mindee 1.0.7 → 1.0.8

package/lib/api/financialDocument.js

@@ -40,9 +40,9 @@ class APIFinancialDocument extends APIObject {
       } = _ref;
       var inputFile = new Input({
         file: input,
-        inputType,
-        filename,
-        cutPdf
+        inputType: inputType,
+        filename: filename,
+        allowCutPdf: cutPdf
       });
       _this.apiToken = inputFile.fileExtension === "pdf" ? _this.invoiceToken : _this.receiptToken;
       yield inputFile.init();

package/lib/api/invoice.js

@@ -43,9 +43,9 @@ class APIInvoice extends APIObject {
 
       var inputFile = new Input({
         file: input,
-        inputType,
-        filename,
-        cutPdf
+        inputType: inputType,
+        filename: filename,
+        allowCutPdf: cutPdf
       });
       yield inputFile.init();
       var url = "v".concat(version, "/predict");

package/lib/api/receipt.js

@@ -43,9 +43,9 @@ class APIReceipt extends APIObject {
 
       var inputFile = new Input({
         file: input,
-        inputType,
-        filename,
-        cutPdf
+        inputType: inputType,
+        filename: filename,
+        allowCutPdf: cutPdf
       });
       yield inputFile.init();
       var url = "v".concat(version, "/predict");

package/lib/api/request.js

@@ -55,9 +55,13 @@ var request = function request(url, method, headers, input) {
         responseBody += chunk;
       });
       res.on("end", function () {
-        resolve(_objectSpread(_objectSpread({}, res), {}, {
-          data: JSON.parse(responseBody)
-        }));
+        try {
+          resolve(_objectSpread(_objectSpread({}, res), {}, {
+            data: JSON.parse(responseBody)
+          }));
+        } catch (error) {
+          console.log(responseBody);
+        }
       });
     });
     req.on("error", err => {

package/lib/inputs.js

@@ -16,21 +16,20 @@ var {
   PDFDocument
 } = require("pdf-lib");
 
-var magic = require("stream-mmmagic");
-
 var concat = require("concat-stream");
 
 var {
   Base64Encode
 } = require("base64-stream");
 
-var ReadableStreamClone = require("readable-stream-clone");
+var fileType = require("file-type");
+
+var ArrayBufferEncode = require("base64-arraybuffer");
 
 class Input {
   /**
    * @param {(String | Buffer)} file - the file that will be read. Either path or base64 string, or a steam
    * @param {String} inputType - the type of input used in file ("base64", "path", "dummy").
-   *                             NB: in case of base64 file, only jpeg file is supported
    *                             NB: dummy is only used for tests purposes
    * @param {String} filename - File name of the input
    * @param {Boolean} cut_pdf: Automatically reconstruct pdf with more than 4 pages
@@ -71,53 +70,60 @@ class Input {
     var _this = this;
 
     return _asyncToGenerator(function* () {
-      if (_this.inputType === "base64") _this.initBase64();else if (_this.inputType === "path") yield _this.initFile();else if (_this.inputType === "stream") yield _this.initStream();else _this.initDummy();
+      if (_this.inputType === "base64") yield _this.initBase64();else if (_this.inputType === "path") yield _this.initFile();else if (_this.inputType === "stream") yield _this.initStream();else _this.initDummy();
     })();
   }
 
   initBase64() {
-    this.fileObject = this.file;
-    this.filepath = undefined;
-    this.fileExtension = undefined;
+    var _this2 = this;
+
+    return _asyncToGenerator(function* () {
+      _this2.fileObject = _this2.file;
+      _this2.filepath = undefined;
+      _this2.fileExtension = undefined;
+
+      if (_this2.allowCutPdf == true) {
+        var typeOfFile = yield fileType.fromBuffer(Buffer.from(_this2.fileObject, "base64"));
+
+        if (typeOfFile === undefined) {
+          console.error("Cannot detect mime type of: ".concat(_this2.fileObject));
+        } else if (typeOfFile.mime === "application/pdf") {
+          yield _this2.cutPdf();
+        }
+      }
+    })();
   }
 
   initFile() {
-    var _this2 = this;
+    var _this3 = this;
 
     return _asyncToGenerator(function* () {
-      _this2.fileObject = yield fs.readFile(_this2.file);
-      _this2.filepath = _this2.file;
-      _this2.filename = _this2.filename || path.basename(_this2.file); // Check if file type is valid
+      _this3.fileObject = yield fs.readFile(_this3.file);
+      _this3.filepath = _this3.file;
+      _this3.filename = _this3.filename || path.basename(_this3.file); // Check if file type is valid
 
