Add pasteWhiteSpaceFormatParser and integrate it into sanitizing context

packages/roosterjs-content-model-core/lib/command/createModelFromHtml/createDomToModelContextForSanitizing.1.ts

@@ -0,0 +1,65 @@
+import { containerSizeFormatParser } from 'roosterjs-content-model-core/lib/override/containerSizeFormatParser';
+import { createDomToModelContext } from 'roosterjs-content-model-dom/lib';
+import { createPasteEntityProcessor } from 'roosterjs-content-model-core/lib/override/pasteEntityProcessor';
+import { createPasteGeneralProcessor } from 'roosterjs-content-model-core/lib/override/pasteGeneralProcessor';
+import { DefaultSanitizingOption } from './createDomToModelContextForSanitizing';
+import { getRootComputedStyleForContext } from 'roosterjs-content-model-core/lib/coreApi/createEditorContext/getRootComputedStyleForContext';
+import { pasteBlockEntityParser } from 'roosterjs-content-model-core/lib/override/pasteCopyBlockEntityParser';
+import { pasteDisplayFormatParser } from 'roosterjs-content-model-core/lib/override/pasteDisplayFormatParser';
+import { pasteTextProcessor } from 'roosterjs-content-model-core/lib/override/pasteTextProcessor';
+import type {
+    ContentModelSegmentFormat,
+    DomToModelOption,
+    DomToModelOptionForSanitizing,
+    DomToModelContext,
+} from 'roosterjs-content-model-types/lib';
+
+/**
+ * @internal
+ */
+
+export function createDomToModelContextForSanitizing(
+    document: Document,
+    defaultFormat?: ContentModelSegmentFormat,
+    defaultOption?: DomToModelOption,
+    additionalSanitizingOption?: Partial<DomToModelOptionForSanitizing>
+): DomToModelContext {
+    const sanitizingOption: DomToModelOptionForSanitizing = {
+        ...DefaultSanitizingOption,
+        ...additionalSanitizingOption,
+    };
+
+    return createDomToModelContext(
+        {
+            defaultFormat,
+            ...getRootComputedStyleForContext(document),
+            experimentalFeatures: [],
+        },
+        defaultOption,
+        {
+            processorOverride: {
+                '#text': pasteTextProcessor,
+                entity: createPasteEntityProcessor(sanitizingOption),
+                '*': createPasteGeneralProcessor(sanitizingOption),
+            },
+            formatParserOverride: {
+                display: pasteDisplayFormatParser,
+                whiteSpace: (format, element, context, defaultStyle) => {
+                    if (element.style.whiteSpace != 'pre') {
+                        context.defaultFormatParsers.whiteSpace?.(
+                            format,
+                            element,
+                            context,
+                            defaultStyle
+                        );
+                    }
+                },
+            },
+            additionalFormatParsers: {
+                container: [containerSizeFormatParser],
+                entity: [pasteBlockEntityParser],
+            },
+        },
+        sanitizingOption
+    );
+}

packages/roosterjs-content-model-core/lib/command/createModelFromHtml/createDomToModelContextForSanitizing.ts

@@ -6,6 +6,7 @@ import { getRootComputedStyleForContext } from '../../coreApi/createEditorContex
 import { pasteBlockEntityParser } from '../../override/pasteCopyBlockEntityParser';
 import { pasteDisplayFormatParser } from '../../override/pasteDisplayFormatParser';
 import { pasteTextProcessor } from '../../override/pasteTextProcessor';
+import { pasteWhiteSpaceFormatParser } from '../../override/pasteWhiteSpaceFormatParser';
 import type {
     ContentModelSegmentFormat,
     DomToModelContext,
@@ -52,6 +53,7 @@ export function createDomToModelContextForSanitizing(
             },
             formatParserOverride: {
                 display: pasteDisplayFormatParser,
+                whiteSpace: pasteWhiteSpaceFormatParser,
             },
             additionalFormatParsers: {
                 container: [containerSizeFormatParser],

packages/roosterjs-content-model-core/lib/command/createModelFromHtml/createModelFromHtml.ts

@@ -1,5 +1,5 @@
 import { convertInlineCss, retrieveCssRules } from './convertInlineCss';
-import { createDomToModelContextForSanitizing } from './createDomToModelContextForSanitizing';
+import { createDomToModelContextForSanitizing } from './createDomToModelContextForSanitizing.1';
 import { createEmptyModel, domToContentModel, parseFormat } from 'roosterjs-content-model-dom';
 import type {
     ContentModelDocument,

packages/roosterjs-content-model-core/lib/command/paste/mergePasteContent.ts

@@ -1,4 +1,4 @@
-import { createDomToModelContextForSanitizing } from '../createModelFromHtml/createDomToModelContextForSanitizing';
+import { createDomToModelContextForSanitizing } from '../createModelFromHtml/createDomToModelContextForSanitizing.1';
 import {
     ChangeSource,
     EmptySegmentFormat,

packages/roosterjs-content-model-core/lib/override/pasteWhiteSpaceFormatParser.ts

@@ -0,0 +1,14 @@
+import type { FormatParser, WhiteSpaceFormat } from 'roosterjs-content-model-types/lib';
+
+const WhiteSpacePre = 'pre';
+
+export const pasteWhiteSpaceFormatParser: FormatParser<WhiteSpaceFormat> = (
+    format,
+    element,
+    context,
+    defaultStyle
+) => {
+    if (element.style.whiteSpace != WhiteSpacePre) {
+        context.defaultFormatParsers.whiteSpace?.(format, element, context, defaultStyle);
+    }
+};

packages/roosterjs-content-model-core/test/overrides/pasteWhiteSpaceFormatParserTest.ts

@@ -0,0 +1,39 @@
+import { pasteWhiteSpaceFormatParser } from '../../lib/override/pasteWhiteSpaceFormatParser';
+import { WhiteSpaceFormat } from 'roosterjs-content-model-types/lib';
+
+describe('pasteWhiteSpaceFormatParser', () => {
+    let format: WhiteSpaceFormat;
+    let element: HTMLElement;
+    let context: any;
+    let defaultStyle: any;
+    let defaultParserSpy: jasmine.Spy;
+
+    beforeEach(() => {
+        format = {};
+        element = document.createElement('div');
+        defaultParserSpy = jasmine.createSpy();
+        context = {
+            defaultFormatParsers: {
+                whiteSpace: defaultParserSpy,
+            },
+        };
+        defaultStyle = {};
+    });
+
+    it('should call default whiteSpace parser when element.style.whiteSpace is not "pre"', () => {
+        element.style.whiteSpace = 'normal';
+        pasteWhiteSpaceFormatParser(format, element, context, defaultStyle);
+        expect(context.defaultFormatParsers.whiteSpace).toHaveBeenCalledWith(
+            format,
+            element,
+            context,
+            defaultStyle
+        );
+    });
+
+    it('should not call default whiteSpace parser when element.style.whiteSpace is "pre"', () => {
+        element.style.whiteSpace = 'pre';
+        pasteWhiteSpaceFormatParser(format, element, context, defaultStyle);
+        expect(context.defaultFormatParsers.whiteSpace).not.toHaveBeenCalled();
+    });
+});