2.0.17

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@pushrocks/smartpdf",
-  "version": "2.0.6",
+  "version": "2.0.17",
   "private": false,
   "description": "create pdfs on the fly",
   "main": "dist_ts/index.js",
@@ -12,22 +12,24 @@
     "build": "(tsbuild --web)"
   },
   "devDependencies": {
-    "@gitzone/tsbuild": "^2.1.25",
+    "@gitzone/tsbuild": "^2.1.28",
-    "@gitzone/tsrun": "^1.2.12",
+    "@gitzone/tsrun": "^1.2.18",
-    "@gitzone/tstest": "^1.0.52",
+    "@gitzone/tstest": "^1.0.60",
-    "@pushrocks/tapbundle": "^3.2.14",
+    "@pushrocks/tapbundle": "^3.2.15",
-    "@types/node": "^14.14.37",
+    "@types/node": "^17.0.8",
     "tslint": "^6.1.3",
     "tslint-config-prettier": "^1.18.0"
   },
   "dependencies": {
-    "@pushrocks/smartfile": "^8.0.9",
+    "@pushrocks/smartfile": "^9.0.5",
-    "@pushrocks/smartnetwork": "^2.0.4",
+    "@pushrocks/smartnetwork": "^2.0.10",
-    "@pushrocks/smartpromise": "^3.1.3",
+    "@pushrocks/smartpromise": "^3.1.6",
-    "@pushrocks/smartpuppeteer": "^1.0.21",
+    "@pushrocks/smartpuppeteer": "^1.0.36",
     "@pushrocks/smartunique": "^3.0.3",
-    "@types/express": "^4.17.11",
+    "@types/express": "^4.17.13",
-    "express": "^4.17.1"
+    "express": "^4.17.2",
+    "pdf-merger-js": "^3.2.1",
+    "pdf2json": "^2.0.0"
   },
   "files": [
     "ts/**/*",

readme.md

@@ -29,7 +29,6 @@ Platform support | [![Supports Windows 10](https://badgen.net/badge/supports%20W
 
 We are always happy for code contributions. If you are not the code contributing type that is ok. Still, maintaining Open Source repositories takes considerable time and thought. If you like the quality of what we do and our modules are useful to you we would appreciate a little monthly contribution: You can [contribute one time](https://lossless.link/contribute-onetime) or [contribute monthly](https://lossless.link/contribute). :)
 
-
 ## Contribution
 
 We are always happy for code contributions. If you are not the code contributing type that is ok. Still, maintaining Open Source repositories takes considerable time and thought. If you like the quality of what we do and our modules are useful to you we would appreciate a little monthly contribution: You can [contribute one time](https://lossless.link/contribute-onetime) or [contribute monthly](https://lossless.link/contribute). :)

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 () => {
@@ -25,18 +29,28 @@ tap.test('should create a pdf from website as single page PDF', async () => {
 });
 
 tap.test('should create a valid PDFResult', async () => {
-  const pdfResult = await testSmartPdf.getFullWebsiteAsSinglePdf(
+  const writePDfToDisk = async (urlArg: string, fileName: string) => {
-    'https://maintainedby.lossless.com'
+    const pdfResult = await testSmartPdf.getFullWebsiteAsSinglePdf(urlArg);
-  );
+    expect(pdfResult.buffer).to.be.instanceOf(Buffer);
-  expect(pdfResult.buffer).to.be.instanceOf(Buffer);
+    const fs = await import('fs');
-  const fs = await import('fs');
 
-  if (!fs.existsSync('.nogit/')) {
+    if (!fs.existsSync('.nogit/')) {
-    fs.mkdirSync('.nogit/');
+      fs.mkdirSync('.nogit/');
-  }
+    }
-  fs.writeFileSync('.nogit/sample.pdf', pdfResult.buffer);
+    fs.writeFileSync(`.nogit/${fileName}`, pdfResult.buffer);
+  };
+  await writePDfToDisk('https://maintainedby.lossless.com/', '1.pdf')
+  await writePDfToDisk('https://rendertron.lossless.one/render/https://lossless.com', '2.pdf')
 });
 
