Added XSD to docs.

Added methods to add parser listeners.
Implemented XML conversion to typed objects.
Simplified parser API.
Implemented async promisable parser.

Author: Viacheslav Dobromyslov

README.md

@@ -37,31 +37,22 @@ Here is a common usage example:
 import {CommerceMlImportParser} from 'commerceml-parser/import-parser';
 import {createReadStream} from "fs";
 
-// Create parser stream for CommerceML catalog import file
-const parserStream = new CommerceMlImportParser().createStream();
+// Create parser for CommerceML catalog import file
+const catalogImportParser = new CommerceMlImportParser();
 
 // Define handler for classifier XML block
-parserStream.on('classifier', data => {
-  console.log('classifier:', JSON.stringify(data));
+catalogImportParser.onClassifier(classifier => {
+  console.log('classifier', JSON.stringify(classifier));
 });
 
-// Define handler for classifier group XML blocks
-parserStream.on('classifierGroup', data => {
-  console.log('classifierGroup:', JSON.stringify(data));
-});
 
-// Define handler for parser error
-parserStream.on('error', error => {
-  console.log('error', error);
-});
-
-// Define handler for file end
-parserStream.on('end', _ => {
-  console.log('Done.');    
+// Define handler for classifier group XML blocks
+catalogImportParser.onClassifierGroup(classifierGroup => {
+  console.log('classifierGroup', JSON.stringify(classifierGroup));
 });
 
