在 SDK 上开放 addQueryString 及 setScwsMulti 相关接口功能

sdk/php/lib/XSIndex.class.php

@@ -303,6 +303,40 @@ public function delSynonym($raw, $synonym = null)
 		return $this;
 	}
 
+	/**
+	 * 设置当前索引库的分词复合等级
+	 * 复合等级是 scws 分词粒度控制的一个重要参数, 是长词细分处理依据, 默认为 3, 值范围 0~15
+	 * 注意: 这个设置仅直对当前索引库有效, 多次调用设置值被覆盖仅最后那次设置有效,
+	 * 而且仅对设置之后提交的索引数据起作用, 如需对以前的索引数据生效请重建索引.
+	 * @param int $level 要设置的分词复合等级
+	 * @return XSIndex 返回自身对象以支持串接操作
+	 * @since 1.4.7
+	 * @throw XSException 出错时抛出异常
+	 */
+	public function setScwsMulti($level)
+	{
+		$level = intval($level);
+		if ($level >= 0 && $level < 16)
+		{
+			$cmd = array('cmd' => CMD_SEARCH_SCWS_SET, 'arg1' => CMD_SCWS_SET_MULTI, 'arg2' => $level);
+			$this->execCommand($cmd);
+		}
+		return $this;
+	}
+
+	/**
+	 * 获取当前索引库的分词复合等级
+	 * @return int 返回当前库的分词复合等级
+	 * @see setScwsMulti
+	 * @since 1.4.7
+	 */
+	public function getScwsMulti()
+	{
+		$cmd = array('cmd' => CMD_SEARCH_SCWS_GET, 'arg1' => CMD_SCWS_GET_MULTI);
+		$res = $this->execCommand($cmd, CMD_OK_INFO);
+		return intval($res->buf);
+	}
+
 	/**
 	 * 开启索引命令提交缓冲区
 	 * 为优化网络性能, 有必要先将本地提交的 add/update/del 等索引变动指令缓存下来, 

sdk/php/lib/XSSearch.class.php

@@ -337,6 +337,26 @@ public function getFacets($field = null)
 		return isset($this->_facets[$field]) ? $this->_facets[$field] : array();
 	}
 
+	/**
+	 * 设置当前搜索语句的分词复合等级
+	 * 复合等级是 scws 分词粒度控制的一个重要参数, 是长词细分处理依据, 默认为 3, 值范围 0~15
+	 * 注意: 这个设置仅直对本次搜索有效, 仅对设置之后的 {@link setQuery} 起作用, 由于 query
+	 * 设计的方式问题, 目前无法支持搜索语句单字切分, 但您可以在模糊检索时设为 0 来关闭复合分词
+	 * @param int $level 要设置的分词复合等级
+	 * @return XSSearch 返回自身对象以支持串接操作
+	 * @since 1.4.7
+	 */
+	public function setScwsMulti($level)
+	{
+		$level = intval($level);
+		if ($level >= 0 && $level < 16)
+		{
+			$cmd = array('cmd' => CMD_SEARCH_SCWS_SET, 'arg1' => CMD_SCWS_SET_MULTI, 'arg2' => $level);
+			$this->execCommand($cmd);
+		}
+		return $this;
+	}
+
 	/**
 	 * 设置搜索结果的数量和偏移
 	 * 用于搜索结果分页, 每次调用 {@link search} 后会还原这2个变量到初始值
@@ -838,7 +858,7 @@ private function clearQuery()
 	 * @param float $scale 权重计算缩放比例, 默认为 1表示不缩放, 其它值范围 0.xx ~ 655.35
 	 * @return string 修正后的搜索语句
 	 */
-	private function addQueryString($query, $addOp = CMD_QUERY_OP_AND, $scale = 1)
+	public function addQueryString($query, $addOp = CMD_QUERY_OP_AND, $scale = 1)
 	{
 		$query = $this->preQueryString($query);
 		$bscale = ($scale > 0 && $scale != 1) ? pack('n', intval($scale * 100)) : '';

