2.0.18

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@pushrocks/smartpdf",
-  "version": "2.0.13",
+  "version": "2.0.18",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "@pushrocks/smartpdf",
-      "version": "2.0.13",
+      "version": "2.0.18",
       "license": "MIT",
       "dependencies": {
         "@pushrocks/smartfile": "^9.0.5",
@@ -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

@@ -1,6 +1,6 @@
 {
   "name": "@pushrocks/smartpdf",
-  "version": "2.0.13",
+  "version": "2.0.18",
   "private": false,
   "description": "create pdfs on the fly",
   "main": "dist_ts/index.js",
@@ -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

@@ -13,7 +13,11 @@ tap.test('should start the instance', async () => {
 });
 
 tap.test('should create a pdf from html string', async () => {
-  await testSmartPdf.getPdfResultForHtmlString('hi');
+  await testSmartPdf.getA4PdfResultForHtmlString('hi');
+});
+
+tap.test('should create a pdf from html string', async () => {
+  await testSmartPdf.getA4PdfResultForHtmlString('hi');
 });
 
 tap.test('should create a pdf from website as A4', async () => {
@@ -36,7 +40,7 @@ tap.test('should create a valid PDFResult', async () => {
     fs.writeFileSync(`.nogit/${fileName}`, pdfResult.buffer);
   };
   await writePDfToDisk('https://maintainedby.lossless.com/', '1.pdf')
-  await writePDfToDisk('https://lossless.com/', '2.pdf')
+  await writePDfToDisk('https://rendertron.lossless.one/render/https://lossless.com', '2.pdf')
 });
 
 tap.test('should combine pdfs', async () => {

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

@@ -63,15 +63,15 @@ export class SmartPdf {
   /**
    * returns a pdf for a given html string;
    */
-  async getPdfResultForHtmlString(htmlStringArg: string): Promise<interfaces.IPdfResult> {
+  async getA4PdfResultForHtmlString(htmlStringArg: string): Promise<interfaces.IPdfResult> {
     await this._readyDeferred.promise;
     const pdfCandidate = new PdfCandidate(htmlStringArg);
     this._candidates[pdfCandidate.pdfId] = pdfCandidate;
     const page = await this.headlessBrowser.newPage();
     await page.setViewport({
-      height: 842,
-      width: 595,
-    })
+      width: 794,
+      height: 1122,
+    });
     const response = await page.goto(`http://localhost:3210/${pdfCandidate.pdfId}`, {
       waitUntil: 'networkidle2',
     });
@@ -84,10 +84,10 @@ export class SmartPdf {
     }
 
     const pdfBuffer = await page.pdf({
-      format: 'a4',
+      width: 794,
+      height: 1122,
       printBackground: true,
       displayHeaderFooter: false,
-      preferCSSPageSize: true,
     });
     await page.close();
     delete this._candidates[pdfCandidate.pdfId];
@@ -96,12 +96,19 @@ export class SmartPdf {
     return {
       id: pdfCandidate.pdfId,
       name: `${pdfCandidate.pdfId}.js`,
+      metadata: {
+        textExtraction: await this.extractTextFromPdfBuffer(pdfBuffer),
+      },
       buffer: pdfBuffer,
     };
   }
 
   async getPdfResultForWebsite(websiteUrl: string): Promise<interfaces.IPdfResult> {
     const page = await this.headlessBrowser.newPage();
+    await page.setViewport({
+      width: 1980,
+      height: 1200,
+    });
     await page.emulateMediaType('screen');
     const response = await page.goto(websiteUrl, { waitUntil: 'networkidle2' });
     const pdfId = plugins.smartunique.shortId();
@@ -112,44 +119,56 @@ export class SmartPdf {
       };
     });
     const pdfBuffer = await page.pdf({
-      format: 'a4',
-      height: documentWidth,
+      height: documentHeight,
       width: documentWidth,
       printBackground: true,
       displayHeaderFooter: false,
-      preferCSSPageSize: true,
     });
     await page.close();
     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();
+    await page.setViewport({
+      width: 1920,
+      height: 1200,
+    });
     page.emulateMediaType('screen');
     const response = await page.goto(websiteUrl, { waitUntil: 'networkidle2' });
     const pdfId = plugins.smartunique.shortId();
     const { documentHeight, documentWidth } = await page.evaluate(() => {
       return {
-        documentHeight: document.height,
-        documentWidth: document.width,
+        documentHeight: document.body.scrollHeight,
+        documentWidth: document.body.clientWidth,
       };
     });
+    await page.setViewport({
+      width: 1920,
+      height: documentHeight,
+    });
     const pdfBuffer = await page.pdf({
-      format: 'a4',
-      height: documentWidth,
-      width: documentWidth,
+      height: documentHeight,
+      width: 1920,
       printBackground: true,
       displayHeaderFooter: false,
-      preferCSSPageSize: true,
+      scale: 1,
+      pageRanges: '1'
     });
     await page.close();
     return {
       id: pdfId,
       name: `${pdfId}.js`,
+      metadata: {
+        textExtraction: await this.extractTextFromPdfBuffer(pdfBuffer),
+      },
       buffer: pdfBuffer,
     };
   }
@@ -161,4 +180,23 @@ 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 };