微信公众号最佳实践 （ 9.7）智能问答，关键词回复，中文分词
智能问答
前面我们使用的都是
基于固定查询指令的回复，这样好处是内容格式统一，方便软件开发人员编写程序做出分析，回复精准的内容给用户- 但在生活中，人们问的内容很随意，甚至千差万别，这时，回复内容想要和用户的问题相匹配，就需要更智能的程序了
关键词回复
我们需先定一个数组，数组中键为关键词，值为对应的回复，当用户输入的文字能匹配到某个关键词时，则回复该关键词对应的内容，我们定义“电话”，“地址”，”微信”,三个关键词，将其作为数组中的键，然后定义相应的值，当用户发送的内容中包含相应的关键词名称时，则回复对应的内容。
代码如下所示：
private function receiveText($object)
{
$content = trim($object->Content);
$keywords = array(
"电话" => "0755-87654321",
"地址" => "广东省深圳市南山区高新南一道飞亚达科技大厦(518057)",
"A型号手机" => "A产品报价 3000元，基本参数为：****，",
"B型号手机" => "A产品报价 4000元，基本参数为：****，",
"C型号手机" => "A产品报价 2000元，基本参数为：****，",
"微信" => "xxxxx"
);
foreach ($keywords as $key=>$value){
$contain = strchr($content, $key);
if (!empty($contain)){
$reply = $value;
break;
}
}
$result = $this->transmitText($object, $reply);
return $result;
}中文分词
除了关键词回复之外，还可以使用中文分词来帮助我们实现智能回复。
中文分词指的是将一个汉字序列切分成一个个单独词。
分词就是将连续的字序列按照一定的规范重新组合成词序列的过程。
我们知道在英文的行文中，单词之间是以空格作为自然分隔符的，而中文只有字、句、和段，能通过简单的分界符来简单划分，唯独词没有一个形式上的分界符，虽然英文也同样存在短语的划分问题，不过在词这一层上，中文要比英文复杂的多，困难的多。
中文分词就是要解决句子中词的划分问题，然后可以提取其中的关键词语进行搜索。
打个比方：“深圳天气怎么样”，这句话可分词为“深圳”，“天气”，“怎么样”三部分，因为它们代表不同词语类别的意思。
新浪云SEA提供了中文分词服务seaSegment,上述句子进行中文分词的代码如下所示
$str="深圳天气怎么样";
$seg = new SaeSegment();
$ret = $seg->segment($str, 1);
print_r($ret);
?>代码运行后，结果如下所示：
Array (
[0] => Array (
[word] => 深圳
[word_tag] => 102
[index] => 0 )
[1] => Array (
[word] => 天气
[word_tag] => 95
[index] => 1 )
[2] => Array (
[word] => 怎么样
[word_tag] => 40
[index] => 2
)
)上述结果中，
- word表示分词
- word_tag表示词性（102代表地点名词，95代表名词）
- index表示词在句子中的位置索引
我们可以把分词功能写成函数，并且将名词作为分类标准，地点作为关键词组合成数组。
相应代码如下：
function sinasegment($str)
{
$seg = new SaeSegment();
$ret = $seg->segment($str, 1);
if ($ret === false){
return;
}
$category = "";
$keyword = "";
foreach ($ret as $key => $value) {
if ($value["word_tag"] == 95){
$category = $value["word"];
}
if ($value["word_tag"] == 102){
$keyword = $value["word"];
}
}
if (!empty($category) && !empty($keyword)){
return array('category'=>$category, 'keyword'=>$keyword);
}else{
return;
}
}
?>
根据获取的功能类别及关键字，我们可以做出根据地点来查询其相应的自动回复，比如天气预报和空气质量，在前面的章节中，我们提到了将语音识别结果并入文本消息结果的查询，基于这些，我们可以开发出非常智能的**语音识别搜索**及**文本搜索**功能.
