XSIndex
包 | XS |
---|---|
继承关系 | class XSIndex » XSServer » XSComponent |
版本 | 1.0.0 |
源代码 | sdk/php/lib/XSIndex.class.php |
XS 索引管理
添加/删除/修改索引数据
Public 属性
名称 | 类型 | 描述 | 定义于 |
---|---|---|---|
project | string | 获取当前项目名称 | XSServer |
respond | XSCommand | 从服务器读取响应指令 | XSServer |
socket | mixed | 获取连接资源描述符 | XSServer |
xs | XS | 服务端关联的 XS 对象 | XSServer |
Public 方法
名称 | 描述 | 定义于 |
---|---|---|
__construct() | 构造函数, 打开连接 | XSServer |
__destruct() | 析构函数 | XSIndex |
__get() | 魔术方法 __get | XSComponent |
__isset() | 魔术方法 __isset | XSComponent |
__set() | 魔术方法 __set | XSComponent |
__unset() | 魔术方法 __unset | XSComponent |
add() | 添加文档到索引中 | XSIndex |
addExdata() | 批量提交索引命令封包数据 | XSIndex |
addSynonym() | 添加同义词 | XSIndex |
beginRebuild() | 开始重建索引 | XSIndex |
clean() | 完全清空索引数据 | XSIndex |
close() | 关闭索引服务端连接 | XSIndex |
closeBuffer() | 提交所有指令并关闭缓冲区 | XSIndex |
del() | 删除索引中的数据 | XSIndex |
delSynonym() | 删除某个同义词 | XSIndex |
endRebuild() | 完成并关闭重建索引 | XSIndex |
execCommand() | 执行服务端指令并获取返回值 | XSServer |
flushIndex() | 强制刷新服务端的当前库的索引缓存 | XSIndex |
flushLogging() | 强制刷新服务端当前项目的搜索日志 | XSIndex |
getProject() | 获取当前项目名称 | XSServer |
getRespond() | 从服务器读取响应指令 | XSServer |
getSocket() | 获取连接资源描述符 | XSServer |
hasRespond() | 判断服务端是否有可读数据 | XSServer |
open() | 打开服务端连接 | XSServer |
openBuffer() | 开启索引命令提交缓冲区 | XSIndex |
reopen() | 重新打开连接 | XSServer |
sendCommand() | 往服务器直接发送指令 (无缓存) | XSServer |
setDb() | 更改存放索引数据的目录 | XSIndex |
setProject() | 设置当前项目 | XSServer |
setTimeout() | 设置服务端超时秒数 | XSServer |
update() | 更新索引文档 | XSIndex |
方法明细
__destruct()
方法
public void __destruct()
|
源码: sdk/php/lib/XSIndex.class.php#L437 (显示)
public function __destruct()
{
if ($this->_rebuild === true)
{
try
{
$this->endRebuild();
}
catch (Exception $e)
{
}
}
parent::__destruct();
}
析构函数 在此自动关闭开启的 rebuild
add()
方法
public XSIndex add(XSDocument $doc)
| ||
$doc | XSDocument | |
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L45 (显示)
public function add(XSDocument $doc)
{
return $this->update($doc, true);
}
添加文档到索引中 特别要注意的是: 系统不会自动检测主键是否冲突, 即便已存在相同主键也会添加进去
参见
addExdata()
方法
public XSIndex addExdata(string $data, bool $check_file=true)
| ||
$data | string | 要提交的命令封包数据, 或存储命令封包的文件路径, 编码必须已经是 UTF-8 |
$check_file | bool | 是否检测参数为文件的情况 |
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L233 (显示)
public function addExdata($data, $check_file = true)
{
if (strlen($data) < 255 && $check_file && file_exists($data) && ($data = file_get_contents($data)) === false)
throw new XSException('Failed to read exdata from file');
// try to check allowed (BUG: check the first cmd only):
// CMD_IMPORT_HEADER, CMD_INDEX_REQUEST, CMD_INDEX_REMOVE, CMD_INDEX_EXDATA
$first = ord(substr($data, 0, 1));
if ($first != CMD_IMPORT_HEADER && $first != CMD_INDEX_REQUEST
&& $first != CMD_INDEX_SYNONYMS
&& $first != CMD_INDEX_REMOVE && $first != CMD_INDEX_EXDATA)
{
throw new XSException('Invalid start command of exdata (CMD:' . $first . ')');
}
// create cmd & execute it
$cmd = array('cmd' => CMD_INDEX_EXDATA, 'buf' => $data);
