Fix history corruption on title updates

package.json

@@ -18,7 +18,7 @@
   },
   "devDependencies": {
     "@types/express": "^4.17.21",
-    "@types/jquery": "^3.5.29",
+    "@types/jquery": "^3.5.30",
     "@types/node": "^20.11.30",
     "eslint": "^8.11.0",
     "eslint-config-etherpad": "^3.0.9",

src/pad/database/hash.ts

@@ -1,8 +1,8 @@
-import HTMLParser, { Node } from "node-html-parser";
 import { SearchEngine, PadType } from "ep_search/setup";
 import { tokenize } from "../static/js/parser";
 import { logPrefix } from "../util/log";
 import { getHashQuery } from "../static/js/hash";
+import { applyReplaceSet, ReplaceSet } from "./text";
 
 const api = require("ep_etherpad-lite/node/db/API");
 const { decode, encode } = require("he");
@@ -14,104 +14,78 @@ export type HashUpdate = {
 
 const MAX_PAGES = 10000;
 
-function replaceHashToken(token: string, oldTitle: string, newTitle: string) {
+function replaceHashToken(
+  token: string,
+  oldTitle: string,
+  newTitle: string,
+  offset: number
+): ReplaceSet | null {
   if (token === `#${oldTitle}`) {
-    return `#${newTitle}`;
+    return {
+      start: offset,
+      ndel: oldTitle.length + 1,
+      text: `#${newTitle}`,
+    };
   }
-  return token;
+  return null;
 }
 
-function replaceHashTokens(token: string, updates: HashUpdate[]) {
+function replaceHashTokens(
+  token: string,
+  updates: HashUpdate[],
+  offset: number
+): ReplaceSet | null {
   for (const update of updates) {
     const { oldTitle, newTitle } = update;
-    const replaced = replaceHashToken(token, oldTitle, newTitle);
-    if (replaced !== token) {
+    const replaced = replaceHashToken(token, oldTitle, newTitle, offset);
+    if (replaced !== null) {
       return replaced;
     }
   }
   return null;
 }
 
-function replaceHash(text: string, updates: HashUpdate[]): string | null {
-  let newtext = "";
+function replaceHash(text: string, updates: HashUpdate[]): ReplaceSet[] {
   let remain = text;
-  let replaced = false;
+  let offset = 0;
+  let replaceSet: ReplaceSet[] = [];
   while (remain.length > 0) {
     const token = tokenize(remain);
-    const replacedToken = replaceHashTokens(token, updates);
+    const replacedToken = replaceHashTokens(token, updates, offset);
     if (replacedToken !== null) {
-      newtext += replacedToken;
-      remain = remain.substring(token.length);
-      replaced = true;
-      continue;
+      replaceSet.push(replacedToken);
     }
-    newtext += token;
     remain = remain.substring(token.length);
+    offset += token.length;
   }
-  if (!replaced) {
-    return null;
-  }
-  return newtext;
-}
-
-function traverseNodes(node: Node, handler: (node: Node) => void) {
-  handler(node);
-  (node.childNodes || []).forEach((child: Node) => {
-    handler(child);
-    traverseNodes(child, handler);
-  });
-}
-
-function replaceHashHtml(html: string, updates: HashUpdate[]) {
-  let html_ = html;
-  const m = html.match(/^\<\!DOCTYPE\s+HTML\>(.+)$/);
-  if (m) {
-    html_ = m[1];
-  }
-  const root = HTMLParser.parse(html_);
-  let replaced = false;
-  traverseNodes(root, (node) => {
-    if (node.nodeType !== 3 /* Node.TEXT_NODE */) {
-      return;
-    }
-    const replacedText = replaceHash(decode(node.rawText), updates);
-    if (replacedText === null) {
-      return;
-    }
-    replaced = true;
-    node.rawText = encode(replacedText);
-  });
-  if (!replaced) {
-    return null;
-  }
-  return root.toString();
+  return replaceSet;
 }
 
 export async function updateHash(padId: string, updates: HashUpdate[]) {
-  const { html } = await api.getHTML(padId);
+  const { text } = await api.getText(padId);
   console.debug(
     logPrefix,
     "Update hash with text",
     padId,
     ", src=",
-    html,
+    text,
     ", updates=",
     updates
   );
-  const newhtml = replaceHashHtml(html, updates);
-  if (newhtml === null) {
-    console.warn(logPrefix, "Hash not found in HTML", updates, html);
+  const replaceSet = replaceHash(text, updates);
+  if (replaceSet.length === 0) {
+    console.warn(logPrefix, "Hash not found in HTML", updates, text);
     return padId;
   }
-  await api.setHTML(padId, newhtml);
+  await applyReplaceSet(padId, replaceSet);
   console.debug(
     logPrefix,
     "Update hash with text",
     padId,
     ", src=",
-    html,
+    text,
     ", dest=",
-    newhtml
+    replaceSet
   );
   return padId;
 }