-      var filetype = _this2.filename.split(".").pop();
+      var filetype = _this3.filename.split(".").pop();
 
-      if (!(filetype in _this2.MIMETYPES)) {
-        errorHandler.throw(new Error("File type is not allowed. It must be ".concat(Object.keys(_this2.MIMETYPES).toString())));
+      if (!(filetype in _this3.MIMETYPES)) {
+        errorHandler.throw(new Error("File type is not allowed. It must be ".concat(Object.keys(_this3.MIMETYPES).toString())));
       }
 
-      _this2.fileExtension = _this2.MIMETYPES[filetype];
+      _this3.fileExtension = _this3.MIMETYPES[filetype];
 
-      if (_this2.fileExtension === "application/pdf" && _this2.allowCutPdf == true) {
-        yield _this2.cutPdf();
+      if (_this3.fileExtension === "application/pdf" && _this3.allowCutPdf == true) {
+        yield _this3.cutPdf();
       }
     })();
   }
 
   initStream() {
-    var _this3 = this;
+    var _this4 = this;
 
     return _asyncToGenerator(function* () {
-      _this3.fileObject = _this3.file;
-      _this3.filename = _this3.filename || "stream";
-      _this3.filepath = undefined; //Copy the ReadableStream
-
-      var stream = new ReadableStreamClone(_this3.fileObject);
-      _this3.fileObject = new ReadableStreamClone(_this3.fileObject);
-      var [mime, output] = yield magic.promise(stream);
-
-      if (mime.type === "application/pdf" && _this3.allowCutPdf == true) {
-        yield _this3.cutPdf();
-      }
+      var file = yield _this4.streamToBase64(_this4.file);
+      _this4.file = file;
+      _this4.inputType = "base64";
+      yield _this4.initBase64();
     })();
   }
 
@@ -137,7 +143,7 @@ class Input {
 
   streamToBase64(stream) {
     return _asyncToGenerator(function* () {
-      return yield new Promise((resolve, reject) => {
+      return new Promise((resolve, reject) => {
         var base64 = new Base64Encode();
 
         var cbConcat = base64 => {
@@ -154,26 +160,26 @@ class Input {
 
 
   cutPdf() {
-    var _this4 = this;
+    var _this5 = this;
 
     return _asyncToGenerator(function* () {
       // convert document to PDFDocument & cut CUT_PDF_SIZE - 1 first pages and last page
-      var pdfDocument;
-
-      if (_this4.filename == "stream") {
-        pdfDocument = yield PDFDocument.load(yield _this4.streamToBase64(_this4.fileObject));
-      } else {
-        pdfDocument = yield PDFDocument.load(_this4.fileObject);
-      }
-
+      var pdfDocument = yield PDFDocument.load(_this5.fileObject, {
+        ignoreEncryption: true
+      });
       var splitedPdfDocument = yield PDFDocument.create();
       var pdfLength = pdfDocument.getPageCount();
-      if (pdfLength <= _this4.CUT_PDF_SIZE) return;
-      var pagesNumbers = [...Array(_this4.CUT_PDF_SIZE - 1).keys(), pdfLength - 1];
+      if (pdfLength <= _this5.CUT_PDF_SIZE) return;
+      var pagesNumbers = [...Array(_this5.CUT_PDF_SIZE - 1).keys(), pdfLength - 1];
       var pages = yield splitedPdfDocument.copyPages(pdfDocument, pagesNumbers);
       pages.forEach(page => splitedPdfDocument.addPage(page));
       var data = yield splitedPdfDocument.save();
-      _this4.fileObject = Buffer.from(data);
+
+      if (_this5.inputType === "path") {
+        _this5.fileObject = Buffer.from(data);
+      } else if (_this5.inputType === "base64") {
+        _this5.fileObject = ArrayBufferEncode.encode(data);
+      }
     })();
   }
 

