Implements support for XLS and XLSX files

README.md

@@ -77,6 +77,8 @@ These formats are conceptually different and only work on a limited basis.  With
 
 Tabular formats:
 - csv
+- xls
+- xlsx
 
 ## Usage
 
@@ -90,6 +92,13 @@ any-json can be used to convert (almost) anything to JSON.
 This version supports:
     cson, csv, hjson, ini, json, json5, toml, yaml
 
+This version has is beta support for:
+    xls, xlsx
+
+The beta formats should work, but since they are new,
+  behavior may change in later releases in response to feedback
+  without requiring a major version update.
+
 command:
     convert    convert between formats (default when none specified)
     combine    combine multiple documents

lib/index.ts

@@ -27,6 +27,43 @@ interface FormatConversion {
   decode(text: string, reviver?: (key: any, value: any) => any): Promise<any>
 }
 
+abstract class AbstractWorkbookConverter implements FormatConversion {
+  abstract readonly name: string;
+  abstract readonly bookType: XLSX.BookType;
+
+  public async encode(value: any) {
+    const book = XLSX.utils.book_new();
+
+    if (Array.isArray(value)) {
+      const sheet = XLSX.utils.json_to_sheet(value);
+      XLSX.utils.book_append_sheet(book, sheet);
+    }
+    else {
+      for (let sheetName of Object.getOwnPropertyNames(value)) {
+        const sheet = XLSX.utils.json_to_sheet(value[sheetName]);
+        XLSX.utils.book_append_sheet(book, sheet, sheetName);
+      }
+    }
+
+    return XLSX.write(book, { WTF: true, bookType: this.bookType, type: "binary" });
+  }
+
+  public async decode(text: string, reviver?: (key: any, value: any) => any): Promise<any> {
+    const book = XLSX.read(text, { type: "binary" });
+    if (book.SheetNames.length === 1) {
+      return XLSX.utils.sheet_to_json(book.Sheets[book.SheetNames[0]], {raw: true, defval: null});
+    }
+
+    const result: any = {};
+
+    for (let sheet of book.SheetNames) {
+      result[sheet] = XLSX.utils.sheet_to_json(book.Sheets[sheet], {raw: true, defval: null })
+    }
+
+    return result;
+  }
+}
+
 class CsonConverter implements FormatConversion {
   readonly name: string = 'cson'
 
@@ -150,6 +187,16 @@ class TomlConverter implements FormatConversion {
   }
 }
 
+class XlsxConverter extends AbstractWorkbookConverter {
+  readonly name = "xlsx"
+  readonly bookType = "xlsx";
+}
+
+class XlsConverter extends AbstractWorkbookConverter {
+  readonly name = "xls"
+  readonly bookType = "xlml";
+}
+
 class XmlConverter implements FormatConversion {
   readonly name: string = 'xml'
 
@@ -183,6 +230,8 @@ const codecs = new Map([
   new JsonConverter(),
   new Json5Converter(),
   new TomlConverter(),
+  new XlsConverter(),
+  new XlsxConverter(),
   new XmlConverter(),
   new YamlConverter()
 ].map(c => [c.name, c] as [string, FormatConversion]))
@@ -205,4 +254,15 @@ export async function encode(value: any, format: string): Promise<string | Buffe
   }
 
   throw new Error("Unknown format " + format + "!");
+}
+
+/**
+ * Gets the anticipated string encoding for the given format.
+ */
+export function getEncoding(format: string) {
+  switch (format) {
+    case "xlsx": return "binary";
+    case "xls": return "binary";
+    default: return "utf8";
+  }
 }

lib/run.ts

@@ -26,14 +26,6 @@ function removeLeadingDot(formatOrExtension: string) {
   else return formatOrExtension;
 }
 
-function getEncoding(format: string) {
-  switch (format) {
-    case "xlsx": return "binary";
-    case "xls": return "binary";
-    default: return "utf8";
-  }
-}
-
 function legacyOptionsParsing(argv: string[]) {
   const options: any = {}
   var unnamedArgs = [] as string[];
@@ -131,6 +123,13 @@ any-json can be used to convert (almost) anything to JSON.
 This version supports:
     cson, csv, hjson, ini, json, json5, toml, yaml
 
+This version has is beta support for:
+    xls, xlsx
+
+The beta formats should work, but since they are new,
+  behavior may change in later releases in response to feedback
+  without requiring a major version update.
+
 command:
     convert    convert between formats (default when none specified)
     combine    combine multiple documents
@@ -180,7 +179,7 @@ for help use 'any-json -?`;
         }
 
         const stdin = process.stdin;
-        const encoding = getEncoding(options.format);
+        const encoding = anyjson.getEncoding(options.format);
 
         stdin.resume();
         if (encoding !== "binary") {
@@ -198,18 +197,19 @@ for help use 'any-json -?`;
       }
 
       const format = options.format || getFormatFromFileName(unnamedArgs[0]);
-      const input = await anyjson.decode(await readFile(unnamedArgs[0], 'utf8'), format);
+      const input = await anyjson.decode(await readFile(unnamedArgs[0], anyjson.getEncoding(format)), format);
 
       return await legacyEncode(options, input);
     }
     // Not legacy parsing continue as usual
   }
 
   async function convert(value: any, options: any, outputFileName?: string) {
-    const result = await anyjson.encode(value, options.output_format || getFormatFromFileName(outputFileName) || "json");
+    const output_format = options.output_format || getFormatFromFileName(outputFileName) || "json";
+    const result = await anyjson.encode(value, output_format);
 
     if (outputFileName) {
-      await util.promisify(fs.writeFile)(outputFileName, result, "utf8")
+      await util.promisify(fs.writeFile)(outputFileName, result, anyjson.getEncoding(output_format))
       return "";
     }
     return result;
@@ -227,8 +227,7 @@ for help use 'any-json -?`;
       const fileName = options._args[0] as string;
       const format = options.input_format || getFormatFromFileName(fileName);
 
