fix(core): update

package-lock.json

@@ -16,7 +16,8 @@
         "@pushrocks/smartunique": "^3.0.3",
         "@types/express": "^4.17.13",
         "express": "^4.17.2",
-        "pdf-merger-js": "^3.2.1"
+        "pdf-merger-js": "^3.2.1",
+        "pdf2json": "^2.0.0"
       },
       "devDependencies": {
         "@gitzone/tsbuild": "^2.1.28",
@@ -10978,6 +10979,33 @@
         "pdfjs": "^2.4.5"
       }
     },
+    "node_modules/pdf2json": {
+      "version": "2.0.0",
+      "resolved": "https://verdaccio.lossless.one/pdf2json/-/pdf2json-2.0.0.tgz",
+      "integrity": "sha512-+FZy7GSvLOLc+zksg0SoMvXqIqcku5lBlEPuYJJkhMWB2x6yfthzEhhSbZc20UheClMPagH/+NXnMRbvQMQR1w==",
+      "bundleDependencies": [
+        "@xmldom/xmldom"
+      ],
+      "license": "Apache-2.0",
+      "dependencies": {
+        "@xmldom/xmldom": "^0.7.5"
+      },
+      "bin": {
+        "pdf2json": "bin/pdf2json"
+      },
+      "engines": {
+        "node": ">=14.18.0",
+        "npm": ">=6.14.15"
+      }
+    },
+    "node_modules/pdf2json/node_modules/@xmldom/xmldom": {
+      "version": "0.7.5",
+      "inBundle": true,
+      "license": "MIT",
+      "engines": {
+        "node": ">=10.0.0"
+      }
+    },
     "node_modules/pdfjs": {
       "version": "2.4.6",
       "resolved": "https://verdaccio.lossless.one/pdfjs/-/pdfjs-2.4.6.tgz",
@@ -23633,6 +23661,20 @@
         "pdfjs": "^2.4.5"
       }
     },
+    "pdf2json": {
+      "version": "2.0.0",
+      "resolved": "https://verdaccio.lossless.one/pdf2json/-/pdf2json-2.0.0.tgz",
+      "integrity": "sha512-+FZy7GSvLOLc+zksg0SoMvXqIqcku5lBlEPuYJJkhMWB2x6yfthzEhhSbZc20UheClMPagH/+NXnMRbvQMQR1w==",
+      "requires": {
+        "@xmldom/xmldom": "^0.7.5"
+      },
+      "dependencies": {
+        "@xmldom/xmldom": {
+          "version": "0.7.5",
+          "bundled": true
+        }
+      }
+    },
     "pdfjs": {
       "version": "2.4.6",
       "resolved": "https://verdaccio.lossless.one/pdfjs/-/pdfjs-2.4.6.tgz",

package.json

@@ -28,7 +28,8 @@
     "@pushrocks/smartunique": "^3.0.3",
     "@types/express": "^4.17.13",
     "express": "^4.17.2",
-    "pdf-merger-js": "^3.2.1"
+    "pdf-merger-js": "^3.2.1",
+    "pdf2json": "^2.0.0"
   },
   "files": [
     "ts/**/*",

test/test.ts

@@ -16,6 +16,10 @@ tap.test('should create a pdf from html string', async () => {
   await testSmartPdf.getPdfResultForHtmlString('hi');
 });
 
+tap.test('should create a pdf from html string', async () => {
+  await testSmartPdf.getPdfResultForHtmlString('hi');
+});
+
 tap.test('should create a pdf from website as A4', async () => {
   await testSmartPdf.getPdfResultForWebsite('https://www.wikipedia.org');
 });

ts/interfaces/interface.pdfresult.ts

@@ -1,5 +1,8 @@
 export interface IPdfResult {
   name: string;
   id: string;
+  metadata: {
+    textExtraction: string;
+  };
   buffer: Buffer;
 }

ts/smartpdf.classes.smartpdf.ts

@@ -70,8 +70,8 @@ export class SmartPdf {
     const page = await this.headlessBrowser.newPage();
     await page.setViewport({
       width: 794,
-      height: 1122
+      height: 1122,
-    })
+    });
     const response = await page.goto(`http://localhost:3210/${pdfCandidate.pdfId}`, {
       waitUntil: 'networkidle2',
     });
@@ -96,6 +96,9 @@ export class SmartPdf {
     return {
       id: pdfCandidate.pdfId,
       name: `${pdfCandidate.pdfId}.js`,
+      metadata: {
+        textExtraction: await this.extractTextFromPdfBuffer(pdfBuffer),
+      },
       buffer: pdfBuffer,
     };
   }
@@ -123,11 +126,14 @@ export class SmartPdf {
     return {
       id: pdfId,
       name: `${pdfId}.js`,
+      metadata: {
+        textExtraction: await this.extractTextFromPdfBuffer(pdfBuffer),
+      },
       buffer: pdfBuffer,
     };
   }
 
-  async getFullWebsiteAsSinglePdf(websiteUrl: string) {
+  async getFullWebsiteAsSinglePdf(websiteUrl: string): Promise<interfaces.IPdfResult> {
     const page = await this.headlessBrowser.newPage();
     page.emulateMediaType('screen');
     const response = await page.goto(websiteUrl, { waitUntil: 'networkidle2' });
@@ -150,6 +156,9 @@ export class SmartPdf {
     return {
       id: pdfId,
       name: `${pdfId}.js`,
+      metadata: {
+        textExtraction: await this.extractTextFromPdfBuffer(pdfBuffer),
+      },
       buffer: pdfBuffer,
     };
   }
@@ -161,4 +170,23 @@ export class SmartPdf {
     }
     return merger.saveAsBuffer();
   }
+
+  public async extractTextFromPdfBuffer(pdfBufferArg: Buffer): Promise<string> {
+    const deferred = plugins.smartpromise.defer<string>();
+    const pdfParser: any = new plugins.pdf2json();
+    pdfParser.on('pdfParser_dataReady', (pdfData: any) => {
+      let finalText = ''
+      for (const page of pdfData.Pages) {
+        for(const text of page.Texts) {
+          for (const letter of text.R) {
+            finalText = finalText + letter.T;
+          }
+          
+        };
+      }
+      deferred.resolve(finalText);
+    });
+    pdfParser.parseBuffer(pdfBufferArg);
+    return deferred.promise;
+  }
 }

ts/smartpdf.plugins.ts

@@ -15,6 +15,8 @@ export { smartfile, smartpromise, smartpuppeteer, smartunique, smartnetwork };
 
 // thirdparty
 import pdfMerger from 'pdf-merger-js';
+// @ts-ignore
+import pdf2json from 'pdf2json';
 import express from 'express';
 
-export { pdfMerger, express };
+export { pdfMerger, pdf2json, express };