dict_find_one.php 7.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338
  1. <?php
  2. require_once "../path.php";
  3. require_once "../public/_pdo.php";
  4. require_once "../public/function.php";
  5. require_once '../ucenter/setting_function.php';
  6. if(isset($_GET["book"])){
  7. $in_book=$_GET["book"];
  8. }
  9. if(isset($_GET["paragraph"])){
  10. $in_para=$_GET["paragraph"];
  11. }
  12. if(isset($_GET["sn"])){
  13. $in_sn=$_GET["sn"];
  14. }
  15. if(isset($_GET["type"])){
  16. $type=$_GET["type"];
  17. }
  18. else{
  19. $type="wbw";
  20. }
  21. if(isset($_GET["dict_name"])){
  22. $dict_name=$_GET["dict_name"];
  23. }
  24. else{
  25. $dict_name="";
  26. }
  27. if($type=="part"){
  28. $lookup_loop=3;
  29. }
  30. else{
  31. $lookup_loop=3;
  32. }
  33. if(isset($_GET["deep"])){
  34. $lookup_loop=$_GET["deep"];
  35. }
  36. else{
  37. $lookup_loop=3;
  38. }
  39. $in_word=$_GET["word"];
  40. if(isset($_GET["debug"])){
  41. $debug=true;;
  42. }
  43. else{
  44. $debug=false;
  45. }
  46. if(mb_strlen($in_word)==0){
  47. exit;
  48. }
  49. function microtime_float()
  50. {
  51. list($usec, $sec) = explode(" ", microtime());
  52. return ((float)$usec + (float)$sec);
  53. }
  54. $time_start = microtime_float();
  55. $user_setting = get_setting();
  56. //open database
  57. global $PDO;
  58. $word_list=str_getcsv($in_word);
  59. $dict_word_spell=array();
  60. $output=array();
  61. $db_file_list=array();
  62. //用户词典
  63. if($dict_name==""){
  64. array_push($db_file_list , _FILE_DB_WBW_);
  65. array_push($db_file_list , _DIR_DICT_SYSTEM_."/sys_regular.db");
  66. array_push($db_file_list , _DIR_DICT_SYSTEM_."/sys_irregular.db");
  67. array_push($db_file_list , _DIR_DICT_SYSTEM_."/union.db");
  68. array_push($db_file_list , _DIR_DICT_SYSTEM_."/comp.db");
  69. array_push($db_file_list , _DIR_DICT_3RD_."/pm.db");
  70. array_push($db_file_list , _DIR_DICT_3RD_."/bhmf.db");
  71. array_push($db_file_list , _DIR_DICT_3RD_."/shuihan.db");
  72. array_push($db_file_list , _DIR_DICT_3RD_."/concise.db");
  73. array_push($db_file_list , _DIR_DICT_3RD_."/uhan_en.db");
  74. }
  75. else{
  76. $dict_list=str_getcsv($dict_name,',');
  77. foreach($dict_list as $dict){
  78. array_push($db_file_list , $dict);
  79. }
  80. }
  81. $_dict_db = array();
  82. foreach($db_file_list as $db_file){
  83. try {
  84. $dbh = new PDO("sqlite:".$db_file, "", "");
  85. $dbh->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_WARNING);
  86. $_dict_db[] = array("file"=>$db_file,"dbh"=>$dbh);
  87. } catch (PDOException $e) {
  88. if($debug){
  89. print "Error!: " . $e->getMessage() . "<br/>";
  90. }
  91. }
  92. }
  93. for($i=0;$i<$lookup_loop;$i++)
  94. {
  95. $parent_list=array();
  96. $strQueryWord="(";//单词查询字串
  97. foreach($word_list as $word){
  98. $word=str_replace("'","’",$word);
  99. $strQueryWord.="'{$word}',";
  100. }
  101. $strQueryWord=mb_substr($strQueryWord, 0,mb_strlen($strQueryWord,"UTF-8")-1,"UTF-8");
  102. $strQueryWord.=")";
  103. if($debug){
  104. echo "<h2>第".($i+1)."轮查询:".count($word_list)."</h2>";
  105. }
  106. foreach($_dict_db as $db_file){
  107. if($debug){
  108. echo "dict connect:{$db_file["file"]}<br>";
  109. }
  110. //PDO_Connect("sqlite:".$db_file);
  111. if($i==0){
  112. $query = "select * from dict where \"pali\" in $strQueryWord ORDER BY rowid DESC";
  113. }
  114. else{
  115. $query = "select * from dict where \"pali\" in $strQueryWord AND ( type <> '.n.' AND type <> '.ti.' AND type <> '.adj.' AND type <> '.pron.' AND type <> '.v.' ) ORDER BY rowid DESC";
  116. }
  117. if($debug){
  118. echo $query."<br>";
  119. }
  120. if($db_file["dbh"]){
  121. $stmt = $db_file["dbh"]->query($query);
  122. if($stmt ){
  123. $Fetch = $stmt->fetchAll(PDO::FETCH_ASSOC);
  124. }
  125. else{
  126. $Fetch = array();
  127. }
  128. }
  129. else{
  130. $Fetch = array();
  131. if($debug){
  132. echo "无效的数据库句柄";
  133. }
  134. }
  135. //$Fetch = PDO_FetchAll($query);
  136. $iFetch=count($Fetch);
  137. if($debug){
  138. echo "count:$iFetch<br>";
  139. }
  140. if($iFetch>0){
  141. foreach($Fetch as $one){