 // Read CommerceML file and feed it to the parser stream
-createReadStream('./data/import0_1_with_nested_groups.xml').pipe(parserStream);
+await catalogImportParser.parse(createReadStream('./data/import0_1_with_nested_groups.xml'));
 ``` 
 
 ## Thanks to

doc/commerceml-2.08.pdf

@@ -3,19 +3,16 @@ import {createReadStream} from "fs";
 
 describe('CommerceMlParser', () => {
   it('is ok', async () => {
-    const stream = new CommerceMlImportParser().createStream();
-    stream.on('classifier', data => {
-      console.log('classifier:', JSON.stringify(data));
+    const catalogImportParser = new CommerceMlImportParser();
+    catalogImportParser.onClassifier(classifier => {
+      console.log('classifier', JSON.stringify(classifier));
     });
 
-    stream.on('classifierGroup', data => {
-      console.log('classifierGroup:', JSON.stringify(data));
+    catalogImportParser.onClassifierGroup(classifierGroup => {
+      console.log('classifierGroup', JSON.stringify(classifierGroup));
     });
 
-    stream.on('error', error => {
-      console.log('error', error);
-    });
-
-    createReadStream('./data/import0_1_with_nested_groups.xml').pipe(stream);
+    await catalogImportParser.parse(createReadStream('./data/import0_1_with_nested_groups.xml'));
+    console.log('Done');
   });
 });

doc/commerceml-2.08.xsd

@@ -1,6 +1,7 @@
 import {createStream, QualifiedTag, SAXStream, Tag} from 'sax';
 import {encodeXML} from 'entities/lib';
 import X2JS from 'x2js';
+import {Readable} from 'stream';
 
 export class CommerceMlCollectRule {
   start: string[] = [];
@@ -11,7 +12,7 @@ export abstract class CommerceMlAbstractParser {
   /**
    * SAX Parser instance.
    */
-  protected parser?: SAXStream;
+  protected parser: SAXStream;
 
   /**
    *
@@ -43,10 +44,7 @@ export abstract class CommerceMlAbstractParser {
    */
   protected collectOpenTags: string[] = [];
 
-  /**
-   * Creates SAX stream for XML parsing.
-   */
-  public createStream(): SAXStream {
+  constructor() {
     this.parser = createStream(true, {
       trim: true,
       normalize: true
@@ -63,7 +61,30 @@ export abstract class CommerceMlAbstractParser {
     this.parser.on('text', (text: string) => {
       this.onText(text);
     });
+  }
+
+  /**
+   * Starts parsing readable stream.
+   * @param readStream
+   */
+  public async parse(readStream: Readable): Promise<void> {
+    return new Promise((resolve, reject) => {
+      this.parser.on('end', () => {
+        resolve();
+      });
+
+      this.parser.on('error', error => {
+        reject(error);
+      });
+
+      readStream.pipe(this.parser);
+    });
+  }
 
+  /**
+   * Returns SAX stream.
+   */
+  public getStream(): SAXStream {
     return this.parser;
   }
 

doc/commerceml_2_10_2.pdf doc/commerceml-2.10.2.pdfdoc/commerceml_2_10_2.pdf renamed to doc/commerceml-2.10.2.pdf

@@ -1,9 +1,245 @@
 import {CommerceMlAbstractParser, CommerceMlCollectRule} from './abstract-parser';
+import {
+  Catalog,
+  Classifier,
+  ClassifierGroup,
+  ClassifierProperty,
+  CommercialInformation,
+  Counterparty, DictionaryValue, Product,
+  PropertyValue, RequisiteValue
+} from './types';
 
 export class CommerceMlImportParser extends CommerceMlAbstractParser {
+  /**
+   * Parses commercial information schemaVersion and creationTimestamp attributes.
+   * @param callback
+   */
+  public onCommercialInformation(callback: (commercialInformation: CommercialInformation) => void): void {
+    this.parser.on('commercialInformation', (data: any) => {
+      const commercialInformation: CommercialInformation = {
+        schemaVersion: data.КоммерческаяИнформация._ВерсияСхемы,
+        creationTimestamp: data.КоммерческаяИнформация._ДатаФормирования
+      };
+
+      callback(commercialInformation);
+    });
+  }
+
+  /**
+   * Parses classifier block header without details.
+   * @param callback
+   */
+  public onClassifier(callback: (classifier: Classifier) => void): void {
+    this.parser.on('classifier', (data: any) => {
+      const classifierXml = data.Классификатор;
+      const classifier: Classifier = {
+        id: classifierXml.Ид,
+        name: classifierXml.Наименование,
+        owner: this.parseCounterpartyXmlData(classifierXml.Владелец)
+      };
+
+      callback(classifier);
+    });
+  }
+
+  /**
+   * Parses classifier groups.
+   * @param callback
+   */
+  public onClassifierGroup(callback: (classifierGroup: ClassifierGroup) => void): void {
+    const processGroup = (groupData: any): ClassifierGroup => {
+      const result: ClassifierGroup = {
+        id: groupData.Ид,
+        name: groupData.Наименование
+      };
+
+      const children: ClassifierGroup[] = [];
+      if (groupData.Группы?.Группа?.length > 0) {
+        for (const child of groupData.Группы.Группа) {
+          children.push(processGroup(child));
+        }
+      }
+
+      if (children.length > 0) {
+        result.groups = children;
+      }
+
+      return result;
+    };
+
+    this.parser.on('classifierGroup', (data: any) => {
+      const classifierGroupXml = data.Группа;
+      const classifierGroup: ClassifierGroup = processGroup(classifierGroupXml);
+      callback(classifierGroup);
+    });
+  }
+
+  /**
+   * Parses classifier properties.
