Add eslint standard with small modifications

- semi always
- comma-dangle always-multiple

.eslintrc.json

@@ -0,0 +1,11 @@
+{
+  "extends": "standard",
+  "parserOptions": {
+    "sourceType": "module",
+    "ecmaVersion": 2017
+  },
+  "rules": {
+    "semi": [2, "always"],
+    "comma-dangle": ["error", "always-multiline"]
+  }
+}

package.json

@@ -24,11 +24,18 @@
     "pdf-text-extract"
   ],
   "scripts": {
-    "doc": "esdoc"
+    "doc": "esdoc",
+    "lint": "eslint src"
   },
   "devDependencies": {
     "esdoc": "^1.1.0",
     "esdoc-node": "^1.0.4",
-    "esdoc-standard-plugin": "^1.0.0"
+    "esdoc-standard-plugin": "^1.0.0",
+    "eslint": "^6.1.0",
+    "eslint-config-standard": "^13.0.1",
+    "eslint-plugin-import": "^2.18.2",
+    "eslint-plugin-node": "^9.1.0",
+    "eslint-plugin-promise": "^4.2.1",
+    "eslint-plugin-standard": "^4.0.0"
   }
 }

src/GoldDigger.js

@@ -1,7 +1,5 @@
 const fs = require('fs');
 const pdf = require('pdfjs-dist');
-const util = require('pdfjs-dist/lib/shared/util');
-const Extract = require('./pdf/Extract');
 const Visitor = require('./pdf/Visitor');
 const Formatter = require('./pdf/Formatter');
 const FileManager = require('./pdf/FileManager');
@@ -18,7 +16,6 @@ class GoldDiggerError extends Error{
  * Code based on pdf.js SVGGraphics
  */
 class GoldDigger {
-
   /**
    * Constructor
    * @param {object} config - configuration
@@ -45,7 +42,7 @@ class GoldDigger {
     const doc = await pdf.getDocument({
       data: data,
     }).promise;
-    return doc
+    return doc;
   }
 
   /**
@@ -63,12 +60,12 @@ class GoldDigger {
     // read pages
     for (let pageNum = 1; pageNum <= doc.numPages; pageNum++) {
       const pageData = await doc.getPage(pageNum);
-      const viewport = pageData.getViewport({ scale: 1.0, });
+      const viewport = pageData.getViewport({ scale: 1.0 });
       if (debug) console.log(`--- BEGIN Page ${pageNum} size: ${viewport.width}x${viewport.height}`);
       const output = await this.digPage(pageData, pageNum);
-      const last = pageNum == doc.numPages;
+      const last = pageNum === doc.numPages;
       this.formatter.format(format, pageData, output, last);
-      if(debug) console.log(`--- END Page ${pageNum} objects : ${output.length}`)
+      if (debug) console.log(`--- END Page ${pageNum} objects : ${output.length}`);
     }
     this.formatter.end(format);
     // save to file
@@ -82,7 +79,6 @@ class GoldDigger {
    * @param {number} pageNum - page number
    */
   async digPage (pageData, pageNum) {
-
     // const text = await page.extractTextContent();
     const operatorList = await pageData.getOperatorList();
     // page.commonObjs, page.objs
@@ -121,8 +117,8 @@ class GoldDigger {
    * (SVGGraphics)
    */
   operatorMapping () {
-    const mapping = {}
-    for(var op in pdf.OPS) {
+    const mapping = {};
+    for (const op in pdf.OPS) {
       mapping[pdf.OPS[op]] = op;
     }
     return mapping;
@@ -141,9 +137,9 @@ class GoldDigger {
     for (let i = 0, ii = fnArray.length; i < ii; i++) {
       const fnId = fnArray[i];
       opList.push({
-        'fnId': fnId,
-        'fn': operatorIdMapping[fnId],
-        'args': argsArray[i],
+        fnId,
+        fn: operatorIdMapping[fnId],
+        args: argsArray[i],
       });
     }
     return this.opListToTree(opList);
@@ -160,7 +156,7 @@ class GoldDigger {
 
     for (const opListElement of opList) {
       if (opListElement.fn === 'save') {
-        opTree.push({ 'fnId': 92, 'fn': 'group', 'items': [], });
+        opTree.push({ fnId: 92, fn: 'group', items: [] });
         tmp.push(opTree);
         opTree = opTree[opTree.length - 1].items;
         continue;
@@ -325,4 +321,4 @@ class GoldDigger {
   }
 }
 
