remove dependency: async

Author: modestysn

bin/pdf2json

@@ -1,6 +1,4 @@
 #!/usr/bin/env node
 
-'use strict';
-
-var P2JCMD = require('../lib/p2jcmd');
+const P2JCMD = require('../lib/p2jcmd');
 new P2JCMD().start();

lib/p2jcmd.js

@@ -1,8 +1,6 @@
-
 const nodeUtil = require("util"),
     fs = require("fs"),
     path = require("path"),
-    async = require("async"),    
     {ParserStream, StringifyStream} = require("./parserstream"),
     pkInfo = require("../package.json"),
     PDFParser = require("../pdfparser");
@@ -96,35 +94,36 @@ class PDFProcessor {
 		this.pdfParser.getAllFieldsTypesStream().pipe(new StringifyStream()).pipe(outputStream);
 	}
 
-	#processAdditionalStreams(outputTasks, callback) {
-		if (PROCESS_FIELDS_CONTENT) {//needs to generate fields.json file
-			outputTasks.push(cbFunc => this.#generateFieldsTypesStream(cbFunc));
-		}
-		if (PROCESS_RAW_TEXT_CONTENT) {//needs to generate content.txt file
-			outputTasks.push(cbFunc => this.#generateRawTextContentStream(cbFunc));
-		}
-		if (PROCESS_MERGE_BROKEN_TEXT_BLOCKS) {//needs to generate json file with merged broken text blocks
-			outputTasks.push(cbFunc => this.#generateMergedTextBlocksStream(cbFunc));
-		}
-
-		if (outputTasks.length > 0) {
-			async.series(outputTasks, (err, results) => {//additional streams process complete
-				if (err) {
-                    this.curCLI.addStatusMsg(err, `[+]=> ${err}`);
-				} else {
-                    results.forEach( r => this.curCLI.addStatusMsg(null, `[+]=> ${r}`));
-				}
-				this.#continue(callback);
-			});
-		}
-		else {
-			this.#continue(callback);
-		}
+	#processAdditionalStreams(callback) {
+        const outputTasks = [];
+        if (PROCESS_FIELDS_CONTENT) {//needs to generate fields.json file
+            outputTasks.push(cbFunc => this.#generateFieldsTypesStream(cbFunc));
+        }
+        if (PROCESS_RAW_TEXT_CONTENT) {//needs to generate content.txt file
+            outputTasks.push(cbFunc => this.#generateRawTextContentStream(cbFunc));
+        }
+        if (PROCESS_MERGE_BROKEN_TEXT_BLOCKS) {//needs to generate json file with merged broken text blocks
+            outputTasks.push(cbFunc => this.#generateMergedTextBlocksStream(cbFunc));
+        }
+
+        let taskId = 0;
+        function sequenceTask() {
+            if (taskId < outputTasks.length) {
+                outputTasks[taskId]((err, ret) => {
+                    this.curCLI.addStatusMsg(err, `[+]=> ${ret}`);
+                    taskId++;
+                    sequenceTask.call(this);
+                });    
+            }
+            else 
+                this.#continue(callback);
+        }
+        sequenceTask.call(this);
 	}
 
 	#onPrimarySuccess(callback) {
 		this.curCLI.addResultCount();
-		this.#processAdditionalStreams([], callback);
+		this.#processAdditionalStreams(callback);
 	}
 
 	#onPrimaryError(err, callback) {

package-lock.json

@@ -47,13 +47,11 @@
     "pdf2json": "./bin/pdf2json"
   },
   "dependencies": {
-    "async": "^3.2.1",
     "@xmldom/xmldom": "^0.7.5",
     "yargs": "^17.2.1"
   },
   "devDependencies": {},
   "bundledDependencies": [
-    "async",
     "@xmldom/xmldom",
     "yargs"
   ],

package.json

@@ -1,7 +1,7 @@
 const fs = require("fs"),
+    { readFile } = require("fs/promises"),
     {EventEmitter} = require("events"),
 	nodeUtil = require("util"),    
-    async = require("async"),
 	PDFJS = require("./lib/pdf"),
     {ParserStream} = require("./lib/parserstream"),
     {kColors, kFontFaces, kFontStyles} = require("./lib/pdfconst");
@@ -21,7 +21,6 @@ class PDFParser extends EventEmitter { // inherit from event emitter
     #password = "";
 
