mindee 1.0.2 → 1.0.7

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

@@ -24,20 +24,24 @@ 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,
+        file: input,
         inputType,
+        filename,
         cutPdf
       });
       _this.apiToken = inputFile.fileExtension === "pdf" ? _this.invoiceToken : _this.receiptToken;

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] : true;
+      var {
+        input,
+        inputType = "path",
+        filename = undefined,
+        version = "2",
+        cutPdf = true,
+        includeWords = false
+      } = _ref;
 
       _superprop_getParse().call(_this);
 
       var inputFile = new Input({
-        file,
+        file: input,
         inputType,
+        filename,
         cutPdf
       });
       yield inputFile.init();
       var url = "v".concat(version, "/predict");
-      return yield _superprop_get_request().call(_this, url, inputFile, version, includeWords);
+      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

@@ -24,28 +24,32 @@ 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,
+        file: input,
         inputType,
+        filename,
         cutPdf
       });
       yield inputFile.init();
       var url = "v".concat(version, "/predict");
-      return yield _superprop_get_request().call(_this, url, inputFile, version, includeWords);
+      return yield _superprop_get_request().call(_this, url, inputFile, includeWords);
     })();
   }
 

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

package/lib/documents/document.js

@@ -77,9 +77,11 @@ class Document {
 
     for (var document of documents) {
       for (var attribute of attributes) {
-        var _document$attribute, _document$attribute2;
+        var _document$attribute;
 
-        if ((document === null || document === void 0 ? void 0 : (_document$attribute = document[attribute]) === null || _document$attribute === void 0 ? void 0 : _document$attribute.probability) > finalDocument[attribute].probability) finalDocument[attribute] = document === null || document === void 0 ? void 0 : (_document$attribute2 = document[attribute]) === null || _document$attribute2 === void 0 ? void 0 : _document$attribute2.probability;
+        if ((document === null || document === void 0 ? void 0 : (_document$attribute = document[attribute]) === null || _document$attribute === void 0 ? void 0 : _document$attribute.probability) > finalDocument[attribute].probability) {
+          finalDocument[attribute] = document === null || document === void 0 ? void 0 : document[attribute];
+        }
       }
     }
 

package/lib/documents/fields/amount.js

@@ -14,7 +14,7 @@ class Amount extends Field {
       prediction,
       valueKey = "amount",
       reconstructed = false,
-      pageNumber = undefined
+      pageNumber = 0
     } = _ref;
     super({
       prediction,

package/lib/documents/fields/date.js

@@ -14,7 +14,7 @@ class DateField extends Field {
       prediction,
       valueKey = "iso",
       reconstructed = false,
-      pageNumber = undefined
+      pageNumber = 0
     } = _ref;
     super({
       prediction,

package/lib/documents/fields/locale.js

@@ -14,7 +14,7 @@ class Locale extends Field {
       prediction,
       valueKey = "value",
       reconstructed = false,
-      pageNumber = undefined
+      pageNumber = 0
     } = _ref;
     super({
       prediction,

package/lib/documents/fields/orientation.js

@@ -14,7 +14,7 @@ class Orientation extends Field {
       prediction,
       valueKey = "degrees",
       reconstructed = false,
-      pageNumber = undefined
+      pageNumber = 0
     } = _ref;
     var orientations = [0, 90, 180, 270];
     super({

package/lib/documents/fields/paymentDetails.js

@@ -26,7 +26,7 @@ class PaymentDetails extends Field {
       routingNumberKey = "routing_number",
       swiftKey = "swift",
       reconstructed = false,
-      pageNumber = undefined
+      pageNumber = 0
     } = _ref;
     super({
       prediction,

package/lib/documents/fields/tax.js

@@ -20,7 +20,7 @@ class Tax extends Field {
       rateKey = "rate",
       codeKey = "code",
       reconstructed = false,
-      pageNumber = undefined
+      pageNumber = 0
     } = _ref;
     super({
       prediction,

package/lib/inputs.js

@@ -16,6 +16,16 @@ 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");
+
 class Input {
   /**
    * @param {(String | Buffer)} file - the file that will be read. Either path or base64 string, or a steam
@@ -77,7 +87,7 @@ class Input {
     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
+      _this2.filename = _this2.filename || path.basename(_this2.file); // Check if file type is valid
 
       var filetype = _this2.filename.split(".").pop();
 
@@ -98,8 +108,16 @@ class Input {
 
     return _asyncToGenerator(function* () {
       _this3.fileObject = _this3.file;
-      _this3.filepath = undefined;
-      _this3.filename = undefined;
+      _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();
+      }
     })();
   }
 
@@ -109,6 +127,29 @@ 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 yield 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 */
 
 
@@ -117,7 +158,14 @@ class Input {
 
     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;
+
+      if (_this4.filename == "stream") {
+        pdfDocument = yield PDFDocument.load(yield _this4.streamToBase64(_this4.fileObject));
+      } else {
+        pdfDocument = yield PDFDocument.load(_this4.fileObject);
+      }
+
       var splitedPdfDocument = yield PDFDocument.create();
       var pdfLength = pdfDocument.getPageCount();
       if (pdfLength <= _this4.CUT_PDF_SIZE) return;

package/mindee/api/financialDocument.js

@@ -0,0 +1,40 @@
+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, filename, 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 +1,48 @@
+const APIObject = require("./object");
+const Input = require("../inputs");
+
+class APIInvoice extends APIObject {
+  constructor(apiToken = undefined) {
+    super(apiToken, "invoice");
+    this.baseUrl = `${this.baseUrl}/invoices/`;
+  }
+
+  /**
+   * @param {Boolean} includeWords: , extract all words into http_response
+   * @param {Boolean} cutPdf: Automatically reconstruct pdf with more than 4 pages
+   * @param {String} inputType: String in {'path', 'stream', 'base64'}
+   * @param {String} file: Receipt filepath (allowed jpg, png, tiff, pdf)
+   * @param {String} version: expense_receipt api version
+   * @returns {Response} Wrapped response with Receipts objects parsed
+   */
+  async parse({
+    input,
+    inputType = "path",
+    filename = undefined,
+    version = "2",
+    cutPdf = true,
+    includeWords = false,
+  }) {
+    super.parse();
+    const inputFile = new Input({ file: input, inputType, filename, cutPdf });
+    await inputFile.init();
+    const url = `v${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 +1,34 @@
+const APIObject = require("./object");
+const Input = require("../inputs");
+
+class APIReceipt extends APIObject {
+  constructor(apiToken = undefined) {
+    super(apiToken, "receipt");
+    this.baseUrl = `${this.baseUrl}/expense_receipts/`;
+  }
+
+  /**
+   * @param {String} file: Receipt filepath (allowed jpg, png, tiff, pdf)
+   * @param {String} inputType: String in {'path', 'stream', 'base64'}
+   * @param {Boolean} includeWords: extract all words into http_response
+   * @param {String} version: expense_receipt api version
+   * @param {Boolean} cutPdf: Automatically reconstruct pdf with more than 4 pages
+   * @returns {Response} Wrapped response with Receipts objects parsed
+   */
+  async parse({
+    input,
+    inputType = "path",
+    filename = undefined,
+    version = "3",
+    cutPdf = true,
+    includeWords = false,
+  }) {
+    super.parse();
+    const inputFile = new Input({ file: input, inputType, filename, cutPdf });
+    await inputFile.init();
+    const url = `v${version}/predict`;
+    return await super._request(url, inputFile, includeWords);
+  }
+}
+
+module.exports = APIReceipt;