MdRender.php 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621
  1. <?php
  2. namespace App\Http\Api;
  3. use Illuminate\Support\Str;
  4. use mustache\mustache;
  5. use App\Models\DhammaTerm;
  6. use App\Models\PaliText;
  7. use App\Models\Channel;
  8. use App\Http\Controllers\CorpusController;
  9. use Illuminate\Support\Facades\Cache;
  10. use App\Tools\RedisClusters;
  11. use Illuminate\Support\Facades\Log;
  12. use App\Tools\Markdown;
  13. define("STACK_DEEP",8);
  14. class MdRender{
  15. /**
  16. * 文字渲染模式
  17. * read 阅读模式
  18. * edit 编辑模式
  19. */
  20. protected $options = [
  21. 'mode' => 'read',
  22. 'channelType'=>'translation',
  23. 'contentType'=>"markdown",
  24. 'format'=>'react',
  25. 'debug'=>[],
  26. 'studioId'=>null,
  27. ];
  28. public function __construct($options=[])
  29. {
  30. foreach ($options as $key => $value) {
  31. $this->options[$key] = $value;
  32. }
  33. }
  34. /**
  35. * 按照{{}}把字符串切分成三个部分。模版之前的,模版,和模版之后的
  36. */
  37. private function tplSplit($tpl){
  38. $before = strpos($tpl,'{{');
  39. if($before === FALSE){
  40. //未找到
  41. return ['data'=>[$tpl,'',''],'error'=>0];
  42. }else{
  43. $pointer = $before;
  44. $stack = array();
  45. $stack[] = $pointer;
  46. $after = substr($tpl,$pointer+2) ;
  47. while (!empty($after) && count($stack)>0 && count($stack)<STACK_DEEP) {
  48. $nextBegin = strpos($after,"{{");
  49. $nextEnd = strpos($after,"}}");
  50. if($nextBegin !== FALSE){
  51. if($nextBegin < $nextEnd){
  52. //有嵌套找到最后一个}}
  53. $pointer = $pointer + 2 + $nextBegin;
  54. $stack[] = $pointer;
  55. $after = substr($tpl,$pointer+2);
  56. }else if($nextEnd !== FALSE){
  57. //无嵌套有结束
  58. $pointer = $pointer + 2 + $nextEnd;
  59. array_pop($stack);
  60. $after = substr($tpl,$pointer+2);
  61. }else{
  62. //无结束符 没找到
  63. break;
  64. }
  65. }else if($nextEnd !== FALSE){
  66. $pointer = $pointer + 2 + $nextEnd;
  67. array_pop($stack);
  68. $after = substr($tpl,$pointer+2);
  69. }else{
  70. //没找到
  71. break;
  72. }
  73. }
  74. if(count($stack)>0){
  75. if(count($stack) === STACK_DEEP){
  76. return ['data'=>[$tpl,'',''],'error'=>2];
  77. }else{
  78. //未关闭
  79. return ['data'=>[$tpl,'',''],'error'=>1];
  80. }
  81. }else{
  82. return ['data'=>
  83. [
  84. substr($tpl,0,$before),
  85. substr($tpl,$before,$pointer-$before+2),
  86. substr($tpl,$pointer+2)
  87. ],
  88. 'error'=>0
  89. ];
  90. }
  91. }
  92. }
  93. private function wiki2xml(string $wiki,$channelId=[]):string{
  94. /**
  95. * 把模版转换为xml
  96. */
  97. $remain = $wiki;
  98. $buffer = array();
  99. do {
  100. $arrWiki = $this->tplSplit($remain);
  101. $buffer[] = $arrWiki['data'][0];
  102. $tpl = $arrWiki['data'][1];
  103. if(!empty($tpl)){
  104. /**
  105. * 处理模版 提取参数
  106. */
  107. $tpl = str_replace("|\n","|",$tpl);
  108. $pattern = "/\{\{(.+?)\|/";
  109. $replacement = '<MdTpl class="tpl" name="$1"><param>';
  110. $tpl = preg_replace($pattern,$replacement,$tpl);
  111. $tpl = str_replace("}}","</param></MdTpl>",$tpl);
  112. $tpl = str_replace("|","</param><param>",$tpl);
  113. /**
  114. * 替换变量名
  115. */
  116. $pattern = "/<param>([a-z]+?)=/";
  117. $replacement = '<param name="$1">';
  118. $tpl = preg_replace($pattern,$replacement,$tpl);
  119. //tpl to react
  120. $tpl = str_replace('<param','<span class="param"',$tpl);
