Refactor Indexer so that writing to the db is done outside.

CMakeLists.txt

@@ -49,10 +49,13 @@ target_include_directories(${bare_name}-inc
 	INTERFACE ${CMAKE_SOURCE_DIR}/include
 )
 
-add_subdirectory(src/scan)
+#Libraries
 add_subdirectory(src/pq)
-add_subdirectory(src/main)
 add_subdirectory(src/common)
+
+#Actions
+add_subdirectory(src/main)
+add_subdirectory(src/scan)
 add_subdirectory(src/delete)
 add_subdirectory(src/query)
 add_subdirectory(src/locate)

src/scan/indexer.cpp

@@ -18,14 +18,12 @@
 #include "indexer.hpp"
 #include "pathname.hpp"
 #include "tiger.hpp"
-#include "dbbackend.hpp"
 #include "dindexer-common/settings.hpp"
 #include "filestats.hpp"
 #include "mimetype.hpp"
 #include "recorddata.hpp"
 #include <algorithm>
 #include <functional>
-#include <vector>
 #include <stdexcept>
 #if defined(WITH_PROGRESS_FEEDBACK)
 #	include <atomic>
@@ -171,12 +169,29 @@ namespace din {
 				parSt.is_symlink
 			);
 		}
+
+		bool file_record_data_lt (const FileRecordData& parLeft, const FileRecordData& parRight) {
+			const FileRecordData& l = parLeft;
+			const FileRecordData& r = parRight;
+			return
+				(l.level < r.level)
+				or (l.level == r.level and l.is_directory and not r.is_directory)
+				or (l.level == r.level and l.is_directory == r.is_directory and l.path < r.path)
+
+				//sort by directory - parent first, children later
+				//(level == o.level and is_dir and not o.is_dir)
+				//or (level == o.level and is_dir == o.is_dir and path < o.path)
+				//or (level > o.level + 1)
+				//or (level + 1 == o.level and is_dir and not o.is_dir and path < o.path)
+				//or (level + 1 == o.level and is_dir and not o.is_dir and path == PathName(o.path).dirname())
+				//or (level == o.level + 1 and not (o.is_dir and not is_dir and o.path == PathName(path).dirname()))
+			;
+		}
 	} //unnamed namespace
 
 	struct Indexer::LocalData {
 		typedef std::vector<FileRecordData> PathList;
 
-		dinlib::SettingsDB db_settings;
 		PathList paths;
 #if defined(WITH_PROGRESS_FEEDBACK)
 		std::atomic<std::size_t> done_count;
@@ -187,25 +202,7 @@ namespace din {
 		bool ignore_read_errors;
 	};
 
-	bool file_record_data_lt (const FileRecordData& parLeft, const FileRecordData& parRight) {
-		const FileRecordData& l = parLeft;
-		const FileRecordData& r = parRight;
-		return
-			(l.level < r.level)
-			or (l.level == r.level and l.is_directory and not r.is_directory)
-			or (l.level == r.level and l.is_directory == r.is_directory and l.path < r.path)
-
-			//sort by directory - parent first, children later
-			//(level == o.level and is_dir and not o.is_dir)
-			//or (level == o.level and is_dir == o.is_dir and path < o.path)
-			//or (level > o.level + 1)
-			//or (level + 1 == o.level and is_dir and not o.is_dir and path < o.path)
-			//or (level + 1 == o.level and is_dir and not o.is_dir and path == PathName(o.path).dirname())
-			//or (level == o.level + 1 and not (o.is_dir and not is_dir and o.path == PathName(path).dirname()))
-		;
-	}
-
-	Indexer::Indexer (const dinlib::Settings& parSettings) :
+	Indexer::Indexer() :
 		m_local_data(new LocalData)
 	{
 #if !defined(NDEBUG)
@@ -225,7 +222,6 @@ namespace din {
 		m_local_data->processing_index = 0;
 #endif
 		m_local_data->file_count = 0;
-		m_local_data->db_settings = parSettings.db;
 	}
 
