src/Controller/Api/OnlyOfficeController.php line 71

Open in your IDE?
  1. <?php
  3. namespace App\Controller\Api;
  5. use App\Service\OnlyOfficeDocBuilderService;
  6. use App\Service\HtmlToDocBuilderConverter;
  7. use Symfony\Bundle\FrameworkBundle\Controller\AbstractController;
  8. use Symfony\Component\HttpFoundation\Request;
  9. use Symfony\Component\HttpFoundation\Response;
  10. use Symfony\Component\Routing\Annotation\Route;
  11. use Psr\Log\LoggerInterface;
  13. /**
  14. * @Route("/")
  15. */
  16. class OnlyOfficeController extends AbstractController
  17. {
  18. private OnlyOfficeDocBuilderService $docBuilder;
  19. private HtmlToDocBuilderConverter $htmlConverter;
  20. private LoggerInterface $logger;
  22. public function __construct(
  23. OnlyOfficeDocBuilderService $docBuilder,
  24. HtmlToDocBuilderConverter $htmlConverter,
  25. LoggerInterface $logger
  26. ) {
  27. $this->docBuilder = $docBuilder;
  28. $this->htmlConverter = $htmlConverter;
  29. $this->logger = $logger;
  30. }
  32. /**
  33. * Convert document to given format
  34. * @Route("/convert-to", name="api_onlyoffice_convert_to", methods={"POST"})
  35. */
  36. public function convertTo(Request $request): Response
  37. {
  38. $data = json_decode($request->getContent(), true);
  40. if (!isset($data['document_url'])) {
  41. return $this->json(['error' => 'document_url is required'], Response::HTTP_BAD_REQUEST);
  42. }
  44. if (!isset($data['output_format'])) {
  45. return $this->json(['error' => 'output_format is required'], Response::HTTP_BAD_REQUEST);
  46. }
  48. try {
  49. $convertedFile = $this->processDocument($data['document_url'], $data['output_format'], [
  50. 'type' => 'convert',
  51. ]);
  53. return new Response($convertedFile);
  54. } catch (\Exception $e) {
  55. return $this->json(['error' => $e->getMessage()], Response::HTTP_INTERNAL_SERVER_ERROR);
  56. }
  57. }
  59. public function escapeDocBuilderString(string $value): string {
  60. return str_replace(
  61. ["\\", "\r\n", "\r", "\n", "'"],
  62. ["\\\\", "\\n", "\\n", "\\n", "\\'"],
  63. $value
  64. );
  65. }
  67. /**
  68. * Convert document to PDF
  69. * @Route("/convert-to-pdf", name="api_onlyoffice_convert_to_pdf", methods={"POST"})
  70. */
  71. public function convertToPdf(Request $request): Response
  72. {
  73. $data = json_decode($request->getContent(), true);
  75. if (!isset($data['document_url'])) {
  76. return $this->json(['error' => 'document_url is required'], Response::HTTP_BAD_REQUEST);
  77. }
  79. try {
  80. $convertedFile = $this->processDocument($data['document_url'], 'pdf', [
  81. 'type' => 'convert',
  82. ]);
  84. return new Response($convertedFile);
  85. } catch (\Exception $e) {
  86. return $this->json(['error' => $e->getMessage()], Response::HTTP_INTERNAL_SERVER_ERROR);
  87. }
  88. }
  90. /**
  91. * Replace labels with text in document
  92. * @Route("/replace-labels-with-text", name="api_onlyoffice_replace_labels_with_text", methods={"POST"})
  93. */
  94. public function replaceLabelsWithText(Request $request): Response
  95. {
  96. $data = json_decode($request->getContent(), true);
  98. if (!isset($data['document_url']) || !isset($data['labels'])) {
  99. return $this->json(
  100. ['error' => 'document_url and labels are required'],
  101. Response::HTTP_BAD_REQUEST
  102. );
  103. }
  105. try {
  106. $resultFile = $this->processDocument($data['document_url'], 'docx', [
  107. 'type' => 'replaceText',
  108. 'labels' => $data['labels']
  109. ]);
  111. return new Response($resultFile);
  112. } catch (\Exception $e) {
  113. return $this->json(['error' => $e->getMessage()], Response::HTTP_INTERNAL_SERVER_ERROR);
  114. }
  115. }
  117. /**
  118. * Replace labels with HTML content in document.