  121. $tpl = str_replace('</param>','</span>',$tpl);
  122. $tpl = $this->xml2tpl($tpl,$channelId);
  123. $buffer[] = $tpl;
  124. }
  125. $remain = $arrWiki['data'][2];
  126. } while (!empty($remain));
  127. $html = implode('' , $buffer);
  128. return $html;
  129. }
  130. private function xmlQueryId(string $xml, string $id):string{
  131. try{
  132. $dom = simplexml_load_string($xml);
  133. }catch(\Exception $e){
  134. Log::error($e);
  135. return "<div></div>";
  136. }
  137. $tpl_list = $dom->xpath('//MdTpl');
  138. foreach ($tpl_list as $key => $tpl) {
  139. foreach ($tpl->children() as $param) {
  140. # 处理每个参数
  141. if($param->getName() === "param"){
  142. foreach($param->attributes() as $pa => $pa_value){
  143. $pValue = $pa_value->__toString();
  144. if($pa === "name" && $pValue === "id"){
  145. if($param->__toString() === $id){
  146. return $tpl->asXML();
  147. }
  148. }
  149. }
  150. }
  151. }
  152. }
  153. return "<div></div>";
  154. }
  155. public static function take_sentence(string $xml):array{
  156. $output = [];
  157. try{
  158. $dom = simplexml_load_string($xml);
  159. }catch(\Exception $e){
  160. Log::error($e);
  161. return $output;
  162. }
  163. $tpl_list = $dom->xpath('//MdTpl');
  164. foreach ($tpl_list as $key => $tpl) {
  165. foreach($tpl->attributes() as $a => $a_value){
  166. if($a==="name"){
  167. if($a_value->__toString() ==="sent"){
  168. foreach ($tpl->children() as $param) {
  169. # 处理每个参数
  170. if($param->getName() === "param"){
  171. $sent = $param->__toString();
  172. if(!empty($sent)){
  173. $output[] = $sent;
  174. break;
  175. }
  176. }
  177. }
  178. }
  179. }
  180. }
  181. }
  182. return $output;
  183. }
  184. private function xml2tpl(string $xml, $channelId=[]):string{
  185. /**
  186. * 解析xml
  187. * 获取模版参数
  188. * 生成react 组件参数
  189. */
  190. try{
  191. //$dom = simplexml_load_string($xml);
  192. $doc = new \DOMDocument();
  193. $xml = str_replace('MdTpl','dfn',$xml);
  194. $xml = mb_convert_encoding($xml, 'HTML-ENTITIES', "UTF-8");
  195. $ok = $doc->loadHTML($xml,LIBXML_NOERROR | LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
  196. }catch(\Exception $e){
  197. Log::error($e);
  198. Log::error($xml);
  199. return "<span>xml解析错误{$e}</span>";
  200. }
  201. if(!$ok){
  202. return "<span>xml解析错误</span>";
  203. }
  204. /*
  205. if(!$dom){
  206. Log::error($xml);
  207. return "<span>xml解析错误</span>";
  208. }
  209. */
  210. //$tpl_list = $dom->xpath('//MdTpl');
  211. $tpl_list = $doc->getElementsByTagName('dfn');
  212. foreach ($tpl_list as $key => $tpl) {
  213. /**
  214. * 遍历 MdTpl 处理参数
  215. */
  216. $props = [];
  217. $tpl_name = '';
  218. foreach($tpl->attributes as $a => $a_value){
  219. if($a_value->nodeName==="name"){
  220. $tpl_name = $a_value->nodeValue;
  221. break;
  222. }
  223. }
  224. $param_id = 0;
  225. $child = $tpl->firstChild;
  226. while ($child) {
  227. # 处理每个参数
  228. if($child->nodeName === "span"){
  229. $param_id++;
  230. $paramName = "";
  231. foreach($child->attributes as $pa => $pa_value){
  232. if($pa_value->nodeName === "name"){
  233. $nodeText = $pa_value->nodeValue;
  234. $props["{$nodeText}"] = $child->nodeValue;
  235. $paramName = $pa_value;
  236. }
  237. }
  238. if(empty($paramName)){
  239. foreach ($child->childNodes as $param_child) {
  240. # code...