sdk/php/lib/xs_cmd.inc.php

@@ -1,5 +1,5 @@
 <?php
-/* Automatically generated at 2013/03/29 11:33 */
+/* Automatically generated at 2013/04/01 14:09 */
 define('CMD_NONE',	0);
 define('CMD_DEFAULT',	CMD_NONE);
 define('CMD_PROTOCOL',	20110707);
@@ -103,6 +103,7 @@
 define('CMD_SCWS_GET_RESULT',	2);
 define('CMD_SCWS_GET_TOPS',	3);
 define('CMD_SCWS_HAS_WORD',	4);
+define('CMD_SCWS_GET_MULTI',	5);
 define('CMD_SCWS_SET_IGNORE',	50);
 define('CMD_SCWS_SET_MULTI',	51);
 define('CMD_SCWS_SET_DUALITY',	52);

sdk/php/tests/bootstrap.php

@@ -4,6 +4,7 @@
 // $Id$
 $prefix = trim(file_get_contents(getenv('HOME') . '/.xs_installed'));
 shell_exec($prefix . '/bin/xs-ctl.sh restart');
+sleep(1);
 
 // global temp ini files
 $GLOBALS['fixIniData'] = array(

sdk/php/tests/lib/XSIndexTest.php

@@ -251,4 +251,60 @@ public function testCustomDict()
 		$this->assertEquals(1, $search->count('subject:测看'));
 		$this->assertEquals(0, $search->count('subject:看'));
 	}
+
+	private function countSubjectTerm($term)
+	{
+		$search = $this->object->xs->search->reopen(true)->setCharset('utf-8');
+		return $search->setQuery(null)->addQueryTerm('subject', $term)->count();
+	}
+
+	public function testScwsMulti()
+	{
+		// objects
+		$index = $this->object;
+		$doc = new XSDocument('utf-8');
+		$doc->pid = 7788;
+		$doc->subject = '管理制度';
+		$doc->message = '中华人民共和国';
+		// default scws
+		$this->assertEquals(3, $index->getScwsMulti());
+		$index->setScwsMulti(16);
+		$this->assertEquals(3, $index->getScwsMulti());
+		$index->setScwsMulti(-1);
+		$this->assertEquals(3, $index->getScwsMulti());
+		$index->update($doc);
+		$index->flushIndex();
+		sleep(2);
+		$this->assertEquals(1, $this->countSubjectTerm('管理制度'));
+		$this->assertEquals(1, $this->countSubjectTerm('管理'));
+		$this->assertEquals(0, $this->countSubjectTerm('管'));
+		$this->assertEquals(0, $this->countSubjectTerm('制'));
+		// multi = 0
+		$index->setScwsMulti(0);
+		$index->update($doc);
+		$index->flushIndex();
+		sleep(2);
+		$this->assertEquals(1, $this->countSubjectTerm('管理制度'));
+		$this->assertEquals(0, $this->countSubjectTerm('管理'));
+		$this->assertEquals(0, $this->countSubjectTerm('管'));
+		$this->assertEquals(0, $this->countSubjectTerm('制'));
+		// multi = 5
+		$index->setScwsMulti(5);
+		$index->update($doc);
+		$index->flushIndex();
+		sleep(2);
+		$this->assertEquals(1, $this->countSubjectTerm('管理制度'));
+		$this->assertEquals(1, $this->countSubjectTerm('管理'));
+		$this->assertEquals(1, $this->countSubjectTerm('管'));
+		$this->assertEquals(0, $this->countSubjectTerm('制'));
+		// multi = 15
+		$index->setScwsMulti(15);
+		$index->update($doc);
+		$index->flushIndex();
+		sleep(2);
+		$this->assertEquals(1, $this->countSubjectTerm('管理制度'));
+		$this->assertEquals(1, $this->countSubjectTerm('管理'));
+		$this->assertEquals(1, $this->countSubjectTerm('管'));
+		$this->assertEquals(1, $this->countSubjectTerm('制'));
+	}
 }

