Detect content type after scanning and save to DB.

src/common/mediatypes.cpp

@@ -39,4 +39,8 @@ namespace dinlib {
 		}
 		return it_ret->second;
 	}
+
+	MediaTypes char_to_media_type (char parMType) {
+		return static_cast<MediaTypes>(parMType);
+	}
 } //namespace dinlib

src/machinery/guess_content_type.cpp

@@ -22,6 +22,7 @@
 #include "dindexer-machinery/set_listing.hpp"
 #include "dindexer-machinery/set_listing_helpers.hpp"
 #include "globbing.hpp"
+#include "pathname.hpp"
 #include <boost/iterator/filter_iterator.hpp>
 #include <boost/iterator/indirect_iterator.hpp>
 #include <boost/range/empty.hpp>
@@ -31,6 +32,7 @@
 #include <ciso646>
 #include <regex>
 #include <utility>
+#include <memory>
 
 namespace mchlib {
 	namespace {
@@ -120,6 +122,9 @@ namespace mchlib {
 	} //unnamed namespace
 
 	ContentTypes guess_content_type (dinlib::MediaTypes parMediaType, const ConstSetListingView& parContent, std::size_t parEntriesCount) {
+		if (boost::empty(parContent))
+			return ContentType_Empty;
+
 		std::vector<EntryChecking> checker_chain {
 			{ 100, &identify_video_dvd, ContentType_VideoDVD },
 			{ 200, &identify_video_cd, ContentType_VideoCD }
@@ -137,6 +142,24 @@ namespace mchlib {
 		}
 		return ContentType_Generic;
 	}
+
+	ContentTypes guess_content_type (dinlib::MediaTypes parMediaType, const std::vector<FileRecordData>& parContent) {
+		if (parContent.empty())
+			return ContentType_Empty;
+
+		//TODO: assert that the first item in the list is the shortest string
+		std::shared_ptr<PathName> pathname(new PathName(parContent.front().abs_path));
+		ConstSetListingView view(parContent.begin(), parContent.end(), pathname->atom_count(), pathname);
+		return guess_content_type(parMediaType, view, parContent.size());
+	}
+
+	char content_type_to_char (mchlib::ContentTypes parCType) {
+		return static_cast<char>(parCType);
+	}
+
+	ContentTypes char_to_content_type (char parCType) {
+		return static_cast<ContentTypes>(parCType);
+	}
 } //namespace mchlib
 
 #endif

src/scan/dbbackend.cpp

@@ -97,11 +97,13 @@ namespace din {
 		uint32_t new_group_id;
 		{
 			auto id_res = conn.query("INSERT INTO \"sets\" "
-				"(\"desc\",\"type\",\"app_name\") "
-				"VALUES ($1, $2, $3) RETURNING \"id\";",
+				"(\"desc\",\"type\",\"app_name\""
+				",\"content_type\") "
+				"VALUES ($1, $2, $3, $4) RETURNING \"id\";",
 				parSetData.name,
-				std::string(1, parSetData.type),
-				parSignature
+				boost::string_ref(&parSetData.type, 1),
+				parSignature,
+				boost::string_ref(&parSetData.content_type, 1)
 			);
 			assert(id_res.size() == 1);
 			assert(id_res[0].size() == 1);

src/scan/main.cpp

@@ -26,6 +26,7 @@
 #include "dindexer-machinery/machinery_info.hpp"
 #include "dindexer-common/common_info.hpp"
 #include "dindexer-common/settings.hpp"
+#include "dindexer-machinery/guess_content_type.hpp"
 #include "commandline.hpp"
 #include "dbbackend.hpp"
 #include <iostream>
@@ -42,7 +43,7 @@
 
 namespace {
 	void run_hash_calculation ( mchlib::Indexer& parIndexer, bool parShowProgress );
-	bool add_to_db ( const std::vector<mchlib::FileRecordData>& parData, const std::string& parSetName, char parType, const dinlib::SettingsDB& parDBSettings, bool parForce=false );
+	bool add_to_db ( const std::vector<mchlib::FileRecordData>& parData, const std::string& parSetName, char parType, char parContent, const dinlib::SettingsDB& parDBSettings, bool parForce=false );
 } //unnamed namespace
 
 int main (int parArgc, char* parArgv[]) {
@@ -117,11 +118,15 @@ int main (int parArgc, char* parArgv[]) {
 		return 1;
 	}
 	else {
+		const auto set_type_casted = dinlib::char_to_media_type(set_type);
+		const mchlib::ContentTypes content = mchlib::guess_content_type(set_type_casted, indexer.record_data());
+		const char content_type = mchlib::content_type_to_char(content);
+
 		run_hash_calculation(indexer, verbose);
 		if (verbose) {
 			std::cout << "Writing to database...\n";
 		}
-		if (not add_to_db(indexer.record_data(), vm["setname"].as<std::string>(), set_type, settings.db)) {
+		if (not add_to_db(indexer.record_data(), vm["setname"].as<std::string>(), set_type, content_type, settings.db)) {
 			std::cerr << "Not written to DB, likely because a set with the same hash already exists\n";
 		}
 	}
@@ -182,7 +187,7 @@ namespace {
 #endif
 	}
 
-	bool add_to_db (const std::vector<mchlib::FileRecordData>& parData, const std::string& parSetName, char parType, const dinlib::SettingsDB& parDBSettings, bool parForce) {
+	bool add_to_db (const std::vector<mchlib::FileRecordData>& parData, const std::string& parSetName, char parType, char parContentType, const dinlib::SettingsDB& parDBSettings, bool parForce) {
 		using mchlib::FileRecordData;
 		using mchlib::SetRecordDataFull;
 		using mchlib::SetRecordData;
@@ -197,7 +202,7 @@ namespace {
 			}
 		}
 
-		SetRecordData set_data {parSetName, parType};
+		SetRecordData set_data {parSetName, parType, parContentType };
 		const auto app_signature = dinlib::dindexer_signature();
 		const auto lib_signature = mchlib::lib_signature();
 		const std::string signature = std::string(app_signature.data(), app_signature.size()) + "/" + std::string(lib_signature.data(), lib_signature.size());