v1.1.1: fix issue #42, #43 and #54

base/shared/util.js

@@ -188,10 +188,10 @@ function error(msg) {
     // Join the arguments into a single string for the lines below.
     msg = [].join.call(arguments, ' ');
   } else {
-    log('Error: ' + msg);
+    //log('Error: ' + msg);
   }
-  log(backtrace());
-  PDFJS.LogManager.notify('error', msg);
+  //log(backtrace());
+  //PDFJS.LogManager.notify('error', msg);
   throw new Error(msg);
 }
 

lib/p2jcmd.js

@@ -36,7 +36,7 @@ let PDF2JSONUtil = (function () {
 
     let _continue = function(callback, err) {
         if (err)
-            console.warn(err);
+            console.error(err);
         if (nodeUtil.isFunction(callback))
             callback(err);
     };
@@ -122,7 +122,7 @@ let PDF2JSONUtil = (function () {
 
                 async.series(outputTasks, function(err, results){
                     if (err) {
-                        console.warn("Error: " + err);
+                        console.error("Error: " + err);
                     } else {
                         console.log("Output files OK", results);
                     }
@@ -181,7 +181,7 @@ let PDF2JSONUtil = (function () {
             retVal = "Input error: input file name doesn't have pdf extention  - " + this.inputFile + ".";
         else {
             this.outputFile = path.basename(this.inputPath, inExtName) + ".json";
-            this.outputPath = this.outputDir + path.sep + this.outputFile;
+            this.outputPath = this.outputDir + this.outputFile;
             if (fs.existsSync(this.outputPath))
                 console.log("Output file will be replaced - " + this.outputPath);
             else {
@@ -291,8 +291,8 @@ let PDFProcessor = (function () {
 
         process.nextTick( () => {
             console.timeEnd(_PRO_TIMER);
-            let exitCode = (this.inputCount === this.successCount) ? 0 : 1;
-            process.exit(exitCode);
+            //let exitCode = (this.inputCount === this.successCount) ? 0 : 1;
+            process.exit(0);
         });
     };
 
@@ -302,7 +302,7 @@ let PDFProcessor = (function () {
 
         this.inputCount = 1;
         this.p2j = new PDF2JSONUtil(inputDir, inputFile, this);
-        this.p2j.processFile( () => this.complete() );
+        this.p2j.processFile( data => this.complete(data) );
     };
 
     cls.prototype.processFiles = function(inputDir, files) {

lib/pdf.js

@@ -188,7 +188,7 @@ let PDFPageParser = (function () {
                     },
                     err => console.error("pdfPage.getAnnotations error:" + err));
             },
-            error => pageViewDrawCallback.call(this, error)
+            err => pageViewDrawCallback.call(this, err)
         );
     };
 
@@ -240,18 +240,25 @@ let PDFJSClass = (function () {
     // inherit from event emitter
 	nodeUtil.inherits(cls, nodeEvents.EventEmitter);
 
+    cls.prototype.raiseErrorEvent = function(errMsg) {
+        console.error(errMsg);
+        process.nextTick( () => this.emit("pdfjs_parseDataError", errMsg));
+        return errMsg;
+    };
+
+    cls.prototype.raiseReadyEvent = function(data) {
+        process.nextTick( () => this.emit("pdfjs_parseDataReady", data) );
+        return data;
+    };
+
+
     cls.prototype.parsePDFData = function(arrayBuffer) {
         this.pdfDocument = null;
 
         let parameters = {password: '', data: arrayBuffer};
         PDFJS.getDocument(parameters).then(
             pdfDocument => this.load(pdfDocument, 1),
-            error => {
-                let errMsg = "An error occurred while parsing the PDF: " + error;
-                nodeUtil.p2jwarn(errMsg);
-                this.emit("pdfjs_parseDataError", errMsg);
-            },
-            progressData => nodeUtil.p2jinfo("Loading progress: " + progressData.loaded / progressData.total + "%")
+            error => this.raiseErrorEvent("An error occurred while parsing the PDF: " + error)
         );
     };
 