相应代码如下：
private function receiveVoice($object)
{
$funcFlag = 0;
if (isset($object->Recognition) && !empty($object->Recognition)){
$content = strval($object->Recognition);
include("segment.php");
$result = sinasegment($content);
if (is_array($result)){
switch ($result['category'])
{
case "天气":
$url = "http://apix.sinaapp.com/weather/?appkey=trialuser&city=".urlencode($result['keyword']);
$output = file_get_contents($url);
$replyContent = json_decode($output, true);
break;
default:
$replyContent = "还不支持这一功能：".$result['category'];
break;
}
}else{
$replyContent = "不能理解你的内容：".$content;
}
}else{
$replyContent = "未开启语音识别功能";
}
if (is_array($replyContent)){
$resultStr = $this->transmitNews($object, $replyContent);
}else{
$resultStr = $this->transmitText($object, $replyContent);
}
return $resultStr;
}segment.php
function sinasegment($str)
{
$seg = new SaeSegment();
$ret = $seg->segment($str, 1);
if ($ret === false){
return;
}
$category = "";
$keyword = "";
foreach ($ret as $key => $value) {
if ($value["word_tag"] == 95){
$category = $value["word"];
}
if ($value["word_tag"] == 102){
$keyword = $value["word"];
}
}
if (!empty($category) && !empty($keyword)){
return array('category'=>$category, 'keyword'=>$keyword);
}else{
return;
}
}
?>
关键词index.php运行图片：
关键词回复 index.php 整体代码如下:
/*
CopyRight 2018 All Rights Reserved
*/
define("TOKEN", "weixin");
$wechatObj = new wechatCallbackapiTest();
if (!isset($_GET['echostr'])) {
$wechatObj->responseMsg();
}else{
$wechatObj->valid();
}
class wechatCallbackapiTest
{
public function valid()
{
$echoStr = $_GET["echostr"];
if($this->checkSignature()){
echo $echoStr;
exit;
}
}
private function checkSignature()
{
$signature = $_GET["signature"];
$timestamp = $_GET["timestamp"];
$nonce = $_GET["nonce"];
$token = TOKEN;
$tmpArr = array($token, $timestamp, $nonce);
sort($tmpArr, SORT_STRING);
$tmpStr = implode( $tmpArr );
$tmpStr = sha1( $tmpStr );
if( $tmpStr == $signature ){
return true;
}else{
return false;
}
}
public function responseMsg()
{
$postStr = $GLOBALS["HTTP_RAW_POST_DATA"];
if (!empty($postStr)){
$this->logger("R ".$postStr);
$postObj = simplexml_load_string($postStr, 'SimpleXMLElement', LIBXML_NOCDATA);
$RX_TYPE = trim($postObj->MsgType);
switch ($RX_TYPE)
{
case "event":
$result = $this->receiveEvent($postObj);
break;
case "text":
$result = $this->receiveText($postObj);
break;
default:
$result = "other msg type: ".$RX_TYPE;
break;
}
$this->logger("T ".$result);
echo $result;
}else {
echo "";
exit;
}
}
private function receiveEvent($object)
{
$content = "";
switch ($object->Event)
{
case "subscribe":
$content = "欢迎关注 德强1012 ";
$content .= isset($object->EventKey)?("\n来自二维码场景 ".$object->EventKey):"";
break;
case "unsubscribe":
$content = "取消关注";
break;
}
$result = $this->transmitText($object, $content);
return $result;
}
private function receiveText($object)
{
$content = trim($object->Content);
$keywords = array(
"电话" => "0755-87654321",
"地址" => "广东省深圳市南山区高新南一道飞亚达科技大厦(518057)",
"A型号手机" => "A产品报价 3000元，基本参数为：****，",
"B型号手机" => "A产品报价 4000元，基本参数为：****，",
"C型号手机" => "A产品报价 2000元，基本参数为：****，",
"微信" => "xxxxx"
);
foreach ($keywords as $key=>$value){
$contain = strchr($content, $key);
if (!empty($contain)){
$reply = $value;
break;
}
}
$result = $this->transmitText($object, $reply);
return $result;
}
private function transmitText($object, $content)
{
$textTpl = "
%s
";
$result = sprintf($textTpl, $object->FromUserName, $object->ToUserName, time(), $content);
return $result;
}
private function transmitNews($object, $arr_item)
{
if(!is_array($arr_item))
return;
$itemTpl = " -
<![CDATA[%s]]>
";
$item_str = "";
foreach ($arr_item as $item)
$item_str .= sprintf($itemTpl, $item['Title'], $item['Description'], $item['PicUrl'], $item['Url']);
$newsTpl = "
%s
%s
$item_str
";
$result = sprintf($newsTpl, $object->FromUserName, $object->ToUserName, time(), count($arr_item));
return $result;
}
private function logger($log_content)
{
}
}
?>中文分词index.php运行图片
中文分词index.php整体代码如下：
说明：方倍工作室接口100，目前免费暂无
他们解释原因如下：
- 接口100免费提供这些年，受到很多用户欢迎，每天访问量非常大，但由于极个别用户使用机器程序抓取数据以及故意使用DDOS恶意攻击，导致服务器不堪重负，维护费用成倍增长。我们决定关停这些接口，有需要的用户，可以联系QQ 1354386063，购买我们的接口源码，或者自行使用第三方收费接口。网页地址
http://www.cnblogs.com/txw1958/p/weixin-api100.html
语音功能公众号显示开通了，不知道为啥没有用，检测中…..