-module.exports = GoldDigger
+module.exports = GoldDigger;

src/pdf/Constraints.js

@@ -5,4 +5,4 @@ const FONT_IDENTITY_MATRIX = [0.001, 0, 0, 0.001, 0, 0];
 module.exports = {
   IDENTITY_MATRIX,
   FONT_IDENTITY_MATRIX,
-}
+};

src/pdf/Extract.js

@@ -18,15 +18,15 @@ class ExtractText {
     const line = new Model.TextFont();
     line.font = page.currentFont;
     // copy from previous line
-    const lastLine = lineList.getLastFontText()
+    const lastLine = lineList.getLastFontText();
     if (lastLine) {
       line.wordSpacing = lastLine.wordSpacing;
       line.charSpacing = lastLine.charSpacing;
     }
-    let startX = page.x;
-    let startY = page.y;
+    const startX = page.x;
+    const startY = page.y;
     // END
-    let partial = "";
+    let partial = '';
     let x = 0;
     for (const glyph of glyphs) {
       if (glyph === null) {
@@ -36,13 +36,13 @@ class ExtractText {
       } else if (util.isNum(glyph)) {
         x += -glyph * line.font.size * 0.001;
         if (!line.font.spaceWidthIsSet && line.isSpace(glyph)) {
-          partial += " ";
+          partial += ' ';
         }
         continue;
       }
       const spacing = (glyph.isSpace ? line.wordSpacing : 0) + line.charSpacing;
       if (spacing > 0) {
-        console.warn(`Not implemented spacing : ${spacing} !`)
+        console.warn(`Not implemented spacing : ${spacing} !`);
       }
       // TODO use glyph font character
       partial += glyph.unicode;
@@ -83,7 +83,7 @@ class ExtractText {
    */
   getFontFamily (name, dependencies) {
     for (let i = 0; i < dependencies.length; i++) {
-      if(dependencies[i].loadedName == name) {
+      if (dependencies[i].loadedName === name) {
         return dependencies[i];
       }
     }
@@ -99,9 +99,9 @@ class ExtractText {
     const fontObj = page.data.commonObjs.get(details[0]);
     const font = new Model.FontObject();
     // calculate space width
-    let spaceKey = -1
-    for(let key in fontObj.toUnicode._map) {
-      if(fontObj.toUnicode._map[key] === " ") {
+    let spaceKey = -1;
+    for (const key in fontObj.toUnicode._map) {
+      if (fontObj.toUnicode._map[key] === ' ') {
         spaceKey = key;
         break;
       }
@@ -111,8 +111,7 @@ class ExtractText {
       font.spaceWidth = fontObj.widths[spaceKey];
     }
     font.setSize(details[1]);
-    font.weight = fontObj.black ? (fontObj.bold ? 'bolder' : 'bold') :
-      (fontObj.bold ? 'bold' : 'normal');
+    font.weight = fontObj.black ? (fontObj.bold ? 'bolder' : 'bold') : (fontObj.bold ? 'bold' : 'normal');
     font.style = fontObj.italic ? 'italic' : 'normal';
     const family = this.getFontFamily(fontObj.loadedName, page.dependencies);
     if (family) {
@@ -126,5 +125,5 @@ class ExtractText {
 }
 
 module.exports = {
-  ExtractText
-}
+  ExtractText,
+};

src/pdf/FileManager.js

@@ -10,7 +10,7 @@ const saveFileAsync = async (fpath, data) => {
   const stream = fs.createWriteStream(fpath);
   await stream.write(data);
   await stream.end();
-}
+};
 