+   * @param callback
+   */
+  public onClassifierProperty(callback: (classifierProperty: ClassifierProperty) => void): void {
+    this.parser.on('classifierProperty', (data: any) => {
+      const propertyXml = data.Свойство;
+
+      const classifierProperty: ClassifierProperty = {
+        id: propertyXml.Ид,
+        name: propertyXml.Наименование,
+        type: propertyXml.ТипЗначений
+      };
+
+      if (propertyXml.ВариантыЗначений?.Справочник?.length > 0) {
+        classifierProperty.dictionaryValues = [];
+        for (const dictionaryValue of propertyXml.ВариантыЗначений.Справочник) {
+          classifierProperty.dictionaryValues.push({
+            id: dictionaryValue.ИдЗначения,
+            value: dictionaryValue.Значение
+          } as DictionaryValue);
+        }
+      }
+
+      callback(classifierProperty);
+    });
+  }
+
+  /**
+   * Parses catalog header without details.
+   * @param callback
+   */
+  public onCatalog(callback: (catalog: Catalog) => void) {
+    this.parser.on('catalog', (data: any) => {
+      const catalogXml = data.Каталог;
+      const catalog: Catalog = {
+        id: catalogXml.Ид,
+        classifierId: catalogXml.ИдКлассификатора,
+        name: catalogXml.Наименование,
+        owner: this.parseCounterpartyXmlData(catalogXml.Владелец),
+        products: []
+      };
+
+      callback(catalog);
+    });
+  }
+
+  /**
+   * Parses catalog products.
+   * @param callback
+   */
+  public onProduct(callback: (product: Product) => void) {
+    this.parser.on('product', (data: any) => {
+      const productXml = data.Товар;
+      const product: Product = {
+        id: productXml.Ид,
+        article: productXml.Артикул,
+        name: productXml.Наименование,
+        baseMeasurementUnit: {
+          code: productXml.БазоваяЕдиница._Код,
+          fullName: productXml.БазоваяЕдиница._НаименованиеПолное,
+          acronym: productXml.БазоваяЕдиница._МеждународноеСокращение
+        },
+        group: productXml.Группы.Ид,
+        description: productXml.Описание
+      };
+
+      if (productXml.СтавкиНалогов?.СтавкаНалога) {
+        if (Array.isArray(productXml.СтавкиНалогов?.СтавкаНалога)) {
+          product.taxRates = [];
+          for (const taxRateXml of productXml.СтавкиНалогов?.СтавкаНалога) {
+            product.taxRates.push({
+              name: taxRateXml.Наименование,
+              rate: taxRateXml.Ставка
+            });
+          }
+        } else {
+          product.taxRates = [{
+            name: productXml.СтавкиНалогов.СтавкаНалога.Наименование,
+            rate: productXml.СтавкиНалогов.СтавкаНалога.Ставка
+          }];
+        }
+      }
+
+      if (productXml.Штрихкод) {
+        product.barcode = productXml.Штрихкод;
+      }
+
+      if (productXml.Изготовитель) {
+        product.manufacturer = {
+          id: productXml.Изготовитель.Ид,
+          name: productXml.Изготовитель.Наименование
+        };
+      }
+
+      if (productXml.ЗначенияСвойств?.ЗначенияСвойства?.length > 0) {
+        product.propertyValues = [];
+        for (const propertyValue of productXml.ЗначенияСвойств.ЗначенияСвойства) {
+          if (Array.isArray(propertyValue.Значение)) {
+            product.propertyValues.push({
+              id: propertyValue.Ид,
+              values: propertyValue.Значение
+            } as PropertyValue);
+          } else {
+            product.propertyValues.push({
+              id: propertyValue.Ид,
+              values: [propertyValue.Значение]
+            } as PropertyValue);
+          }
+        }
+      }
+
+      if (productXml.ЗначенияРеквизитов?.ЗначениеРеквизита?.length > 0) {
+        product.requisiteValues = [];
+        for (const requisiteValue of productXml.ЗначенияРеквизитов.ЗначениеРеквизита ?? []) {
+          if (Array.isArray(requisiteValue.Значение)) {
+            product.requisiteValues.push({
+              name: requisiteValue.Наименование,
+              values: requisiteValue.Значение
+            } as RequisiteValue);
+          } else {
+            product.requisiteValues.push({
+              name: requisiteValue.Наименование,
+              values: [requisiteValue.Значение]
+            } as RequisiteValue);
+          }
+        }
+      }
+
+      callback(product);
+    });
+  }
+
+  /**
+   * Helper method to parse counterparty XML data.
+   * @param xmlData
+   */
+  protected parseCounterpartyXmlData(xmlData: any): Counterparty {
+    const counterparty: Counterparty = {
+      id: xmlData.Ид,
+      name: xmlData.Наименование
+    };
+
+    // Detect company info or person info
+    if (xmlData.ОфициальноеНаименование) {
+      counterparty.companyInfo = {
+        officialName: xmlData.ОфициальноеНаименование,
+        inn: xmlData.ИНН,
+        kpp: xmlData.КПП,
+        okpo: xmlData.ОКПО
+      };
+    } else {
+      counterparty.personInfo = {
+        fullName: xmlData.ПолноеНаименование
+      };
+    }
+
+    return counterparty;
+  }
+
+  /**
+   * Parser rules.
+   */
   protected getCollectRules(): {[key: string]: CommerceMlCollectRule} {
     return {
-      commercialInfo: {
+      commercialInformation: {
         start: ['КоммерческаяИнформация']
       },
       classifier: {