mindee 1.0.3 → 1.0.8

package/examples/financialDocument.js

@@ -5,7 +5,7 @@ const mindeeClient = new Client();
 
 // parsing receipt from picture
 mindeeClient.financialDocument
-  .parse("./documents/receipts/receipt.jpg")
+  .parse({ input: "./documents/receipts/receipt.jpg" })
   .then((res) => {
     console.log("Success !");
     console.log(res.financialDocuments);
@@ -21,7 +21,7 @@ const base64 = fs.readFileSync("./documents/receipts/receipt.jpg", {
   encoding: "base64",
 });
 mindeeClient.financialDocument
-  .parse(base64, "base64")
+  .parse({ input: base64, inputType: "base64" })
   .then((res) => {
     console.log("Success !");
     console.log(res.financialDocuments);

package/examples/invoice.js

@@ -6,11 +6,25 @@ const mindeeClient = new Client();
 
 // parsing invoice from pdf
 mindeeClient.invoice
-  .parse("./documents/invoices/invoice.pdf")
+  .parse({ input: "./documents/invoices/invoice.pdf" })
   .then((res) => {
     console.log("Success !");
     console.log(res.invoices);
     console.log(res.invoice);
+    console.log(res.documentType);
+  })
+  .catch((err) => {
+    console.error(err);
+  });
+
+// parsing credit note from pdf
+mindeeClient.invoice
+  .parse({ input: "./documents/invoices/credit_note.pdf" })
+  .then((res) => {
+    console.log("Success !");
+    console.log(res.invoices);
+    console.log(res.invoice);
+    console.log(res.documentType);
   })
   .catch((err) => {
     console.error(err);
@@ -18,9 +32,10 @@ mindeeClient.invoice
 
 // // parsing invoice from multiple page pdf
 mindeeClient.invoice
-  .parse("./documents/invoices/invoice_6p.pdf")
+  .parse({ input: "./documents/invoices/invoice_6p.pdf" })
   .then((res) => {
     console.log("Success !");
+    console.log(res.documentType);
     console.log(res.invoices);
     console.log(res.invoice);
   })
@@ -33,9 +48,10 @@ const base64 = fs.readFileSync("./documents/invoices/invoice.pdf", {
   encoding: "base64",
 });
 mindeeClient.invoice
-  .parse(base64, "base64")
+  .parse({ input: base64, inputType: "base64" })
   .then((res) => {
     console.log("Success !");
+    console.log(res.documentType);
     console.log(res.invoices);
     console.log(res.invoice);
   })
@@ -46,9 +62,10 @@ mindeeClient.invoice
 // parsing invoice from stream
 const stream = fs.createReadStream("./documents/invoices/invoice.pdf");
 mindeeClient.invoice
-  .parse(stream, "stream")
+  .parse({ input: stream, inputType: "stream" })
   .then((res) => {
     console.log("Success !");
+    console.log(res.documentType);
     console.log(res.invoices);
     console.log(res.invoice);
   })

package/examples/receipt.js

@@ -6,7 +6,7 @@ const mindeeClient = new Client();
 
 // parsing receipt from picture
 mindeeClient.receipt
-  .parse("./documents/receipts/receipt.jpg")
+  .parse({ input: "./documents/receipts/receipt.jpg" })
   .then((res) => {
     console.log("Success !");
     console.log(res.receipts);
@@ -21,7 +21,7 @@ const base64 = fs.readFileSync("./documents/receipts/receipt.jpg", {
   encoding: "base64",
 });
 mindeeClient.receipt
-  .parse(base64, "base64")
+  .parse({ input: base64, inputType: "base64" })
   .then((res) => {
     console.log("Success !");
     console.log(res.receipts);
@@ -34,7 +34,7 @@ mindeeClient.receipt
 // parsing receipt from stream
 const stream = fs.createReadStream("./documents/receipts/receipt.jpg");
 mindeeClient.receipt
-  .parse(stream, "stream")
+  .parse({ input: stream, inputType: "stream" })
   .then((res) => {
     console.log("Success !");
     console.log(res.receipts);

package/lib/api/financialDocument.js

@@ -24,21 +24,25 @@ class APIFinancialDocument extends APIObject {
    */
 
 
-  parse(file) {
-    var _arguments = arguments,
-        _superprop_getParse = () => super.parse,
+  parse(_ref) {
+    var _superprop_getParse = () => super.parse,
         _superprop_get_request = () => super._request,
         _this = this;
 
