Add tesseract and foreign object rendering (opensearch-project#86) (o…

…pensearch-project#87) Signed-off-by: Joshua Li <[email protected]> (cherry picked from commit 503ee051dd909a970aaf0864114624d805a1a2cf) Co-authored-by: Joshua Li <[email protected]>
kavilla · Apr 11, 2023 · b746a0e · b746a0e
1 parent dd7221e
commit b746a0e
Show file tree

Hide file tree

Showing 11 changed files with 543 additions and 70 deletions.
diff --git a/NOTICE.txt b/NOTICE.txt
@@ -1,2 +1,5 @@
 OpenSearch (https://opensearch.org/)
 Copyright OpenSearch Contributors
+
+This product includes software developed by
+naptha (https://github.com/naptha/tesseract.js/)
diff --git a/common/tesseract/.gitignore b/common/tesseract/.gitignore
@@ -0,0 +1,2 @@
+*
+!.gitignore
diff --git a/package.json b/package.json
@@ -14,7 +14,7 @@
     "cypress:run": "cypress run",
     "cypress:open": "cypress open",
     "plugin-helpers": "node ../../scripts/plugin_helpers",
-    "postinstall": "node ./scripts/patch-html2canvas.js"
+    "postinstall": "node ./scripts/postinstall.js"
   },
   "dependencies": {
     "babel-polyfill": "^6.26.0",
@@ -38,7 +38,8 @@
     "react-router-dom": "^5.3.0",
     "react-toast-notifications": "^2.4.0",
     "set-interval-async": "1.0.33",
-    "showdown": "^1.9.1"
+    "showdown": "^1.9.1",
+    "tesseract.js": "^4.0.2"
   },
   "devDependencies": {
     "@elastic/eslint-import-resolver-kibana": "link:../../packages/osd-eslint-import-resolver-opensearch-dashboards",

diff --git a/public/components/context_menu/context_menu_ui.js b/public/components/context_menu/context_menu_ui.js
diff --git a/public/components/visual_report/assets/report_styles.ts b/public/components/visual_report/assets/report_styles.ts
@@ -8,6 +8,7 @@ html,
 body {
   margin: 0;
   padding: 0;
+  padding-top: 0px;
 }
 
 iframe, embed, object {

diff --git a/public/components/visual_report/generate_report.ts b/public/components/visual_report/generate_report.ts
@@ -6,6 +6,7 @@
 import createDOMPurify from 'dompurify';
 import html2canvas from 'html2canvas';
 import jsPDF from 'jspdf';
+import { createWorker } from 'tesseract.js';
 import { v1 as uuidv1 } from 'uuid';
 import { ReportSchemaType } from '../../../server/model';
 import { uiSettingsService } from '../utils/settings_service';
@@ -57,15 +58,16 @@ const removeNonReportElements = (
   reportSource: VISUAL_REPORT_TYPE
 ) => {
   // remove buttons
-  doc.querySelectorAll("button[class^='euiButton']:not(.visLegend__button)").forEach((e) => e.remove());
+  doc
+    .querySelectorAll("button[class^='euiButton']:not(.visLegend__button)")
+    .forEach((e) => e.remove());
   // remove top navBar
   doc.querySelectorAll("[class^='euiHeader']").forEach((e) => e.remove());
   // remove visualization editor
   if (reportSource === VISUAL_REPORT_TYPE.visualization) {
     doc.querySelector('[data-test-subj="splitPanelResizer"]')?.remove();
     doc.querySelector('.visEditor__collapsibleSidebar')?.remove();
   }
-  doc.body.style.paddingTop = '0px';
 };
 
 const addReportHeader = (doc: Document, header: string) => {
@@ -96,8 +98,10 @@ const addReportFooter = (doc: Document, footer: string) => {
 
 const addReportStyle = (doc: Document, style: string) => {
   const styleElement = document.createElement('style');
+  styleElement.className = 'reportInjectedStyles';
   styleElement.innerHTML = style;
   doc.getElementsByTagName('head')[0].appendChild(styleElement);
+  doc.body.style.paddingTop = '0px';
 };
 
 const computeHeight = (height: number, header: string, footer: string) => {
@@ -115,6 +119,7 @@ const computeHeight = (height: number, header: string, footer: string) => {
 
 export const generateReport = async (id: string, forceDelay = 15000) => {
   const http = uiSettingsService.getHttpClient();
+  const useForeignObjectRendering = uiSettingsService.get('reporting:useFOR');
   const DOMPurify = createDOMPurify(window);
 
   const report = await http.get<ReportSchemaType>(
@@ -154,6 +159,26 @@ export const generateReport = async (id: string, forceDelay = 15000) => {
   }
   await timeout(forceDelay);
 