  119. * Unlike replaceLabelsWithText, this method interprets HTML content and generates
  120. * structured document elements (paragraphs, headings, tables, lists, formatted text, etc.)
  121. *
  122. * Supported HTML elements:
  123. * - Block elements: <p>, <h1>, <h2>, <h3>, <div>
  124. * - Text formatting: <strong>, <b>, <em>, <i>, <u>
  125. * - Lists: <ul>, <ol>, <li>
  126. * - Tables: <table>, <tr>, <td>, <th>, <thead>, <tbody>
  127. * - Media: <img> (with src, width, height attributes)
  128. * - Line breaks: <br>
  129. *
  130. * @Route("/replace-labels-html-with-text", name="api_onlyoffice_replace_labels_html_with_text", methods={"POST"})
  131. */
  132. public function replaceLabelsHtmlWithText(Request $request): Response
  133. {
  134. $data = json_decode($request->getContent(), true);
  136. if (!isset($data['document_url']) || !isset($data['labels'])) {
  137. return $this->json(
  138. ['error' => 'document_url and labels are required'],
  139. Response::HTTP_BAD_REQUEST
  140. );
  141. }
  143. try {
  144. $processedLabels = [];
  145. foreach ($data['labels'] as $labelKey => $labelValue) {
  146. $labelValue = $this->normalizeLabelValue($labelValue);
  147. $processedLabels[$labelKey] = [
  148. 'content' => $labelValue,
  149. 'docbuilder' => $this->htmlConverter->htmlToDocBuilder($labelValue)
  150. ];
  151. }
  153. $resultFile = $this->processDocument($data['document_url'], 'docx', [
  154. 'type' => 'replaceHtml',
  155. // 'labels' => $data['labels']
  156. 'labels' => $processedLabels,
  157. ]);
  159. return new Response($resultFile);
  160. } catch (\Exception $e) {
  161. return $this->json(['error' => $e->getMessage()], Response::HTTP_INTERNAL_SERVER_ERROR);
  162. }
  163. }
  165. /**
  166. * Replace labels with text in document
  167. * @Route("/replace-labels-with-image", name="api_onlyoffice_replace_labels_with_image", methods={"POST"})
  168. */
  169. public function replaceLabelsWithImage(Request $request): Response
  170. {
  171. $data = json_decode($request->getContent(), true);
  173. if (!isset($data['document_url']) || !isset($data['image'])) {
  174. return $this->json(
  175. ['error' => 'document_url and image are required'],
  176. Response::HTTP_BAD_REQUEST
  177. );
  178. }
  180. try {
  181. $resultFile = $this->processDocument($data['document_url'], 'docx', [
  182. 'type' => 'replaceImage',
  183. 'label' => $data['label'],
  184. 'image' => $data['image'],
  185. 'width' => intval($data['width'] ?? 220),
  186. 'height' => intval($data['height'] ?? 250)
  187. ]);
  189. return new Response($resultFile);
  190. } catch (\Exception $e) {
  191. return $this->json(['error' => $e->getMessage()], Response::HTTP_INTERNAL_SERVER_ERROR);
  192. }
  193. }
  195. /**
  196. * Add watermark to document
  197. * @Route("/add-watermark", name="api_onlyoffice_add_watermark", methods={"POST"})
  198. */
  199. public function addWatermark(Request $request): Response
  200. {
  201. $data = json_decode($request->getContent(), true);
  203. if (!isset($data['document_url']) || !isset($data['watermark'])) {
  204. return $this->json(
  205. ['error' => 'document_url and watermark are required'],
  206. Response::HTTP_BAD_REQUEST
