php关键词替换的类(避免重复替换，保留与还原原始链接)

3. /*

4. * 关键词匹配类
5. * @author ylx
6. * @packet mipang
7. * 使用实例
8. * $str = "绿壳蛋鸡撒范德萨下一年，下一年1的洒落开房间卢卡斯地方军";
9. * $key = new KeyReplace($str,array("xxxx"=>"sadf","下一年1"=>'http://baidu.com',"下一年"=>'google.com'));
10. * echo $key->getResultText();
11. * echo $key->getRuntime();
12. */
13. class KeyReplace
14. {
15. private $keys = array();
16. private $text = "";
17. private $runtime = 0;
18. private $url = true;
19. private $stopkeys = array();
20. private $all = false;
21. /**
22. * @access public
23. * @param string $text 指定被处理的文章
24. * @param array $keys 指定字典词组array(key=>url,...) url可以是数组，如果是数组将随机替换其中的一个
25. * @param array $stopkeys 指定停止词array(key,...) 这里面的词将不会被处理
26. * @param boolean $url true 表示替换成链接否则只替换
27. * @param boolean $all true 表示替换所有找到的词，否则只替换第一次
28. */
29. public function __construct($text='',$keys=array(),$url=true,$stopkeys=array(),$all=false) {
30. $this->keys = $keys;
31. $this->text = $text;
32. $this->url = $url;
33. $this->stopkeys = $stopkeys;
34. $this->all = $all;
35. }

36. /**

37. * 获取处理好的文章
38. * @access public
39. * @return string text
40. */
41. public function getResultText() {
42. $start = microtime(true);
43. $keys = $this->hits_keys();

44. $keys_tmp = array_keys($keys);

45. function cmp($a, $b){

46. if (mb_strlen($a) == mb_strlen($b)) {
47. return 0;
48. }
49. return (mb_strlen($a) }

50. usort($keys_tmp,"cmp");

51. foreach($keys_tmp as $key){

52. if(is_array($keys[$key])){

53. $url = $keys[$key][rand(0,count($keys[$key])-1)];
54. }else
55. $url = $keys[$key];

56. $this->text = $this->r_s($this->text,$key,$url);

57. }

58. $this->runtime = microtime(true)-$start;

59. return $this->text;

60. }
61. /**
62. * 获取处理时间
63. * @access public
64. * @return float
65. */
66. public function getRuntime() {

67. return $this->runtime;

68. }

69. /**

70. * 设置关键词
71. * @access public
72. * @param array $keys array(key=>url,...)
73. */
74. public function setKeys($keys) {

75. $this->keys = $keys;

76. }

77. /**
78. * 设置停止词
79. * @access public
80. * @param array $keys array(key,...)
81. */
82. public function setStopKeys($keys) {

83. $this->stopkeys = $keys;

84. }

85. /**
86. * 设置文章
87. * @access public
88. * @param string $text
89. */
90. public function setText($text) {

91. $this->text = $text;

92. }

93. /**

94. * 用来找到字符串里面命中的关键词
95. * @access public
96. * @return array $keys 返回匹配到的词array(key=>url,...)
97. */
98. public function hits_keys(){
99. $ar = $this->keys;
100. $ar = $ar?$ar:array();
101. $result=array();
102. $str = $this->text;
103. foreach($ar as $k=>$url){
104. $k = trim($k);
105. if(!$k)
106. continue;
107. if(strpos($str,$k)!==false && !in_array($k,$this->stopkeys)){
108. $result[$k] = $url;
109. }
110. }
111. return $result?$result:array();
112. }

113. /**

114. * 用来找到字符串里面命中的停止词
115. * @access public
116. * @return array $keys 返回匹配到的词array(key,...)
117. */
118. public function hits_stop_keys(){
119. $ar = $this->stopkeys;
120. $ar = $ar?$ar:array();
121. $result=array();
122. $str = $this->text;
123. foreach($ar as $k){
124. $k = trim($k);
125. if(!$k)
126. continue;
127. if(strpos($str,$k)!==false && in_array($k,$this->stopkeys)){
128. $result[] = $k;
129. }
130. }
131. return $result?$result:array();
132. }

133. /**

134. * 处理替换过程
135. * @access private
136. * @param string $text 被替换者
137. * @param string $key 关键词
138. * @param string $url 链接
139. * @return string $text 处理好的文章
140. */
141. private function r_s($text,$key,$url){

142. $tmp = $text;

143. $stop_keys = $this->hits_stop_keys();

144. $stopkeys = $tags = $a = array();

145. if(preg_match_all("#]+>[^]*>#su",$tmp,$m)){
146. $a=$m[0];

147. foreach($m[0] as $k=>$z){

148. $z = preg_replace("#\##s","\#",$z);

149. $tmp = preg_replace('#'.$z.'#s',"[_a".$k."_]",$tmp,1);

150. }

151. };

152. if(preg_match_all("#]+>#s",$tmp,$m)){

153. $tags = $m[0];
154. foreach($m[0] as $k=>$z){
155. $z = preg_replace("#\##s","\#",$z);
156. $tmp = preg_replace('#'.$z.'#s',"[_tag".$k."_]",$tmp,1);
157. }
158. }
159. if(!empty($stop_keys)){
160. if(preg_match_all("#".implode("|",$stop_keys)."#s",$tmp,$m)){
161. $stopkeys = $m[0];
162. foreach($m[0] as $k=>$z){
163. $z = preg_replace("#\##s","\#",$z);
164. $tmp = preg_replace('#'.$z.'#s',"[_s".$k."_]",$tmp,1);
165. }
166. }
167. }
168. $key1 = preg_replace("#([\#\(\)\[\]\*])#s","\\\\$1",$key);

169. if($this->url)

170. $tmp = preg_replace("#(?!\[_s|\[_a|\[_|\[_t|\[_ta|\[_tag)".$key1."(?!ag\d+_\]|g\d+_\]|\d+_\]|s\d+_\]|_\])#us",''.$key.'',$tmp,$this->all?-1:1);
171. else
172. $tmp = preg_replace("#(?!\[_s|\[_a|\[_|\[_t|\[_ta|\[_tag)".$key1."(?!ag\d+_\]|g\d+_\]|\d+_\]|s\d+_\]|_\])#us",$url,$tmp,$this->all?-1:1);

173. if(!empty($a)){

174. foreach($a as $n=>$at){

175. $tmp = str_replace("[_a".$n."_]",$at,$tmp);

176. }

177. }

178. if(!empty($tags)){

179. foreach($tags as $n=>$at){

180. $tmp = str_replace("[_tag".$n."_]",$at,$tmp);

181. }

182. }

183. if(!empty($stopkeys)){

184. foreach($stopkeys as $n=>$at){

185. $tmp = str_replace("[_s".$n."_]",$at,$tmp);

186. }

187. }

188. return $tmp;
189. }
190. }

复制代码