Detect content type after scanning and save to DB.

dindexer.sql.in

@@ -4,7 +4,7 @@
 
 -- Dumped from database version 9.4.5
 -- Dumped by pg_dump version 9.4.5
--- Started on 2016-01-29 20:18:23 CET
+-- Started on 2016-02-22 08:40:56 CET
 
 SET statement_timeout = 0;
 SET lock_timeout = 0;
@@ -22,7 +22,7 @@ CREATE EXTENSION IF NOT EXISTS plpgsql WITH SCHEMA pg_catalog;
 
 
 --
--- TOC entry 2044 (class 0 OID 0)
+-- TOC entry 2045 (class 0 OID 0)
 -- Dependencies: 176
 -- Name: EXTENSION plpgsql; Type: COMMENT; Schema: -; Owner: 
 --
@@ -112,7 +112,7 @@ CREATE TABLE files (
 ALTER TABLE files OWNER TO @DB_OWNER_NAME@;
 
 --
--- TOC entry 2045 (class 0 OID 0)
+-- TOC entry 2046 (class 0 OID 0)
 -- Dependencies: 172
 -- Name: CONSTRAINT chk_hash_0 ON files; Type: COMMENT; Schema: public; Owner: @DB_OWNER_NAME@
 --
@@ -136,7 +136,7 @@ CREATE SEQUENCE files_id_seq
 ALTER TABLE files_id_seq OWNER TO @DB_OWNER_NAME@;
 
 --
--- TOC entry 2046 (class 0 OID 0)
+-- TOC entry 2047 (class 0 OID 0)
 -- Dependencies: 173
 -- Name: files_id_seq; Type: SEQUENCE OWNED BY; Schema: public; Owner: @DB_OWNER_NAME@
 --
@@ -156,6 +156,7 @@ CREATE TABLE sets (
     disk_number integer DEFAULT 0 NOT NULL,
     creation timestamp with time zone DEFAULT now() NOT NULL,
     app_name character varying NOT NULL,
+    content_type character(1) DEFAULT 'G'::bpchar NOT NULL,
     CONSTRAINT chk_sets_type CHECK (((((((((type = 'C'::bpchar) OR (type = 'D'::bpchar)) OR (type = 'V'::bpchar)) OR (type = 'B'::bpchar)) OR (type = 'F'::bpchar)) OR (type = 'H'::bpchar)) OR (type = 'Z'::bpchar)) OR (type = 'O'::bpchar)))
 );
 
@@ -163,7 +164,7 @@ CREATE TABLE sets (
 ALTER TABLE sets OWNER TO @DB_OWNER_NAME@;
 
 --
--- TOC entry 2047 (class 0 OID 0)
+-- TOC entry 2048 (class 0 OID 0)
 -- Dependencies: 174
 -- Name: COLUMN sets.type; Type: COMMENT; Schema: public; Owner: @DB_OWNER_NAME@
 --
@@ -193,7 +194,7 @@ CREATE SEQUENCE sets_id_seq
 ALTER TABLE sets_id_seq OWNER TO @DB_OWNER_NAME@;
 
 --
--- TOC entry 2048 (class 0 OID 0)
+-- TOC entry 2049 (class 0 OID 0)
 -- Dependencies: 175
 -- Name: sets_id_seq; Type: SEQUENCE OWNED BY; Schema: public; Owner: @DB_OWNER_NAME@
 --
@@ -218,7 +219,7 @@ ALTER TABLE ONLY sets ALTER COLUMN id SET DEFAULT nextval('sets_id_seq'::regclas
 
 
 --
--- TOC entry 1921 (class 2606 OID 16414)
+-- TOC entry 1922 (class 2606 OID 16414)
 -- Name: pk_files_id; Type: CONSTRAINT; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -227,7 +228,7 @@ ALTER TABLE ONLY files
 
 
 --
--- TOC entry 1925 (class 2606 OID 16416)
+-- TOC entry 1926 (class 2606 OID 16416)
 -- Name: pk_sets_id; Type: CONSTRAINT; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -236,7 +237,7 @@ ALTER TABLE ONLY sets
 
 
 --
--- TOC entry 1923 (class 2606 OID 16418)
+-- TOC entry 1924 (class 2606 OID 16418)
 -- Name: uniq_item; Type: CONSTRAINT; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -245,7 +246,7 @@ ALTER TABLE ONLY files
 
 
 --
--- TOC entry 1917 (class 1259 OID 16419)
+-- TOC entry 1918 (class 1259 OID 16419)
 -- Name: fki_files_sets; Type: INDEX; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -253,7 +254,7 @@ CREATE INDEX fki_files_sets ON files USING btree (group_id);
 
 
 --
--- TOC entry 1918 (class 1259 OID 16420)
+-- TOC entry 1919 (class 1259 OID 16420)
 -- Name: idx_mimetype; Type: INDEX; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -261,7 +262,7 @@ CREATE INDEX idx_mimetype ON files USING btree (mimetype, charset);
 
 
 --
--- TOC entry 1919 (class 1259 OID 16421)
+-- TOC entry 1920 (class 1259 OID 16421)
 -- Name: idx_paths; Type: INDEX; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -269,7 +270,7 @@ CREATE INDEX idx_paths ON files USING btree (path);
 
 
 --
--- TOC entry 1927 (class 2620 OID 16422)
+-- TOC entry 1928 (class 2620 OID 16422)
 -- Name: triggerupcasehash; Type: TRIGGER; Schema: public; Owner: @DB_OWNER_NAME@
 --
 