  241. if($param_child->nodeType ===3){
  242. $props["{$param_id}"] = $param_child->nodeValue;
  243. }
  244. }
  245. }
  246. }
  247. $child = $child->nextSibling;
  248. }
  249. /**
  250. * 生成模版参数
  251. *
  252. */
  253. //TODO 判断$channelId里面的是否都是uuid
  254. $channelInfo = Channel::whereIn('uid',$channelId)->get();
  255. $tplRender = new TemplateRender($props,
  256. $channelInfo,
  257. $this->options['mode'],
  258. $this->options['format'],
  259. $this->options['studioId'],
  260. $this->options['debug']);
  261. $tplProps = $tplRender->render($tpl_name);
  262. if($this->options['format']==='react' && $tplProps){
  263. $props = $doc->createAttribute("props");
  264. $props->nodeValue = $tplProps['props'];
  265. $tpl->appendChild($props);
  266. $attTpl = $doc->createAttribute("tpl");
  267. $attTpl->nodeValue = $tplProps['tpl'];
  268. $tpl->appendChild($attTpl);
  269. $htmlElement = $doc->createElement($tplProps['tag']);
  270. $htmlElement->nodeValue=$tplProps['html'];
  271. $tpl->appendChild($htmlElement);
  272. }
  273. }
  274. $html = $doc->saveHTML();
  275. $html = str_replace(['<dfn','</dfn>'],['<MdTpl','</MdTpl>'],$html);
  276. switch ($this->options['format']) {
  277. case 'react':
  278. return trim($html);
  279. break;
  280. case 'unity':
  281. if($tplProps){
  282. return "{{"."{$tplProps['tpl']}|{$tplProps['props']}"."}}";
  283. }else{
  284. return '';
  285. }
  286. break;
  287. case 'html':
  288. if(isset($tplProps)){
  289. if(is_array($tplProps)){
  290. return '';
  291. }else{
  292. return $tplProps;
  293. }
  294. }else{
  295. Log::error('tplProps undefine');
  296. return '';
  297. }
  298. break;
  299. case 'text':
  300. case 'simple':
  301. if(isset($tplProps)){
  302. if(is_array($tplProps)){
  303. return '';
  304. }else{
  305. return $tplProps;
  306. }
  307. }else{
  308. Log::error('tplProps undefine');
  309. return '';
  310. }
  311. break;
  312. case 'tex':
  313. if(isset($tplProps)){
  314. if(is_array($tplProps)){
  315. return '';
  316. }else{
  317. return $tplProps;
  318. }
  319. }else{
  320. Log::error('tplProps undefine');
  321. return '';
  322. }
  323. break;
  324. default:
  325. return '';
  326. break;
  327. }
  328. }
  329. private function markdown2wiki(string $markdown): string{
  330. //$markdown = mb_convert_encoding($markdown,'UTF-8','UTF-8');
  331. $markdown = iconv('UTF-8','UTF-8//IGNORE',$markdown);
  332. /**
  333. * nissaya
  334. * aaa=bbb\n
  335. * {{nissaya|aaa|bbb}}
  336. */
  337. if($this->options['channelType']==='nissaya'){
  338. if($this->options['contentType'] === "json"){
  339. $json = json_decode($markdown);
  340. $nissayaWord = [];
  341. foreach ($json as $word) {
  342. if(count($word->sn) === 1){
  343. //只输出第一层级
  344. $str = "{{nissaya|";
  345. if(isset($word->word->value)){
  346. $str .= $word->word->value;
  347. }
  348. $str .= "|";
  349. if(isset($word->meaning->value)){
  350. $str .= $word->meaning->value;
  351. }
  352. $str .= "}}";
  353. $nissayaWord[] = $str;
  354. }
  355. }
  356. $markdown = implode('',$nissayaWord);
  357. }else if($this->options['contentType'] === "markdown"){
  358. $lines = explode("\n",$markdown);
  359. $newLines = array();
  360. foreach ($lines as $line) {
  361. if(strstr($line,'=') === FALSE){
  362. $newLines[] = $line;
  363. }else{
  364. $nissaya = explode('=',$line);