 	Indexer::~Indexer() noexcept {
@@ -297,28 +293,6 @@ namespace din {
 #endif
 	}
 
-	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) {
-			const auto& first_hash = m_local_data->paths.front().hash;
-			FileRecordData itm;
-			SetRecordDataFull set;
-			const bool already_in_db = read_from_db(itm, set, m_local_data->db_settings, first_hash);
-			if (already_in_db) {
-				return false;
-			}
-		}
-
-		PathName base_path(m_local_data->paths.front().path);
-
-		SetRecordData set_data {parSetName, parType};
-		write_to_db(m_local_data->db_settings, m_local_data->paths, set_data);
-		return true;
-	}
-
 	bool Indexer::add_path (const char* parPath, const fastf::FileStats& parStats) {
 		m_local_data->paths.push_back(
 			make_file_record_data(parPath, parStats));
@@ -386,4 +360,11 @@ namespace din {
 	void Indexer::ignore_read_errors (bool parIgnore) {
 		m_local_data->ignore_read_errors = parIgnore;
 	}
+
+	const std::vector<FileRecordData>& Indexer::record_data() const {
+#if defined(WITH_PROGRESS_FEEDBACK)
+		assert(m_local_data->done_count == m_local_data->file_count);
+#endif
+		return m_local_data->paths;
+	}
 } //namespace din

src/scan/indexer.hpp

@@ -20,6 +20,7 @@
 
 #include <memory>
 #include <string>
+#include <vector>
 
 #if !defined(NDEBUG)
 #	define INDEXER_VERBOSE
@@ -40,9 +41,11 @@ namespace dinlib {
 } //namespace dinlib
 
 namespace din {
+	struct FileRecordData;
+
 	class Indexer {
 	public:
-		explicit Indexer ( const dinlib::Settings& parSettings );
+		Indexer ( void );
 		Indexer ( Indexer&& ) = default;
 		Indexer ( const Indexer& ) = delete;
 		~Indexer ( void ) noexcept;
@@ -60,9 +63,9 @@ namespace din {
 		std::condition_variable& step_notify ( void );
 #endif
 		void calculate_hash ( void );
-		bool add_to_db ( const std::string& parSetName, char parType, bool parForce=false ) const;
 		bool empty ( void ) const;
 		void ignore_read_errors ( bool parIgnore );
+		const std::vector<FileRecordData>& record_data ( void ) const;
 
 	private:
 		struct LocalData;

src/scan/main.cpp

@@ -19,6 +19,13 @@
 #	undef WITH_PROGRESS_FEEDBACK
 #endif
 
+#include "recorddata.hpp"
+#include "dindexerConfig.h"
+#include "filesearcher.hpp"
+#include "indexer.hpp"
+#include "dindexer-common/settings.hpp"
+#include "commandline.hpp"
+#include "dbbackend.hpp"
 #include <iostream>
 #include <iomanip>
 #include <ciso646>
@@ -30,14 +37,10 @@
 #	include <mutex>
 #	include <condition_variable>
 #endif
-#include "dindexerConfig.h"
-#include "filesearcher.hpp"
-#include "indexer.hpp"
-#include "dindexer-common/settings.hpp"
-#include "commandline.hpp"
 
 namespace {
 	void run_hash_calculation ( din::Indexer& parIndexer, bool parShowProgress );
+	bool add_to_db ( const std::vector<din::FileRecordData>& parData, const std::string& parSetName, char parType, const dinlib::SettingsDB& parDBSettings, bool parForce=false );
 } //unnamed namespace
 
 int main (int parArgc, char* parArgv[]) {
@@ -96,7 +99,7 @@ int main (int parArgc, char* parArgv[]) {
 
 	std::cout << "constructing...\n";
 
-	din::Indexer indexer(settings);
+	din::Indexer indexer;
 	indexer.ignore_read_errors(vm.count("ignore-errors") > 0);
 	fastf::FileSearcher searcher(search_path);
 	fastf::FileSearcher::ConstCharVecType ext, ignore;
@@ -116,7 +119,7 @@ int main (int parArgc, char* parArgv[]) {
 		if (verbose) {
 			std::cout << "Writing to database...\n";
 		}
-		if (not indexer.add_to_db(vm["setname"].as<std::string>(), set_type)) {
+		if (not add_to_db(indexer.record_data(), vm["setname"].as<std::string>(), set_type, settings.db)) {
 			std::cerr << "Not written to DB, likely because a set with the same hash already exists\n";
 		}
 	}
@@ -177,4 +180,23 @@ namespace {
 #endif
 	}
 
+	bool add_to_db (const std::vector<din::FileRecordData>& parData, const std::string& parSetName, char parType, const dinlib::SettingsDB& parDBSettings, bool parForce) {
+		using din::FileRecordData;
+		using din::SetRecordDataFull;
+		using din::SetRecordData;
+
+		if (not parForce) {
+			const auto& first_hash = parData.front().hash;
+			FileRecordData itm;
+			SetRecordDataFull set;
+			const bool already_in_db = read_from_db(itm, set, parDBSettings, first_hash);
+			if (already_in_db) {
+				return false;
+			}
+		}
+
+		SetRecordData set_data {parSetName, parType};
+		write_to_db(parDBSettings, parData, set_data);
+		return true;
+	}
 } //unnamed namespace