     return _asyncToGenerator(function* () {
-      var inputType = _arguments.length > 1 && _arguments[1] !== undefined ? _arguments[1] : "path";
-      var version = _arguments.length > 2 && _arguments[2] !== undefined ? _arguments[2] : "3";
-      var cutPdf = _arguments.length > 3 && _arguments[3] !== undefined ? _arguments[3] : true;
-      var includeWords = _arguments.length > 4 && _arguments[4] !== undefined ? _arguments[4] : false;
+      var {
+        input,
+        inputType = "path",
+        filename = undefined,
+        version = "3",
+        cutPdf = true,
+        includeWords = false
+      } = _ref;
       var inputFile = new Input({
-        file,
-        inputType,
-        cutPdf
+        file: input,
+        inputType: inputType,
+        filename: filename,
+        allowCutPdf: cutPdf
       });
       _this.apiToken = inputFile.fileExtension === "pdf" ? _this.invoiceToken : _this.receiptToken;
       yield inputFile.init();

package/lib/api/invoice.js

@@ -24,30 +24,49 @@ class APIInvoice extends APIObject {
    */
 
 
-  parse(file) {
-    var _arguments = arguments,
-        _superprop_getParse = () => super.parse,
+  parse(_ref) {
+    var _superprop_getParse = () => super.parse,
         _superprop_get_request = () => super._request,
         _this = this;
 
     return _asyncToGenerator(function* () {
-      var inputType = _arguments.length > 1 && _arguments[1] !== undefined ? _arguments[1] : "path";
-      var version = _arguments.length > 2 && _arguments[2] !== undefined ? _arguments[2] : "2";
-      var cutPdf = _arguments.length > 3 && _arguments[3] !== undefined ? _arguments[3] : true;
-      var includeWords = _arguments.length > 4 && _arguments[4] !== undefined ? _arguments[4] : false;
+      var {
+        input,
+        inputType = "path",
+        filename = undefined,
+        version = "2",
+        cutPdf = true,
+        includeWords = false
+      } = _ref;
 
       _superprop_getParse().call(_this);
 
       var inputFile = new Input({
-        file,
-        inputType,
-        cutPdf
+        file: input,
+        inputType: inputType,
+        filename: filename,
+        allowCutPdf: cutPdf
       });
       yield inputFile.init();
       var url = "v".concat(version, "/predict");
       return yield _superprop_get_request().call(_this, url, inputFile, includeWords);
     })();
   }
+  /** 
+    @param {String} inputFile - Input object
+    @param {} response - HTTP response
+    @param {Document} documentType - Document class in {"Receipt", "Invoice", "Financial_document"}
+    @returns {Response}
+  */
+
+
+  wrapResponse(inputFile, response, documentType) {
+    var _result$httpResponse$, _result$httpResponse$2, _result$httpResponse$3, _result$httpResponse$4, _result$httpResponse$5;
+
+    var result = super.wrapResponse(inputFile, response, documentType);
+    result.documentType = ((_result$httpResponse$ = result.httpResponse.data) === null || _result$httpResponse$ === void 0 ? void 0 : (_result$httpResponse$2 = _result$httpResponse$.predictions) === null || _result$httpResponse$2 === void 0 ? void 0 : (_result$httpResponse$3 = _result$httpResponse$2[0]) === null || _result$httpResponse$3 === void 0 ? void 0 : (_result$httpResponse$4 = _result$httpResponse$3.document_type) === null || _result$httpResponse$4 === void 0 ? void 0 : (_result$httpResponse$5 = _result$httpResponse$4.value) === null || _result$httpResponse$5 === void 0 ? void 0 : _result$httpResponse$5.toLowerCase()) || result.documentType;
+    return result;
+  }
 
 }
 

package/lib/api/receipt.js

@@ -24,24 +24,28 @@ class APIReceipt extends APIObject {
    */
 
 
-  parse(file) {
-    var _arguments = arguments,
-        _superprop_getParse = () => super.parse,
+  parse(_ref) {
+    var _superprop_getParse = () => super.parse,
         _superprop_get_request = () => super._request,
         _this = this;
 
     return _asyncToGenerator(function* () {
-      var inputType = _arguments.length > 1 && _arguments[1] !== undefined ? _arguments[1] : "path";
-      var version = _arguments.length > 2 && _arguments[2] !== undefined ? _arguments[2] : "3";
-      var cutPdf = _arguments.length > 3 && _arguments[3] !== undefined ? _arguments[3] : true;
-      var includeWords = _arguments.length > 4 && _arguments[4] !== undefined ? _arguments[4] : false;
+      var {
+        input,
+        inputType = "path",
+        filename = undefined,
+        version = "3",
+        cutPdf = true,
+        includeWords = false
+      } = _ref;
 
       _superprop_getParse().call(_this);
 
       var inputFile = new Input({
-        file,
-        inputType,
-        cutPdf
+        file: input,
+        inputType: inputType,
+        filename: filename,
+        allowCutPdf: cutPdf
       });
       yield inputFile.init();
       var url = "v".concat(version, "/predict");

package/lib/api/request.js

@@ -25,17 +25,13 @@ var request = function request(url, method, headers, input) {
     var body;
     headers["User-Agent"] = "mindee-node/".concat(sdkVersion, " node/").concat(process.version);
 
