feat & test: support rich text

* support rich text string
* add rich text test case
* add saxTrim option for keep blank in string or not

Author: dankang

README.md

@@ -24,6 +24,7 @@ Options
 |---|---|---|
 |verbose|true|throw additional exceptions, if `false` - then pass empty string in that places|
 |formatting|true|should cells with combined formats be formatted or not|
+|saxTrim|true|whether or not to trim text and comment nodes|
 
 -------
 ```javascript

lib/workbook.js

@@ -313,12 +313,17 @@ XlsxStreamReaderWorkBook.prototype._parseSharedStrings = function (nodeData) {
   var nodeObjValue = nodeData.pop()
   var nodeObjName = nodeData.pop()
 
-  if (nodeObjName && nodeObjName.name === 't') {
-    self.workBookSharedStrings.push(nodeObjValue)
-  } else {
-    if (nodeObjValue && typeof nodeObjValue === 'object' && nodeObjValue.hasOwnProperty('name') && nodeObjValue.name === 't') {
+  var isSharedStringItem = !!nodeData.find((n) => n && n.name === 'si') // <si> tag
+  if (isSharedStringItem) {
+    if (nodeObjName && nodeObjName.name === 't') {
+      self.workBookSharedStrings.push(nodeObjValue)
+    } else {
       self.workBookSharedStrings.push('')
     }
+  } else {
+    if (nodeObjName && nodeObjName.name === 't') {
+      self.workBookSharedStrings[self.workBookSharedStrings.length - 1] += nodeObjValue
+    }
   }
 }
 

lib/xlsx-stream-reader.js

@@ -19,6 +19,10 @@ function XlsxStreamReader (options) {
     options = {}
   }
 
+  if (typeof options.saxTrim === 'undefined') {
+    options.saxTrim = true
+  }
+
   if (typeof options.verbose === 'undefined') {
     options.verbose = true
   }
@@ -30,10 +34,10 @@ function XlsxStreamReader (options) {
   Object.defineProperty(this, 'options', {
     value: {
       saxStrict: true,
-      saxTrim: true,
       saxNormalize: true,
       saxPosition: true,
       saxStrictEntities: true,
+      saxTrim: options.saxTrim,
       verbose: options.verbose,
       formatting: options.formatting
     },

test/richtext.xlsx

@@ -72,4 +72,19 @@ describe('The xslx stream parser', function () {
       workSheetReader.process()
     })
   })
+  it('support rich-text', function (done) {
+    const workBookReader = new XlsxStreamReader({saxTrim: false})
+    fs.createReadStream(path.join(__dirname, 'richtext.xlsx')).pipe(workBookReader)
+    const rows = []
+    workBookReader.on('worksheet', function (workSheetReader) {
+      workSheetReader.on('end', function () {
+        assert(rows[0][1] === 'This is only one sentence.')
+        done()
+      })
+      workSheetReader.on('row', function (r) {
+        rows.push(r.values)
+      })
+      workSheetReader.process()
+    })
+  })
 })