 /**
  * Make directory if not exists in given path
@@ -20,18 +20,17 @@ const mkdirNotExists = (path) => {
   if (!fs.existsSync(path)) {
     fs.mkdirSync(path);
   }
-}
+};
+
 /**
  * Reads directory
  * @param {string} path - directory path
  * @returns {string[]} - directory listing
  */
-const readdirSync = (path) => {
-  return fs.readdirSync(path)
-}
+const readdirSync = (path) => fs.readdirSync(path);
 
 module.exports = {
   saveFileAsync,
   mkdirNotExists,
   readdirSync,
-}
+};

src/pdf/Formatter.js

@@ -4,7 +4,6 @@ const f = require('./formatters');
  * Formats PDF to desired output
  */
 class Formatter {
-
   /**
    * Constructor
    * @param config
@@ -18,8 +17,8 @@ class Formatter {
       json: new f.FormatterJSON(),
       xml: new f.FormatterXML(),
       text: new f.FormatterText(),
-    }
-    this.data = "";
+    };
+    this.data = '';
   }
 
   /**

src/pdf/Visitor.js

@@ -13,7 +13,6 @@ const FN_IMAGE = ['paintJpegXObject', 'paintImageXObject', 'paintInlineImageXObj
  * Visits pdf.OPT.* methods using pdf page data
  */
 class Visitor {
-
   /**
    * Constructor
    * @param {object} config - application configuration
@@ -59,4 +58,4 @@ class Visitor {
   }
 }
 
-module.exports = Visitor
+module.exports = Visitor;

src/pdf/formatters/FormatterJSON.js

@@ -9,12 +9,12 @@ class FormatterJSON {
    * @returns {string}
    */
   start (doc, metadata) {
-    const meta = JSON.stringify(metadata)
+    const meta = JSON.stringify(metadata);
     return `{
       "pages_count": ${doc.numPages},
       "metadata": ${meta},
       "pages": {
-    `
+    `;
   }
 
   /**
@@ -47,7 +47,7 @@ class FormatterJSON {
       x: textLine.x,
       y: textLine.y,
       width: textLine.width,
-    }
+    };
     textLine.getData().forEach(textFont => {
       const txtFontOut = this.formatTextFont(textFont);
       txtLineOut.text.push(txtFontOut);
@@ -73,7 +73,7 @@ class FormatterJSON {
       text: textFont.getText(),
       charSpacing: textFont.charSpacing,
       wordSpacing: textFont.wordSpacing,
-    }
+    };
   }
 
   /**
@@ -89,11 +89,11 @@ class FormatterJSON {
       const txtObjOut = this.formatTextObject(textObject);
       txtData.push(txtObjOut);
     });
-    let output = {
-      "data": txtData,
-    }
-    const out = JSON.stringify(output)// pretty print (output, null, 4)
-    return `"${page.pageIndex}": ${out}${last ? '': ','}`
+    const output = {
+      data: txtData,
+    };
+    const out = JSON.stringify(output); // pretty print (output, null, 4)
+    return `"${page.pageIndex}": ${out}${last ? '' : ','}`;
   }
 
   /**
@@ -103,7 +103,7 @@ class FormatterJSON {
   end () {
     return `}
     }
-    `
+    `;
   }
 }
 

src/pdf/formatters/FormatterText.js

@@ -9,7 +9,7 @@ class FormatterText {
    * @returns {string}
    */
   start (doc, metadata) {
-    return ''
+    return '';
   }
 
   /**
@@ -69,9 +69,8 @@ class FormatterText {
    * @returns {string}
    */
   end () {
-    return ''
+    return '';
   }
-
 }
 
 module.exports = FormatterText;

src/pdf/formatters/FormatterXML.js

@@ -11,7 +11,7 @@ class FormatterXML {
   start (doc, metadata) {
     return `<?xml version="1.0" encoding="UTF-8"?>
     <document>
-    `
+    `;
   }
 
   /**
@@ -23,7 +23,7 @@ class FormatterXML {
    */
   format (page, data, last) {
     const output = '';
-    return output
+    return output;
   }
 