  207. );
  208. }
  210. try {
  211. $resultFile = $this->processDocument($data['document_url'], 'docx', [
  212. 'type' => 'watermark',
  213. 'watermark' => $data['watermark'],
  214. 'watermark_type' => $data['watermark_type'] ?? 'text',
  215. ]);
  217. return new Response($resultFile);
  218. } catch (\Exception $e) {
  219. return $this->json(['error' => $e->getMessage()], Response::HTTP_INTERNAL_SERVER_ERROR);
  220. }
  221. }
  223. private function processDocument(string $documentUrl, $outputFormat, array $options): string
  224. {
  226. $outName = 'result_' . bin2hex(random_bytes(6)) . '.' . $outputFormat;
  228. // Script para OnlyOffice DocBuilder
  229. // $script = "builder.SetTmpFolder('DocBuilderTemp');\nbuilder.OpenFile('$documentUrl', '');\nvar doc = Api.GetDocument();\n";
  231. // Force language to avoid locale issues in generated documents (e.g. "1st", "2nd", etc. in English instead of "1º", "2º" in Spanish)
  232. $script = "builder.SetTmpFolder('DocBuilderTemp');\n
  233. builder.OpenFile('$documentUrl', '');\n
  234. var doc = Api.GetDocument();\n
  235. var LANG = 'es-ES';\n
  236. try { doc.GetDefaultTextPr().SetLanguage(LANG); } catch(e) {}\n
  237. try { doc.GetCore().SetLanguage(LANG); } catch(e) {}\n
  238. function applyLangToParagraphs(paragraphs, lang) {\n
  239. if (!paragraphs) return;\n
  240. for (var i = 0; i < paragraphs.length; i++) {\n
  241. var p = paragraphs[i];\n
  242. try {\n
  243. var pr = p.GetParaPr();\n
  244. if (pr && pr.GetTextPr) { pr.GetTextPr().SetLanguage(lang); }\n
  245. } catch(e) {}\n
  246. var count = 0;\n
  247. try { count = p.GetElementsCount(); } catch(e) {}\n
  248. for (var j = 0; j < count; j++) {\n
  249. try {\n
  250. var el = p.GetElement(j);\n
  251. if (el && el.SetLanguage) { el.SetLanguage(lang); }\n
  252. } catch(e) {}\n
  253. }\n
  254. }\n
  255. }\n
  256. applyLangToParagraphs(doc.GetAllParagraphs(), LANG);\n
  257. try {\n
  258. var sections = [doc.GetFinalSection()];\n
  259. for (var s = 0; s < sections.length; s++) {\n
  260. var sec = sections[s];\n
  261. var hdrTypes = ['default','even','first'];\n
  262. for (var h = 0; h < hdrTypes.length; h++) {\n
  263. try {\n
  264. var hdr = sec.GetHeader(hdrTypes[h], false);\n
  265. if (hdr) applyLangToParagraphs(hdr.GetAllParagraphs(), LANG);\n
  266. } catch(e) {}\n
  267. try {\n
  268. var ftr = sec.GetFooter(hdrTypes[h], false);\n
  269. if (ftr) applyLangToParagraphs(ftr.GetAllParagraphs(), LANG);\n
  270. } catch(e) {}\n
  271. }\n
  272. }\n
  273. } catch(e) {}\n";
  275. // Force Lang in ordinals list to avoid locale issues in generated documents (e.g. "1st", "2nd", etc. in English instead of "1º", "2º" in Spanish)
  276. // $script = "builder.SetTmpFolder('DocBuilderTemp');\n";
  277. // $script .= "builder.OpenFile('$documentUrl', '');\n";
  278. // $script .= "var doc = Api.GetDocument();\n";
  279. // $script .= "doc.SetLanguage('es-ES');\n";
  281. // $script .= "try {\n";
  282. // $script .= " var paragraphs = doc.GetAllParagraphs();\n";
  283. // $script .= " for (var i = 0; i < paragraphs.length; i++) {\n";
  284. // $script .= " var p = paragraphs[i];\n";
  285. // $script .= " var t = p.GetText();\n";
  286. // $script .= " }\n";