+  // Style changes onclone does not work with foreign object rendering enabled.
+  // Additionally increase span width to prevent text being truncated
+  if (useForeignObjectRendering) {
+    document
+      .querySelectorAll<HTMLSpanElement>('span:not([data-html2canvas-ignore])')
+      .forEach((el) => {
+        if (!el.closest('.globalFilterItem'))
+          el.style.width = el.offsetWidth + 30 + 'px';
+      });
+    document
+      .querySelectorAll<HTMLSpanElement>(
+        'span.globalFilterItem:not([data-html2canvas-ignore])'
+      )
+      .forEach((el) => (el.style.width = el.offsetWidth + 5 + 'px'));
+    addReportHeader(document, header);
+    addReportFooter(document, footer);
+    addReportStyle(document, reportingStyle);
+    await timeout(1000);
+  }
+
   const width = document.documentElement.scrollWidth;
   const height = computeHeight(
     document.documentElement.scrollHeight,
@@ -170,40 +195,87 @@ export const generateReport = async (id: string, forceDelay = 15000) => {
     imageTimeout: 30000,
     useCORS: true,
     removeContainer: false,
+    allowTaint: true,
+    foreignObjectRendering: useForeignObjectRendering,
     onclone: function (documentClone) {
       removeNonReportElements(documentClone, reportSource);
-      addReportHeader(documentClone, header);
-      addReportFooter(documentClone, footer);
-      addReportStyle(documentClone, reportingStyle);
+      if (!useForeignObjectRendering) {
+        addReportHeader(documentClone, header);
+        addReportFooter(documentClone, footer);
+        addReportStyle(documentClone, reportingStyle);
+      }
     },
-  }).then(function (canvas) {
-    // TODO remove this and 'removeContainer: false' when https://github.com/niklasvh/html2canvas/pull/2949 is merged
-    document
-      .querySelectorAll<HTMLIFrameElement>('.html2canvas-container')
-      .forEach((e) => {
-        const iframe = e.contentWindow;
-        if (e) {
-          e.src = 'about:blank';
-          if (iframe) {
-            iframe.document.write('');
-            iframe.document.clear();
-            iframe.close();
+  })
+    .then(async function (canvas) {
+      // TODO remove this and 'removeContainer: false' when https://github.com/niklasvh/html2canvas/pull/2949 is merged
+      document
+        .querySelectorAll<HTMLIFrameElement>('.html2canvas-container')
+        .forEach((e) => {
+          const iframe = e.contentWindow;
+          if (e) {
+            e.src = 'about:blank';
+            if (iframe) {
+              iframe.document.write('');
+              iframe.document.clear();
+              iframe.close();
+            }
+            e.remove();
           }
-          e.remove();
-        }
-      });
+        });
 
-    if (format === 'png') {
-      const link = document.createElement('a');
-      link.download = fileName;
-      link.href = canvas.toDataURL();
-      link.click();
-    } else {
-      const orient = canvas.width > canvas.height ? 'landscape' : 'portrait';
-      const pdf = new jsPDF(orient, 'px', [canvas.width, canvas.height]);
-      pdf.addImage(canvas, 'JPEG', 0, 0, canvas.width, canvas.height);
-      pdf.save(fileName);
-    }
-    return true;
-  });
+      if (format === 'png') {
+        const link = document.createElement('a');
+        link.download = fileName;
+        link.href = canvas.toDataURL();
+        link.click();
+      } else if (uiSettingsService.get('reporting:useOcr')) {
+        const worker = await createWorker({
+          workerPath: '../api/reporting/tesseract.js/dist/worker.min.js',
+          langPath: '../api/reporting/tesseract-lang-data',
+          corePath: '../api/reporting/tesseract.js-core/tesseract-core.wasm.js',
+        });
+        await worker.loadLanguage('eng');
+        await worker.initialize('eng');
+        const {
+          data: { text, pdf },
+        } = await worker
+          .recognize(canvas.toDataURL(), { pdfTitle: fileName }, { pdf: true })
+          .catch((e) => console.error('recognize', e));
+        await worker.terminate();
+
+        const blob = new Blob([new Uint8Array(pdf)], {
+          type: 'application/pdf',
+        });
+        const link = document.createElement('a');
+        if (link.download !== undefined) {
+          const url = URL.createObjectURL(blob);
+          link.setAttribute('href', url);
+          link.setAttribute('download', fileName);
+          link.style.visibility = 'hidden';
+          document.body.appendChild(link);
+          link.click();
+          document.body.removeChild(link);
+        }
+      } else {
+        const orient = canvas.width > canvas.height ? 'landscape' : 'portrait';
+        const pdf = new jsPDF(orient, 'px', [canvas.width, canvas.height]);
+        pdf.addImage(canvas, 'JPEG', 0, 0, canvas.width, canvas.height);
+        pdf.save(fileName);
+      }
+      return true;
+    })
+    .finally(() => {
+      if (useForeignObjectRendering) {
+        document
+          .querySelectorAll<HTMLSpanElement>(
+            'span:not(.data-html2canvas-ignore)'
+          )
+          .forEach((el) => (el.style.width = ''));
+        document.querySelectorAll('.reportWrapper').forEach((e) => e.remove());
+        document
+          .querySelectorAll('.reportInjectedStyles')
+          .forEach((e) => e.remove());
+        document.body.style.paddingTop = '';
+      }
+    });
 };
diff --git a/scripts/patch-html2canvas.js → scripts/postinstall.js b/scripts/patch-html2canvas.js → scripts/postinstall.js
@@ -5,6 +5,8 @@
 