sdk/php/tests/lib/XSSearchTest.php

@@ -547,12 +547,12 @@ public function testSetDb()
 		$this->assertInstanceOf('XSException', $e2);
 		$this->assertEquals(CMD_ERR_XAPIAN, $e2->getCode());
 	}
-	
+
 	public function testCustomDict()
 	{
 		$index = self::$xs->index;
 		$search = self::$xs->search;
-		
+
 		// without custom dict
 		$index->setCustomDict('');
 		$query = $search->reopen(true)->getQuery('去测测看');
@@ -567,4 +567,17 @@ public function testCustomDict()
 		$query = $search->reopen(true)->getQuery('去测测看');
 		$this->assertEquals('Xapian::Query((去:(pos=1) AND (测测看:(pos=2) SYNONYM (测测:(pos=90) AND 测看:(pos=91)))))', $query);
 	}
+
+	public function testScwsMulti()
+	{
+		$search = self::$xs->search;
+		// default scws		
+		$this->assertEquals(array('管理制度', '管理', '制度'), $search->terms('管理制度'));
+		// multi = 0
+		$search->setScwsMulti(0);
+		$this->assertEquals(array('管理制度'), $search->terms('管理制度'));
+		// multi = 2
+		$search->setScwsMulti(2);
+		$this->assertEquals(array('管理制度', '管理', '理制', '制度'), $search->terms('管理制度'));
+	}
 }

sdk/php/util/Indexer.php

@@ -37,11 +37,12 @@
 // file & database
 $file = XSUtil::getOpt(null, 'file', true);
 $db = XSUtil::getOpt('d', 'db');
+$scws_multi = XSUtil::getOpt(null, 'scws-multi');
 
 // help message
 if (XSUtil::getOpt('h', 'help') !== null || !is_string($project)
-	|| (!$custom_dict && !$stop_rebuild && !$flush && !$flush_log 
-		&& !$info && !$clean && !$source && !$add_synonym && !$del_synonym))
+	|| (!$custom_dict && !$stop_rebuild && !$flush && !$flush_log
+	&& !$info && !$clean && !$source && !$add_synonym && !$del_synonym && !$scws_multi))
 {
 	$version = PACKAGE_NAME . '/' . PACKAGE_VERSION;
 	echo <<<EOF
@@ -79,6 +80,8 @@
     --del-synonym=<raw1[:synonym1[,raw2[:synonym2]]]...>
                  删除一个或多个同义词, 多个之间用半角逗号分隔, 原词和同义词之间用冒号分隔
                  省略同义词则表示删除该原词的所有同义词
+    --scws-multi[=level]
+                 查看或设置搜索语句的 scws 复合分词等级（值：0-15，默认为 3）
     --rebuild    使用平滑重建方式导入数据，必须与 --source 配合使用
     --stop-rebuild 强制中止没未完成的索引重建状态 (慎用)
     --clean      清空库内当前的索引数据
@@ -161,6 +164,14 @@
 	if ($db !== null)
 		$index->setDb($db);
 
+	// scws multi
+	if ($scws_multi !== null && $scws_multi !== true)
+	{
+		$index->setScwsMulti($scws_multi);
+		if (!empty($source))
+			$scws_multi = null;
+	}
+
 	// special actions
 	if ($info !== null)
 	{
@@ -218,6 +229,11 @@
 			}
 		}
 	}
+	else if ($scws_multi !== null)
+	{
+		$level = $index->getScwsMulti();
+		echo "当前索引库的 scws 复合分词等级为：$level\n";
+	}
 	else
 	{
 		// clean

sdk/php/util/Quest.php

@@ -13,14 +13,15 @@
 require_once dirname(__FILE__) . '/XSUtil.class.php';
 
