Documentation update / move formatter, visitor to formatters, visitors

src/GoldDigger.js

@@ -19,8 +19,15 @@ class GoldDiggerError extends Error{
  */
 class GoldDigger {
 
+  /**
+   * Constructor
+   * @param {object} config - configuration
+   */
   constructor(config) {
     this.config = config;
+    /**
+     * @type {Formatter}
+     */
     this.formatter = new Formatter(config);
   }
 
@@ -72,7 +79,7 @@ class GoldDigger {
   /**
    * Process page
    * @param pageData - pdf page
-   * @param pageNum - page number
+   * @param {number} pageNum - page number
    */
   async digPage(pageData, pageNum) {
 
@@ -172,9 +179,7 @@ class GoldDigger {
    * Process pdf file format
    * Based on (SVGGraphics)
    * @param opTree - pdf tree of information
-   * @param page - pdf page
-   * @param dependencies - loadDependencies data
-   * @returns {Array} PDFObject array
+   * @param {Visitor} visitor - class for parsing incoming tags
    */
   executeOpTree(opTree, visitor) {
     const debug = visitor.debug;

src/pdf/Extract.js

@@ -8,8 +8,8 @@ const util = require('pdfjs-dist/lib/shared/util');
 class ExtractText {
   /**
    * Return text from glyphs array
-   * @param glyphs - glyphs from pdf.OPS.showText, pdf.OPS.showSpacedText
-   * @param page - PdfPage object @see PdfPage
+   * @param {array} glyphs - glyphs from pdf.OPS.showText, pdf.OPS.showSpacedText
+   * @param {PdfPage} page - pdf page object
    */
   showText(glyphs, page) {
     // MOVED from VisitorText
@@ -77,7 +77,7 @@ class ExtractText {
 
   /**
    * Find font family inside loadedDependencies based on font name
-   * @param name - font name
+   * @param {string} name - font name
    * @param dependencies - pdf document data
    * @returns {*} font from dependencies if found otherwise null
    * (probably need to warn user for missing font inside document)
@@ -92,9 +92,9 @@ class ExtractText {
   }
 
   /**
-   * Gets FontObject from page information when pdf.OPS.setFont
+   * Gets {@link FontObject} from page information when pdf.OPS.setFont
    * @param details - arguments from pdf.OPS.setFont
-   * @param page - pdf page @see PdfPage
+   * @param {PdfPage} page - pdf page
    */
   setFont(details, page) {
     const fontObj = page.data.commonObjs.get(details[0]);

src/pdf/FileManager.js

@@ -21,7 +21,11 @@ const mkdirNotExists = (path) => {
     fs.mkdirSync(path);
   }
 }
-
+/**
+ * Reads directory
+ * @param {string} path - directory path
+ * @returns {string[]} - directory listing
+ */
 const readdirSync = (path) => {
   return fs.readdirSync(path)
 }

src/pdf/Formatter.js

@@ -1,18 +1,23 @@
-const FormatterJSON = require('./formatter/FormatterJSON');
-const FormatterXML = require('./formatter/FormatterXML');
-const FormatterText = require('./formatter/FormatterText');
+const f = require('./formatters');
 
 /**
  * Formats PDF to desired output
  */
 class Formatter {
 
+  /**
+   * Constructor
+   * @param config
+   */
   constructor(config) {
     this.debug = config.debug;
+    /**
+     * @type {{json: FormatterJSON, xml: FormatterXML, text: FormatterText}}
+     */
     this.formatters = {
-      json: new FormatterJSON(),
-      xml: new FormatterXML(),
-      text: new FormatterText(),
+      json: new f.FormatterJSON(),
+      xml: new f.FormatterXML(),
+      text: new f.FormatterText(),
     }
     this.data = "";
   }

src/pdf/Visitor.js

@@ -1,4 +1,4 @@
-const v = require('./visitor/index');
+const v = require('./visitors');
 const Model = require('./model');
 
 const FN_TEXT = ['beginText', 'setFont', 'showText',
@@ -6,7 +6,7 @@ const FN_TEXT = ['beginText', 'setFont', 'showText',
   'setLeading', 'setLeadingMoveText', 'setCharSpacing',
   'setWordSpacing', 'setHScale', 'setTextMatrix',
   'setTextRise', 'setTextRenderingMode', 'nextLine'];
-const FN_XOBJECT = ['setTextMatrix', 'paintFormXObjectBegin', 'paintFormXObjectEnd'];
+const FN_XOBJECT = ['paintFormXObjectBegin', 'paintFormXObjectEnd'];
 const FN_IMAGE = ['paintJpegXObject', 'paintImageXObject', 'paintInlineImageXObject', 'paintImageMaskXObject'];
 