   /**
@@ -31,7 +31,7 @@ class FormatterXML {
    * @returns {string}
    */
   end () {
-    return '</document>'
+    return '</document>';
   }
 }
 

src/pdf/formatters/index.js

@@ -5,5 +5,5 @@ const FormatterXML = require('./FormatterXML');
 module.exports = {
   FormatterXML,
   FormatterJSON,
-  FormatterText
-}
+  FormatterText,
+};

src/pdf/model/FontObject.js

@@ -57,4 +57,4 @@ class FontObject {
     }
   }
 }
-module.exports = FontObject
+module.exports = FontObject;

src/pdf/model/PdfObject.js

@@ -19,4 +19,4 @@ class PdfObject {
   }
 }
 
-module.exports = PdfObject
+module.exports = PdfObject;

src/pdf/model/PdfPage.js

@@ -26,8 +26,8 @@ class PdfPage extends PdfObject {
     this.textHScale = 1;
     this.textRise = 0;
 
-    this.currentObject;
-    this.currentFont;
+    this.currentObject = null;
+    this.currentFont = null;
   }
 
   /**
@@ -35,7 +35,7 @@ class PdfPage extends PdfObject {
    * @param {TextObject} obj
    */
   setCurrentObject (obj) {
-    this.currentObject = obj
+    this.currentObject = obj;
     this.objectList.push(obj);
   }
 }

src/pdf/model/TextObject.js

@@ -8,6 +8,9 @@ const TextLine = require('./text/TextLine');
  * with multiple lines in pdf document
  */
 class TextObject extends PdfObject {
+  /**
+   * Constructor
+   */
   constructor () {
     super();
     this.textMatrix = Constraints.IDENTITY_MATRIX;
@@ -32,7 +35,7 @@ class TextObject extends PdfObject {
    * @returns {TextLine}
    */
   getLine () {
-    return this._textLines[this._textLines.length -1]
+    return this._textLines[this._textLines.length - 1];
   }
 
   /**

src/pdf/model/index.js

@@ -12,4 +12,4 @@ module.exports = {
   TextFont,
   TextLine,
   PdfPage,
-}
+};

src/pdf/model/text/TextFont.js

@@ -5,10 +5,13 @@ const PdfObject = require('./../PdfObject');
  * @extends {PdfObject}
  */
 class TextFont extends PdfObject {
+  /**
+   * Constructor
+   */
   constructor () {
     super();
     this.font = null;
-    this._text = "";
+    this._text = '';
     this.charSpacing = 0;
     this.wordSpacing = 0;
     this._tolerance = 5;

src/pdf/model/text/TextLine.js

@@ -1,16 +1,19 @@
 const PdfObject = require('./../PdfObject');
-const TextFont = require('./../text/TextFont');
 
 /**
  * Represents text line in pdf file
  * @extends {PdfObject}
  */
 class TextLine extends PdfObject {
+  /**
+   * Constructor
+   */
   constructor () {
     super();
     this._textFonts = [];
     this.width = 0;
   }
+
   /**
    * Adds line with font to text
    * @param {TextFont} line
@@ -25,16 +28,17 @@ class TextLine extends PdfObject {
    */
   getLastFontText () {
     if (this._textFonts.length > 0) {
-      return this._textFonts[this._textFonts.length - 1]
+      return this._textFonts[this._textFonts.length - 1];
     }
-    return null
+    return null;
   }
+
   /**
    * Output line text to the console
    */
   printText () {
-    let txt = "";
-    this._textFonts.forEach((el) => txt += el.getText());
+    let txt = '';
+    this._textFonts.forEach((el) => { txt += el.getText(); });
     console.log(txt);
   }
 

src/pdf/visitors/VisitorImage.js

@@ -7,7 +7,11 @@ const FileManager = require('../FileManager');
  * @extends {VisitorBase}
  */
 class VisitorImage extends VisitorBase {
-
+  /**
+   * Constructor
+   * @param config
+   * @param {PdfPage} page
+   */
   constructor (config, page) {
     super(config, page);
     FileManager.mkdirNotExists(`${this.config.outputDir}/img`);
@@ -19,11 +23,12 @@ class VisitorImage extends VisitorBase {
   paintJpegXObject (args) {
     if (this.config.debug) console.log('paintJpegXObject');
     // if (this.config.skip) return;
-    const objId = args[1], w = args[1], h = args[2];
-    console.log(objId);
+    const objId = args[1];
+    const w = args[1];
+    const h = args[2];
+    console.log(objId, w, h);
   }
 