package/mindee/api/financialDocument.js

@@ -24,7 +24,12 @@ class APIFinancialDocument extends APIObject {
     cutPdf = true,
     includeWords = false,
   }) {
-    const inputFile = new Input({ file: input, inputType, filename, cutPdf });
+    const inputFile = new Input({
+      file: input,
+      inputType: inputType,
+      filename: filename,
+      allowCutPdf: cutPdf,
+    });
     this.apiToken =
       inputFile.fileExtension === "pdf" ? this.invoiceToken : this.receiptToken;
     await inputFile.init();

package/mindee/api/invoice.js

@@ -24,7 +24,12 @@ class APIInvoice extends APIObject {
     includeWords = false,
   }) {
     super.parse();
-    const inputFile = new Input({ file: input, inputType, filename, cutPdf });
+    const inputFile = new Input({
+      file: input,
+      inputType: inputType,
+      filename: filename,
+      allowCutPdf: cutPdf,
+    });
     await inputFile.init();
     const url = `v${version}/predict`;
     return await super._request(url, inputFile, includeWords);

package/mindee/api/receipt.js

@@ -24,7 +24,12 @@ class APIReceipt extends APIObject {
     includeWords = false,
   }) {
     super.parse();
-    const inputFile = new Input({ file: input, inputType, filename, cutPdf });
+    const inputFile = new Input({
+      file: input,
+      inputType: inputType,
+      filename: filename,
+      allowCutPdf: cutPdf,
+    });
     await inputFile.init();
     const url = `v${version}/predict`;
     return await super._request(url, inputFile, includeWords);

package/mindee/api/request.js

@@ -38,10 +38,14 @@ const request = (url, method, headers, input, includeWords = false) => {
       });
 
       res.on("end", function () {
-        resolve({
-          ...res,
-          data: JSON.parse(responseBody),
-        });
+        try {
+          resolve({
+            ...res,
+            data: JSON.parse(responseBody),
+          });
+        } catch (error) {
+          console.log(responseBody);
+        }
       });
     });
 

package/mindee/index.js

@@ -3,6 +3,7 @@ const logger = require("./logger");
 const APIReceipt = require("./api/receipt");
 const APIInvoice = require("./api/invoice");
 const APIFinancialDocument = require("./api/financialDocument");
+
 class Client {
   /**
    * @param {string} receiptToken - Receipt Expense Token from Mindee dashboard

package/mindee/inputs.js

@@ -2,10 +2,10 @@ const fs = require("fs").promises;
 const errorHandler = require("./errors/handler");
 const path = require("path");
 const { PDFDocument } = require("pdf-lib");
-const magic = require("stream-mmmagic");
 const concat = require("concat-stream");
 const { Base64Encode } = require("base64-stream");
-const ReadableStreamClone = require("readable-stream-clone");
+const fileType = require("file-type");
+const ArrayBufferEncode = require("base64-arraybuffer");
 
 class Input {
   MIMETYPES = {
@@ -21,7 +21,6 @@ class Input {
   /**
    * @param {(String | Buffer)} file - the file that will be read. Either path or base64 string, or a steam
    * @param {String} inputType - the type of input used in file ("base64", "path", "dummy").
-   *                             NB: in case of base64 file, only jpeg file is supported
    *                             NB: dummy is only used for tests purposes
    * @param {String} filename - File name of the input
    * @param {Boolean} cut_pdf: Automatically reconstruct pdf with more than 4 pages
@@ -44,16 +43,28 @@ class Input {
   }
 
   async init() {
-    if (this.inputType === "base64") this.initBase64();
+    if (this.inputType === "base64") await this.initBase64();
     else if (this.inputType === "path") await this.initFile();
     else if (this.inputType === "stream") await this.initStream();
     else this.initDummy();
   }
 