 /**
@@ -14,12 +14,30 @@ const FN_IMAGE = ['paintJpegXObject', 'paintImageXObject', 'paintInlineImageXObj
  */
 class Visitor {
 
+  /**
+   * Constructor
+   * @param {object} config - application configuration
+   * @param data - pdf data
+   * @param dependencies - pdf loaded dependencies
+   */
   constructor (config, data, dependencies) {
     this.config = config;
     this.config.skip = false;
+    /**
+     * @type {PdfPage}
+     */
     this.page = new Model.PdfPage(data, dependencies);
+    /**
+     * @type {VisitorText}
+     */
     this.txt = new v.VisitorText(config, this.page);
+    /**
+     * @type {VisitorXObject}
+     */
     this.xobject = new v.VisitorXObject(config, this.page);
+    /**
+     * @type {VisitorImage}
+     */
     this.image = new v.VisitorImage(config, this.page);
   }
 

src/pdf/formatter/FormatterXML.js

@@ -1,19 +0,0 @@
-
-class FormatterXML {
-  start(doc, metadata) {
-    return `<?xml version="1.0" encoding="UTF-8"?>
-    <document>
-    `
-  }
-
-  format(page, data) {
-    const output = '';
-    return output
-  }
-
-  end() {
-    return '</document>'
-  }
-}
-
-module.exports = FormatterXML;

src/pdf/formatters/FormatterJSON.js

@@ -2,6 +2,12 @@
  * Format PDF into json data
  */
 class FormatterJSON {
+  /**
+   * See {@link Formatter}
+   * @param doc
+   * @param metadata
+   * @returns {string}
+   */
   start(doc, metadata) {
     const meta = JSON.stringify(metadata)
     return `{
@@ -12,8 +18,8 @@ class FormatterJSON {
   }
 
   /**
-   * Formats {TextObject} to JSON serializable object
-   * @param textObject {TextObject}
+   * Formats {@link TextObject} to JSON serializable object
+   * @param  {TextObject} textObject
    * @returns {object}
    */
   formatTextObject(textObject) {
@@ -26,8 +32,8 @@ class FormatterJSON {
   }
 
   /**
-   * Formats {TextLine} to JSON serializable object
-   * @param textLine {TextLine}
+   * Formats {@link TextLine} to JSON serializable object
+   * @param {TextLine} textLine
    * @returns {object}
    */
   formatTextLine(textLine) {
@@ -47,8 +53,8 @@ class FormatterJSON {
   }
 
   /**
-   * Formats {TextFont} to JSON serializable object
-   * @param textFont {TextFont}
+   * Formats {@link TextFont} to JSON serializable object
+   * @param {TextFont} textFont
    * @returns {object}
    */
   formatTextFont(textFont) {
@@ -67,7 +73,7 @@ class FormatterJSON {
   }
 
   /**
-   * See {Formatter}
+   * See {@link Formatter}
    * @param page
    * @param data
    * @param last
@@ -87,7 +93,7 @@ class FormatterJSON {
   }
 
   /**
-   * See {Formatter}
+   * See {@link Formatter}
    * @returns {string}
    */
   end() {

src/pdf/formatters/FormatterText.js

@@ -1,7 +1,9 @@
-
+/**
+ * Format PDF into text data
+ */
 class FormatterText {
   /**
-   * See {Formatter}
+   * See {@link Formatter}
    * @param doc
    * @param metadata
    * @returns {string}
@@ -11,8 +13,8 @@ class FormatterText {
   }
 
   /**
-   * Format {TextObject} to string
-   * @param textObject {TextObject}
+   * Format {@link TextObject} to string
+   * @param {TextObject} textObject
    * @returns {string}
    */
   formatTextObject(textObject) {
@@ -25,8 +27,8 @@ class FormatterText {
   }
 
   /**
-   * Format {TextLine} to string
-   * @param textLine
+   * Format {@link TextLine} to string
+   * @param {TextLine} textLine
    * @returns {string}
    */
   formatTextLine(textLine) {
@@ -38,8 +40,8 @@ class FormatterText {
   }
 