$this->execCommand($cmd, CMD_OK_RQST_FINISHED);
return $this;
}
批量提交索引命令封包数据 把多个命令封包内容连续保存为文件或变量, 然后一次性提交以减少网络开销提升性能
addSynonym()
方法
(自版本 v1.3.0 起可用)
public XSIndex addSynonym(string $raw, string $synonym)
| ||
$raw | string | 需要同义的原词, 英文词汇支持用空格分开多个单词并强制被转换为小写 |
$synonym | string | 同义词条, 最小语素, 勿带空格等分隔符 |
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L262 (显示)
public function addSynonym($raw, $synonym)
{
$raw = strval($raw);
$synonym = strval($synonym);
if ($raw !== '' && $synonym !== '')
{
$cmd = new XSCommand(CMD_INDEX_SYNONYMS, CMD_INDEX_SYNONYMS_ADD, 0, $raw, $synonym);
if ($this->_bufSize > 0)
$this->appendBuffer(strval($cmd));
else
$this->execCommand($cmd, CMD_OK_RQST_FINISHED);
}
return $this;
}
添加同义词
beginRebuild()
方法
public XSIndex beginRebuild()
| ||
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L336 (显示)
public function beginRebuild()
{
$this->execCommand(array('cmd' => CMD_INDEX_REBUILD, 'arg1' => 0), CMD_OK_DB_REBUILD);
$this->_rebuild = true;
return $this;
}
开始重建索引 此后所有的索引更新指令将写到临时库, 而不是当前搜索库, 重建完成后调用 endRebuild 实现平滑重建索引, 重建过程仍可搜索旧的索引库, 如直接用 clean 清空数据, 则会导致重建过程搜索到不全的数据
参见
clean()
方法
public XSIndex clean()
| ||
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L32 (显示)
public function clean()
{
$this->execCommand(CMD_INDEX_CLEAN_DB, CMD_OK_DB_CLEAN);
return $this;
}
完全清空索引数据 如果当前数据库处于重建过程中将禁止清空
参见
close()
方法
public void close($ioerr=false)
| ||
$ioerr |
源码: sdk/php/lib/XSIndex.class.php#L412 (显示)
public function close($ioerr = false)
{
$this->closeBuffer();
parent::close($ioerr);
}
关闭索引服务端连接
closeBuffer()
方法
public XSIndex closeBuffer()
| ||
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L323 (显示)
public function closeBuffer()
{
return $this->openBuffer(0);
}
提交所有指令并关闭缓冲区 若未曾打开缓冲区, 调用本方法是无意义的
参见
del()
方法
public XSIndex del(mixed $term, string $field=NULL)
| ||
$term | mixed | 单个主键或指定字段的索引词, 或多个组成的数组, 编码与 xs 默认字符集一致 |
$field | string | 索引词所属的字段名称, 默认不指定则为主键字段 (类型为ID) |
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L198 (显示)
public function del($term, $field = null)
{
// get field
$field = $field === null ? $this->xs->getFieldId() : $this->xs->getField($field);
// get commands
$cmds = array();
$terms = is_array($term) ? array_unique($term) : array($term);
$terms = XS::convert($terms, 'UTF-8', $this->xs->getDefaultCharset());
foreach ($terms as $term)
{
$cmds[] = new XSCommand(CMD_INDEX_REMOVE, 0, $field->vno, strtolower($term));
}
// combine multi commands into exdata
if ($this->_bufSize > 0)
$this->appendBuffer(implode('', $cmds));
else if (count($cmds) == 1)
$this->execCommand($cmds[0], CMD_OK_RQST_FINISHED);
else
{
$cmd = array('cmd' => CMD_INDEX_EXDATA, 'buf' => implode('', $cmds));
$this->execCommand($cmd, CMD_OK_RQST_FINISHED);
}
return $this;
}
删除索引中的数据
$index->del('123'); // 删除主键为 123 的记录 $index->del(array('123', '789', '456')); // 删除主键为 123, 789, 456 的记录 $index->del('abc', 'field'); // 删除字段 field 上带有索引词 abc 的所有记录 $index->del(array('abc', 'def'), 'field'); // 删除字段 field 上带有索引词 abc 或 def 的所有记录
delSynonym()
方法
(自版本 v1.3.0 起可用)
