Revert "refactor: Remove deprecated normalizeWhitespace option (#614)"

This reverts commit 018ddf9.

Author: fb55

readme.md

@@ -76,6 +76,77 @@ When the parser is used in a non-streaming fashion, `endIndex` is an integer
 indicating the position of the end of the node in the document.
 The default value is `false`.
 
+## Option: `normalizeWhitespace` _(deprecated)_
+
+Replace all whitespace with single spaces.
+The default value is `false`.
+
+**Note:** Enabling this might break your markup.
+
+For the following examples, this HTML will be used:
+
+```html
+<font> <br />this is the text <font></font></font>
+```
+
+### Example: `normalizeWhitespace: true`
+
+```javascript
+[
+    {
+        type: "tag",
+        name: "font",
+        children: [
+            {
+                data: " ",
+                type: "text",
+            },
+            {
+                type: "tag",
+                name: "br",
+            },
+            {
+                data: "this is the text ",
+                type: "text",
+            },
+            {
+                type: "tag",
+                name: "font",
+            },
+        ],
+    },
+];
+```
+
+### Example: `normalizeWhitespace: false`
+
+```javascript
+[
+    {
+        type: "tag",
+        name: "font",
+        children: [
+            {
+                data: "\n\t",
+                type: "text",
+            },
+            {
+                type: "tag",
+                name: "br",
+            },
+            {
+                data: "this is the text\n",
+                type: "text",
+            },
+            {
+                type: "tag",
+                name: "font",
+            },
+        ],
+    },
+];
+```
+
 ---
 
 License: BSD-2-Clause

src/__fixtures__/16-normalize_whitespace.json

@@ -0,0 +1,47 @@
+{
+    "name": "Normalize whitespace",
+    "options": {
+        "normalizeWhitespace": true
+    },
+    "html": "Line one\n<br>\t  \r\n\f  <br>\nline two<font><br> x </font>",
+    "expected": [
+        {
+            "data": "Line one ",
+            "type": "text"
+        },
+        {
+            "type": "tag",
+            "name": "br",
+            "attribs": {}
+        },
+        {
+            "data": " ",
+            "type": "text"
+        },
+        {
+            "type": "tag",
+            "name": "br",
+            "attribs": {}
+        },
+        {
+            "data": " line two",
+            "type": "text"
+        },
+        {
+            "type": "tag",
+            "name": "font",
+            "attribs": {},
+            "children": [
+                {
+                    "type": "tag",
+                    "name": "br",
+                    "attribs": {}
+                },
+                {
+                    "data": " x ",
+                    "type": "text"
+                }
+            ]
+        }
+    ]
+}

src/index.ts

@@ -12,6 +12,8 @@ import {
 
 export * from "./node";
 
+const reWhitespace = /\s+/g;
+
 export interface DomHandlerOptions {
     /**
      * Add a `startIndex` property to nodes.
@@ -31,6 +33,16 @@ export interface DomHandlerOptions {
      */
     withEndIndices?: boolean;
 
+    /**
+     * Replace all whitespace with single spaces.
+     *
+     * **Note:** Enabling this might break your markup.
+     *
+     * @default false
+     * @deprecated
+     */
+    normalizeWhitespace?: boolean;
+
     /**
      * Treat the markup as XML.
      *
@@ -41,6 +53,7 @@ export interface DomHandlerOptions {
 
 // Default options
 const defaultOpts: DomHandlerOptions = {
+    normalizeWhitespace: false,
     withStartIndices: false,
     withEndIndices: false,
     xmlMode: false,
@@ -153,14 +166,26 @@ export class DomHandler {
     }
 
     public ontext(data: string): void {
+        const { normalizeWhitespace } = this.options;
         const { lastNode } = this;
 
         if (lastNode && lastNode.type === ElementType.Text) {
-            lastNode.data += data;
+            if (normalizeWhitespace) {
+                lastNode.data = (lastNode.data + data).replace(
+                    reWhitespace,
+                    " "
+                );
+            } else {
+                lastNode.data += data;
+            }
             if (this.options.withEndIndices) {
                 lastNode.endIndex = this.parser!.endIndex;
             }
         } else {
+            if (normalizeWhitespace) {
+                data = data.replace(reWhitespace, " ");
+            }
+
             const node = new Text(data);
             this.addNode(node);
             this.lastNode = node;