  287. // $script .= "} catch(e) {\n";
  288. // $script .= " builder.WriteLog(e.message);\n";
  289. // $script .= "}\n";
  291. // Test set double force by builder
  292. // $script .= "var oElements = doc.GetAllElements('paragraph');\n";
  293. // $script .= "for (var i = 0; i < oElements.length; i++) {\n";
  294. // $script .= " var oPar = oElements[i];\n";
  295. // $script .= " var oRuns = oPar.GetAllElements('run');\n";
  296. // $script .= " for (var j = 0; j < oRuns.length; j++) {\n";
  297. // $script .= " oRuns[j].SetLanguage('es-ES');\n";
  298. // $script .= " }\n";
  299. // $script .= "}\n";
  301. switch ($options['type']) {
  302. case 'replaceText':
  303. $labels = $options['labels'];
  304. foreach ($labels as $key => $value) {
  305. $value = $this->normalizeLabelValue($value);
  306. if (str_contains($value, "\n") || str_contains($value, "\r")) {
  307. $lines = preg_split('/\r\n|\r|\n/', $value);
  308. $script .= "var found = doc.Search('" . $this->escapeDocBuilderString($key) . "');\n";
  309. $script .= "if (found && found.length > 0) {\n";
  310. $script .= " for (var fi = found.length - 1; fi >= 0; fi--) {\n";
  311. $script .= " var _content = [];\n";
  312. foreach ($lines as $line) {
  313. $script .= " var _p = Api.CreateParagraph();\n";
  314. $script .= " _p.AddText('" . $this->escapeDocBuilderString($line) . "');\n";
  315. $script .= " _content.push(_p);\n";
  316. }
  317. $script .= " found[fi].Select();\n";
  318. $script .= " doc.InsertContent(_content);\n";
  319. $script .= " }\n";
  320. $script .= "}\n";
  321. } else {
  322. $script .= "doc.SearchAndReplace({searchString: '" . $this->escapeDocBuilderString($key) . "', replaceString: '" . $this->escapeDocBuilderString($value) . "', matchCase: false});\n";
  323. }
  324. }
  325. break;
  326. case 'replaceImage':
  327. $label = $options['label'];
  328. $image = $options['image'];
  329. $width = $options['width'];
  330. $height = $options['height'];
  331. $wEMU = intval(round($width * 3600));
  332. $hEMU = intval(round($height * 3600));
  333. $script .= "var found = doc.Search('" . addslashes($label) . "');\nif (found && found.length>0){ found[0].Select();\nvar p = Api.CreateParagraph();\nvar img = Api.CreateImage('" . addslashes($image) . "', $wEMU, $hEMU);\nimg.SetWrappingStyle('inline');\np.AddDrawing(img);\ndoc.InsertContent([p]);\n}\n";
  334. break;
  335. case 'watermark':
  336. $watermarkType = $options['watermark_type'] ?? 'text';
  337. if ($watermarkType === 'text') {
  338. $texto = $options['watermark'];
  339. $script .= "doc.InsertWatermark('$texto', true);\n";
  340. } else {
  341. $image = $options['watermark'];
  342. $script .= "var sec=doc.GetFinalSection();\nvar wTw=sec.GetPageWidth();\nvar hTw=sec.GetPageHeight();\nvar EMU_PER_TWIP=635;\nvar wEMU=wTw*EMU_PER_TWIP;\nvar hEMU=hTw*EMU_PER_TWIP;\ndoc.InsertWatermark('WATERMARK',false);\nvar ws=doc.GetWatermarkSettings();\nws.SetType('image');\nws.SetImageURL('" . addslashes($image) . "');\nws.SetImageSize(wEMU,hEMU);\nws.SetDirection('horizontal');\nws.SetOpacity(100);\ndoc.SetWatermarkSettings(ws);\n";
  343. }
  344. break;
  345. case 'replaceHtml':
  346. $labels = $options['labels'];
