Add word count in document info modal (#738)

Co-authored-by: Tilman Vatteroth <git@tilmanvatteroth.de>

src/components/render-page/iframe-editor-to-renderer-communicator.ts

@@ -27,6 +27,7 @@ export class IframeEditorToRendererCommunicator extends IframeCommunicator<
   private onRendererReadyHandler?: () => void
   private onImageClickedHandler?: (details: ImageDetails) => void
   private onHeightChangeHandler?: (height: number) => void
+  private onWordCountCalculatedHandler?: (words: number) => void
 
   public onHeightChange(handler?: (height: number) => void): void {
     this.onHeightChangeHandler = handler
@@ -60,6 +61,10 @@ export class IframeEditorToRendererCommunicator extends IframeCommunicator<
     this.onSetScrollStateHandler = handler
   }
 
+  public onWordCountCalculated(handler?: (words: number) => void): void {
+    this.onWordCountCalculatedHandler = handler
+  }
+
   public sendSetBaseConfiguration(baseConfiguration: BaseConfiguration): void {
     this.sendMessageToOtherSide({
       type: RenderIframeMessageType.SET_BASE_CONFIGURATION,
@@ -91,6 +96,12 @@ export class IframeEditorToRendererCommunicator extends IframeCommunicator<
     })
   }
 
+  public sendGetWordCount(): void {
+    this.sendMessageToOtherSide({
+      type: RenderIframeMessageType.GET_WORD_COUNT
+    })
+  }
+
   protected handleEvent(event: MessageEvent<RendererToEditorIframeMessage>): boolean | undefined {
     const renderMessage = event.data
     switch (renderMessage.type) {
@@ -118,6 +129,9 @@ export class IframeEditorToRendererCommunicator extends IframeCommunicator<
       case RenderIframeMessageType.ON_HEIGHT_CHANGE:
         this.onHeightChangeHandler?.(renderMessage.height)
         return false
+      case RenderIframeMessageType.ON_WORD_COUNT_CALCULATED:
+        this.onWordCountCalculatedHandler?.(renderMessage.words)
+        return false
     }
   }
 }

src/components/render-page/iframe-markdown-renderer.tsx

@@ -14,6 +14,7 @@ import { ImageClickHandler } from '../markdown-renderer/replace-components/image
 import { useImageClickHandler } from './hooks/use-image-click-handler'
 import { MarkdownDocument } from './markdown-document'
 import { useIFrameRendererToEditorCommunicator } from '../editor-page/render-context/iframe-renderer-to-editor-communicator-context-provider'
+import { countWords } from './word-counter'
 
 export const IframeMarkdownRenderer: React.FC = () => {
   const [markdownContent, setMarkdownContent] = useState('')
@@ -22,10 +23,24 @@ export const IframeMarkdownRenderer: React.FC = () => {
 
   const iframeCommunicator = useIFrameRendererToEditorCommunicator()
 
+  const countWordsInRenderedDocument = useCallback(() => {
+    const documentContainer = document.querySelector('.markdown-body')
+    if (!documentContainer) {
+      iframeCommunicator?.sendWordCountCalculated(0)
+      return
+    }
+    const wordCount = countWords(documentContainer)
+    iframeCommunicator?.sendWordCountCalculated(wordCount)
+  }, [iframeCommunicator])
+
   useEffect(() => iframeCommunicator?.onSetBaseConfiguration(setBaseConfiguration), [iframeCommunicator])
   useEffect(() => iframeCommunicator?.onSetMarkdownContent(setMarkdownContent), [iframeCommunicator])
   useEffect(() => iframeCommunicator?.onSetDarkMode(setDarkMode), [iframeCommunicator])
   useEffect(() => iframeCommunicator?.onSetScrollState(setScrollState), [iframeCommunicator, scrollState])
+  useEffect(
+    () => iframeCommunicator?.onGetWordCount(countWordsInRenderedDocument),
+    [iframeCommunicator, countWordsInRenderedDocument]
+  )
 
   const onTaskCheckedChange = useCallback(
     (lineInMarkdown: number, checked: boolean) => {

src/components/render-page/iframe-renderer-to-editor-communicator.ts

@@ -23,6 +23,7 @@ export class IframeRendererToEditorCommunicator extends IframeCommunicator<
   private onSetDarkModeHandler?: (darkModeActivated: boolean) => void
   private onSetScrollStateHandler?: (scrollState: ScrollState) => void
   private onSetBaseConfigurationHandler?: (baseConfiguration: BaseConfiguration) => void
+  private onGetWordCountHandler?: () => void
 
   public onSetBaseConfiguration(handler?: (baseConfiguration: BaseConfiguration) => void): void {
     this.onSetBaseConfigurationHandler = handler
@@ -40,6 +41,10 @@ export class IframeRendererToEditorCommunicator extends IframeCommunicator<
     this.onSetScrollStateHandler = handler
   }
 