src/pad/database/text.ts

@@ -0,0 +1,59 @@
+const api = require("ep_etherpad-lite/node/db/API");
+const padMessageHandler = require("ep_etherpad-lite/node/handler/PadMessageHandler");
+const CustomError = require('ep_etherpad-lite/node/utils/customError');
+const padManager = require('ep_etherpad-lite/node/db/PadManager');
+
+// based on https://github.com/ether/etherpad-lite/blob/develop/src/node/db/API.ts
+// ... This is required to use spliceText, which is not exported in the API
+
+// gets a pad safe
+const getPadSafe = async (padID: string|object, shouldExist: boolean, text?:string, authorId:string = '') => {
+  // check if padID is a string
+  if (typeof padID !== 'string') {
+    throw new CustomError('padID is not a string', 'apierror');
+  }
+
+  // check if the padID maches the requirements
+  if (!padManager.isValidPadId(padID)) {
+    throw new CustomError('padID did not match requirements', 'apierror');
+  }
+
+  // check if the pad exists
+  const exists = await padManager.doesPadExists(padID);
+
+  if (!exists && shouldExist) {
+    // does not exist, but should
+    throw new CustomError('padID does not exist', 'apierror');
+  }
+
+  if (exists && !shouldExist) {
+    // does exist, but shouldn't
+    throw new CustomError('padID does already exist', 'apierror');
+  }
+
+  // pad exists, let's get it
+  return padManager.getPad(padID, text, authorId);
+};
+
+export type ReplaceSet = {
+  start: number;
+  ndel: number;
+  text: string;
+};
+
+export async function applyReplaceSet(
+  padID: string,
+  replaceSet: ReplaceSet[],
+  authorId: string = ""
+) {
+  const pad = await getPadSafe(padID, true);
+
+  const sortedReplaceSet = new Array<ReplaceSet>(...replaceSet)
+    .sort((a, b) => a.start - b.start)
+    .reverse();
+  for (const replaceSet of sortedReplaceSet) {
+    const { start, ndel, text } = replaceSet;
+    await pad.spliceText(start, ndel, text, authorId);
+  }
+  await padMessageHandler.updatePadClients(pad);
+}

src/pad/database/title.ts

@@ -1,9 +1,9 @@
-import HTMLParser, { Node } from "node-html-parser";
 import { SearchEngine, PadType } from "ep_search/setup";
 import removeMdBase from "remove-markdown";
 import { searchHashes, updateHash } from "./hash";
 import { logPrefix } from "../util/log";
 import { escapeForText } from "../static/js/result";
+import { applyReplaceSet, ReplaceSet } from "./text";
 
 const api = require("ep_etherpad-lite/node/db/API");
 const db = require("ep_etherpad-lite/node/db/DB").db;
@@ -43,51 +43,21 @@ export function extractTitle(padData: PadType) {
   return removeMd(lines[0]);
 }
 
-function traverseNodes(node: Node, handler: (node: Node) => void) {
-  handler(node);
-  (node.childNodes || []).forEach((child: Node) => {
-    handler(child);
-    traverseNodes(child, handler);
-  });
-}
-
-function replaceTitle(text: string, oldtitle: string, newtitle: string) {
-  return text.replace(oldtitle, newtitle);
-}
-
-function replaceTitleHtml(
-  html: string,
+function replaceTitle(
+  text: string,
   oldtitle: string,
-  newtitle: string
-): string | null {
-  let html_ = html;
-  const m = html.match(/^\<\!DOCTYPE\s+HTML\>(.+)$/);
-  if (m) {
-    html_ = m[1];
-  }
-  const root = HTMLParser.parse(html_);
-  let replaced = false;
-  traverseNodes(root, (node) => {
-    if (node.nodeType !== 3 /* Node.TEXT_NODE */) {
-      return;
-    }
-    if (replaced) {
-      return;
-    }
-    const decodedText = decode(node.rawText);
-    if (!decodedText.includes(oldtitle)) {
-      return;
-    }
-    replaced = true;
-    const replacedText = replaceTitle(decodedText, oldtitle, newtitle);
-    console.info(logPrefix, "Title replaced", decodedText, replacedText);
-    node.rawText = encode(replacedText);
-  });
-  if (!replaced) {
-    console.warn(logPrefix, "Title not found in HTML", oldtitle, html);
+  newtitle: string,
+  offset: number = 0
+): ReplaceSet | null {
+  const pos = text.indexOf(oldtitle);
+  if (pos === -1) {
     return null;
   }
-  return root.toString();
+  return {
+    start: pos + offset,
+    ndel: oldtitle.length,
+    text: newtitle,
+  };
 }
 
 async function updateTitleContent(
@@ -97,9 +67,9 @@ async function updateTitleContent(
   newTitle: string
 ): Promise<TitleUpdateResult> {
   console.info(logPrefix, "Update title", pad.id, oldTitle, newTitle);
-  const { html } = await api.getHTML(pad.id);
-  const replacedHtml = replaceTitleHtml(html, oldTitle, newTitle);
-  if (replacedHtml === null) {
+  const { text } = await api.getText(pad.id);
+  const replaceSet = replaceTitle(text, oldTitle, newTitle);
+  if (replaceSet === null) {
     console.warn(logPrefix, "Title not found in HTML", oldTitle, newTitle);
     const updates = await searchHashes(searchEngine, oldTitle);
     return {
@@ -110,7 +80,7 @@ async function updateTitleContent(
       },
     };
   }
-  await api.setHTML(pad.id, replacedHtml);
+  await applyReplaceSet(pad.id, [replaceSet]);
 
   const updates = await searchHashes(searchEngine, oldTitle);
   return {

src/pad/templates/index.html

@@ -17,6 +17,9 @@
 <script src="https://cdnjs.cloudflare.com/ajax/libs/require.js/2.3.6/require.min.js">
 </script>
 
+<script src="https://code.jquery.com/jquery-3.7.1.min.js" integrity="sha256-/JqT3SQfawRcv/BIHPThkBvs0OEvtFFmqPF/lYI/Cxo=" crossorigin="anonymous">
+</script>
+
 <script type="text/javascript">
   $(document).ready(function(){
     require(['index/index'], function(index) {