 // check arguments
-XSUtil::parseOpt(array('p', 'q', 'c', 'd', 's', 'project', 'query', 'db', 'limit', 'charset', 'sort', 'add-weight'));
+XSUtil::parseOpt(array('p', 'q', 'c', 'd', 's', 'project', 'query', 'db', 'limit', 'charset', 'sort', 'add-weight', 'scws-multi'));
 $project = XSUtil::getOpt('p', 'project', true);
 $query = XSUtil::getOpt('q', 'query', true);
 $hot = XSUtil::getOpt(null, 'hot');
 $synonyms = XSUtil::getOpt(null, 'list-synonyms');
 $terms = XSUtil::getOpt(null, 'terms');
 $weights = XSUtil::getOpt(null, 'add-weight');
-$info = XSUtil::getopt(null, 'info');
+$info = XSUtil::getOpt(null, 'info');
+$scws_multi = XSUtil::getOpt(null, 'scws-multi');
 
 // magick output charset
 $charset = XSUtil::getOpt('c', 'charset');
@@ -55,6 +56,8 @@
     -s <field1[,field2[,...]] 指定排序字段，在字段前加上 ~ 符号表示逆序
     --fuzzy      将搜索默认设为模糊搜索
     --synonym    开启自动同义词搜索功能
+    --scws-multi=<level>
+                 查看或设置搜索语句的 scws 复合分词等级（值：0-15，默认为 3）
     --add-weight=<[field1:]word1[:weight1][,[field2:]word2[:weight2]]>
                  添加搜索权重词汇，词与次数之间用半角冒号分隔
     --hot[=total|last|cur]
@@ -114,6 +117,8 @@
 			$search->addDb(trim($dbs[$i]));
 		}
 	}
+	if ($scws_multi !== null)
+		$search->setScwsMulti($scws_multi);
 
 	if ($hot !== null)
 	{

src/conn.c

@@ -225,7 +225,8 @@ int conn_quit(XS_CONN *conn, int res)
 	close(CONN_FD());
 
 	debug_free(conn);
-	conn_server.num_burst--;
+	if (conn_server.num_burst > 0)
+		conn_server.num_burst--;
 	return CMD_RES_QUIT;
 }
 

src/global.h

@@ -16,6 +16,7 @@
 #define	SEARCH_LOG_DB		"log_db"
 #define	DEFAULT_BACKLOG		63			// default backlog for listen()
 #define	MAX_EXPAND_LEN		15
+#define DEFAULT_SCWS_MULTI	3			// default scws multi level
 
 #ifdef HAVE_MM
 

src/import.cc

@@ -93,7 +93,7 @@ static scws_t load_user_scws(int multi, char *dbpath)
 	scws_set_dict(s, SCWS_ETCDIR "/dict.utf8.xdb", SCWS_XDICT_MEM);
 	scws_add_dict(s, SCWS_ETCDIR "/" CUSTOM_DICT_FILE, SCWS_XDICT_TXT);
 	scws_add_dict(s, ptr, SCWS_XDICT_TXT);
-	scws_set_multi(s, multi << 12);
+	scws_set_multi(s, (multi << 12) & SCWS_MULTI_MASK);
 	return s;
 }
 

src/import.h

@@ -10,7 +10,6 @@
 #define	DEFAULT_STEMMER				"english"	// default stemmer
 #define	DEFAULT_COMMIT_NUMBER		10000		// document numbers
 #define	DEFAULT_COMMIT_SIZE			256			// MB
-#define DEFAULT_SCWS_MULTI			3			// default scws multi level
 