  365. $meaning = array_slice($nissaya,1);
  366. $meaning = implode('=',$meaning);
  367. $newLines[] = "{{nissaya|{$nissaya[0]}|{$meaning}}}";
  368. }
  369. }
  370. $markdown = implode("\n",$newLines);
  371. }
  372. }
  373. //$markdown = preg_replace("/\n\n/","<div></div>",$markdown);
  374. /**
  375. * 处理 mermaid
  376. */
  377. if(strpos($markdown,"```mermaid") !== false){
  378. $lines = explode("\n",$markdown);
  379. $newLines = array();
  380. $mermaidBegin = false;
  381. $mermaidString = array();
  382. foreach ($lines as $line) {
  383. if($line === "```mermaid"){
  384. $mermaidBegin = true;
  385. $mermaidString = [];
  386. continue;
  387. }
  388. if($mermaidBegin){
  389. if($line === "```"){
  390. $newLines[] = "{{mermaid|".base64_encode(\json_encode($mermaidString))."}}";
  391. $mermaidBegin = false;
  392. }else{
  393. $mermaidString[] = $line;
  394. }
  395. }else{
  396. $newLines[] = $line;
  397. }
  398. }
  399. $markdown = implode("\n",$newLines);
  400. }
  401. /**
  402. * 替换换行符
  403. * react 无法处理 <br> 替换为<div></div>代替换行符作用
  404. */
  405. //$markdown = str_replace('<br>','<div></div>',$markdown);
  406. /**
  407. * markdown -> html
  408. */
  409. /*
  410. $html = MdRender::fixHtml($html);
  411. */
  412. #替换术语
  413. $pattern = "/\[\[(.+?)\]\]/";
  414. $replacement = '{{term|$1}}';
  415. $markdown = preg_replace($pattern,$replacement,$markdown);
  416. #替换句子模版
  417. $pattern = "/\{\{([0-9].+?)\}\}/";
  418. $replacement = '{{sent|$1}}';
  419. $markdown = preg_replace($pattern,$replacement,$markdown);
  420. /**
  421. * 替换多行注释
  422. * ```
  423. * bla
  424. * bla
  425. * ```
  426. * {{note|
  427. * bla
  428. * bla
  429. * }}
  430. */
  431. if(strpos($markdown,"```\n") !== false){
  432. $lines = explode("\n",$markdown);
  433. $newLines = array();
  434. $noteBegin = false;
  435. $noteString = array();
  436. foreach ($lines as $line) {
  437. if($noteBegin){
  438. if($line === "```"){
  439. $newLines[] = "}}";
  440. $noteBegin = false;
  441. }else{
  442. $newLines[] = $line;
  443. }
  444. }else{
  445. if($line === "```"){
  446. $noteBegin = true;
  447. $newLines[] = "{{note|";
  448. continue;
  449. }else{
  450. $newLines[] = $line;
  451. }
  452. }
  453. }
  454. if($noteBegin){
  455. $newLines[] = "}}";
  456. }
  457. $markdown = implode("\n",$newLines);
  458. }
  459. /**
  460. * 替换单行注释
  461. * `bla bla`
  462. * {{note|bla}}
  463. */
  464. $pattern = "/`(.+?)`/";
  465. $replacement = '{{note|$1}}';
  466. $markdown = preg_replace($pattern,$replacement,$markdown);
  467. return $markdown;
  468. }
  469. private function markdownToHtml($markdown){
  470. $markdown = str_replace('MdTpl','mdtpl',$markdown);
  471. $markdown = str_replace(['<param','</param>'],['<span','</span>'],$markdown);
  472. $html = Markdown::render($markdown);
  473. if($this->options['format']==='react'){
  474. $html = $this->fixHtml($html);
  475. }
  476. $html = str_replace('<hr>','<hr />',$html);
  477. //给H1-6 添加uuid
  478. for ($i=1; $i<7 ; $i++) {
  479. if(strpos($html,"<h{$i}>")===false){
  480. continue;
  481. }
  482. $output = array();
  483. $input = $html;
  484. $hPos = strpos($input,"<h{$i}>");
  485. while ($hPos !== false) {
  486. $output[] = substr($input,0,$hPos);
  487. $output[] = "<h{$i} id='".Str::uuid()."'>";
  488. $input = substr($input,$hPos+4);
  489. $hPos = strpos($input,"<h{$i}>");
