peek at docs without unarchiving

services/docstore/app.js

@@ -54,6 +54,7 @@ app.get('/project/:project_id/ranges', HttpController.getAllRanges)
 app.get('/project/:project_id/doc/:doc_id', HttpController.getDoc)
 app.get('/project/:project_id/doc/:doc_id/deleted', HttpController.isDocDeleted)
 app.get('/project/:project_id/doc/:doc_id/raw', HttpController.getRawDoc)
+app.get('/project/:project_id/doc/:doc_id/peek', HttpController.peekDoc)
 // Add 64kb overhead for the JSON encoding, and double the size to allow for ranges in the json payload
 app.post(
   '/project/:project_id/doc/:doc_id',

services/docstore/app/js/DocManager.js

@@ -124,6 +124,70 @@ module.exports = DocManager = {
     )
   },
 
+  // returns the doc without any version information
+  _peekRawDoc(project_id, doc_id, callback) {
+    MongoManager.findDoc(
+      project_id,
+      doc_id,
+      {
+        lines: true,
+        rev: true,
+        deleted: true,
+        version: true,
+        ranges: true,
+        inS3: true,
+      },
+      (err, doc) => {
+        if (err) return callback(err)
+        if (doc == null) {
+          return callback(
+            new Errors.NotFoundError(
+              `No such doc: ${doc_id} in project ${project_id}`
+            )
+          )
+        }
+        if (doc && !doc.inS3) {
+          return callback(null, doc)
+        }
+        // skip the unarchiving to mongo when getting a doc
+        DocArchive.getDoc(project_id, doc_id, function (err, archivedDoc) {
+          if (err != null) {
+            logger.err(
+              { err, project_id, doc_id },
+              'error getting doc from archive'
+            )
+            return callback(err)
+          }
+          doc = _.extend(doc, archivedDoc)
+          callback(null, doc)
+        })
+      }
+    )
+  },
+
+  // get the doc from mongo if possible, or from the persistent store otherwise,
+  // without unarchiving it (avoids unnecessary writes to mongo)
+  peekDoc(project_id, doc_id, callback) {
+    DocManager._peekRawDoc(project_id, doc_id, (err, doc) => {
+      if (err) {
+        return callback(err)
+      }
+      MongoManager.WithRevCheck(
+        doc,
+        MongoManager.getDocVersion,
+        function (error, version) {
+          // If the doc has been modified while we were retrieving it, we
+          // will get a DocModified error
+          if (error != null) {
+            return callback(error)
+          }
+          doc.version = version
+          return callback(err, doc)
+        }
+      )
+    })
+  },
+
   getDocLines(project_id, doc_id, callback) {
     if (callback == null) {
       callback = function (err, doc) {}

services/docstore/app/js/Errors.js

@@ -4,7 +4,10 @@ const { Errors } = require('@overleaf/object-persistor')
 
 class Md5MismatchError extends OError {}
 
+class DocModifiedError extends OError {}
+
 module.exports = {
   Md5MismatchError,
+  DocModifiedError,
   ...Errors,
 }

services/docstore/app/js/HttpController.js

@@ -44,6 +44,22 @@ module.exports = HttpController = {
     })
   },
 
+  peekDoc(req, res, next) {
+    const { project_id } = req.params
+    const { doc_id } = req.params
+    logger.log({ project_id, doc_id }, 'peeking doc')
+    DocManager.peekDoc(project_id, doc_id, function (error, doc) {
+      if (error) {
+        return next(error)
+      }
+      if (doc == null) {
+        return res.sendStatus(404)
+      } else {
+        return res.json(HttpController._buildDocView(doc))
+      }
+    })
+  },
+
   isDocDeleted(req, res, next) {
     const { doc_id: docId, project_id: projectId } = req.params
     DocManager.isDocDeleted(projectId, docId, function (error, deleted) {

services/docstore/app/js/MongoManager.js

@@ -15,6 +15,7 @@ const { db, ObjectId } = require('./mongodb')
 const logger = require('logger-sharelatex')
 const metrics = require('@overleaf/metrics')
 const Settings = require('@overleaf/settings')
+const { DocModifiedError } = require('./Errors')
 const { promisify } = require('util')
 
 module.exports = MongoManager = {
@@ -178,6 +179,45 @@ module.exports = MongoManager = {
     )
   },
 