-    if (input.inputType === "path") {
+    if (["path", "stream"].includes(input.inputType)) {
       var fileParams = {
         filename: input.filename
       };
       form.append("file", input.fileObject, fileParams);
       if (includeWords) form.append("include_mvision", "true");
       headers = _objectSpread(_objectSpread({}, headers), form.getHeaders());
-    } else if (input.inputType === "stream") {
-      form.append("file", input.fileObject);
-      if (includeWords) form.append("include_mvision", "true");
-      headers = _objectSpread(_objectSpread({}, headers), form.getHeaders());
     } else if (input.inputType === "base64") {
       var body_obj = {
         file: input.fileObject
@@ -59,9 +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

@@ -16,11 +16,20 @@ var {
   PDFDocument
 } = require("pdf-lib");
 
+var concat = require("concat-stream");
+
+var {
+  Base64Encode
+} = require("base64-stream");
+
+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
@@ -61,45 +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 = 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.filepath = undefined;
-      _this3.filename = undefined;
+      var file = yield _this4.streamToBase64(_this4.file);
+      _this4.file = file;
+      _this4.inputType = "base64";
+      yield _this4.initBase64();
     })();
   }
 
@@ -109,23 +133,53 @@ class Input {
     this.filepath = "";
     this.fileExtension = "";
   }
+  /**
+   * Convert ReadableStream to Base64 encoded String
+   *
+   * @param {*} stream ReadableStream to encode
+   * @returns Base64 encoded String
+   */
+
+
+  streamToBase64(stream) {
+    return _asyncToGenerator(function* () {
+      return new Promise((resolve, reject) => {
+        var base64 = new Base64Encode();
+
+        var cbConcat = base64 => {
+          resolve(base64);
+        };
+
+        stream.pipe(base64).pipe(concat(cbConcat)).on("error", error => {
+          reject(error);
+        });
+      });
+    })();
+  }
   /** Cut PDF if pages > 5 */
 
 
   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 = 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

@@ -0,0 +1,45 @@
+const APIObject = require("./object");
+const Input = require("../inputs");
+
+class APIFinancialDocument extends APIObject {
+  constructor(invoiceToken, receiptToken) {
+    super(undefined, "financialDocument");
+    this.invoiceToken = invoiceToken;
+    this.receiptToken = receiptToken;
+  }
+
+  /**
+   * @param {String} file: Receipt filepath (allowed jpg, png, tiff, pdf)
+   * @param {String} inputType: String in {'path', 'stream', 'base64'}
+   * @param {Boolean} includeWords: extract all words into http_response
+   * @param {String} version: expense_receipt api version
+   * @param {Boolean} cutPdf: Automatically reconstruct pdf with more than 4 pages
+   * @returns {Response} Wrapped response with Receipts objects parsed
+   */
+  async parse({
+    input,
+    inputType = "path",
+    filename = undefined,
+    version = "3",
+    cutPdf = true,
+    includeWords = false,
+  }) {
+    const inputFile = new Input({
+      file: input,
+      inputType: inputType,
+      filename: filename,
+      allowCutPdf: cutPdf,
+    });
+    this.apiToken =
+      inputFile.fileExtension === "pdf" ? this.invoiceToken : this.receiptToken;
+    await inputFile.init();
+    const url =
+      inputFile.fileExtension === "pdf"
+        ? "/invoices/v1/predict"
+        : "/expense_receipts/v3/predict";
+    super.parse();
+    return await super._request(url, inputFile, version, includeWords);
+  }
+}
+
+module.exports = APIFinancialDocument;

package/mindee/api/index.js

@@ -0,0 +1,5 @@
+exports.APIObject = require("./object");
+exports.APIReceipt = require("./receipt");
+exports.Response = require("./response");
+exports.Request = require("./response");
+exports.Invoice = require("./invoice");