  490. }
  491. $output[] = $input;
  492. $html = implode('',$output);
  493. }
  494. $html = str_replace('mdtpl','MdTpl',$html);
  495. return $html;
  496. }
  497. private function fixHtml($html) {
  498. $doc = new \DOMDocument();
  499. libxml_use_internal_errors(true);
  500. $html = mb_convert_encoding($html, 'HTML-ENTITIES', "UTF-8");
  501. $doc->loadHTML('<span>'.$html.'</span>',LIBXML_NOERROR | LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
  502. $fixed = $doc->saveHTML();
  503. $fixed = mb_convert_encoding($fixed, "UTF-8", 'HTML-ENTITIES');
  504. return $fixed;
  505. }
  506. public static function init(){
  507. $GLOBALS["MdRenderStack"] = 0;
  508. }
  509. public function convert($markdown,$channelId=[],$queryId=null){
  510. if(isset($GLOBALS["MdRenderStack"]) && is_numeric($GLOBALS["MdRenderStack"])){
  511. $GLOBALS["MdRenderStack"]++;
  512. }else{
  513. $GLOBALS["MdRenderStack"] = 1;
  514. }
  515. if($GLOBALS["MdRenderStack"]<3){
  516. $output = $this->_convert($markdown,$channelId,$queryId);
  517. }else{
  518. $output = $markdown;
  519. }
  520. $GLOBALS["MdRenderStack"]--;
  521. return $output;
  522. }
  523. private function _convert($markdown,$channelId=[],$queryId=null){
  524. if(empty($markdown)){
  525. switch ($this->options['format']) {
  526. case 'react':
  527. return "<span></span>";
  528. break;
  529. default:
  530. return "";
  531. break;
  532. }
  533. }
  534. $wiki = $this->markdown2wiki($markdown);
  535. $html = $this->wiki2xml($wiki,$channelId);
  536. if(!is_null($queryId)){
  537. $html = $this->xmlQueryId($html, $queryId);
  538. }
  539. $html = $this->markdownToHtml($html);
  540. //后期处理
  541. $output = '';
  542. switch ($this->options['format']) {
  543. case 'react':
  544. //生成可展开组件
  545. $html = str_replace("<div/>","<div></div>",$html);
  546. $pattern = '/<li><div>(.+?)<\/div><\/li>/';
  547. $replacement = '<li><MdTpl name="toggle" tpl="toggle" props=""><div>$1</div></MdTpl></li>';
  548. $output = preg_replace($pattern,$replacement,$html);
  549. break;
  550. case 'text':
  551. case 'simple':
  552. $html = strip_tags($html);
  553. $output = htmlspecialchars_decode($html,ENT_QUOTES);
  554. //$output = html_entity_decode($html);
  555. break;
  556. case 'tex':
  557. $html = strip_tags($html);
  558. $output = htmlspecialchars_decode($html,ENT_QUOTES);
  559. //$output = html_entity_decode($html);
  560. break;
  561. case 'unity':
  562. $html = str_replace(['<strong>','</strong>','<em>','</em>'],['[%b%]','[%/b%]','[%i%]','[%/i%]'],$html);
  563. $html = strip_tags($html);
  564. $html = str_replace(['[%b%]','[%/b%]','[%i%]','[%/i%]'],['<b>','</b>','<i>','</i>'],$html);
  565. $output = htmlspecialchars_decode($html,ENT_QUOTES);
  566. break;
  567. case 'html':
  568. $output = htmlspecialchars_decode($html,ENT_QUOTES);
  569. break;
  570. }
  571. return $output;
  572. }
  573. /**
  574. * string[] $channelId
  575. */
  576. public static function render($markdown,$channelId,$queryId=null,$mode='read',$channelType='translation',$contentType="markdown",$format='react'){
  577. $mdRender = new MdRender(
  578. [
  579. 'mode'=>$mode,
  580. 'channelType'=>$channelType,
  581. 'contentType'=>$contentType,
  582. 'format'=>$format
  583. ]);
  584. $output = $mdRender->convert($markdown,$channelId,$queryId);
  585. return $output;
  586. }
  587. }