     #context = null; // service context object, only used in Web Service project; null in command line
-    #fq = null; //async queue for reading files
 
     #pdfFilePath = null; //current PDF file to load and parse, null means loading/parsing not started
     #pdfFileMTime = null; // last time the current pdf was modified, used to recognize changes and ignore cache
@@ -37,9 +36,6 @@ class PDFParser extends EventEmitter { // inherit from event emitter
         // private
         // service context object, only used in Web Service project; null in command line
         this.#context = context;
-        this.#fq = async.queue( (task, callback) => {
-            fs.readFile(task.path, callback);
-        }, 1);
 
         this.#pdfFilePath = null; //current PDF file to load and parse, null means loading/parsing not started
         this.#pdfFileMTime = null; // last time the current pdf was modified, used to recognize changes and ignore cache
@@ -104,37 +100,34 @@ class PDFParser extends EventEmitter { // inherit from event emitter
 		return false;
 	}
 
-	#processPDFContent(err, data) {
-		nodeUtil.p2jinfo("Load PDF file status:" + (!!err ? "Error!" : "Success!") );
-		if (err) {
-			this.#data = null;
-			this.emit("pdfParser_dataError", err);
-		}
-		else {
-			PDFParser.#binBuffer[this.binBufferKey] = data;
-			this.#startParsingPDF();
-		}
-	};
-
 	//public APIs
     createParserStream() {
         return new ParserStream(this, {objectMode: true, bufferSize: 64 * 1024});
     }
 
-	loadPDF(pdfFilePath, verbosity) {
+	async loadPDF(pdfFilePath, verbosity) {
 		nodeUtil.verbosity(verbosity || 0);
 		nodeUtil.p2jinfo("about to load PDF file " + pdfFilePath);
 
 		this.#pdfFilePath = pdfFilePath;
-		this.#pdfFileMTime = fs.statSync(pdfFilePath).mtimeMs;
-		if (this.#processFieldInfoXML) {
-			this.#PDFJS.tryLoadFieldInfoXML(pdfFilePath);
-		}
-
-		if (this.#processBinaryCache())
-			return;
 
-		this.#fq.push({path: pdfFilePath}, this.#processPDFContent.bind(this));
+		try {
+            this.#pdfFileMTime = fs.statSync(pdfFilePath).mtimeMs;
+            if (this.#processFieldInfoXML) {
+                this.#PDFJS.tryLoadFieldInfoXML(pdfFilePath);
+            }
+
+            if (this.#processBinaryCache())
+                return;
+        
+            PDFParser.#binBuffer[this.binBufferKey] = await readFile(pdfFilePath);
+            nodeUtil.p2jinfo(`Load OK: ${pdfFilePath}`);
+            this.#startParsingPDF();
+        }
+        catch(err) {
+            nodeUtil.p2jerror(`Load Failed: ${pdfFilePath} - ${err}`);
+            this.emit("pdfParser_dataError", err);
+        }
 	}
 
 	// Introduce a way to directly process buffers without the need to write it to a temporary file

pdfparser.js

@@ -888,7 +888,7 @@ In order to support this auto merging capability, text block objects have an add
   * More test coverage, 4 more test scripts added, see _package.json_ for details
   * Easier access to dictionaries, including color, font face and font style, see Dictionary reference section for details
   * Refactor to ES6 class for major entry modules
-  * Dependency is removed: lodash.
+  * Dependencies removed: lodash, async
   * Upgrade to Node v14.18.0 LTSs
 
 ### Install on Ubuntu