Implement searching for sets by name.

Searching for sets also prints the total count of files and
directories in that set.

dindexer.sql.in

@@ -4,7 +4,7 @@
 
 -- Dumped from database version 9.4.5
 -- Dumped by pg_dump version 9.4.5
--- Started on 2016-01-06 01:55:35 GMT
+-- Started on 2016-01-07 15:10:56 GMT
 
 SET statement_timeout = 0;
 SET lock_timeout = 0;
@@ -22,7 +22,7 @@ CREATE EXTENSION IF NOT EXISTS plpgsql WITH SCHEMA pg_catalog;
 
 
 --
--- TOC entry 2040 (class 0 OID 0)
+-- TOC entry 2041 (class 0 OID 0)
 -- Dependencies: 178
 -- Name: EXTENSION plpgsql; Type: COMMENT; Schema: -; Owner: 
 --
@@ -32,6 +32,23 @@ COMMENT ON EXTENSION plpgsql IS 'PL/pgSQL procedural language';
 
 SET search_path = public, pg_catalog;
 
+--
+-- TOC entry 192 (class 1255 OID 31654)
+-- Name: str_match_partial(text, text, boolean); Type: FUNCTION; Schema: public; Owner: @DB_OWNER_NAME@
+--
+
+CREATE FUNCTION str_match_partial(text, text, boolean) RETURNS boolean
+    LANGUAGE sql IMMUTABLE
+    AS $_$SELECT CASE WHEN $3 THEN
+LOWER($1) LIKE LOWER('%'||regexp_replace($2, '(\%_)', '\\1')||'%')
+ELSE
+$1 LIKE '%'||regexp_replace($2, '(\%_)', '\\1')||'%'
+END;
+$_$;
+
+
+ALTER FUNCTION public.str_match_partial(text, text, boolean) OWNER TO @DB_OWNER_NAME@;
+
 --
 -- TOC entry 191 (class 1255 OID 31290)
 -- Name: upcase_hash(); Type: FUNCTION; Schema: public; Owner: @DB_OWNER_NAME@
@@ -81,7 +98,7 @@ CREATE TABLE files (
 ALTER TABLE files OWNER TO @DB_OWNER_NAME@;
 
 --
--- TOC entry 2041 (class 0 OID 0)
+-- TOC entry 2042 (class 0 OID 0)
 -- Dependencies: 175
 -- Name: CONSTRAINT chk_hash_0 ON files; Type: COMMENT; Schema: public; Owner: @DB_OWNER_NAME@
 --
@@ -105,7 +122,7 @@ CREATE SEQUENCE files_id_seq
 ALTER TABLE files_id_seq OWNER TO @DB_OWNER_NAME@;
 
 --
--- TOC entry 2042 (class 0 OID 0)
+-- TOC entry 2043 (class 0 OID 0)
 -- Dependencies: 174
 -- Name: files_id_seq; Type: SEQUENCE OWNED BY; Schema: public; Owner: @DB_OWNER_NAME@
 --
@@ -131,7 +148,7 @@ CREATE TABLE sets (
 ALTER TABLE sets OWNER TO @DB_OWNER_NAME@;
 
 --
--- TOC entry 2043 (class 0 OID 0)
+-- TOC entry 2044 (class 0 OID 0)
 -- Dependencies: 177
 -- Name: COLUMN sets.type; Type: COMMENT; Schema: public; Owner: @DB_OWNER_NAME@
 --
@@ -161,7 +178,7 @@ CREATE SEQUENCE sets_id_seq
 ALTER TABLE sets_id_seq OWNER TO @DB_OWNER_NAME@;
 
 --
--- TOC entry 2044 (class 0 OID 0)
+-- TOC entry 2045 (class 0 OID 0)
 -- Dependencies: 176
 -- Name: sets_id_seq; Type: SEQUENCE OWNED BY; Schema: public; Owner: @DB_OWNER_NAME@
 --
@@ -170,7 +187,7 @@ ALTER SEQUENCE sets_id_seq OWNED BY sets.id;
 
 
 --
--- TOC entry 1904 (class 2604 OID 31284)
+-- TOC entry 1905 (class 2604 OID 31284)
 -- Name: id; Type: DEFAULT; Schema: public; Owner: @DB_OWNER_NAME@
 --
 
@@ -178,7 +195,7 @@ ALTER TABLE ONLY files ALTER COLUMN id SET DEFAULT nextval('files_id_seq'::regcl
 
 
 --
--- TOC entry 1908 (class 2604 OID 31414)
+-- TOC entry 1909 (class 2604 OID 31414)
 -- Name: id; Type: DEFAULT; Schema: public; Owner: @DB_OWNER_NAME@
 --
 
@@ -186,7 +203,7 @@ ALTER TABLE ONLY sets ALTER COLUMN id SET DEFAULT nextval('sets_id_seq'::regclas
 
 
 --
--- TOC entry 1917 (class 2606 OID 31289)
+-- TOC entry 1918 (class 2606 OID 31289)
 -- Name: pk_files_id; Type: CONSTRAINT; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -195,7 +212,7 @@ ALTER TABLE ONLY files
 