public XSIndex delSynonym(string $raw, string $synonym=NULL)
| ||
$raw | string | 需要同义的原词, 英文词汇支持用空格分开多个单词并强制被转换为小写 |
$synonym | string | 要删除的同义词条, 默认 null 表示删除原词下的所有同义词 |
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L285 (显示)
public function delSynonym($raw, $synonym = null)
{
$raw = strval($raw);
$synonym = $synonym === null ? '' : strval($synonym);
if ($raw !== '')
{
$cmd = new XSCommand(CMD_INDEX_SYNONYMS, CMD_INDEX_SYNONYMS_DEL, 0, $raw, $synonym);
if ($this->_bufSize > 0)
$this->appendBuffer(strval($cmd));
else
$this->execCommand($cmd, CMD_OK_RQST_FINISHED);
}
return $this;
}
删除某个同义词
endRebuild()
方法
public XSIndex endRebuild()
| ||
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L349 (显示)
public function endRebuild()
{
if ($this->_rebuild === true)
{
$this->_rebuild = false;
$this->execCommand(array('cmd' => CMD_INDEX_REBUILD, 'arg1' => 1), CMD_OK_DB_REBUILD);
}
return $this;
}
完成并关闭重建索引 重建完成后调用, 用重建好的索引数据代替旧的索引数据
参见
flushIndex()
方法
public bool flushIndex()
| ||
{return} | bool | 刷新成功返回 true, 失败则返回 false |
源码: sdk/php/lib/XSIndex.class.php#L394 (显示)
public function flushIndex()
{
try
{
$this->execCommand(CMD_INDEX_COMMIT, CMD_OK_DB_COMMITED);
}
catch (XSException $e)
{
if ($e->getCode() === CMD_ERR_BUSY || $e->getCode() === CMD_ERR_RUNNING)
return false;
throw $e;
}
return true;
}
强制刷新服务端的当前库的索引缓存
flushLogging()
方法
public bool flushLogging()
| ||
{return} | bool | 刷新成功返回 true, 失败则返回 false |
源码: sdk/php/lib/XSIndex.class.php#L375 (显示)
public function flushLogging()
{
try
{
$this->execCommand(CMD_FLUSH_LOGGING, CMD_OK_LOG_FLUSHED);
}
catch (XSException $e)
{
if ($e->getCode() === CMD_ERR_BUSY)
return false;
throw $e;
}
return true;
}
强制刷新服务端当前项目的搜索日志
openBuffer()
方法
public XSIndex openBuffer(int $size=4)
| ||
$size | int | 缓冲区大小, 单位: MB 默认为 4MB |
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L308 (显示)
public function openBuffer($size = 4)
{
if ($this->_buf !== '')
$this->addExdata($this->_buf, false);
$this->_bufSize = intval($size) << 20;
$this->_buf = '';
return $this;
}
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, 当总大小达到参数指定的 size 时或调用 closeBuffer 时再真正提交到服务器 注意: 此举常用于需要大批量更新索引时, 此外重复调用本函数是无必要的
setDb()
方法
public XSIndex setDb(string $name)
| ||
$name | string | 数据库名称 |
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L365 (显示)
public function setDb($name)
{
$this->execCommand(array('cmd' => CMD_INDEX_SET_DB, 'buf' => $name), CMD_OK_DB_CHANGED);
return $this;
}
更改存放索引数据的目录 默认索引数据保存到服务器上的 db 目录, 通过此方法修改数据目录名
update()
方法
public XSIndex update(XSDocument $doc, bool $add=false)
| ||
$doc | XSDocument | |
$add | bool | 是否为新增文档, 已有数据中不存在同一主键的其它数据 |
{return} | XSIndex | 返回自身对象以支持串接操作 |
源码: sdk/php/lib/XSIndex.class.php#L58 (显示)
public function update(XSDocument $doc, $add = false)
{
// before submit
if ($doc->beforeSubmit($this) === false)
return $this;
// check primary key of document
$fid = $this->xs->getFieldId();
$key = $doc->f($fid);
if ($key === null || $key === '')
throw new XSException('Missing value of primarky key (FIELD:' . $fid . ')');
