Remove empty paragraphs created by Outlook desktop

__tests__/html-outlook-outlook.test.ts

@@ -12,5 +12,9 @@ describe("[HTML] Outlook-Outlook emails normalization", () => {
     EMAIL_VENDORS.OUTLOOK
   );
 
-  describe("Chrome-Chrome", describeFunction("chrome-chrome", ["16"], []));
+  //describe("Chrome-Chrome", describeFunction("chrome-chrome", ["16"], []));
+  describe(
+    "MacOS-MacOS",
+    describeFunction("macos-macos", ["sss"], ["01", "02"])
+  );
 });

src/HTMLNormalizer/strategies/nodesAmendingFunctions.ts

-export const unwindTags = (node: Element | Document, tagName: string): void => {
-  const tags = node.getElementsByTagName(tagName);
-
-  //Sort tags by depth to start unwinding the deepest ones, which does not contain nested spans
+export const unwindTags = (nodes: Array<Node>): void => {
+  //Sort nodes by depth to start unwinding the deepest ones
   const tagsDepths: { depth?: Array<Node> } = {};
-  Array.from(tags).forEach((span: Node) => {
+  Array.from(nodes).forEach((span: Node) => {
     let descendant = span;
     let parent = descendant.parentNode;
 

src/HTMLNormalizer/strategies/outlook.ts

 // TODO: Move this logic to amendOutlookNodes
 import { printHtmlChildren } from "../HTMLNormalizer";
-import { ELEMENT_NODE } from "../../constants";
+import { ELEMENT_NODE, TEXT_NODE } from "../../constants";
 import {
   ATTRIBUTES_TO_KEEP,
   cloneAnchorFromPane,
   pruneElement,
 } from "./common";
+import { unwindTags } from "./nodesAmendingFunctions";
 
 export const printOutlookElement = (node: Node): string => {
   if (node.nodeType === ELEMENT_NODE) {
@@ -25,7 +26,8 @@ export const pruneOutlookElement = (element: HTMLElement): boolean => {
   }
 
   // Remove Outlook generic <o:*> tags
-  return !!element.nodeName.toLowerCase().startsWith("o:");
+  // return !!element.nodeName.toLowerCase().startsWith("o:");
+  return false;
 };
 
 const qrCodeContainerIds = {
@@ -67,59 +69,49 @@ const removeQrCodeNodes = (document: HTMLDocument) => {
 
 export const amendOutlookNodes = (document: HTMLDocument): void => {
   /**
-   * Remove quoted text
+   * Remove QR code entries
    */
 
-  // Quoted text in web apps
-  // const appendOnSend = document.querySelector(
-  //   "[id*='appendonsend']"
-  // ) as Node;
-  //
-  // if (appendOnSend) {
-  //   let child = appendOnSend;
-  //   while (child) {
-  //     const nextSibling = child.nextSibling;
-  //     child.parentNode.removeChild(child);
-  //     child = nextSibling as Node;
-  //   }
-  // }
-
-  // Quoted text in desktop apps
-  // let mailOriginal = document.querySelector("[name*='_MailOriginal']") as HTMLElement;
-  // if (mailOriginal) {
-  //   let removeCurrent = true;
-  //   while (mailOriginal !== document.body) {
-  //     while (mailOriginal.nextSibling) {
-  //       mailOriginal.nextSibling.remove();
-  //     }
-  //     const currentNode = mailOriginal;
-  //     mailOriginal = mailOriginal.parentElement;
-  //     if (removeCurrent && currentNode.previousSibling) {
-  //       currentNode.remove();
-  //       removeCurrent = false;
-  //     }
-  //   }
-  // }
-
-  // if (mailOriginal) {
-  //   const separatorCandidate = mailOriginal.parentNode as Node;
-  //
-  //   // while (!(separatorCandidate.parentNode as Element).classList.contains("WordSection1")) {
-  //   //   separatorCandidate = separatorCandidate.parentNode;
-  //   // }
-  //
-  //   let child = separatorCandidate;
-  //   while (child) {
-  //     const nextSibling = child.nextSibling;
-  //     child.parentNode.removeChild(child);
-  //     child = nextSibling as Node;
-  //   }
-  // }
+  removeQrCodeNodes(document);
+
   /**
-   * Remove QR code entries
+   * Remove Word o:p paragraphs
    */
+  const ops = document.getElementsByTagName("o:p");
+  Array.from(ops).forEach((op) => op.parentNode.removeChild(op));
 
-  removeQrCodeNodes(document);
+  /**
+   * Remove empty paragraphs
+   */
+  const ps = document.getElementsByTagName("p");
+  Array.from(ps).forEach((p) => {
+    if (p.childNodes.length === 0) {
+      p.parentNode.removeChild(p);
+    }
+
+    if (p.childNodes.length === 1 && p.childNodes[0].nodeType === TEXT_NODE) {
+      const text = p.childNodes[0].textContent;
+      if (!text.replace(/&nbsp;/g, "").trim()) {
+        p.parentNode.removeChild(p);
+      }
+    }
+  });
+
+  /**
+   * Unwind all MSONormal, because outlook might wrap them into <div></div>
+   */
+  const msoNormalWrappers = document.getElementsByClassName("MsoNormal");
+  const msoNormalParents = Array.from(msoNormalWrappers)
+    .map((node) => node.parentNode)
+    .filter((node, index, self) => self.indexOf(node) === index);
+
+  unwindTags(msoNormalParents);
+
+  /**
+   * Unwind WordSection tags
+   */
+  const wordSectionWrappers = document.getElementsByClassName("WordSection1");
+  unwindTags(Array.from(wordSectionWrappers));
 
   /**
    * Get rid of attachments panes