-  initBase64() {
+  async initBase64() {
     this.fileObject = this.file;
     this.filepath = undefined;
     this.fileExtension = undefined;
+
+    if (this.allowCutPdf == true) {
+      const typeOfFile = await fileType.fromBuffer(
+        Buffer.from(this.fileObject, "base64")
+      );
+
+      if (typeOfFile === undefined) {
+        console.error(`Cannot detect mime type of: ${this.fileObject}`);
+      } else if (typeOfFile.mime === "application/pdf") {
+        await this.cutPdf();
+      }
+    }
   }
 
   async initFile() {
@@ -80,19 +91,11 @@ class Input {
   }
 
   async initStream() {
-    this.fileObject = this.file;
-    this.filename = this.filename || "stream";
-    this.filepath = undefined;
-
-    //Copy the ReadableStream
-    const stream = new ReadableStreamClone(this.fileObject);
-    this.fileObject = new ReadableStreamClone(this.fileObject);
+    const file = await this.streamToBase64(this.file);
 
-    const [mime, output] = await magic.promise(stream);
-
-    if (mime.type === "application/pdf" && this.allowCutPdf == true) {
-      await this.cutPdf();
-    }
+    this.file = file;
+    this.inputType = "base64";
+    await this.initBase64();
   }
 
   initDummy() {
@@ -109,7 +112,7 @@ class Input {
    * @returns Base64 encoded String
    */
   async streamToBase64(stream) {
-    return await new Promise((resolve, reject) => {
+    return new Promise((resolve, reject) => {
       const base64 = new Base64Encode();
 
       const cbConcat = (base64) => {
@@ -128,16 +131,9 @@ class Input {
   /** Cut PDF if pages > 5 */
   async cutPdf() {
     // convert document to PDFDocument & cut CUT_PDF_SIZE - 1 first pages and last page
-    let pdfDocument;
-
-    if (this.filename == "stream") {
-      pdfDocument = await PDFDocument.load(
-        await this.streamToBase64(this.fileObject)
-      );
-    } else {
-      pdfDocument = await PDFDocument.load(this.fileObject);
-    }
-
+    let pdfDocument = await PDFDocument.load(this.fileObject, {
+      ignoreEncryption: true,
+    });
     const splitedPdfDocument = await PDFDocument.create();
     const pdfLength = pdfDocument.getPageCount();
     if (pdfLength <= this.CUT_PDF_SIZE) return;
@@ -148,7 +144,11 @@ class Input {
     const pages = await splitedPdfDocument.copyPages(pdfDocument, pagesNumbers);
     pages.forEach((page) => splitedPdfDocument.addPage(page));
     const data = await splitedPdfDocument.save();
-    this.fileObject = Buffer.from(data);
+    if (this.inputType === "path") {
+      this.fileObject = Buffer.from(data);
+    } else if (this.inputType === "base64") {
+      this.fileObject = ArrayBufferEncode.encode(data);
+    }
   }
 }
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mindee",
-  "version": "1.0.7",
+  "version": "1.0.8",
   "description": "Mindee API SDK for Node.js",
   "main": "mindee/index.js",
   "license": "GPL-3.0",
@@ -34,19 +34,18 @@
     "prettier": "^2.2.0"
   },
   "dependencies": {
+    "base64-arraybuffer": "^1.0.1",
     "base64-stream": "^1.0.0",
     "concat-stream": "^2.0.0",
+    "file-type": "^16.5.3",
     "form-data": "^3.0.0",
-    "pdf-lib": "^1.13.0",
-    "readable-stream-clone": "^0.0.7",
-    "stream-mmmagic": "^2.3.0"
+    "pdf-lib": "^1.13.0"
   },
   "keywords": [
     "javascript",
     "mindee",
     "api",
     "SDK",
-    "nodejs",
-    "OCR"
+    "nodejs"
   ]
 }