   /**
-   * Format {TextFont} to string
-   * @param textFont {TextFont}
+   * Format {@link TextFont} to string
+   * @param {TextFont} textFont
    * @returns {string|TextFont|*}
    */
   formatTextFont(textFont) {
@@ -47,7 +49,7 @@ class FormatterText {
   }
 
   /**
-   * See {Formatter}
+   * See {@link Formatter}
    * @param page
    * @param data
    * @param last
@@ -63,7 +65,7 @@ class FormatterText {
   }
 
   /**
-   * See {Formatter}
+   * See {@link Formatter}
    * @returns {string}
    */
   end() {

src/pdf/formatters/FormatterXML.js

@@ -0,0 +1,38 @@
+/**
+ * Format PDF into xml data
+ */
+class FormatterXML {
+  /**
+   * See {@link Formatter}
+   * @param doc
+   * @param metadata
+   * @returns {string}
+   */
+  start(doc, metadata) {
+    return `<?xml version="1.0" encoding="UTF-8"?>
+    <document>
+    `
+  }
+
+  /**
+   * See {@link Formatter}
+   * @param page
+   * @param data
+   * @param last
+   * @returns {string}
+   */
+  format(page, data, last) {
+    const output = '';
+    return output
+  }
+  
+  /**
+   * See {@link Formatter}
+   * @returns {string}
+   */
+  end() {
+    return '</document>'
+  }
+}
+
+module.exports = FormatterXML;

src/pdf/model/FontObject.js

@@ -2,21 +2,51 @@
  * Store font information
  */
 class FontObject {
+  /**
+   * Constructor
+   */
   constructor() {
+    /**
+     * @type {number}
+     */
     this.size = 0;
+    /**
+     * @type {number}
+     */
     this.sizeScale = 1;
+    /**
+     * @type {string}
+     */
     this.weight = 'normal';
+    /**
+     * @type {string}
+     */
     this.style = null;
+    /**
+     * @type {string}
+     */
     this.family = null;
+    /**
+     * @type {number}
+     */
     this.direction = 1;
+    /**
+     * @type {boolean}
+     */
     this.vertical = false;
+    /**
+     * @type {boolean}
+     */
     this.spaceWidthIsSet = false;
+    /**
+     * @type {number}
+     */
     this.spaceWidth = 250;
   }
 
   /**
    * Set font size based on number if it's negative it will be other font direction
-   * @param size
+   * @param {number} size
    */
   setSize(size) {
     if (size < 0) {

src/pdf/model/PdfObject.js

@@ -2,8 +2,19 @@
  * Main class for handling pdf data
  */
 class PdfObject {
+  /**
+   * Constructor
+   */
   constructor () {
+    /**
+     * x position of element
+     * @type {number}
+     */
     this.x = 0;
+    /**
+     * y position of element
+     * @type {number}
+     */
     this.y = 0;
   }
 }

src/pdf/model/PdfPage.js

@@ -1,6 +1,15 @@
 const PdfObject = require('./PdfObject');
 
+/**
+ * Holds PDF page information
+ * @extends {PdfObject}
+ */
 class PdfPage extends PdfObject {
+  /**
+   * Constructor
+   * @param {object} data - pdf page data opcodes
+   * @param {object} dependencies - pdf loaded resources
+   */
   constructor (data, dependencies) {
     super();
     this.data = data;
@@ -21,6 +30,10 @@ class PdfPage extends PdfObject {
     this.currentFont;
   }
 
+  /**
+   * Set current object and add it to objectList
+   * @param {TextObject} obj
+   */
   setCurrentObject(obj) {
     this.currentObject = obj
     this.objectList.push(obj);

src/pdf/model/TextObject.js

@@ -18,7 +18,7 @@ class TextObject extends PdfObject {
   }
 
   /**
-   * Create new {TextLine} adds it to array and returns it as value
+   * Create new {@link TextLine} adds it to array and returns it as value
    * @returns {TextLine}
    */
   newLine() {
@@ -28,7 +28,7 @@ class TextObject extends PdfObject {
   }
 
   /**
-   * Return last {TextLine} from text array
+   * Return last {@link TextLine} from text array
    * @returns {TextLine}
    */
   getLine() {
@@ -36,7 +36,7 @@ class TextObject extends PdfObject {
   }
 
