Merge pull request #23212 from overleaf/em-docupdater-send-doc-hash

Send doc hash with history updates from docupdater

GitOrigin-RevId: 119475d4198c6603cecd4fd579a64ff4448261ce

services/document-updater/app/js/UpdateManager.js

@@ -14,7 +14,7 @@ const DocumentManager = require('./DocumentManager')
 const RangesManager = require('./RangesManager')
 const SnapshotManager = require('./SnapshotManager')
 const Profiler = require('./Profiler')
-const { isInsert, isDelete, getDocLength } = require('./Utils')
+const { isInsert, isDelete, getDocLength, computeDocHash } = require('./Utils')
 
 /**
  * @import { DeleteOp, InsertOp, Op, Ranges, Update, HistoryUpdate } from "./types"
@@ -162,6 +162,7 @@ const UpdateManager = {
         projectHistoryId,
         lines,
         ranges,
+        updatedDocLines,
         historyRangesSupport
       )
 
@@ -290,8 +291,9 @@ const UpdateManager = {
    * @param {HistoryUpdate[]} updates
    * @param {string} pathname
    * @param {string} projectHistoryId
-   * @param {string[]} lines
-   * @param {Ranges} ranges
+   * @param {string[]} lines - document lines before updates were applied
+   * @param {Ranges} ranges - ranges before updates were applied
+   * @param {string[]} newLines - document lines after updates were applied
    * @param {boolean} historyRangesSupport
    */
   _adjustHistoryUpdatesMetadata(
@@ -300,6 +302,7 @@ const UpdateManager = {
     projectHistoryId,
     lines,
     ranges,
+    newLines,
     historyRangesSupport
   ) {
     let docLength = getDocLength(lines)
@@ -363,6 +366,12 @@ const UpdateManager = {
         delete update.meta.tc
       }
     }
+
+    if (historyRangesSupport && updates.length > 0) {
+      const lastUpdate = updates[updates.length - 1]
+      lastUpdate.meta ??= {}
+      lastUpdate.meta.doc_hash = computeDocHash(newLines)
+    }
   },
 }
 

services/document-updater/app/js/Utils.js

@@ -1,4 +1,5 @@
 // @ts-check
+const { createHash } = require('node:crypto')
 const _ = require('lodash')
 
 /**
@@ -79,6 +80,27 @@ function addTrackedDeletesToContent(content, trackedChanges) {
   return result
 }
 
+/**
+ * Compute the content hash for a doc
+ *
+ * This hash is sent to the history to validate updates.
+ *
+ * @param {string[]} lines
+ * @return {string} the doc hash
+ */
+function computeDocHash(lines) {
+  const hash = createHash('sha1')
+  if (lines.length > 0) {
+    for (const line of lines.slice(0, lines.length - 1)) {
+      hash.update(line)
+      hash.update('\n')
+    }
+    // The last line doesn't end with a newline
+    hash.update(lines[lines.length - 1])
+  }
+  return hash.digest('hex')
+}
+
 /**
  * checks if the given originOrSource should be treated as a source or origin
  * TODO: remove this hack and remove all "source" references
@@ -102,5 +124,6 @@ module.exports = {
   isComment,
   addTrackedDeletesToContent,
   getDocLength,
+  computeDocHash,
   extractOriginOrSource,
 }

services/document-updater/app/js/types.ts

@@ -84,6 +84,7 @@ export type HistoryUpdate = {
     pathname?: string
     doc_length?: number
     history_doc_length?: number
+    doc_hash?: string
     tc?: boolean
     user_id?: string
   }

services/document-updater/test/unit/js/UpdateManager/UpdateManagerTests.js

@@ -1,5 +1,4 @@
-// @ts-check
-
+const { createHash } = require('node:crypto')
 const sinon = require('sinon')
 const { expect } = require('chai')
 const SandboxedModule = require('sandboxed-module')
@@ -399,7 +398,9 @@ describe('UpdateManager', function () {
           this.historyUpdates,
           this.pathname,
           this.projectHistoryId,
-          this.lines
+          this.lines,
+          this.ranges,
+          this.updatedDocLines
         )
       })
 
@@ -526,6 +527,7 @@ describe('UpdateManager', function () {
   describe('_adjustHistoryUpdatesMetadata', function () {
     beforeEach(function () {
       this.lines = ['some', 'test', 'data']
+      this.updatedDocLines = ['after', 'updates']
       this.historyUpdates = [
         {
           v: 42,
@@ -570,6 +572,7 @@ describe('UpdateManager', function () {
         this.pathname,
         this.projectHistoryId,
         this.lines,
+        this.updatedDocLines,
         this.ranges,
         false
       )
@@ -632,6 +635,7 @@ describe('UpdateManager', function () {
         this.projectHistoryId,
         this.lines,
         this.ranges,
+        this.updatedDocLines,
         true
       )
       this.historyUpdates.should.deep.equal([
@@ -685,6 +689,7 @@ describe('UpdateManager', function () {
           meta: {
             pathname: this.pathname,
             doc_length: 21, // 23 - 'so'
+            doc_hash: stringHash(this.updatedDocLines.join('\n')),
             history_doc_length: 28, // 30 - 'so'
           },
         },
@@ -699,6 +704,7 @@ describe('UpdateManager', function () {
         this.projectHistoryId,
         [],
         {},
+        ['foobar'],
         false
       )
       this.historyUpdates.should.deep.equal([
@@ -822,3 +828,9 @@ describe('UpdateManager', function () {
     })
   })
 })
+
+function stringHash(s) {
+  const hash = createHash('sha1')
+  hash.update(s)
+  return hash.digest('hex')
+}

services/document-updater/test/unit/js/UtilsTests.js

@@ -1,5 +1,6 @@
 // @ts-check
 
+const { createHash } = require('node:crypto')
 const { expect } = require('chai')
 const Utils = require('../../../app/js/Utils')
 
@@ -24,4 +25,30 @@ describe('Utils', function () {
       expect(result).to.equal('the quick brown fox jumps over the lazy dog')
     })
   })
+
+  describe('computeDocHash', function () {
+    it('computes the hash for an empty doc', function () {
+      const actual = Utils.computeDocHash([])
+      const expected = stringHash('')
+      expect(actual).to.equal(expected)
+    })
+
+    it('computes the hash for a single-line doc', function () {
+      const actual = Utils.computeDocHash(['hello'])
+      const expected = stringHash('hello')
+      expect(actual).to.equal(expected)
+    })
+
+    it('computes the hash for a multiline doc', function () {
+      const actual = Utils.computeDocHash(['hello', 'there', 'world'])
+      const expected = stringHash('hello\nthere\nworld')
+      expect(actual).to.equal(expected)
+    })
+  })
 })
+
+function stringHash(s) {
+  const hash = createHash('sha1')
+  hash.update(s)
+  return hash.digest('hex')
+}