+  getDocRev(doc_id, callback) {
+    db.docs.findOne(
+      {
+        _id: ObjectId(doc_id.toString()),
+      },
+      {
+        projection: { rev: 1 },
+      },
+      function (err, doc) {
+        if (err != null) {
+          return callback(err)
+        }
+        callback(null, doc && doc.rev)
+      }
+    )
+  },
+
+  // Helper method  to support optimistic locking. Call the provided method for
+  // an existing doc and return the result if the rev in mongo is unchanged when
+  // checked afterwards. If the rev has changed, return a DocModifiedError.
+  WithRevCheck(doc, method, callback) {
+    method(doc._id, function (err, result) {
+      if (err) return callback(err)
+      MongoManager.getDocRev(doc._id, function (err, currentRev) {
+        if (err) return callback(err)
+        if (doc.rev !== currentRev) {
+          return callback(
+            new DocModifiedError('doc rev has changed', {
+              doc_id: doc._id,
+              rev: doc.rev,
+              currentRev,
+            })
+          )
+        }
+        return callback(null, result)
+      })
+    })
+  },
+
   destroyDoc(doc_id, callback) {
     db.docs.deleteOne(
       {

services/docstore/test/acceptance/js/GettingDocsFromArchiveTest.js

@@ -0,0 +1,99 @@
+/* eslint-disable
+    camelcase,
+    handle-callback-err,
+    no-unused-vars,
+*/
+// TODO: This file was created by bulk-decaffeinate.
+// Fix any style issues and re-enable lint.
+/*
+ * decaffeinate suggestions:
+ * DS101: Remove unnecessary use of Array.from
+ * DS102: Remove unnecessary code created because of implicit returns
+ * DS207: Consider shorter variations of null checks
+ * Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
+ */
+process.env.BACKEND = 'gcs'
+const Settings = require('@overleaf/settings')
+const { expect } = require('chai')
+const { db, ObjectId } = require('../../../app/js/mongodb')
+const async = require('async')
+const DocstoreApp = require('./helpers/DocstoreApp')
+const DocstoreClient = require('./helpers/DocstoreClient')
+const { Storage } = require('@google-cloud/storage')
+const Persistor = require('../../../app/js/PersistorManager')
+const Streamifier = require('streamifier')
+
+function uploadContent(path, json, callback) {
+  const stream = Streamifier.createReadStream(JSON.stringify(json))
+  Persistor.sendStream(Settings.docstore.bucket, path, stream)
+    .then(() => callback())
+    .catch(callback)
+}
+
+describe('Getting A Doc from Archive', function () {
+  before(function (done) {
+    return DocstoreApp.ensureRunning(done)
+  })
+
+  before(async function () {
+    const storage = new Storage(Settings.docstore.gcs.endpoint)
+    await storage.createBucket(Settings.docstore.bucket)
+    await storage.createBucket(`${Settings.docstore.bucket}-deleted`)
+  })
+
+  describe('archiving a single doc', function () {
+    before(function (done) {
+      this.project_id = ObjectId()
+      this.timeout(1000 * 30)
+      this.doc = {
+        _id: ObjectId(),
+        lines: ['foo', 'bar'],
+        ranges: {},
+        version: 2,
+      }
+      DocstoreClient.createDoc(
+        this.project_id,
+        this.doc._id,
+        this.doc.lines,
+        this.doc.version,
+        this.doc.ranges,
+        error => {
+          if (error) {
+            return done(error)
+          }
+          DocstoreClient.archiveDocById(
+            this.project_id,
+            this.doc._id,
+            (error, res) => {
+              this.res = res
+              if (error) {
+                return done(error)
+              }
+              done()
+            }
+          )
+        }
+      )
+    })
+
+    it('should successully archive the doc', function (done) {
+      this.res.statusCode.should.equal(204)
+      done()
+    })
+
+    it('should get the doc lines and version', function (done) {
+      return DocstoreClient.peekDoc(
+        this.project_id,
+        this.doc._id,
+        {},
+        (error, res, doc) => {
+          res.statusCode.should.equal(200)
+          doc.lines.should.deep.equal(this.doc.lines)
+          doc.version.should.equal(this.doc.version)
+          doc.ranges.should.deep.equal(this.doc.ranges)
+          return done()
+        }
+      )
+    })
+  })
+})

services/docstore/test/acceptance/js/helpers/DocstoreClient.js

@@ -60,6 +60,20 @@ module.exports = DocstoreClient = {
     )
   },
 
+  peekDoc(project_id, doc_id, qs, callback) {
+    if (callback == null) {
+      callback = function (error, res, body) {}
+    }
+    return request.get(
+      {
+        url: `http://localhost:${settings.internal.docstore.port}/project/${project_id}/doc/${doc_id}/peek`,
+        json: true,
+        qs,
+      },
+      callback
+    )
+  },
+
   isDocDeleted(project_id, doc_id, callback) {
     request.get(
       {

services/docstore/test/unit/js/MongoManagerTests.js

@@ -17,6 +17,7 @@ const modulePath = require('path').join(
 )
 const { ObjectId } = require('mongodb')
 const { assert } = require('chai')
+const Errors = require('../../../app/js/Errors')
 
 describe('MongoManager', function () {
   beforeEach(function () {
@@ -28,6 +29,7 @@ describe('MongoManager', function () {
         },
         '@overleaf/metrics': { timeAsyncMethod: sinon.stub() },
         '@overleaf/settings': { max_deleted_docs: 42 },
+        './Errors': { Errors },
       },
     })
     this.project_id = ObjectId().toString()