dict_find_auto.php 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519
  1. <?php
  2. include("../log/pref_log.php");
  3. require_once "../path.php";
  4. require_once "../public/_pdo.php";
  5. require_once "../public/function.php";
  6. require_once '../ucenter/setting_function.php';
  7. require_once "../redis/function.php";
  8. $redis = redis_connect();
  9. global $error;
  10. $error = array();
  11. set_error_handler(function(int $number, string $message) {
  12. global $error;
  13. $error[] = "Handler captured error $number: '$message'" . PHP_EOL ;
  14. });
  15. $user_setting = get_setting();
  16. if (isset($_GET["book"])) {
  17. $in_book = $_GET["book"];
  18. }
  19. if (isset($_GET["para"])) {
  20. $in_para = $_GET["para"];
  21. }
  22. $para_list = str_getcsv($in_para);
  23. $strQueryPara = "("; //单词查询字串
  24. foreach ($para_list as $para) {
  25. $strQueryPara .= "'{$para}',";
  26. }
  27. $strQueryPara = mb_substr($strQueryPara, 0, mb_strlen($strQueryPara, "UTF-8") - 1, "UTF-8");
  28. $strQueryPara .= ")";
  29. if (isset($_GET["debug"])) {
  30. $debug = true;
  31. } else {
  32. $debug = false;
  33. }
  34. function microtime_float()
  35. {
  36. list($usec, $sec) = explode(" ", microtime());
  37. return ((float) $usec + (float) $sec);
  38. }
  39. $time_start = microtime_float();
  40. //open database
  41. global $PDO;
  42. //查询单词表
  43. $db_file = _DIR_PALICANON_TEMPLET_ . "/p" . $in_book . "_tpl.db3";
  44. PDO_Connect("sqlite:{$db_file}");
  45. $query = "SELECT paragraph,wid,real FROM \"main\" WHERE (\"paragraph\" in " . $strQueryPara . " ) and \"real\"<>\"\" and \"type\"<>'.ctl.' ";
  46. if ($debug) {
  47. echo "filename:" . $db_file . "<br>";
  48. echo $query . "<br>";
  49. }
  50. $FetchAllWord = PDO_FetchAll($query);
  51. $iFetch = count($FetchAllWord);
  52. if ($iFetch == 0) {
  53. echo json_encode(array(), JSON_UNESCAPED_UNICODE);
  54. exit;
  55. }
  56. $voc_list = array();
  57. foreach ($FetchAllWord as $word) {
  58. $voc_list[$word["real"]] = 1;
  59. }
  60. if ($debug) {
  61. echo "单词表共计:" . count($voc_list) . "词<br>";
  62. }
  63. //查询单词表结束
  64. $word_list = array();
  65. foreach ($voc_list as $word => $value) {
  66. array_push($word_list, $word);
  67. }
  68. $lookup_loop = 2;
  69. $dict_word_spell = array();
  70. $output = array();
  71. $db_file_list = array();
  72. //字典列表
  73. /*
  74. array_push($db_file_list, array(_FILE_DB_WBW1_, " ORDER BY rowid DESC"));
  75. array_push($db_file_list, array(_DIR_DICT_SYSTEM_ . "/sys_regular.db", " ORDER BY confidence DESC"));
  76. array_push($db_file_list, array(_DIR_DICT_SYSTEM_ . "/sys_irregular.db", ""));
  77. array_push($db_file_list, array(_DIR_DICT_SYSTEM_ . "/union.db", ""));
  78. array_push($db_file_list, array(_DIR_DICT_SYSTEM_ . "/comp.db", ""));
  79. array_push($db_file_list, array(_DIR_DICT_3RD_ . "/pm.db", ""));
  80. array_push($db_file_list, array(_DIR_DICT_3RD_ . "/bhmf.db", ""));
  81. array_push($db_file_list, array(_DIR_DICT_3RD_ . "/shuihan.db", ""));
  82. array_push($db_file_list, array(_DIR_DICT_3RD_ . "/concise.db", ""));
  83. array_push($db_file_list, array(_DIR_DICT_3RD_ . "/uhan_en.db", ""));
  84. */
  85. $db_file_list[] = array("","wbwdict://new/".$_COOKIE["userid"],true);
  86. $db_file_list[] = array(_FILE_DB_TERM_,"dict://term",true);
  87. $db_file_list[] = array(_FILE_DB_WBW1_,"dict://user",true);