-      // TODO: Will need to check for binary files (see `getEncoding`)
-      const fileContents = await readFile(fileName, "utf8")
+      const fileContents = await readFile(fileName, anyjson.getEncoding(format))
 
       const parsed = await anyjson.decode(fileContents, format)
 
@@ -242,9 +241,8 @@ for help use 'any-json -?`;
 
       const items = await Promise.all(
         options._args.map(async fileName => {
-          // TODO: Will need to check for binary files (see `getEncoding`)
           const format = options.input_format || getFormatFromFileName(fileName);
-          const fileContents = await readFile(fileName, 'utf8') as string
+          const fileContents = await readFile(fileName, anyjson.getEncoding(format)) as string
           return await anyjson.decode(fileContents, format)
         })
       )
@@ -263,7 +261,6 @@ for help use 'any-json -?`;
 
       const format = options.input_format || getFormatFromFileName(fileName);
 
-      // TODO: Will need to check for binary files (see `getEncoding`)
       const fileContents = await readFile(fileName, "utf8")
 
       const array = await anyjson.decode(fileContents, format) as any[]
@@ -289,7 +286,7 @@ for help use 'any-json -?`;
 
 if (require.main === module) {
   main(process.argv)
-    .then(s => console.log(s))
+    .then(s => process.stdout.write(s as string, 'binary'))
     .catch(error => {
       console.error(error);
       process.exit(2);

package.json

@@ -49,7 +49,7 @@
     "typescript": "^2.4.2",
     "util.promisify": "^1.0.0",
     "xlsjs": "^0.7.5",
-    "xlsx": "^0.8.1",
+    "xlsx": "^0.11.3",
     "xml2js": "0.4.17"
   },
   "devDependencies": {

test/fixtures/in/multi-sheet-workbook.json

@@ -0,0 +1,22 @@
+{
+    "Quantity": [
+        {
+            "Name": "Widget 1",
+            "Quantity": 5
+        },
+        {
+            "Name": "Widget 2",
+            "Quantity": 3
+        }
+    ],
+    "Prices": [
+        {
+            "Name": "Widget A",
+            "Price": 3
+        },
+        {
+            "Name": "Widget B",
+            "Price": 0.26
+        }
+    ]
+}

test/fixtures/in/product-set.xls.json

@@ -0,0 +1,18 @@
+[
+    {
+        "id": 2,
+        "name": "An ice sculpture",
+        "price": 12.5,
+        "tags": null,
+        "dimensions": null,
+        "warehouseLocation": null
+    },
+    {
+        "id": 3,
+        "name": "A blue mouse",
+        "price": 25.5,
+        "tags": null,
+        "dimensions": null,
+        "warehouseLocation": null
+    }
+]

test/fixtures/in/product-set.xlsx.json

@@ -0,0 +1,18 @@
+[
+    {
+        "id": 2,
+        "name": "An ice sculpture",
+        "price": 12.5,
+        "tags": null,
+        "dimensions": null,
+        "warehouseLocation": null
+    },
+    {
+        "id": 3,
+        "name": "A blue mouse",
+        "price": 25.5,
+        "tags": null,
+        "dimensions": null,
+        "warehouseLocation": null
+    }
+]

test/index.test.ts

@@ -70,18 +70,43 @@ suite('problematic-formats', () => {
   })
 });
 
-suite('tabular-formats', () => {
 
-  const tabular_formats = [
-    "csv",
-    // "xls",
-    // "xlsx"
-  ]
+suite('tabular-formats', () => {
+  const tabular_formats = ["csv"];
 
   testEncode(tabular_formats);
   testDecode(tabular_formats);
 });
 
+suite('multi-sheet workbooks', () => {
+
+  const multisheet_formats = [
+    "xls",
+    "xlsx"
+  ];
+
+  testDecode(multisheet_formats);
+
+  for (const format of multisheet_formats) {
+    test(`decode ${format}`, async () => {
+      const expected = JSON.parse(await readInputFixture(`multi-sheet-workbook.json`));
+      const contents = await readOutputFixture(`multi-sheet-workbook.${format}`, anyjson.getEncoding(format))
+      const actual = await anyjson.decode(contents, format);
+      return assert.deepEqual(actual, expected)
+    });
+
+    test(`encode ${format}`, async () => {
+      const input = JSON.parse(await readInputFixture(`multi-sheet-workbook.json`));
+      const actual = await anyjson.encode(input, format)
+      const expected = await readOutputFixture(`multi-sheet-workbook.${format}`, anyjson.getEncoding(format))
+
+      return assert.deepEqual(
+        await anyjson.decode(actual as string, format),
+        await anyjson.decode(expected, format));
+    });
+  }
+});
+
 function testEncode(formats: string[]) {
   suite('encode', function () {
     suite('product-set', () => {