  347. foreach ($labels as $key => $value) {
  348. $content = $this->normalizeLabelValue($value['content']);
  349. $hasNewlines = str_contains($content, "\n") || str_contains($content, "\r");
  351. if ($this->shouldUseInlineReplacement($content)) {
  352. $inlineText = $this->htmlToInlineText($content);
  353. if (str_contains($inlineText, "\n") || str_contains($inlineText, "\r")) {
  354. $script .= "var found = doc.Search('" . $this->escapeDocBuilderString($key) . "');\n";
  355. $script .= "if (found && found.length > 0) {\n";
  356. $script .= " for (var fi = found.length - 1; fi >= 0; fi--) {\n";
  357. $script .= " try {\n";
  358. $script .= " found[fi].Delete();\n";
  359. $script .= " found[fi].AddText('" . $this->escapeDocBuilderString($inlineText) . "');\n";
  360. $script .= " } catch (e) {\n";
  361. $script .= " found[fi].Select();\n";
  362. $script .= " doc.SearchAndReplace({searchString: '" . $this->escapeDocBuilderString($key) . "', replaceString: '" . $this->escapeDocBuilderString($inlineText) . "', matchCase: false});\n";
  363. $script .= " }\n";
  364. $script .= " }\n";
  365. $script .= "}\n";
  366. } else {
  367. $script .= "doc.SearchAndReplace({searchString: '" . $this->escapeDocBuilderString($key) . "', replaceString: '" . $this->escapeDocBuilderString($inlineText) . "', matchCase: false});\n";
  368. }
  369. continue;
  370. }
  372. if (strip_tags($content) === $content && !$hasNewlines) {
  373. // $script .= "doc.SearchAndReplace({searchString: '" . addslashes($key) . "', replaceString: '" . addslashes($content) . "', matchCase: false});\n";
  374. $script .= "doc.SearchAndReplace({searchString: '" . $this->escapeDocBuilderString($key) . "', replaceString: '" . $this->escapeDocBuilderString($content) . "', matchCase: false});\n";
  375. } else {
  376. $script .= "var found = doc.Search('" . $this->escapeDocBuilderString($key) . "');\n";
  377. $script .= "if (found && found.length > 0) {\n";
  378. $script .= " for (var fi = found.length - 1; fi >= 0; fi--) {\n";
  379. $script .= " var range = found[fi];\n";
  380. $script .= " var _content = [];\n";
  381. $script .= " var _inlinePara = Api.CreateParagraph();\n";
  382. $script .= " var _inlineUsed = false;\n";
  383. $script .= " var __para = null;\n";
  384. $script .= " var __cell = null;\n";
  385. $script .= " try { __para = range.GetParagraph(0); } catch (e) {}\n";
  386. $script .= " if (!__para) { try { __para = range.GetParagraph(); } catch (e) {} }\n";
  387. $script .= " if (__para) { try { __cell = __para.GetParentTableCell(); } catch (e) {} }\n";
  388. $script .= " var oPara = null;\n";
  389. $script .= " if (__cell && __para) {\n";
  390. $script .= " oPara = __para;\n";
  391. $script .= " try { range.Delete(); } catch (e) {}\n";
  392. $script .= " } else {\n";
  393. $script .= " oPara = {\n";
  394. $script .= " AddElement: function(el) { _inlineUsed = true; _inlinePara.AddElement(el); },\n";
  395. $script .= " GetParentTableCell: function() {\n";
  396. $script .= " return {\n";
  397. $script .= " GetContent: function() {\n";
  398. $script .= " return { Push: function(el) { _content.push(el); } };\n";
  399. $script .= " }\n";
  400. $script .= " };\n";
  401. $script .= " }\n";