  88. $db_file_list[] = array( _DIR_DICT_SYSTEM_ . "/sys_regular.db","dict://regular",true);
  89. $db_file_list[] = array( _DIR_DICT_SYSTEM_ . "/sys_irregular.db","dict://irregular",true);
  90. $db_file_list[] = array( _DIR_DICT_SYSTEM_ . "/union.db","dict://union",true);
  91. $db_file_list[] = array( _DIR_DICT_SYSTEM_ . "/comp.db","dict://comp",true);
  92. $db_file_list[] = array( _DIR_DICT_3RD_ . "/pm.db","dict://pm",true);
  93. $db_file_list[] = array( _DIR_DICT_3RD_ . "/bhmf.db","dict://bhmf",true);
  94. $db_file_list[] = array( _DIR_DICT_3RD_ . "/shuihan.db","dict://shuihan",true);
  95. $db_file_list[] = array( _DIR_DICT_3RD_ . "/concise.db","dict://concise",true);
  96. $db_file_list[] = array( _DIR_DICT_3RD_ . "/uhan_en.db","dict://uhan_en",true);
  97. $_dict_db = array();
  98. foreach ($db_file_list as $db_file) {
  99. try {
  100. if ($redis && !empty($db_file[1])) {
  101. $dbh=null;
  102. }
  103. else{
  104. $dbh = new PDO("sqlite:" . $db_file[0], "", "");
  105. $dbh->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_WARNING);
  106. $dbh->query("PRAGMA synchronous = OFF");
  107. $dbh->query("PRAGMA journal_mode = WAL");
  108. $dbh->query("PRAGMA foreign_keys = ON");
  109. $dbh->query("PRAGMA busy_timeout = 5000");
  110. }
  111. $_dict_db[] = array("file" => $db_file[0], "dbh" => $dbh,"redis"=>$db_file[1],"static"=>$db_file[2]);
  112. } catch (PDOException $e) {
  113. if ($debug) {
  114. print "Error!: " . $e->getMessage() . "<br/>";
  115. }
  116. }
  117. }
  118. $lookuped=array();
  119. for ($i = 0; $i < $lookup_loop; $i++) {
  120. $parent_list = array();
  121. # 记录已经查过的词,下次就不查了
  122. $newWordList = array();
  123. foreach ($word_list as $lsWord) {
  124. if(!isset($lookuped[$lsWord]) && !empty($lsWord)){
  125. $newWordList[]=$lsWord;
  126. $lookuped[$lsWord]=1;
  127. }
  128. }
  129. if(count($newWordList)==0){
  130. break;
  131. }
  132. $word_list = $newWordList;
  133. # 记录已经查过的词结束
  134. $strQueryWord = "("; //单词查询字串
  135. foreach ($word_list as $word) {
  136. $word = str_replace("'", "’", $word);
  137. $strQueryWord .= "'{$word}',";
  138. }
  139. $strQueryWord = mb_substr($strQueryWord, 0, mb_strlen($strQueryWord, "UTF-8") - 1, "UTF-8");
  140. $strQueryWord .= ")";
  141. if ($debug) {
  142. echo "<h2>第{$i}轮查询:$strQueryWord</h2>";
  143. }
  144. foreach ($_dict_db as $db) {
  145. $db_file = $db["file"];
  146. if ($debug) {
  147. echo "dict:$db_file<br>";
  148. }
  149. $strOrderby = $db["file"][1];
  150. if ($i == 0) {
  151. $query = "select * from dict where \"pali\" in {$strQueryWord} AND ( type <> '.n:base.' AND type <> '.ti:base.' AND type <> '.adj:base.' AND type <> '.pron:base.' AND type <> '.v:base.' AND type <> '.part.' ) " . $strOrderby;
  152. } else {
  153. $query = "select * from dict where \"pali\" in {$strQueryWord} " . $strOrderby;
  154. }
  155. if ($debug) {
  156. echo $query . "<br>";
  157. }
  158. $Fetch = array();
  159. if ($redis && !empty($db["redis"])) {
  160. if ($debug) {
  161. echo "<spen style='color:green;'>redis</spen>:{$db["redis"]}<br>";
  162. }
  163. foreach ($word_list as $word) {
  164. $wordData = $redis->hGet($db["redis"],$word);
  165. if($wordData){
  166. if(!empty($wordData)){
  167. $arrWord = json_decode($wordData,true);
  168. foreach ($arrWord as $one) {
  169. # code...