+  public onGetWordCount(handler?: () => void): void {
+    this.onGetWordCountHandler = handler
+  }
+
   public sendRendererReady(): void {
     this.sendMessageToOtherSide({
       type: RenderIframeMessageType.RENDERER_READY
@@ -95,6 +100,13 @@ export class IframeRendererToEditorCommunicator extends IframeCommunicator<
     })
   }
 
+  public sendWordCountCalculated(words: number): void {
+    this.sendMessageToOtherSide({
+      type: RenderIframeMessageType.ON_WORD_COUNT_CALCULATED,
+      words
+    })
+  }
+
   protected handleEvent(event: MessageEvent<EditorToRendererIframeMessage>): boolean | undefined {
     const renderMessage = event.data
     switch (renderMessage.type) {
@@ -110,6 +122,9 @@ export class IframeRendererToEditorCommunicator extends IframeCommunicator<
       case RenderIframeMessageType.SET_BASE_CONFIGURATION:
         this.onSetBaseConfigurationHandler?.(renderMessage.baseConfiguration)
         return false
+      case RenderIframeMessageType.GET_WORD_COUNT:
+        this.onGetWordCountHandler?.()
+        return false
     }
   }
 }

src/components/render-page/rendering-message.ts

@@ -17,7 +17,9 @@ export enum RenderIframeMessageType {
   ON_SET_FRONTMATTER = 'ON_SET_FRONTMATTER',
   IMAGE_CLICKED = 'IMAGE_CLICKED',
   ON_HEIGHT_CHANGE = 'ON_HEIGHT_CHANGE',
-  SET_BASE_CONFIGURATION = 'SET_BASE_CONFIGURATION'
+  SET_BASE_CONFIGURATION = 'SET_BASE_CONFIGURATION',
+  GET_WORD_COUNT = 'GET_WORD_COUNT',
+  ON_WORD_COUNT_CALCULATED = 'ON_WORD_COUNT_CALCULATED'
 }
 
 export interface RendererToEditorSimpleMessage {
@@ -40,6 +42,10 @@ export interface SetBaseUrlMessage {
   baseConfiguration: BaseConfiguration
 }
 
+export interface GetWordCountMessage {
+  type: RenderIframeMessageType.GET_WORD_COUNT
+}
+
 export interface ImageClickedMessage {
   type: RenderIframeMessageType.IMAGE_CLICKED
   details: ImageDetails
@@ -76,11 +82,17 @@ export interface OnHeightChangeMessage {
   height: number
 }
 
+export interface OnWordCountCalculatedMessage {
+  type: RenderIframeMessageType.ON_WORD_COUNT_CALCULATED
+  words: number
+}
+
 export type EditorToRendererIframeMessage =
   | SetMarkdownContentMessage
   | SetDarkModeMessage
   | SetScrollStateMessage
   | SetBaseUrlMessage
+  | GetWordCountMessage
 
 export type RendererToEditorIframeMessage =
   | RendererToEditorSimpleMessage
@@ -90,6 +102,7 @@ export type RendererToEditorIframeMessage =
   | SetScrollStateMessage
   | ImageClickedMessage
   | OnHeightChangeMessage
+  | OnWordCountCalculatedMessage
 
 export enum RendererType {
   DOCUMENT,

src/components/render-page/word-counter.ts

@@ -0,0 +1,50 @@
+/*
+ * SPDX-FileCopyrightText: 2021 The HedgeDoc developers (see AUTHORS file)
+ *
+ * SPDX-License-Identifier: AGPL-3.0-only
+ */
+
+import wordsCount from 'words-count'
+
+/** List of HTML tag names that should not be counted. */
+const EXCLUDED_TAGS = ['img', 'pre', 'nav']
+/** List of class names that should not be counted. */
+const EXCLUDED_CLASSES = ['katex-mathml']
+
+/**
+ * Checks whether the given node is an excluded HTML tag and therefore should be
+ * excluded from counting.
+ * @param node The node to test.
+ * @return true if the node should be excluded, false otherwise.
+ */
+const isExcludedTag = (node: Element | ChildNode): boolean => {
+  return EXCLUDED_TAGS.includes(node.nodeName.toLowerCase())
+}
+
+/**
+ * Checks whether the given node is a HTML element with an excluded class name,
+ * so that it should be excluded.
+ * @param node The node to test.
+ * @return true if the node should be excluded, false otherwise.
+ */
+const isExcludedClass = (node: Element | ChildNode): boolean => {
+  return EXCLUDED_CLASSES.some((excludedClass) => (node as HTMLElement).classList?.contains(excludedClass))
+}
+
+/**
+ * Counts the words of the given node while ignoring empty nodes and excluded
+ * nodes. Child nodes will recursively counted as well.
+ * @param node The node whose content's words should be counted.
+ * @return The number of words counted in this node and its children.
+ */
+export const countWords = (node: Element | ChildNode): number => {
+  if (!node.textContent || isExcludedTag(node) || isExcludedClass(node)) {
+    return 0
+  }
+  if (!node.hasChildNodes()) {
+    return wordsCount(node.textContent)
+  }
+  return [...node.childNodes].reduce((words, childNode) => {
+    return words + countWords(childNode)
+  }, 0)
+}