   /**
-   * Return array of {TextLine}
+   * Return array of {@link TextLine}
    * @returns {Array}
    */
   getData() {

src/pdf/model/text/TextFont.js

@@ -2,6 +2,7 @@ const PdfObject = require('./../PdfObject');
 
 /**
  * Represents Font information in pdf file
+ * @extends {PdfObject}
  */
 class TextFont extends PdfObject {
   constructor() {
@@ -25,6 +26,11 @@ class TextFont extends PdfObject {
     }
   }
 
+  /**
+   * Heuristic method to check if glyph number is space
+   * @param {number} glyph - numeric size
+   * @returns {boolean} - flag true if it's space false if it's not
+   */
   isSpace(glyph) {
     if(-glyph >= this.font.spaceWidth) {
       return true;
@@ -44,17 +50,6 @@ class TextFont extends PdfObject {
   getText() {
     return this._text;
   }
-
-  /**
-   * Check if provided {FontObject} is equal to existing one
-   * @param font {FontObject}
-   * @returns {boolean}
-   */
-  equals(font) {
-    return this.font === font.font
-      && this.charSpacing === font.charSpacing
-      && this.wordSpacing === font.wordSpacing;
-  }
 }
 
 module.exports = TextFont;

src/pdf/model/text/TextLine.js

@@ -3,6 +3,7 @@ const TextFont = require('./../text/TextFont');
 
 /**
  * Represents text line in pdf file
+ * @extends {PdfObject}
  */
 class TextLine extends PdfObject {
   constructor() {
@@ -14,14 +15,14 @@ class TextLine extends PdfObject {
 
   /**
    * Adds line with font to text
-   * @param line @see {TextFont}
+   * @param {TextFont} line
    */
   addTextFont(line) {
     this._textFonts.push(line);
   }
 
   /**
-   * Get last TextFont
+   * Get last {@link TextFont}
    * @returns {TextFont}
    */
   getLastFontText() {
@@ -40,8 +41,8 @@ class TextLine extends PdfObject {
   }
 
   /**
-   * Get Array of all {TextFont} data inside this text line
-   * @returns {Array} of {TextFont}
+   * Get Array of all {@link TextFont} data inside this text line
+   * @returns {Array} of {@link TextFont}
    */
   getData() {
     return this._textFonts;

src/pdf/visitors/VisitorBase.js

@@ -5,7 +5,7 @@ class VisitorBase {
   /**
    * Constructor
    * @param config - configuration
-   * @param page - PdfPage @see PdfPage
+   * @param {PdfPage} page
    */
   constructor(config, page) {
     this.config = config;

src/pdf/visitors/VisitorImage.js

@@ -4,6 +4,7 @@ const FileManager = require('../FileManager');
 
 /**
  * Visit image objects in pdf files
+ * @extends {VisitorBase}
  */
 class VisitorImage extends VisitorBase {
 

src/pdf/visitors/VisitorText.js

@@ -4,6 +4,7 @@ const VisitorBase = require('./VisitorBase');
 
 /**
  * Visits text data while parsing pdf
+ * @extends {VisitorBase}
  */
 class VisitorText extends VisitorBase {
   constructor(config, page) {
@@ -138,7 +139,10 @@ class VisitorText extends VisitorBase {
     if (this.debug) console.log('setTextRise');
     if (this.config.skip) return;
   }
-
+  
+  /**
+   * pdf.OPS.setTextRenderingMode
+   */
   setTextRenderingMode(args) {
     if (this.debug) console.log('setTextRenderingMode');
     if (this.config.skip) return;

src/pdf/visitors/VisitorXObject.js

@@ -3,22 +3,10 @@ const VisitorBase = require('./VisitorBase');
 /**
  * Visits XObject data when parsing pdf
  * Currently only sets flag skip for XObjectBegin
+ * @extends {VisitorBase}
  */
 class VisitorXObject extends VisitorBase {
 
-  /**
-   * pdf.OPS.setTextMatrix
-   */
-  setTextMatrix(args) {
-    if (this.config.debug) console.log('setTextMatrix');
-    if (this.config.skip) return;
-    /*const a = args[0], b = args[1], c = args[2], d = args[3], e = args[4], f = args[5];
-    const el = this.currentObject.getLine();
-    // new text font
-    el.newText();
-    el.textMatrix = [a, b, c, d, e, f];*/
-  }
-
   /**
    * pdf.OPS.paintFormXObjectBegin
    */