Merge pull request #569 from sharelatex/bg-compile-from-redis

compile from redis

services/web/app/coffee/Features/Compile/ClsiManager.coffee

@@ -3,17 +3,31 @@ async = require "async"
 Settings = require "settings-sharelatex"
 request = require('request')
 Project = require("../../models/Project").Project
+ProjectGetter = require("../Project/ProjectGetter")
 ProjectEntityHandler = require("../Project/ProjectEntityHandler")
 logger = require "logger-sharelatex"
 Url = require("url")
 ClsiCookieManager = require("./ClsiCookieManager")
+ClsiStateManager = require("./ClsiStateManager")
 _ = require("underscore")
 async = require("async")
 ClsiFormatChecker = require("./ClsiFormatChecker")
+DocumentUpdaterHandler = require "../DocumentUpdater/DocumentUpdaterHandler"
+Metrics = require('metrics-sharelatex')
 
 module.exports = ClsiManager =
 
-	sendRequest: (project_id, user_id, options = {}, callback = (error, status, outputFiles, clsiServerId, validationProblems) ->) ->
+	sendRequest: (project_id, user_id, options = {}, callback) ->
+		ClsiManager.sendRequestOnce project_id, user_id, options, (error, status, result...) ->
+			return callback(error) if error?
+			if status is 'conflict'
+				options = _.clone(options)
+				options.syncType = "full" #  force full compile
+				ClsiManager.sendRequestOnce project_id, user_id, options, callback # try again
+			else
+				callback(error, status, result...)
+
+	sendRequestOnce: (project_id, user_id, options = {}, callback = (error, status, outputFiles, clsiServerId, validationProblems) ->) ->
 		ClsiManager._buildRequest project_id, options, (error, req) ->
 			return callback(error) if error?
 			logger.log project_id: project_id, "sending compile to CLSI"
@@ -28,7 +42,7 @@ module.exports = ClsiManager =
 				if error?
 					logger.err err:error, project_id:project_id, "error sending request to clsi"
 					return callback(error)
-				logger.log project_id: project_id, outputFilesLength: response?.outputFiles?.length, status: response?.status, "received compile response from CLSI"
+				logger.log project_id: project_id, outputFilesLength: response?.outputFiles?.length, status: response?.status, compile_status: response?.compile?.status, "received compile response from CLSI"
 				ClsiCookieManager._getServerId project_id, (err, clsiServerId)->
 					if err?
 						logger.err err:err, project_id:project_id, "error getting server id"
@@ -85,6 +99,8 @@ module.exports = ClsiManager =
 				callback null, body
 			else if response.statusCode == 413
 				callback null, compile:status:"project-too-large"
+			else if response.statusCode == 409
+				callback null, compile:status:"conflict"
 			else
 				error = new Error("CLSI returned non-success code: #{response.statusCode}")
 				logger.error err: error, project_id: project_id, "CLSI returned failure code"
@@ -101,56 +117,116 @@ module.exports = ClsiManager =
 		return outputFiles
 
 	VALID_COMPILERS: ["pdflatex", "latex", "xelatex", "lualatex"]
+
 	_buildRequest: (project_id, options={}, callback = (error, request) ->) ->
-		Project.findById project_id, {compiler: 1, rootDoc_id: 1, imageName: 1}, (error, project) ->
+		ProjectGetter.getProject project_id, {compiler: 1, rootDoc_id: 1, imageName: 1, rootFolder:1}, (error, project) ->
 			return callback(error) if error?
 			return callback(new Errors.NotFoundError("project does not exist: #{project_id}")) if !project?
-
 			if project.compiler not in ClsiManager.VALID_COMPILERS
 				project.compiler = "pdflatex"
 