  170. if(count($one)==14){
  171. $Fetch[] = array("id"=>$one[0],
  172. "pali"=>$one[1],
  173. "type"=>$one[2],
  174. "gramma"=>$one[3],
  175. "parent"=>$one[4],
  176. "mean"=>$one[5],
  177. "note"=>$one[6],
  178. "parts"=>$one[7],
  179. "partmean"=>$one[8],
  180. "status"=>$one[9],
  181. "confidence"=>$one[10],
  182. "dict_name"=>$one[12],
  183. "lang"=>$one[13]
  184. );
  185. }
  186. else{
  187. $Fetch[] = array("id"=>$one[0],
  188. "pali"=>$one[1],
  189. "type"=>$one[2],
  190. "gramma"=>$one[3],
  191. "parent"=>$one[4],
  192. "mean"=>$one[5],
  193. "note"=>$one[6],
  194. "parts"=>$one[7],
  195. "partmean"=>"",
  196. "status"=>$one[8],
  197. "confidence"=>$one[9],
  198. "dict_name"=>$one[10],
  199. "lang"=>$one[12]
  200. );
  201. }
  202. }
  203. }
  204. }
  205. else{
  206. # 没找到就不找了
  207. }
  208. }
  209. }
  210. else{
  211. try {
  212. //$Fetch = PDO_FetchAll($query);
  213. $stmt = $db["dbh"]->query($query);
  214. if ($stmt) {
  215. $Fetch = $stmt->fetchAll(PDO::FETCH_ASSOC);
  216. } else {
  217. $Fetch = array();
  218. if ($debug) {
  219. echo "无效的Statement句柄";
  220. }
  221. }
  222. } catch (Exception $e) {
  223. if ($debug) {
  224. echo 'Caught exception: ', $e->getMessage(), "\n";
  225. }
  226. continue;
  227. }
  228. }
  229. $iFetch = count($Fetch);
  230. if ($debug) {
  231. echo "count:{$iFetch}<br>";
  232. }
  233. if ($iFetch > 0) {
  234. foreach ($Fetch as $one) {
  235. $id = $one["id"];
  236. if (isset($one["guid"])) {
  237. $guid = $one["guid"];
  238. } else {
  239. $guid = "";
  240. }
  241. if (isset($one["lang"])) {
  242. $language = $one["lang"];
  243. } else if (isset($one["language"])) {
  244. $language = $one["language"];
  245. } else {
  246. $language = "en";
  247. }
  248. $pali = $one["pali"];
  249. $dict_word_spell["{$pali}"] = 1;
  250. $type = $one["type"];
  251. $gramma = $one["gramma"];
  252. $parent = $one["parent"];
  253. if (inLangSetting($language, $user_setting["dict.lang"])) {
  254. $mean = $one["mean"];
  255. } else {
  256. $mean = "";
  257. }
  258. if (isset($one["note"])) {
  259. $note = $one["note"];
  260. } else {
  261. $note = "";
  262. }
  263. if (isset($one["parts"])) {
  264. $parts = $one["parts"];
  265. } else if (isset($one["factors"])) {
  266. $parts = $one["factors"];
  267. } else {
  268. $parts = "";
  269. }
  270. if (isset($one["partmean"])) {
  271. $partmean = $one["partmean"];
  272. } else if (isset($one["factormean"])) {
  273. $partmean = $one["factormean"];
  274. } else {
  275. $partmean = "";
  276. }
  277. if (inLangSetting($language, $user_setting["dict.lang"]) == false) {
  278. $partmean = "";
  279. }
  280. if (isset($one["part_id"])) {
  281. $part_id = $one["part_id"];
  282. } else {
  283. $part_id = "";
  284. }
  285. if (isset($one["status"])) {
  286. $status = $one["status"];
  287. } else {
  288. $status = "";
  289. }
  290. if (isset($one["dict_name"])) {
  291. $dict_name = $one["dict_name"];
  292. } else {
  293. $dict_name = "";
  294. }
  295. array_push($output, array(
  296. "id" => $id,
  297. "guid" => $guid,
  298. "pali" => $pali,
  299. "type" => $type,
  300. "gramma" => $gramma,
  301. "parent" => $parent,
  302. "mean" => $mean,
  303. "note" => $note,
  304. "parts" => $parts,
  305. "part_id" => $part_id,
  306. "partmean" => $partmean,
  307. "status" => $status,
  308. "dict_name" => $dict_name,
  309. "language" => $language,
  310. ));
  311. if (!empty($parent)) {
  312. if ($pali != $parent) {
  313. $parent_list[$one["parent"]] = 1;
  314. }
  315. }
  316. if ($type != "part") {
  317. if (isset($one["factors"])) {
  318. $parts = str_getcsv($one["factors"], '+');
  319. foreach ($parts as $x) {
  320. if (!empty($x)) {
  321. if ($x != $pali) {
  322. $parent_list[$x] = 1;
  323. }
  324. }