define("TOKEN", "weixin");
$wechatObj = new wechatCallbackapiTest();
if (!isset($_GET['echostr'])) {
$wechatObj->responseMsg();
}else{
$wechatObj->valid();
}
class wechatCallbackapiTest
{
public function valid()
{
$echoStr = $_GET["echostr"];
if($this->checkSignature()){
echo $echoStr;
exit;
}
}
private function checkSignature()
{
$signature = $_GET["signature"];
$timestamp = $_GET["timestamp"];
$nonce = $_GET["nonce"];
$token = TOKEN;
$tmpArr = array($token, $timestamp, $nonce);
sort($tmpArr);
$tmpStr = implode( $tmpArr );
$tmpStr = sha1( $tmpStr );
if( $tmpStr == $signature ){
return true;
}else{
return false;
}
}
public function responseMsg()
{
$postStr = $GLOBALS["HTTP_RAW_POST_DATA"];
if (!empty($postStr)){
$postObj = simplexml_load_string($postStr, 'SimpleXMLElement', LIBXML_NOCDATA);
$RX_TYPE = trim($postObj->MsgType);
switch ($RX_TYPE)
{
case "text":
$resultStr = $this->receiveText($postObj);
break;
case "event":
$resultStr = $this->receiveEvent($postObj);
break;
case "voice":
$resultStr = $this->receiveVoice($postObj);
break;
default:
$resultStr = "";
break;
}
echo $resultStr;
}else {
echo "";
exit;
}
}
private function receiveText($object)
{
if (isset($object->Recognition)){
$content = strval($object->Recognition);
$mediaid = trim($object->MediaID);
}else{
$content = trim($object->Content);
}
if (isset($_SERVER['HTTP_APPNAME'])){
include("segment.php");
$result = sinasegment($content);
if (is_array($result)){
switch ($result['category'])
{
case "天气":
//$url = "http://api100.duapp.com/weather/?appkey=book97&city=".urlencode($result['keyword']);
//$output = file_get_contents($url);
//$replyContent = json_decode($output, true);
//$replyContent = $result['keyword'];
$city = $result['keyword'];
include("weather2.php");
$content = getWeatherInfo($city);
$result = $this->transmitNews($object, $content);
return $result;
break;
case "空气":
//$url = "http://api100.duapp.com/airquality/?appkey=book97&city=".urlencode($result['keyword']);
//$output = file_get_contents($url);
//$replyContent = json_decode($output, true);
$keyword = $result['keyword'];
include("airquality.php");
$content = getAirQualityChina($keyword);
if(is_array($content)){
$result = $this->transmitNews($object, $content);
}else{
$result = $this->transmitText($object, $content);
}
return $result;
default:
$replyContent = "还不支持这一功能：".$result['category'];
break;
}
}else{
$replyContent = "不能理解你的内容：".$content;
}
}else{
$replyContent = "只能运行在SAE环境！";
}
if (is_array($replyContent)){
$resultStr = $this->transmitNews($object, $replyContent);
}else{
$resultStr = $this->transmitText($object, $replyContent);
}
return $resultStr;
}
private function receiveEvent($object)
{
$replyContent = "";
switch ($object->Event)
{
case "subscribe":
$replyContent = "欢迎关注 德强1012";
case "unsubscribe":
break;
default:
break;
}
if (is_array($replyContent)){
$resultStr = $this->transmitNews($object, $replyContent);
}else{
$resultStr = $this->transmitText($object, $replyContent);
}
return $resultStr;
}
private function receiveVoice($object)
{
$funcFlag = 0;
if (isset($object->Recognition) && !empty($object->Recognition)){
$content = strval($object->Recognition);
include("segment.php");
$result = sinasegment($content);
if (is_array($result)){
switch ($result['category'])
{
case "天气":
// $url = "http://apix.sinaapp.com/weather/?appkey=trialuser&city=".urlencode($result['keyword']);
//$output = file_get_contents($url);
//$replyContent = json_decode($output, true);
$city = $result['keyword'];
include("weather2.php");
$content = getWeatherInfo($city);
$result = $this->transmitNews($object, $content);
return $result;
//$replyContent = $result['keyword'];
break;
default:
$replyContent = "还不支持这一功能：".$result['category'];
break;
}
}else{
$replyContent = "不能理解你的内容：".$content;
}
}else{
$replyContent = "未开启语音识别功能";
}
if (is_array($replyContent)){
$resultStr = $this->transmitNews($object, $replyContent);
}else{
$resultStr = $this->transmitText($object, $replyContent);
}
return $resultStr;
}
private function transmitText($object, $content, $flag = 0)
{
$textTpl = "
%s
%d
";
$resultStr = sprintf($textTpl, $object->FromUserName, $object->ToUserName, time(), $content, $flag);
return $resultStr;
}
private function transmitNews($object, $arr_item)
{
if(!is_array($arr_item))
return;
$itemTpl = " -
<![CDATA[%s]]>
";
$item_str = "";
foreach ($arr_item as $item)
$item_str .= sprintf($itemTpl, $item['Title'], $item['Description'], $item['PicUrl'], $item['Url']);
$newsTpl = "
%s
%s
$item_str
";
$result = sprintf($newsTpl, $object->FromUserName, $object->ToUserName, time(), count($arr_item));
return $result;
}
}
?>空气质量airquality.php整体代码如下：
function getAirQualityChina($city)
{
//http://www.pm25.in/api/querys/aqi_details.json?stations=no&city=%E6%B7%B1%E5%9C%B3&token=5j1znBVAsnSf5xQyNQyq
$url = "http://www.pm25.in/api/querys/aqi_details.json?stations=no&city=".urlencode($city)."&token=5j1znBVAsnSf5xQyNQyq";
$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
$output = curl_exec($ch);
curl_close($ch);
$cityAir = json_decode($output, true);
if (isset($cityAir['error'])){
return $cityAir['error'];
}else{
$result = "空气质量指数(AQI)：".$cityAir[0]['aqi']."\n".