@@ -277,7 +278,7 @@ CREATE TRIGGER triggerupcasehash BEFORE INSERT OR UPDATE ON files FOR EACH ROW E
 
 
 --
--- TOC entry 1926 (class 2606 OID 16423)
+-- TOC entry 1927 (class 2606 OID 16423)
 -- Name: fk_files_sets; Type: FK CONSTRAINT; Schema: public; Owner: @DB_OWNER_NAME@
 --
 
@@ -286,7 +287,7 @@ ALTER TABLE ONLY files
 
 
 --
--- TOC entry 2043 (class 0 OID 0)
+-- TOC entry 2044 (class 0 OID 0)
 -- Dependencies: 6
 -- Name: public; Type: ACL; Schema: -; Owner: postgres
 --
@@ -297,7 +298,7 @@ GRANT ALL ON SCHEMA public TO postgres;
 GRANT ALL ON SCHEMA public TO PUBLIC;
 
 
--- Completed on 2016-01-29 20:18:23 CET
+-- Completed on 2016-02-22 08:40:57 CET
 
 --
 -- PostgreSQL database dump complete

include/dindexer-common/compatibility.h

@@ -0,0 +1,62 @@
+/* Copyright 2015, 2016, Michele Santullo
+ * This file is part of "dindexer".
+ *
+ * "dindexer" is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * "dindexer" is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with "dindexer".  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef id45CDD1DAEF4F42968E3C89F68FDDA9BC
+#define id45CDD1DAEF4F42968E3C89F68FDDA9BC
+
+#if defined(__GNUC__)
+#	if defined(__clang__)
+#		if !defined(__has_attribute)
+			//Fall back to version number comparing
+#		else
+#			if __has_attribute(flatten)
+#				define a_flatten __attribute__((flatten))
+#			else
+#				define a_flatten
+#			endif
+#			if __has_attribute(always_inline)
+#				define a_always_inline __attribute__((always_inline))
+#			else
+#				define a_always_inline
+#			endif
+#			if __has_attribute(pure)
+#				define a_pure __attribute__((pure))
+#			else
+#				define a_pure
+#			endif
+#			if __has_attribute(deprecated)
+#				define a_deprecated __attribute__((deprecated))
+#			else
+#				define a_deprecated
+#endif
+#		endif
+#	else
+		//Fix here if you get warnings about unsupported attributes on your compiler
+#		define a_flatten __attribute__((flatten))
+#		define a_always_inline __attribute__((always_inline))
+#		define a_pure __attribute__((pure))
+#		define a_deprecated __attribute__((deprecated))
+#	endif
+#else
+#	warning "Unsupported compiler, please fill this section or file a bug"
+#	define a_flatten
+#	define a_always_inline
+#	define a_pure
+#	define a_deprecated
+#endif
+
+#endif

include/dindexer-common/mediatypes.hpp

@@ -18,6 +18,7 @@
 #ifndef id700AFD0F33634ACC88079BB8853A9E13
 #define id700AFD0F33634ACC88079BB8853A9E13
 
+#include "dindexer-common/compatibility.h"
 #include <string>
 
 namespace dinlib {
@@ -33,6 +34,7 @@ namespace dinlib {
 	};
 
 	const std::string& media_type_to_str ( MediaTypes parType );
+	MediaTypes char_to_media_type ( char parMType ) a_pure;
 } //namespace dinlib
 
 #endif

include/dindexer-machinery/guess_content_type.hpp

@@ -20,21 +20,27 @@
 
 #include "dindexer-common/mediatypes.hpp"
 #include "dindexer-machinery/recorddata.hpp"
+#include "dindexer-common/compatibility.h"
 #include <vector>
 
 namespace mchlib {
 	enum ContentTypes {
-		ContentType_Generic,
-		ContentType_Backup,
-		ContentType_VideoDVD,
-		ContentType_VideoBD,
-		ContentType_VideoCD,
-		ContentType_Unknown
+		ContentType_Empty = 'E',
+		ContentType_Generic = 'G',
+		ContentType_Backup = 'A',
+		ContentType_VideoDVD = 'D',
+		ContentType_VideoBD = 'B',
+		ContentType_VideoCD = 'C',
+		ContentType_Unknown = 'U'
 	};
 
 	template <bool> class SetListingView;
 
 	ContentTypes guess_content_type ( dinlib::MediaTypes parMediaType, const SetListingView<true>& parContent, std::size_t parEntriesCount=0 );
+	ContentTypes guess_content_type ( dinlib::MediaTypes parMediaType, const std::vector<FileRecordData>& parContent );
+
+	char content_type_to_char ( ContentTypes parCType ) a_pure;
+	ContentTypes char_to_content_type ( char parCType ) a_pure;
 } //namespace mchlib
 
 #endif

include/dindexer-machinery/recorddata.hpp

@@ -73,15 +73,15 @@ namespace mchlib {
 		bool hash_valid;
 	};
 
-	struct SetRecordDataFull {
-		std::string name;
-		uint32_t disk_number;
+	struct SetRecordData {
+		boost::string_ref name;
 		char type;
+		char content_type;
 	};
 
-	struct SetRecordData {
-		const boost::string_ref name;
-		const char type;
+	struct SetRecordDataFull : public SetRecordData {
+		std::string name;
+		uint32_t disk_number;
 	};
 } //namespace mchlib
 

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());