 // @ts-check
 // workaround for Safari support before https://github.com/niklasvh/html2canvas/pull/2911 is merged
+const https = require('https');
+const fs = require('fs');
 const replace = require('replace-in-file');
 
 const options = {
@@ -31,3 +33,16 @@ try {
     error
   );
 }
+
+// download tesseract model
+const modelFile = fs.createWriteStream(__dirname + '/../common/tesseract/eng.traineddata.gz');
+https.get(
+  'https://raw.githubusercontent.com/naptha/tessdata/gh-pages/4.0.0_best/eng.traineddata.gz',
+  function (response) {
+    response.pipe(modelFile);
+    modelFile.on('finish', () => {
+      modelFile.close();
+      console.log('Downloaded eng.traineddata.gz for tesseract.js');
+    });
+  }
+);
diff --git a/server/plugin.ts b/server/plugin.ts
@@ -3,23 +3,24 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
+import { schema } from '@osd/config-schema';
 import {
-  PluginInitializerContext,
   CoreSetup,
   CoreStart,
-  Plugin,
-  Logger,
   ILegacyClusterClient,
+  Logger,
+  Plugin,
+  PluginInitializerContext,
 } from '../../../src/core/server';
 import opensearchReportsPlugin from './backend/opensearch-reports-plugin';
+import { NotificationsPlugin } from './clusters/notificationsPlugin';
+import { buildConfig, ReportingConfigType } from './config';
+import { ReportingConfig } from './config/config';
+import registerRoutes from './routes';
 import {
   ReportsDashboardsPluginSetup,
   ReportsDashboardsPluginStart,
 } from './types';
-import registerRoutes from './routes';
-import { NotificationsPlugin } from './clusters/notificationsPlugin';
-import { buildConfig, ReportingConfigType } from './config';
-import { ReportingConfig } from './config/config';
 
 export interface ReportsPluginRequestContext {
   logger: Logger;
@@ -49,6 +50,23 @@ export class ReportsDashboardsPlugin
   public async setup(core: CoreSetup) {
     this.logger.debug('reports-dashboards: Setup');
 
+    core.uiSettings.register({
+      'reporting:useOcr': {
+        name: 'Reporting use OCR on PDF',
+        value: false,
+        description:
+          'Whether to run optical character recognition on PDF reports to make text selectable',
+        schema: schema.boolean(),
+      },
+      'reporting:useFOR': {
+        name: 'Reporting use ForeignObject rendering',
+        value: true,
+        description:
+          'Whether to use ForeignObject rendering when generating reports. If it causes issues, try disabling this option.',
+        schema: schema.boolean(),
+      },
+    });
+
     try {
       const config = await buildConfig(
         this.initializerContext,

diff --git a/server/routes/index.ts b/server/routes/index.ts
@@ -8,6 +8,7 @@ import registerReportDefinitionRoute from './reportDefinition';
 import registerReportSourceRoute from './reportSource';
 import registerMetricRoute from './metric';
 import registerNotificationRoute from './notifications';
+import registerTesseractRoute from './tesseract';
 import { IRouter } from '../../../../src/core/server';
 import { ReportingConfig } from 'server/config/config';
 
@@ -17,4 +18,5 @@ export default function (router: IRouter, config: ReportingConfig) {
   registerReportSourceRoute(router);
   registerMetricRoute(router);
   registerNotificationRoute(router);
+  registerTesseractRoute(router);
 }