 
 --
--- TOC entry 1921 (class 2606 OID 31420)
+-- TOC entry 1922 (class 2606 OID 31420)
 -- Name: pk_sets_id; Type: CONSTRAINT; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -204,7 +221,7 @@ ALTER TABLE ONLY sets
 
 
 --
--- TOC entry 1919 (class 2606 OID 31294)
+-- TOC entry 1920 (class 2606 OID 31294)
 -- Name: uniq_item; Type: CONSTRAINT; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -213,7 +230,7 @@ ALTER TABLE ONLY files
 
 
 --
--- TOC entry 1913 (class 1259 OID 31426)
+-- TOC entry 1914 (class 1259 OID 31426)
 -- Name: fki_files_sets; Type: INDEX; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -221,7 +238,7 @@ CREATE INDEX fki_files_sets ON files USING btree (group_id);
 
 
 --
--- TOC entry 1914 (class 1259 OID 31651)
+-- TOC entry 1915 (class 1259 OID 31651)
 -- Name: idx_mimetype; Type: INDEX; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -229,7 +246,7 @@ CREATE INDEX idx_mimetype ON files USING btree (mimetype, charset);
 
 
 --
--- TOC entry 1915 (class 1259 OID 31292)
+-- TOC entry 1916 (class 1259 OID 31292)
 -- Name: idx_paths; Type: INDEX; Schema: public; Owner: @DB_OWNER_NAME@; Tablespace: 
 --
 
@@ -237,7 +254,7 @@ CREATE INDEX idx_paths ON files USING btree (path);
 
 
 --
--- TOC entry 1923 (class 2620 OID 31291)
+-- TOC entry 1924 (class 2620 OID 31291)
 -- Name: triggerupcasehash; Type: TRIGGER; Schema: public; Owner: @DB_OWNER_NAME@
 --
 
@@ -245,7 +262,7 @@ CREATE TRIGGER triggerupcasehash BEFORE INSERT OR UPDATE ON files FOR EACH ROW E
 
 
 --
--- TOC entry 1922 (class 2606 OID 31421)
+-- TOC entry 1923 (class 2606 OID 31421)
 -- Name: fk_files_sets; Type: FK CONSTRAINT; Schema: public; Owner: @DB_OWNER_NAME@
 --
 
@@ -254,7 +271,7 @@ ALTER TABLE ONLY files
 
 
 --
--- TOC entry 2039 (class 0 OID 0)
+-- TOC entry 2040 (class 0 OID 0)
 -- Dependencies: 8
 -- Name: public; Type: ACL; Schema: -; Owner: postgres
 --
@@ -265,7 +282,7 @@ GRANT ALL ON SCHEMA public TO postgres;
 GRANT ALL ON SCHEMA public TO PUBLIC;
 
 
--- Completed on 2016-01-06 01:55:37 GMT
+-- Completed on 2016-01-07 15:11:05 GMT
 
 --
 -- PostgreSQL database dump complete

src/locate/commandline.cpp

@@ -27,6 +27,7 @@ namespace din {
 		po::options_description set_options(ACTION_NAME " options");
 		set_options.add_options()
 			("case-insensitive,i", "Disable case sensitivity during search")
+			("set,s", "Look for matching sets instead of files")
 			//("option,o", po::value<std::string>()->default_value("default_value"), "Help message")
 			//("option2", po::value<int>(), "Help message")
 		;

src/locate/main.cpp

@@ -29,6 +29,15 @@ namespace din {
 		parStream << parItem.group_id << '\t' << parItem.id << '\t' << parItem.path;
 		return parStream;
 	}
+
+	std::ostream& operator<< (std::ostream& parStream, const LocatedSet& parItem) {
+		const decltype(parItem.dir_count) one = 1;
+		const auto dircount = std::max(parItem.dir_count, one) - one;
+
+		parStream << parItem.id << "\t\"" << parItem.desc << "\"\t" <<
+			'\t' << parItem.files_count << '\t' << dircount;
+		return parStream;
+	}
 } //namespace din
 
 int main (int parArgc, char* parArgv[]) {
@@ -59,7 +68,13 @@ int main (int parArgc, char* parArgv[]) {
 		}
 	}
 
-	const auto results = din::locate_in_db(settings.db, vm["substring"].as<std::string>(), not not vm.count("case-insensitive"));
+	if (vm.count("set")) {
-	std::copy(results.begin(), results.end(), std::ostream_iterator<din::LocatedItem>(std::cout, "\n"));
+		const auto results = din::locate_sets_in_db(settings.db, vm["substring"].as<std::string>(), not not vm.count("case-insensitive"));
+		std::copy(results.begin(), results.end(), std::ostream_iterator<din::LocatedSet>(std::cout, "\n"));
+	}
+	else {
+		const auto results = din::locate_in_db(settings.db, vm["substring"].as<std::string>(), not not vm.count("case-insensitive"));
+		std::copy(results.begin(), results.end(), std::ostream_iterator<din::LocatedItem>(std::cout, "\n"));
+	}
 	return 0;
 }