package/mindee/api/invoice.js

@@ -0,0 +1,53 @@
+const APIObject = require("./object");
+const Input = require("../inputs");
+
+class APIInvoice extends APIObject {
+  constructor(apiToken = undefined) {
+    super(apiToken, "invoice");
+    this.baseUrl = `${this.baseUrl}/invoices/`;
+  }
+
+  /**
+   * @param {Boolean} includeWords: , extract all words into http_response
+   * @param {Boolean} cutPdf: Automatically reconstruct pdf with more than 4 pages
+   * @param {String} inputType: String in {'path', 'stream', 'base64'}
+   * @param {String} file: Receipt filepath (allowed jpg, png, tiff, pdf)
+   * @param {String} version: expense_receipt api version
+   * @returns {Response} Wrapped response with Receipts objects parsed
+   */
+  async parse({
+    input,
+    inputType = "path",
+    filename = undefined,
+    version = "2",
+    cutPdf = true,
+    includeWords = false,
+  }) {
+    super.parse();
+    const inputFile = new Input({
+      file: input,
+      inputType: inputType,
+      filename: filename,
+      allowCutPdf: cutPdf,
+    });
+    await inputFile.init();
+    const url = `v${version}/predict`;
+    return await super._request(url, inputFile, includeWords);
+  }
+
+  /** 
+    @param {String} inputFile - Input object
+    @param {} response - HTTP response
+    @param {Document} documentType - Document class in {"Receipt", "Invoice", "Financial_document"}
+    @returns {Response}
+  */
+  wrapResponse(inputFile, response, documentType) {
+    let result = super.wrapResponse(inputFile, response, documentType);
+    result.documentType =
+      result.httpResponse.data?.predictions?.[0]?.document_type?.value?.toLowerCase() ||
+      result.documentType;
+    return result;
+  }
+}
+
+module.exports = APIInvoice;

package/mindee/api/object.js

@@ -0,0 +1,82 @@
+const Response = require("./response");
+const errorHandler = require("../errors/handler");
+const request = require("./request");
+
+/**
+ * Base class for APIs (APIReceipt & APIInvoice)
+ *  @param {String} apiToken - Token of the API used for parsing document
+ *  @param {String} apiName - Name of the API used for parsing document
+ */
+class APIObject {
+  constructor(apiToken = undefined, apiName = "") {
+    this.apiToken = apiToken;
+    this.baseUrl = "https://api.mindee.net/products";
+    this.apiName = apiName;
+  }
+
+  /**
+   */
+  parse() {
+    if (!this.apiToken) {
+      errorHandler.throw({
+        error: new Error(
+          `Missing API token for ${this.apiName}. \
+          Have you create a mindee Client with a ${this.apiName}Token in parameters ?`
+        ),
+      });
+    }
+  }
+
+  /** 
+    @param {String} url - API url for request
+    @param {Input} inputFile - input file for API
+    @param {Boolean} includeWords - Include Mindee vision words in Response
+    @returns {Response}
+  */
+  async _request(url, inputFile, includeWords = false) {
+    const headers = {
+      "X-Inferuser-Token": this.apiToken,
+    };
+    const response = await request(
+      `${this.baseUrl}${url}`,
+      "POST",
+      headers,
+      inputFile,
+      includeWords
+    );
+    return this.wrapResponse(inputFile, response, this.apiName);
+  }
+
+  /** 
+    @param {String} inputFile - Input object
+    @param {} response - HTTP response
+    @param {Document} documentType - Document class in {"Receipt", "Invoice", "Financial_document"}
+    @returns {Response}
+  */
+  wrapResponse(inputFile, response, documentType) {
+    if (response.statusCode != 200) {
+      const errorMessage = JSON.stringify(response.data, null, 4);
+      errorHandler.throw(
+        new Error(
+          `${this.apiName} API ${response.statusCode} HTTP error: ${errorMessage}`
+        ),
+        false
+      );
+      return new Response({
+        httpResponse: response,
+        documentType: documentType,
+        document: undefined,
+        input: inputFile,
+        error: true,
+      });
+    }
+    return new Response({
+      httpResponse: response,
+      documentType: documentType,
+      document: inputFile,
+      input: inputFile,
+    });
+  }
+}
+
+module.exports = APIObject;

package/mindee/api/receipt.js

@@ -0,0 +1,39 @@
+const APIObject = require("./object");
+const Input = require("../inputs");
+
+class APIReceipt extends APIObject {
+  constructor(apiToken = undefined) {
+    super(apiToken, "receipt");
+    this.baseUrl = `${this.baseUrl}/expense_receipts/`;
+  }
+
+  /**
+   * @param {String} file: Receipt filepath (allowed jpg, png, tiff, pdf)
+   * @param {String} inputType: String in {'path', 'stream', 'base64'}
+   * @param {Boolean} includeWords: extract all words into http_response
+   * @param {String} version: expense_receipt api version
+   * @param {Boolean} cutPdf: Automatically reconstruct pdf with more than 4 pages
+   * @returns {Response} Wrapped response with Receipts objects parsed
+   */
+  async parse({
+    input,
+    inputType = "path",
+    filename = undefined,
+    version = "3",
+    cutPdf = true,
+    includeWords = false,
+  }) {
+    super.parse();
+    const inputFile = new Input({
+      file: input,
+      inputType: inputType,
+      filename: filename,
+      allowCutPdf: cutPdf,
+    });
+    await inputFile.init();
+    const url = `v${version}/predict`;
+    return await super._request(url, inputFile, includeWords);
+  }
+}
+
+module.exports = APIReceipt;