  325. }
  326. }
  327. }
  328. }
  329. }
  330. $PDO = null;
  331. }
  332. /*
  333. if($i==0){
  334. //自动查找单词词干
  335. $word_base=getPaliWordBase($in_word);
  336. foreach($word_base as $x=>$infolist){
  337. foreach($infolist as $gramma){
  338. array_push($output,
  339. array("pali"=>$in_word,
  340. "type"=>$gramma["type"],
  341. "gramma"=>$gramma["gramma"],
  342. "mean"=>"",
  343. "parent"=>$x,
  344. "parts"=>$gramma["parts"],
  345. "partmean"=>"",
  346. "language"=>"en",
  347. "dict_name"=>"auto",
  348. "status"=>128
  349. ));
  350. $part_list=str_getcsv($gramma["parts"],"+");
  351. foreach($part_list as $part){
  352. $parent_list[$part]=1;
  353. }
  354. }
  355. }
  356. }
  357. */
  358. if ($debug) {
  359. echo "parent:" . count($parent_list) . "<br>";
  360. //print_r($parent_list)."<br>";
  361. }
  362. if (count($parent_list) == 0) {
  363. break;
  364. } else {
  365. $word_list = array();
  366. foreach ($parent_list as $x => $value) {
  367. array_push($word_list, $x);
  368. }
  369. }
  370. }
  371. //查询结束
  372. //删除无效数据
  373. $newOutput = array();
  374. foreach ($output as $value) {
  375. if ($value["dict_name"] == "auto") {
  376. if (isset($dict_word_spell["{$value["parent"]}"])) {
  377. array_push($newOutput, $value);
  378. }
  379. } else {
  380. array_push($newOutput, $value);
  381. }
  382. }
  383. if ($debug) {
  384. echo "<textarea width=\"100%\" >";
  385. echo json_encode($newOutput, JSON_UNESCAPED_UNICODE);
  386. echo "</textarea>";
  387. }
  388. if ($debug) {
  389. echo "生成:" . count($output) . "<br>";
  390. echo "有效:" . count($newOutput) . "<br>";
  391. }
  392. //开始匹配
  393. $counter = 0;
  394. $output = array();
  395. foreach ($FetchAllWord as $word) {
  396. $pali = $word["real"];
  397. $type = "";
  398. $gramma = "";
  399. $mean = "";
  400. $parent = "";
  401. $parts = "";
  402. $partmean = "";
  403. foreach ($newOutput as $dictword) {
  404. if ($dictword["pali"] == $pali) {
  405. if ($type == "" && $gramma == "") {
  406. $type = $dictword["type"];
  407. $gramma = $dictword["gramma"];
  408. }
  409. if (trim($mean) == "") {
  410. $mean = str_getcsv($dictword["mean"], "$")[0];
  411. }
  412. if ($parent == "") {
  413. $parent = $dictword["parent"];
  414. }
  415. if ($parts == "") {
  416. $parts = $dictword["parts"];
  417. }
  418. if ($partmean == "") {
  419. $partmean = $dictword["partmean"];
  420. }
  421. }
  422. }
  423. if ($mean == "" && $parent != "") {
  424. foreach ($newOutput as $parentword) {
  425. if ($parentword["pali"] == $parent) {
  426. if ($parentword["mean"] != "") {
  427. $mean = trim(str_getcsv($parentword["mean"], "$")[0]);
  428. if ($mean != "") {
  429. break;
  430. }
  431. }
  432. }
  433. }
  434. }
  435. if ($type != "" ||
  436. $gramma != "" ||
  437. $mean != "" ||
  438. $parent != "" ||
  439. $parts != "" ||
  440. $partmean != "") {
  441. $counter++;
  442. }
  443. array_push($output,
  444. array("book" => $in_book,
  445. "paragraph" => $word["paragraph"],
  446. "num" => $word["wid"],
  447. "pali" => $word["real"],
  448. "type" => $type,
  449. "gramma" => $gramma,
  450. "mean" => $mean,
  451. "parent" => $parent,
  452. "parts" => $parts,
  453. "partmean" => $partmean,
  454. "status" => 3,
  455. ));
  456. }
  457. if ($debug) {
  458. echo "<textarea width=\"100%\" >";
  459. }
  460. echo json_encode($output, JSON_UNESCAPED_UNICODE);
  461. if ($debug) {
  462. echo "</textarea>";
  463. }
  464. if ($debug) {
  465. echo "匹配" . (($counter / count($FetchAllWord)) * 100) . "<br>";
  466. foreach ($output as $result) {
  467. //echo "{$result["pali"]}-{$result["mean"]}-{$result["parent"]}<br>";
  468. }
  469. $queryTime = (microtime_float() - $time_start) * 1000;
  470. echo "<div >搜索时间:$queryTime 毫秒</div>";
  471. }
  472. PrefLog();