"空气质量等级：".$cityAir[0]['quality']."\n".
"细颗粒物(PM2.5)：".$cityAir[0]['pm2_5']."\n".
"可吸入颗粒物(PM10)：".$cityAir[0]['pm10']."\n".
"一氧化碳(CO)：".$cityAir[0]['co']."\n".
"二氧化氮(NO2)：".$cityAir[0]['no2']."\n".
"二氧化硫(SO2)：".$cityAir[0]['so2']."\n".
"臭氧(O3)：".$cityAir[0]['o3']."\n".
"更新时间：".preg_replace("/([a-zA-Z])/i", " ", $cityAir[0]['time_point']);
$aqiArray = array();
$aqiArray[] = array("Title" =>$cityAir[0]['area']."空气质量", "Description" =>$result, "PicUrl" =>"", "Url" =>"");
return $aqiArray;
}
}
?>天气查询weather2.php整体代码如下：
//var_dump(getWeatherInfo("桃江"));
function getWeatherInfo($cityName)
{
if ($cityName == "" || (strstr($cityName, "+"))){
return "发送天气加城市，例如'天气深圳'";
}
//1a3cde429f38434f1811a75e1a90310c
$ak = '1a3cde429f38434f1811a75e1a90310c';
$sk = '';
$url = 'http://api.map.baidu.com/telematics/v3/weather?ak=%s&location=%s&output=%s&sn=%s';
$uri = '/telematics/v3/weather';
$location = $cityName;
$output = 'json';
$querystring_arrays = array(
'ak' => $ak,
'location' => $location,
'output' => $output
);
$querystring = http_build_query($querystring_arrays);
$sn = md5(urlencode($uri.'?'.$querystring.$sk));
$targetUrl = sprintf($url, $ak, urlencode($location), $output, $sn);
// var_dump($targetUrl);
$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $targetUrl);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
$result = curl_exec($ch);
curl_close($ch);
$result = json_decode($result, true);
if ($result["error"] != 0){
return $result["status"];
}
$curHour = (int)date('H',time());
$weather = $result["results"][0];
$weatherArray[] = array("Title" =>$weather['currentCity']."天气预报", "Description" =>"", "PicUrl" =>"", "Url" =>"");
for ($i = 0; $i < count($weather["weather_data"]); $i++) {
$weatherArray[] = array("Title"=>
$weather["weather_data"][$i]["date"]."\n".
$weather["weather_data"][$i]["weather"]." ".
$weather["weather_data"][$i]["wind"]." ".
$weather["weather_data"][$i]["temperature"],
"Description"=>"",
"PicUrl"=>(($curHour >= 6) && ($curHour < 18))?$weather["weather_data"][$i]["dayPictureUrl"]:$weather["weather_data"][$i]["nightPictureUrl"], "Url"=>"");
}
return $weatherArray;
}
?>分词功能函数segment.php整体代码如下：
// $str="深圳天气怎么样";
//$result = sinasegment($str);
// print_r($result) ;
function sinasegment($str)
{
$seg = new SaeSegment();
$ret = $seg->segment($str, 1);
if ($ret === false){
return;
}
$category = "";
$keyword = "";
foreach ($ret as $key => $value) {
if ($value["word_tag"] == 95){
$category = $value["word"];
}
if ($value["word_tag"] == 102){
$keyword = $value["word"];
}
}
if (!empty($category) && !empty($keyword)){
return array('category'=>$category, 'keyword'=>$keyword);
}else{
return;
}
}
?>