Add image data information to output xml, json formatters

src/pdf/formatters/FormatterJSON.js

@@ -1,3 +1,5 @@
+const Model = require('../model');
+
 /**
  * Format PDF into json data
  */
@@ -43,6 +45,20 @@ class FormatterJSON {
     return txtObjOut;
   }
 
+  /**
+   * Format image object
+   * @param {ImageObject} imageObject
+   */
+  formatImageObject (imageObject) {
+    return {
+      x: imageObject.x,
+      y: imageObject.y,
+      width: imageObject.width,
+      height: imageObject.height,
+      name: imageObject.name,
+    };
+  }
+
   /**
    * Formats {@link TextLine} to JSON serializable object
    * @param {TextLine} textLine
@@ -92,14 +108,20 @@ class FormatterJSON {
    */
   format (page, data, last) {
     const txtData = [];
-    data.forEach(textObject => {
-      const txtObjOut = this.formatTextObject(textObject);
-      txtData.push(txtObjOut);
+    data.forEach(pdfObject => {
+      if (pdfObject instanceof Model.TextObject) {
+        const txtObjOut = this.formatTextObject(pdfObject);
+        txtData.push(txtObjOut);
+      } else if (pdfObject instanceof Model.ImageObject) {
+        const imgObjOut = this.formatImageObject(pdfObject);
+        txtData.push(imgObjOut);
+      } else {
+        console.warn(`Not recognised object ${pdfObject}`);
+      }
     });
-    const output = {
+    const out = JSON.stringify({
       data: txtData,
-    };
-    const out = JSON.stringify(output); // pretty print (output, null, 4)
+    }); // pretty print (output, null, 4)
     return `"${page.pageIndex}": ${out}${last ? '' : ','}`;
   }
 

src/pdf/formatters/FormatterText.js

@@ -1,3 +1,4 @@
+const Model = require('../model');
 /**
  * Format PDF into text data
  */
@@ -64,9 +65,14 @@ class FormatterText {
    */
   format (page, data, last) {
     let output = '';
-    data.forEach(textObject => {
-      const txtObjOut = this.formatTextObject(textObject);
-      output += txtObjOut;
+    data.forEach(pdfObject => {
+      if (pdfObject instanceof Model.TextObject) {
+        output += this.formatTextObject(pdfObject);
+      } else if (pdfObject instanceof Model.ImageObject) {
+        // TODO add config to save image position
+      } else {
+        console.warn(`Not recognised object ${pdfObject}`);
+      }
     });
     return output;
   }

src/pdf/formatters/FormatterXML.js

@@ -1,3 +1,5 @@
+const Model = require('../model');
+
 /**
  * Format PDF into xml data
  */
@@ -30,10 +32,22 @@ class FormatterXML {
     lines.forEach(textLine => {
       txtObjOut += this.formatTextLine(textLine);
     });
-    txtObjOut += '</object>';
+    txtObjOut += '</object>\n';
     return txtObjOut;
   }
 
+  /**
+   * Format image object
+   * @param {ImageObject} imageObject
+   */
+  formatImageObject (imageObject) {
+    return `<image 
+    x="${imageObject.x}" 
+    y="${imageObject.y}" 
+    width="${imageObject.width}" 
+    height="${imageObject.height}">${imageObject.name}</image>\n`;
+  }
+
   /**
    * Formats {@link TextLine} to xml object
    * @param {TextLine} textLine
@@ -76,8 +90,14 @@ class FormatterXML {
    */
   format (page, data, last) {
     let out = '<data>\n';
-    data.forEach(textObject => {
-      out += this.formatTextObject(textObject) + '\n';
+    data.forEach(pdfObject => {
+      if (pdfObject instanceof Model.TextObject) {
+        out += this.formatTextObject(pdfObject);
+      } else if (pdfObject instanceof Model.ImageObject) {
+        out += this.formatImageObject(pdfObject);
+      } else {
+        console.warn(`Not recognised object ${pdfObject}`);
+      }
     });
     out += '</data>';
     return out;

src/pdf/model/ImageObject.js

@@ -0,0 +1,43 @@
+const PdfObject = require('./PdfObject');
+
+/**
+ * Information about images
+ */
+class ImageObject extends PdfObject {
+  /**
+   * Constructor
+   */
+  constructor () {
+    super();
+    /**
+     * @type {string} saved image name
+     */
+    this.name = '';
+    /**
+     * @type {number} document image width
+     */
+    this.width = 0;
+    /**
+     * @type {number} document image height
+     */
+    this.height = 0;
+  }
+
+  /**
+   * Fill properties about this object
+   * @param {string} name
+   * @param {number} x
+   * @param {number} y
+   * @param {number} width
+   * @param {number} height
+   */
+  fill (name, x, y, width, height) {
+    this.name = name;
+    this.x = x;
+    this.y = y;
+    this.width = width;
+    this.height = height;
+  }
+}
+
+module.exports = ImageObject;

src/pdf/model/PdfPage.js

@@ -33,6 +33,14 @@ class PdfPage extends PdfObject {
     this.fonts = {};
   }
 
+  /**
+   * Add image to object list
+   * @param {ImageObject} image
+   */
+  addImage (image) {
+    this.objectList.push(image);
+  }
+
   /**
    * Set current object and add it to objectList
    * @param {TextObject} obj

src/pdf/model/index.js

@@ -1,5 +1,6 @@
 const FontObject = require('./FontObject');
 const PdfObject = require('./PdfObject');
+const ImageObject = require('./ImageObject');
 const PdfPage = require('./PdfPage');
 const TextObject = require('./TextObject');
 const TextFont = require('./text/TextFont');
@@ -12,4 +13,5 @@ module.exports = {
   TextFont,
   TextLine,
   PdfPage,
+  ImageObject,
 };

src/pdf/visitors/VisitorImage.js

@@ -1,4 +1,5 @@
 const VisitorBase = require('./VisitorBase');
+const Model = require('../model');
 const pdfjs = require('../../pdfjs');
 const FileManager = require('../FileManager');
 
@@ -59,7 +60,11 @@ class VisitorImage extends VisitorBase {
     // TODO imlement mask
     const mask = false;
     const imgBinary = pdfjs.convertImgDataToPng(imgData, this.forceDataSchema, !!mask);
-    const fpath = `${this.config.outputDir}/img/page.${this.page.data.pageIndex}.${args[1]}.png`;
+    const fname = `page.${this.page.data.pageIndex}.${args[1]}.png`;
+    const fpath = `${this.config.outputDir}/img/${fname}`;
+    const image = new Model.ImageObject();
+    image.fill(fname, this.page.x, this.page.y, imgData.width, imgData.height);
+    this.page.addImage(image);
     await FileManager.saveFileAsync(fpath, imgBinary);
   }
 }