  142. $id = $one["id"];
  143. if(isset($one["guid"])){
  144. $guid = $one["guid"];
  145. }
  146. else{
  147. $guid = "";
  148. }
  149. if(isset($one["lang"])){
  150. $language = $one["lang"];
  151. }
  152. else if(isset($one["language"])){
  153. $language = $one["language"];
  154. }
  155. else{
  156. $language = "en";
  157. }
  158. $pali = $one["pali"];
  159. $dict_word_spell["{$pali}"]=1;
  160. $type = $one["type"];
  161. $gramma = $one["gramma"];
  162. $parent = $one["parent"];
  163. //$mean = $one["mean"];
  164. if(inLangSetting($language,$user_setting["dict.lang"])){
  165. $mean = $one["mean"];
  166. }
  167. else{
  168. $mean = "";
  169. }
  170. $note = $one["note"];
  171. if(isset($one["factors"])){
  172. $parts = $one["factors"];
  173. }
  174. else if(isset($one["parts"])){
  175. $parts = $one["parts"];
  176. }
  177. else{
  178. $parts = "";
  179. }
  180. if(isset($one["factormean"])){
  181. $partmean = $one["factormean"];
  182. }
  183. else if(isset($one["partmean"])){
  184. $partmean = $one["partmean"];
  185. }
  186. else{
  187. $partmean = "";
  188. }
  189. if(inLangSetting($language,$user_setting["dict.lang"])==false){
  190. $partmean = "";
  191. }
  192. $status = $one["status"];
  193. if(isset($one["confidence"])){
  194. $confidence = $one["confidence"];
  195. }
  196. else{
  197. $confidence = 100;
  198. }
  199. if(isset($one["dict_name"])){
  200. $dict_name = $one["dict_name"];
  201. }
  202. else{
  203. $dict_name = "";
  204. }
  205. array_push($output,array(
  206. "id"=>$id,
  207. "guid"=>$guid,
  208. "pali"=>$pali,
  209. "type"=>$type,
  210. "gramma"=>$gramma,
  211. "parent"=>$parent,
  212. "mean"=>$mean,
  213. "note"=>$note,
  214. "parts"=>$parts,
  215. "partmean"=>$partmean,
  216. "status"=>$status,
  217. "confidence"=>$confidence,
  218. "dict_name"=>$dict_name,
  219. "language"=>$language
  220. ));
  221. //将语基插入下次查询的列表
  222. if(!empty($parent)){
  223. if($pali != $parent){
  224. $parent_list[$parent]=1;
  225. }
  226. }
  227. //将拆分插入下次查询的列表
  228. if($type!=".part."){
  229. if(!empty($parts)){
  230. $wordparts=str_getcsv($parts,'+');
  231. foreach($wordparts as $x){
  232. if(!empty($x)){
  233. if($x != $pali){
  234. $parent_list[$x]=1;
  235. }
  236. }
  237. }
  238. }
  239. }
  240. }
  241. }
  242. }
  243. /*
  244. if($i==0){
  245. //自动查找单词词干
  246. $word_base=getPaliWordBase($in_word);
  247. foreach($word_base as $x=>$infolist){
  248. foreach($infolist as $gramma){
  249. array_push($output,
  250. array("pali"=>$in_word,
  251. "parent"=>$x,
  252. "type"=>$gramma["type"],
  253. "gramma"=>$gramma["gramma"],
  254. "parts"=>$gramma["parts"],
  255. "partmean"=>"",
  256. "mean"=>"",
  257. "language"=>"en",
  258. "dict_name"=>"auto",
  259. "status"=>128
  260. ));
  261. $part_list=str_getcsv($gramma["parts"],"+");
  262. foreach($part_list as $part){
  263. $parent_list[$part]=1;
  264. }
  265. }
  266. }
  267. }
  268. */
  269. if($debug){
  270. echo "parent:".count($parent_list)."<br>";
  271. print_r($parent_list)."<br>";
  272. }
  273. if(count($parent_list)==0){
  274. break;
  275. }
  276. else{
  277. $word_list=array();
  278. foreach($parent_list as $x=>$value){
  279. array_push($word_list,$x);
  280. }
  281. }
  282. }
  283. //删除无效数据
  284. $newOutput = array();
  285. foreach($output as $value){
  286. if($value["dict_name"]=="auto"){
  287. if(isset($dict_word_spell["{$value["parent"]}"])){
  288. array_push($newOutput,$value);
  289. }
  290. }
  291. else
  292. {
  293. array_push($newOutput,$value);
  294. }
  295. }
  296. if($debug){
  297. echo "<textarea width=\"100%\" >";
  298. }
  299. echo json_encode($newOutput, JSON_UNESCAPED_UNICODE);
  300. if($debug){
  301. echo "</textarea>";
  302. }
  303. if($debug){
  304. echo "生成:".count($output)."<br>";
  305. echo "有效:".count($newOutput)."<br>";
  306. foreach($newOutput as $result){
  307. echo "{$result["pali"]}-{$result["parent"]}-{$result["mean"]}<br>";
  308. }
  309. $queryTime=(microtime_float()-$time_start)*1000;
  310. echo "<div >搜索时间:$queryTime 毫秒</div>";
  311. }
  312. ?>