@@ -289,7 +296,7 @@ let PDFJSClass = (function () {
 
         pagesPromise.then(
 		    promisedPages => this.parsePage(promisedPages, 0, 1.5),
-	        error => console.log("pagesPromise error: " + error)
+            error => this.raiseErrorEvent("pagesPromise error: " + error)
         );
 
         pdfDocument.getMetadata().then(
@@ -298,7 +305,7 @@ let PDFJSClass = (function () {
                 this.metadata = data.metadata;
 				this.parseMetaData();
             },
-	        error => console.log("pdfDocument.getMetadata error: " + error)
+            error => this.raiseErrorEvent("pdfDocument.getMetadata error: " + error)
         );
     };
 
@@ -325,7 +332,7 @@ let PDFJSClass = (function () {
             formAttr.Parent = _getMetaDataInt(metadata, 'pdfx:parent');
         }
 
-        this.emit("pdfjs_parseDataReady", {Transcoder: _PARSER_SIG, Agency:pdfTile, Id: formAttr});
+        this.raiseReadyEvent({Transcoder: _PARSER_SIG, Agency:pdfTile, Id: formAttr});
     };
 
     cls.prototype.parsePage = function(promisedPages, id, scale) {
@@ -337,7 +344,7 @@ let PDFJSClass = (function () {
         function continueOnNextPage() {
             nodeUtil.p2jinfo("complete parsing page:" + (id+1));
             if (id === (this.pdfDocument.numPages - 1) ) {
-                this.emit("pdfjs_parseDataReady", {Pages:this.pages, Width: this.pageWidth});
+	            this.raiseReadyEvent({Pages:this.pages, Width: this.pageWidth});
             }
             else {
                 process.nextTick(() => this.parsePage(promisedPages, ++id, scale));
@@ -369,14 +376,14 @@ let PDFJSClass = (function () {
 	                        nodeUtil.p2jinfo("complete parsing raw text content:" + (id+1));
 	                        continueOnNextPage.call(this);
 	                    },
-		                error => console.error("pdfPage.getTextContent error: " + error)
+                        error => this.raiseErrorEvent("pdfPage.getTextContent error: " + error)
 	                );
 	            }
 	            else {
 		            continueOnNextPage.call(this);
 	            }
 	        },
-	        errMsg => this.emit("pdfjs_parseDataError", errMsg)
+	        errMsg => this.raiseErrorEvent("parsePage error:" + errMsg)
         );
     };
 

lib/pdffont.js

@@ -216,6 +216,7 @@ let PDFFont = (function PFPFontClosure() {
         if (!this.bold) {
             this.bold = this.typeName.indexOf("bold") >= 0 || this.typeName.indexOf("black") >= 0;
         }
+        this.italic = fontObj.italic; // fix https://github.com/modesty/pdf2json/issues/42
 
         let typeName = this.subType;
         if (fontObj.isSerifFont) {

package.json

@@ -1,6 +1,6 @@
 {
   "name": "pdf2json",
-  "version": "1.0.9",
+  "version": "1.1.1",
   "description": "A PDF file parser that converts PDF binaries to text based JSON, powered by porting a fork of PDF.JS to Node.js",
   "keywords": [
     "pdf",
@@ -27,7 +27,8 @@
   },
   "main": "./pdfparser.js",
   "scripts": {
-    "test": "cd ./test && sh p2j.forms.sh"
+    "test": "cd ./test && sh p2j.forms.sh",
+    "test-misc": "node pdf2json.js -f ./test/pdf/misc/ -o ./test/target/misc/ -c -m"
   },
   "engines": {
     "node": ">=4.0"

readme.md

@@ -752,6 +752,13 @@ Or, from `npm scripts`:
 	npm test
 ````
 
+Some testing PDFs are provided by bug reporters, like the "unsupported encryption" ([#43](https://github.com/modesty/pdf2json/issues/43)), "read property num from undefined" ([#26](https://github.com/modesty/pdf2json/issues/26)), and "excessive line breaks in text content" ([#28](https://github.com/modesty/pdf2json/issues/28)), their PDFs are all stored in `test/pdf/misc` directory. To run tests against these community contributed PDFs, run commandline:
+
+````
+	npm run-script test-misc
+````
+
+
 ## Upgrading to ~v1.0.x
 
 If you have an early version of pdf2json, please remove your local `node_modules` directory and re-run `npm install` to upgrade to [email protected].