 #define	DEFAULT_ARCHIVE_THRESHOLD	100000		// default threshold value to archive
 

src/indexd.c

@@ -218,7 +218,9 @@ static void db_import_call(XS_DB *db, XS_USER *user)
 	// fork child process to run the import
 	if ((pid = fork()) == 0)
 	{
-		EXTERNAL_CALL(xs_import, "xs-import", "-Q", dbpath, sndfile);
+		char arg[16];
+		sprintf(arg, "-m%d", db->scws_multi);
+		EXTERNAL_CALL(xs_import, "xs-import", "-Q", arg, dbpath, sndfile);
 	}
 	else if (pid > 0)
 	{
@@ -1222,6 +1224,21 @@ static int index_zcmd_exec(XS_CONN *conn)
 		case CMD_DOC_INDEX:
 			rc = CMD_RES_CONT | CMD_RES_SAVE;
 			break;
+			// scws multi
+		case CMD_SEARCH_SCWS_SET:
+			if (cmd->arg1 == CMD_SCWS_SET_MULTI && get_conn_wdb(conn) != NULL)
+				conn->wdb->scws_multi = (short) cmd->arg2;
+			break;
+		case CMD_SEARCH_SCWS_GET:
+			if (cmd->arg1 != CMD_SCWS_GET_MULTI || get_conn_wdb(conn) == NULL)
+				rc = CMD_RES_UNIMP;
+			else
+			{
+				char buf[8];
+				sprintf(buf, "%d", conn->wdb->scws_multi);
+				rc = CONN_RES_OK2(INFO, buf);
+			}
+			break;
 			// others, passed to next handler
 		default:
 			rc = CMD_RES_NEXT;

src/logging.cc

@@ -91,7 +91,7 @@ static scws_t load_user_scws(int multi, char *dbpath)
 	scws_set_dict(s, SCWS_ETCDIR "/dict.utf8.xdb", SCWS_XDICT_MEM);
 	scws_add_dict(s, SCWS_ETCDIR "/" CUSTOM_DICT_FILE, SCWS_XDICT_TXT);
 	scws_add_dict(s, ptr, SCWS_XDICT_TXT);
-	scws_set_multi(s, multi << 12);
+	scws_set_multi(s, (multi << 12) & SCWS_MULTI_MASK);
 	return s;
 }
 

src/task.cc

@@ -645,9 +645,18 @@ static int zcmd_task_default(XS_CONN *conn)
 		}
 			break;
 		case CMD_SEARCH_SCWS_SET:
+			if (cmd->arg1 == CMD_SCWS_SET_MULTI)
+			{
+				scws_t scws = (scws_t) zarg->qp->get_scws();
+				if (scws != NULL)
+				{
+					scws_set_multi(scws, (cmd->arg2 << 12) & SCWS_MULTI_MASK);
+					log_debug_conn("change scws multi level (MODE:%d)", cmd->arg2);
+				}
+			}
+			break;
 		case CMD_SEARCH_SCWS_GET:
 			rc = CMD_RES_UNIMP;
-			break;
 		default:
 			rc = CMD_RES_NEXT;
 			break; // passed to next
@@ -1887,7 +1896,7 @@ static int zcmd_scws_set(XS_CONN *conn)
 	}
 	else if (cmd->arg1 == CMD_SCWS_SET_MULTI)
 	{
-		scws_set_multi(scws, (cmd->arg2 << 12));
+		scws_set_multi(scws, (cmd->arg2 << 12) & SCWS_MULTI_MASK);
 	}
 	else if (cmd->arg1 == CMD_SCWS_SET_IGNORE)
 	{
@@ -1909,6 +1918,12 @@ static int zcmd_scws_get(XS_CONN *conn)
 	{
 		return CONN_RES_OK2(INFO, SCWS_VERSION);
 	}
+	else if (cmd->arg1 == CMD_SCWS_GET_MULTI)
+	{
+		char buf[8];
+		sprintf(buf, "%d", scws->mode & SCWS_MULTI_MASK);
+		return CONN_RES_OK2(INFO, buf);
+	}
 	else if (cmd->arg1 == CMD_SCWS_HAS_WORD)
 	{
 		int count;