+tap.test('should combine pdfs', async () => {
+  const fs = await import('fs');
+  const buffer1 = fs.readFileSync('.nogit/1.pdf');
+  const buffer2 = fs.readFileSync('.nogit/2.pdf');
+  fs.writeFileSync(`.nogit/combined.pdf`, await testSmartPdf.mergePdfBuffers([buffer1, buffer2]));
+
+})
+
 tap.test('should be able to close properly', async () => {
   await testSmartPdf.stop();
 });

ts/interfaces/interface.pdfresult.ts

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

ts/smartpdf.classes.pdfcandidate.ts

@@ -1,8 +1,11 @@
 import * as plugins from './smartpdf.plugins';
 
 export class PdfCandidate {
-  pdfId = plugins.smartunique.shortId();
+  public htmlString: string;
-  doneDeferred = plugins.smartpromise.defer();
+  public pdfId = plugins.smartunique.shortId();
+  public doneDeferred = plugins.smartpromise.defer();
 
-  constructor(public htmlString) {}
+  constructor(htmlStringArg: string) {
+    this.htmlString = htmlStringArg;
+  }
 }

ts/smartpdf.classes.smartpdf.ts

@@ -3,7 +3,7 @@ import * as paths from './smartpdf.paths';
 import { Server } from 'http';
 import { PdfCandidate } from './smartpdf.classes.pdfcandidate';
 
-declare const document;
+declare const document: any;
 
 import * as interfaces from './interfaces';
 
@@ -19,7 +19,7 @@ export class SmartPdf {
     this._readyDeferred = new plugins.smartpromise.Deferred();
   }
 
-  async start(headlessBrowserArg?) {
+  async start(headlessBrowserArg?: plugins.smartpuppeteer.puppeteer.Browser) {
     // lets set the external browser in case one is provided
     this.headlessBrowser = headlessBrowserArg;
     // setup puppeteer
@@ -63,11 +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({
+      width: 794,
+      height: 1122,
+    });
     const response = await page.goto(`http://localhost:3210/${pdfCandidate.pdfId}`, {
       waitUntil: 'networkidle2',
     });
@@ -80,10 +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];
@@ -92,12 +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();
@@ -108,45 +119,79 @@ export class SmartPdf {
       };
     });
     const pdfBuffer = await page.pdf({
-      format: 'A4',
+      height: documentHeight,
-      height: documentWidth,
       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,
+        documentHeight: document.body.scrollHeight,
-        documentWidth: document.width,
+        documentWidth: document.body.clientWidth,
       };
     });
     const pdfBuffer = await page.pdf({
-      format: 'A4',
+      height: documentHeight,
-      height: documentWidth,
+      width: 1920,
-      width: documentWidth,
       printBackground: true,
       displayHeaderFooter: false,
-      preferCSSPageSize: true,
+      scale: 1
     });
     await page.close();
     return {
       id: pdfId,
       name: `${pdfId}.js`,
+      metadata: {
+        textExtraction: await this.extractTextFromPdfBuffer(pdfBuffer),
+      },
       buffer: pdfBuffer,
     };
   }
+
+  public async mergePdfBuffers(pdfBuffers: Buffer[]): Promise<Buffer> {
+    const merger = new plugins.pdfMerger();
+    for (const buffer of pdfBuffers) {
+      merger.add(buffer);
+    }
+    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

@@ -14,6 +14,9 @@ import * as smartunique from '@pushrocks/smartunique';
 export { smartfile, smartpromise, smartpuppeteer, smartunique, smartnetwork };
 
 // thirdparty
+import pdfMerger from 'pdf-merger-js';
+// @ts-ignore
+import pdf2json from 'pdf2json';
 import express from 'express';
 
-export { express };
+export { pdfMerger, pdf2json, express };