-
   /**
    * pdf.OPS.paintImageXObject
    */
@@ -54,7 +59,7 @@ 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}.${args[1]}.png`
+    const fpath = `${this.config.outputDir}/img/page.${this.page.data.pageIndex}.${args[1]}.png`;
     await FileManager.saveFileAsync(fpath, imgBinary);
   }
 }

src/pdf/visitors/VisitorText.js

@@ -18,12 +18,11 @@ class VisitorText extends VisitorBase {
   beginText (args) {
     if (this.config.debug) console.log('beginText');
     if (this.config.skip) return;
-    if(!this.page.currentObject || !("textMatrixScale" in this.page.currentObject)) {
+    if (!this.page.currentObject || !('textMatrixScale' in this.page.currentObject)) {
       this.page.setCurrentObject(new Model.TextObject());
       // SHOULD determine if new line while extracting text cause it can begin in any time
       this.page.currentObject.newLine();
     }
-
   }
 
   /**
@@ -39,9 +38,10 @@ class VisitorText extends VisitorBase {
    * pdf.OPS.setLeadingMoveText
    */
   setLeadingMoveText (args) {
-    if(debug) console.log('setLeadingMoveText');
+    if (this.config.debug) console.log('setLeadingMoveText');
     if (this.config.skip) return;
-    const x = args[0], y = args[1];
+    const x = args[0];
+    const y = args[1];
     this.page.leading = -y;
     this.moveText(x, y);
   }
@@ -52,23 +52,23 @@ class VisitorText extends VisitorBase {
   setFont (args) {
     if (this.config.debug) console.log('setFont');
     if (this.config.skip) return;
-    this.txt.setFont(args, this.page)
+    this.txt.setFont(args, this.page);
   }
 
   /**
    * pdf.OPS.showText
    */
   showText (args) {
-    if (this.config.debug) console.log("showText");
+    if (this.config.debug) console.log('showText');
     if (this.config.skip) return;
-    this.txt.showText(args[0], this.page)
+    this.txt.showText(args[0], this.page);
   }
 
   /**
    * pdf.OPS.showSpacedText
    */
   showSpacedText (args) {
-    if (this.config.debug) console.log("showSpacedText");
+    if (this.config.debug) console.log('showSpacedText');
     if (this.config.skip) return;
     this.txt.setText(args[0], this.page);
   }
@@ -83,8 +83,8 @@ class VisitorText extends VisitorBase {
       this.page.x = args[0];
       this.page.y = args[1];
     } else {
-      this.page.x = args[0]
-      this.page.y += args[1]
+      this.page.x = args[0];
+      this.page.y += args[1];
     }
   }
 
@@ -92,16 +92,16 @@ class VisitorText extends VisitorBase {
    * pdf.OPS.endText
    */
   endText (args) {
-    if (this.debug) console.log('endText');
-    if (this.config.skip) return;
+    if (this.config.debug) console.log('endText');
+    // if (this.config.skip) return;
   }
 
   /**
    * pdf.OPS.setCharSpacing
    */
   setCharSpacing (args) {
-    if (this.debug) console.log('setCharSpacing');
-    if (this.config.skip) return;
+    if (this.config.debug) console.log('setCharSpacing');
+    // if (this.config.skip) return;
   }
 
   /**
@@ -109,24 +109,29 @@ class VisitorText extends VisitorBase {
    */
   setWordSpacing (args) {
     if (this.debug) console.log('setWordSpacing');
-    if (this.config.skip) return;
+    // if (this.config.skip) return;
   }
 
   /**
    * pdf.OPS.setHScale
    */
   setHScale (args) {
-    if (this.debug) console.log('setHScale');
-    if (this.config.skip) return;
+    if (this.config.debug) console.log('setHScale');
+    // if (this.config.skip) return;
   }
 
