 		void calculate_hash ( void );
-		void add_to_db ( const std::string& parSetName, char parType ) const;
+		bool add_to_db ( const std::string& parSetName, char parType, bool parForce=false ) const;
 		bool empty ( void ) const;
 
 	private:

src/main.cpp

@@ -94,7 +94,9 @@ int main (int parArgc, char* parArgv[]) {
 		if (verbose) {
 			std::cout << "Writing to database...\n";
 		}
-		indexer.add_to_db(vm["setname"].as<std::string>(), vm["type"].as<char>());
+		if (not indexer.add_to_db(vm["setname"].as<std::string>(), vm["type"].as<char>())) {
+			std::cerr << "Not written to DB, likely because a set with the same hash already exists\n";
+		}
 	}
 	return 0;
 }