  402. $script .= " };\n";
  403. $script .= " }\n";
  404. $script .= " " . $value['docbuilder'] . "\n";
  405. $script .= " if (!__cell) {\n";
  406. $script .= " if (_inlineUsed) { _content.unshift(_inlinePara); }\n";
  407. $script .= " range.Select();\n";
  408. $script .= " if (_content.length > 0) { doc.InsertContent(_content); }\n";
  409. $script .= " }\n";
  410. $script .= " }\n";
  411. $script .= "}\n";
  412. }
  413. }
  414. break;
  415. default:
  416. break;
  417. }
  419. // Cerrar y guardar el documento
  420. $script .= "builder.SaveFile('$outputFormat', '$outName');\nbuilder.CloseFile();";
  422. $this->logger->debug('Generated DocBuilder script', [
  423. 'type' => $options['type'],
  424. 'script' => $script
  425. ]);
  427. try {
  428. // Build the document and upload to Google Cloud Storage
  429. $urlScript = $this->docBuilder->construirSubirDocumento($script);
  430. $doc = $this->docBuilder->enviarDocBuilder($urlScript, $outName, $outputFormat);
  432. return $doc['urls'][$outName];
  433. } catch (\Exception $e) {
  434. $this->logger->error('Error al convertir documento', [
  435. 'error' => $e->getMessage(),
  436. 'documentUrl' => $documentUrl,
  437. 'outputFormat' => $outputFormat,
  438. 'trace' => $e->getTraceAsString()
  439. ]);
  440. throw new \RuntimeException('Failed to convert document: ' . $e->getMessage());
  441. }
  442. }
  444. private function shouldUseInlineReplacement(string $content): bool
  445. {
  446. $lowerContent = mb_strtolower($content);
  448. // Block tags usually create new paragraphs/structures in DOCX and can break inline sentences.
  449. $hasBlockTags = (bool) preg_match('/<\s*(p|div|section|article|table|thead|tbody|tr|td|th|ul|ol|li|h1|h2|h3|h4|h5|h6)\b/', $lowerContent);
  450. if ($hasBlockTags) {
  451. return false;
  452. }
  454. // Inline-like content (including <br>) is safer via SearchAndReplace to preserve paragraph flow.
  455. return strip_tags($content) !== $content || str_contains($content, "\n") || str_contains($content, "\r");
  456. }
  458. private function htmlToInlineText(string $content): string
  459. {
  460. $textWithBreaks = preg_replace('/<\s*br\s*\/?>/i', "\n", $content) ?? $content;
  461. $text = strip_tags($textWithBreaks);
  462. $text = html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8');
  464. // Normalize whitespace but preserve line breaks coming from <br>.
  465. $text = str_replace(["\r\n", "\r"], "\n", $text);
  466. $text = preg_replace('/[^\S\n]+/u', ' ', $text) ?? $text;
  467. // Trim only horizontal whitespace around line breaks, without collapsing consecutive \n.
  468. $text = preg_replace('/\n[^\S\n]+/u', "\n", $text) ?? $text;
  469. $text = preg_replace('/[^\S\n]+\n/u', "\n", $text) ?? $text;
  471. return trim($text);
  472. }
  474. private function normalizeLabelValue(string $value): string
  475. {
  476. $decoded = html_entity_decode($value, ENT_QUOTES | ENT_HTML5, 'UTF-8');
  478. // Decodificar iterativamente hasta que no haya más entidades
  479. while ($decoded !== $value) {
  480. $value = $decoded;
  481. $decoded = html_entity_decode($value, ENT_QUOTES | ENT_HTML5, 'UTF-8');
  482. }
  484. return $decoded;
  485. }
  486. }