// request cmd
$cmd = new XSCommand(CMD_INDEX_REQUEST, CMD_INDEX_REQUEST_ADD);
if ($add !== true)
{
$cmd->arg1 = CMD_INDEX_REQUEST_UPDATE;
$cmd->arg2 = $fid->vno;
$cmd->buf = $key;
}
$cmds = array($cmd);
// document cmds
foreach ($this->xs->getAllFields() as $field) /* @var $field XSFieldMeta */
{
// value
if (($value = $doc->f($field)) !== null)
{
$varg = $field->isNumeric() ? CMD_VALUE_FLAG_NUMERIC : 0;
$value = $field->val($value);
if (!$field->hasCustomTokenizer())
{
// internal tokenizer
$wdf = $field->weight | ($field->withPos() ? CMD_INDEX_FLAG_WITHPOS : 0);
if ($field->hasIndexMixed())
$cmds[] = new XSCommand(CMD_DOC_INDEX, $wdf, XSFieldScheme::MIXED_VNO, $value);
if ($field->hasIndexSelf())
{
$wdf |= $field->isNumeric() ? 0 : CMD_INDEX_FLAG_SAVEVALUE;
$cmds[] = new XSCommand(CMD_DOC_INDEX, $wdf, $field->vno, $value);
}
// add value
if (!$field->hasIndexSelf() || $field->isNumeric())
$cmds[] = new XSCommand(CMD_DOC_VALUE, $varg, $field->vno, $value);
}
else
{
// add index
if ($field->hasIndex())
{
$terms = $field->getCustomTokenizer()->getTokens($value, $doc);
// self: [bool term, NOT weight, NOT stem, NOT pos]
if ($field->hasIndexSelf())
{
foreach ($terms as $term)
{
$term = strtolower($term);
$cmds[] = new XSCommand(CMD_DOC_TERM, 1, $field->vno, $term);
}
}
// mixed: [use default tokenizer]
if ($field->hasIndexMixed())
{
$mtext = implode(' ', $terms);
$cmds[] = new XSCommand(CMD_DOC_INDEX, $field->weight, XSFieldScheme::MIXED_VNO, $mtext);
}
}
// add value
$cmds[] = new XSCommand(CMD_DOC_VALUE, $varg, $field->vno, $value);
}
}
// process add terms
if (($terms = $doc->getAddTerms($field)) !== null)
{
// ignore weight for bool index
$wdf1 = $field->isBoolIndex() ? 0 : CMD_INDEX_FLAG_CHECKSTEM;
foreach ($terms as $term => $wdf)
{
$term = strtolower($term);
$wdf2 = $field->isBoolIndex() ? 1 : $wdf * $field->weight;
while ($wdf2 > XSFieldMeta::MAX_WDF)
{
$cmds[] = new XSCommand(CMD_DOC_TERM, $wdf1 | XSFieldMeta::MAX_WDF, $field->vno, $term);
$wdf2 -= XSFieldMeta::MAX_WDF;
}
$cmds[] = new XSCommand(CMD_DOC_TERM, $wdf1 | $wdf2, $field->vno, $term);
}
}
// process add text
if (($text = $doc->getAddIndex($field)) !== null)
{
if (!$field->hasCustomTokenizer())
{
$wdf = $field->weight | ($field->withPos() ? CMD_INDEX_FLAG_WITHPOS : 0);
$cmds[] = new XSCommand(CMD_DOC_INDEX, $wdf, $field->vno, $text);
}
else
{
// NOT pos
$wdf = $field->isBoolIndex() ? 1 : ($field->weight | CMD_INDEX_FLAG_CHECKSTEM);
$terms = $field->getCustomTokenizer()->getTokens($text, $doc);
foreach ($terms as $term)
{
$term = strtolower($term);
$cmds[] = new XSCommand(CMD_DOC_TERM, $wdf, $field->vno, $term);
}
}
}
}
// submit cmd
$cmds[] = new XSCommand(CMD_INDEX_SUBMIT);
// execute cmd
if ($this->_bufSize > 0)
$this->appendBuffer(implode('', $cmds));
else
{
for ($i = 0; $i < count($cmds) - 1; $i++)
$this->execCommand($cmds[$i]);
$this->execCommand($cmds[$i], CMD_OK_RQST_FINISHED);
}
// after submit
$doc->afterSubmit($this);
return $this;
}
更新索引文档 该方法相当于先根据主键删除已存在的旧文档, 然后添加该文档 如果你能明确认定是新文档, 则建议使用 add