src/locate/postgre_locate.cpp

@@ -26,6 +26,33 @@
 namespace din {
 	namespace {
 		const int g_max_results = 200;
+
+		pq::Connection make_pq_conn ( const dinlib::SettingsDB& parDB, bool parOpen=true );
+
+		pq::Connection make_pq_conn (const dinlib::SettingsDB& parDB, bool parOpen) {
+			auto conn = pq::Connection(std::string(parDB.username), std::string(parDB.password), std::string(parDB.dbname), std::string(parDB.address), parDB.port);
+			if (parOpen) {
+				conn.connect();
+			}
+			return std::move(conn);
+		}
+
+		std::vector<LocatedSet> sets_result_to_vec (pq::ResultSet&& parResult) {
+			using boost::lexical_cast;
+
+			std::vector<LocatedSet> retval;
+			retval.reserve(parResult.size());
+			for (const auto& record : parResult) {
+				retval.push_back(LocatedSet{
+					record["desc"],
+					lexical_cast<decltype(LocatedSet::id)>(record["id"]),
+					lexical_cast<decltype(LocatedSet::files_count)>(record["file_count"]),
+					lexical_cast<decltype(LocatedSet::dir_count)>(record["dir_count"])
+				});
+			}
+
+			return std::move(retval);
+		}
 	} //unnamed namespace
 
 	std::vector<LocatedItem> locate_in_db (const dinlib::SettingsDB& parDB, const std::string& parSearch, bool parCaseInsensitive) {
@@ -33,8 +60,7 @@ namespace din {
 		using boost::string_ref;
 		namespace ba = boost::algorithm;
 
-		pq::Connection conn(std::string(parDB.username), std::string(parDB.password), std::string(parDB.dbname), std::string(parDB.address), parDB.port);
+		auto conn = make_pq_conn(parDB);
-		conn.connect();
 
 		const auto clean_string_with_quotes = conn.escaped_literal(parSearch);
 		const auto clean_string = string_ref(clean_string_with_quotes).substr(1, clean_string_with_quotes.size() - 2);
@@ -64,4 +90,34 @@ namespace din {
 
 		return std::move(retval);
 	}
+
+	std::vector<LocatedSet> locate_sets_in_db (const dinlib::SettingsDB& parDB, const std::string& parSearch, bool parCaseInsensitive) {
+		auto conn = make_pq_conn(parDB);
+
+		const std::string query = std::string("SELECT \"id\", \"desc\", "
+			"(SELECT COUNT(*) FROM \"files\" WHERE \"group_id\"=\"sets\".\"id\" AND NOT \"is_directory\") as \"file_count\", "
+			"(SELECT COUNT(*) FROM \"files\" WHERE \"group_id\"=\"sets\".\"id\" AND \"is_directory\") as \"dir_count\" "
+			"FROM \"sets\" WHERE str_match_partial(\"desc\", $1, $2) LIMIT "
+		) + std::to_string(g_max_results) + ";";
+
+		auto result = conn.query(query, parSearch, parCaseInsensitive);
+		return sets_result_to_vec(std::move(result));
+	}
+
+	std::vector<LocatedSet> locate_sets_in_db (const dinlib::SettingsDB& parDB, const std::string& parSearch, const std::vector<uint32_t>& parSets, bool parCaseInsensitive) {
+		if (parSets.empty()) {
+			return locate_sets_in_db(parDB, parSearch, parCaseInsensitive);
+		}
+
+		auto conn = make_pq_conn(parDB);
+
+		const std::string query = std::string("SELECT \"id\", \"desc\", "
+			"(SELECT COUNT(*) FROM \"files\" WHERE \"group_id\"=\"sets\".\"id\" AND NOT \"is_directory\") as \"file_count\", "
+			"(SELECT COUNT(*) FROM \"files\" WHERE \"group_id\"=\"sets\".\"id\" AND \"is_directory\") as \"dir_count\" "
+			"FROM \"sets\" WHERE \"id\" = ANY($1) AND str_match_partial(\"desc\", $3, $2) LIMIT "
+		) + std::to_string(g_max_results) + ";";
+
+		auto result = conn.query(query, parSearch, parCaseInsensitive, parSets);
+		return sets_result_to_vec(std::move(result));
+	}
 } //namespace din

src/locate/postgre_locate.hpp

@@ -30,7 +30,16 @@ namespace din {
 		uint32_t group_id;
 	};
 
+	struct LocatedSet {
+		std::string desc;
+		uint32_t id;
+		uint32_t files_count;
+		uint32_t dir_count;
+	};
+
 	std::vector<LocatedItem> locate_in_db ( const dinlib::SettingsDB& parDB, const std::string& parSearch, bool parCaseInsensitive );
+	std::vector<LocatedSet> locate_sets_in_db ( const dinlib::SettingsDB& parDB, const std::string& parSearch, bool parCaseInsensitive );
+	std::vector<LocatedSet> locate_sets_in_db ( const dinlib::SettingsDB& parDB, const std::string& parSearch, const std::vector<uint32_t>& parSets, bool parCaseInsensitive );
 } //namespace din
 
 #endif