   /**
    * pdf.OPS.setTextMatrix
    */
   setTextMatrix (args) {
-    if (this.debug) console.log('setWordSpacing');
+    if (this.config.debug) console.log('setWordSpacing');
     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 a = args[0];
+    const b = args[1];
+    const c = args[2];
+    const d = args[3];
+    const e = args[4];
+    const f = args[5];
     this.page.currentObject.textMatrix = this.page.currentObject.lineMatrix = [a, b, c, d, e, f];
     this.page.x = e;
     this.page.y = f;
@@ -136,25 +141,25 @@ class VisitorText extends VisitorBase {
    * pdf.OPS.setTextRise
    */
   setTextRise (args) {
-    if (this.debug) console.log('setTextRise');
-    if (this.config.skip) return;
+    if (this.config.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;
+    if (this.config.debug) console.log('setTextRenderingMode');
+    // if (this.config.skip) return;
   }
 
   /**
    * pdf.OPS.nextLine
    */
   nextLine (args) {
-    if (this.debug) console.log('nextLine');
-    if (this.config.skip) return;
+    if (this.config.debug) console.log('nextLine');
+    // if (this.config.skip) return;
   }
 }
 
-module.exports = VisitorText
+module.exports = VisitorText;

src/pdf/visitors/VisitorXObject.js

@@ -6,7 +6,6 @@ const VisitorBase = require('./VisitorBase');
  * @extends {VisitorBase}
  */
 class VisitorXObject extends VisitorBase {
-
   /**
    * pdf.OPS.paintFormXObjectBegin
    */
@@ -26,4 +25,4 @@ class VisitorXObject extends VisitorBase {
   }
 }
 
-module.exports = VisitorXObject
+module.exports = VisitorXObject;

src/pdf/visitors/index.js

@@ -4,4 +4,4 @@ const VisitorImage = require('./VisitorImage');
 
 module.exports = {
   VisitorText, VisitorXObject, VisitorImage,
-}
+};

src/pdfjs.js

@@ -6,7 +6,7 @@ const ImageKind = util.ImageKind;
 /**
  * See pdf.js/src/display/svg.js
  */
-const convertImgDataToPng = (function() {
+const convertImgDataToPng = (() => {
   const PNG_HEADER =
     new Uint8Array([0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a]);
   const CHUNK_WRAPPER_SIZE = 12;
@@ -99,14 +99,15 @@ const convertImgDataToPng = (function() {
         input = literals;
       } else {
         // eslint-disable-next-line no-undef
-        input = new Buffer(literals);
+        // input = new Buffer(literals);
+        input = Buffer.from(literals);
       }
       // const output = __non_webpack_require__('zlib')
       const output = zlib
-        .deflateSync(input, { level: 9, });
+        .deflateSync(input, { level: 9 });
       return output instanceof Uint8Array ? output : new Uint8Array(output);
     } catch (e) {
-      warn('Not compressing PNG because zlib.deflateSync is unavailable: ' + e);
+      console.warn('Not compressing PNG because zlib.deflateSync is unavailable: ' + e);
     }
 
     return deflateSyncUncompressed(literals);
@@ -182,7 +183,8 @@ const convertImgDataToPng = (function() {
 
     // prefix every row with predictor 0
     const literals = new Uint8Array((1 + lineSize) * height);
-    let offsetLiterals = 0, offsetBytes = 0;
+    let offsetLiterals = 0;
+    let offsetBytes = 0;
     for (let y = 0; y < height; ++y) {
       literals[offsetLiterals++] = 0; // no prediction
       literals.set(bytes.subarray(offsetBytes, offsetBytes + lineSize),
@@ -215,7 +217,7 @@ const convertImgDataToPng = (function() {
       colorType, // color type
       0x00, // compression method
       0x00, // filter method
-      0x00 // interlace method
+      0x00, // interlace method
     ]);
     const idat = deflateSync(literals);
 
@@ -236,8 +238,7 @@ const convertImgDataToPng = (function() {
   }
 
   return function convertImgDataToPng (imgData, forceDataSchema, isMask) {
-    const kind = (imgData.kind === undefined ?
-      ImageKind.GRAYSCALE_1BPP : imgData.kind);
+    const kind = (imgData.kind === undefined ? ImageKind.GRAYSCALE_1BPP : imgData.kind);
     return encode(imgData, kind, forceDataSchema, isMask);
   };
 })();