+			if options.incrementalCompilesEnabled or options.syncType? # new way, either incremental or full
+				timer = new Metrics.Timer("editor.compile-getdocs-redis")
+				ClsiManager.getContentFromDocUpdaterIfMatch project_id, project, (error, projectStateHash, docUpdaterDocs) ->
+					timer.done()
+					return callback(error) if error?
+					logger.log project_id: project_id, projectStateHash: projectStateHash, docs: docUpdaterDocs?, "checked project state"
+					# see if we can send an incremental update to the CLSI
+					if docUpdaterDocs? and options.syncType isnt "full"
+						# Workaround: for now, always flush project to mongo on compile
+						# until we have automatic periodic flushing on the docupdater
+						# side, to prevent documents staying in redis too long.
+						DocumentUpdaterHandler.flushProjectToMongo project_id, (error) ->
+							return callback(error) if error?
+							Metrics.inc "compile-from-redis"
+							ClsiManager._buildRequestFromDocupdater project_id, options, project, projectStateHash, docUpdaterDocs, callback
+					else
+						Metrics.inc "compile-from-mongo"
+						ClsiManager._buildRequestFromMongo project_id, options, project, projectStateHash, callback
+			else # old way, always from mongo
+				timer = new Metrics.Timer("editor.compile-getdocs-mongo")
+				ClsiManager._getContentFromMongo project_id, (error, docs, files) ->
+					timer.done()
+					return callback(error) if error?
+					ClsiManager._finaliseRequest project_id, options, project, docs, files, callback
+
+	getContentFromDocUpdaterIfMatch: (project_id, project, callback = (error, projectStateHash, docs) ->) ->
+		ClsiStateManager.computeHash project, (error, projectStateHash) ->
+			return callback(error) if error?
+			DocumentUpdaterHandler.getProjectDocsIfMatch project_id, projectStateHash, (error, docs) ->
+				return callback(error) if error?
+				callback(null, projectStateHash, docs)
+
+	_buildRequestFromDocupdater: (project_id, options, project, projectStateHash, docUpdaterDocs, callback = (error, request) ->) ->
+		ProjectEntityHandler.getAllDocPathsFromProject project, (error, docPath) ->
+				return callback(error) if error?
+				docs = {}
+				for doc in docUpdaterDocs or []
+					path = docPath[doc._id]
+					docs[path] = doc
+				# send new docs but not files as those are already on the clsi
+				options = _.clone(options)
+				options.syncType = "incremental"
+				options.syncState = projectStateHash
+				ClsiManager._finaliseRequest project_id, options, project, docs, [], callback
+
+	_buildRequestFromMongo: (project_id, options, project, projectStateHash, callback = (error, request) ->) ->
+		ClsiManager._getContentFromMongo project_id, (error, docs, files) ->
+			return callback(error) if error?
+			options = _.clone(options)
+			options.syncType = "full"
+			options.syncState = projectStateHash
+			ClsiManager._finaliseRequest project_id, options, project, docs, files, callback
+
+	_getContentFromMongo: (project_id, callback = (error, docs, files) ->) ->
+		DocumentUpdaterHandler.flushProjectToMongo project_id, (error) ->
+			return callback(error) if error?
 			ProjectEntityHandler.getAllDocs project_id, (error, docs = {}) ->
 				return callback(error) if error?
 				ProjectEntityHandler.getAllFiles project_id, (error, files = {}) ->
 					return callback(error) if error?
-
-					resources = []
-					rootResourcePath = null
-					rootResourcePathOverride = null
-
-					for path, doc of docs
-						path = path.replace(/^\//, "") # Remove leading /
-						resources.push
-							path:    path
-							content: doc.lines.join("\n")
-						if project.rootDoc_id? and doc._id.toString() == project.rootDoc_id.toString()
-							rootResourcePath = path
-						if options.rootDoc_id? and doc._id.toString() == options.rootDoc_id.toString()
-							rootResourcePathOverride = path
-
-					rootResourcePath = rootResourcePathOverride if rootResourcePathOverride?
-					if !rootResourcePath?
-						logger.warn {project_id}, "no root document found, setting to main.tex"
-						rootResourcePath = "main.tex"
-
-					for path, file of files
-						path = path.replace(/^\//, "") # Remove leading /
-						resources.push
-							path:     path
-							url:      "#{Settings.apis.filestore.url}/project/#{project._id}/file/#{file._id}"
-							modified: file.created?.getTime()
-
-					callback null, {
-						compile:
-							options:
-								compiler: project.compiler
-								timeout: options.timeout
-								imageName: project.imageName
-								draft: !!options.draft
-								check: options.check
-							rootResourcePath: rootResourcePath
-							resources: resources
-					}
+					callback(null, docs, files)
+
+	_finaliseRequest: (project_id, options, project, docs, files, callback = (error, params) -> ) ->
+		resources = []
+		rootResourcePath = null
+		rootResourcePathOverride = null
+
+		for path, doc of docs
+			path = path.replace(/^\//, "") # Remove leading /
+			resources.push
+				path:    path
+				content: doc.lines.join("\n")
+			if project.rootDoc_id? and doc._id.toString() == project.rootDoc_id.toString()
+				rootResourcePath = path
+			if options.rootDoc_id? and doc._id.toString() == options.rootDoc_id.toString()
+				rootResourcePathOverride = path
+
+		rootResourcePath = rootResourcePathOverride if rootResourcePathOverride?
+		if !rootResourcePath?
+			logger.warn {project_id}, "no root document found, setting to main.tex"
+			rootResourcePath = "main.tex"
+
+		for path, file of files
+			path = path.replace(/^\//, "") # Remove leading /
+			resources.push
+				path:     path
+				url:      "#{Settings.apis.filestore.url}/project/#{project._id}/file/#{file._id}"
+				modified: file.created?.getTime()
+
+		callback null, {
+			compile:
+				options:
+					compiler: project.compiler
+					timeout: options.timeout
+					imageName: project.imageName
+					draft: !!options.draft
+					check: options.check
+					syncType: options.syncType
+					syncState: options.syncState
+				rootResourcePath: rootResourcePath
+				resources: resources
+		}
 
 	wordCount: (project_id, user_id, file, options, callback = (error, response) ->) ->
 		ClsiManager._buildRequest project_id, options, (error, req) ->

services/web/app/coffee/Features/Compile/ClsiStateManager.coffee

@@ -0,0 +1,31 @@
+Settings = require "settings-sharelatex"
+logger = require "logger-sharelatex"
+crypto = require "crypto"
+ProjectEntityHandler = require "../Project/ProjectEntityHandler"
+
+# The "state" of a project is a hash of the relevant attributes in the
+# project object in this case we only need the rootFolder.
+#
+# The idea is that it will change if any doc or file is
+# created/renamed/deleted, and also if the content of any file (not
+# doc) changes.
+#
+# When the hash changes the full set of files on the CLSI will need to
+# be updated.  If it doesn't change then we can overwrite changed docs
+# in place on the clsi, getting them from the docupdater.
+#
+# The docupdater is responsible for setting the key in redis, and
+# unsetting it if it removes any documents from the doc updater.
+
+buildState = (s) ->
+	return crypto.createHash('sha1').update(s, 'utf8').digest('hex')
+
+module.exports = ClsiStateManager =
+
+	computeHash: (project, callback = (err, hash) ->) ->
+		ProjectEntityHandler.getAllEntitiesFromProject project, (err, docs, files) ->
+			fileList = ("#{f.file._id}:#{f.file.rev}:#{f.file.created}:#{f.path}" for f in files or [])
+			docList = ("#{d.doc._id}:#{d.path}" for d in docs or [])
+			sortedEntityList = [docList..., fileList...].sort()
+			hash = buildState(sortedEntityList.join("\n"))
+			callback(null, hash)

services/web/app/coffee/Features/Compile/CompileController.coffee

@@ -30,6 +30,8 @@ module.exports = CompileController =
 			options.draft = req.body.draft
 		if req.body?.check in ['validate', 'error', 'silent']
 			options.check = req.body.check
+		if req.body?.incrementalCompilesEnabled
+			options.incrementalCompilesEnabled = true
 		logger.log {options:options, project_id:project_id, user_id:user_id}, "got compile request"
 		CompileManager.compile project_id, user_id, options, (error, status, outputFiles, clsiServerId, limits, validationProblems) ->
 			return next(error) if error?

services/web/app/coffee/Features/Compile/CompileManager.coffee

@@ -1,7 +1,6 @@
 Settings = require('settings-sharelatex')
 RedisWrapper = require("../../infrastructure/RedisWrapper")
 rclient = RedisWrapper.client("clsi_recently_compiled")
-DocumentUpdaterHandler = require "../DocumentUpdater/DocumentUpdaterHandler"
 Project = require("../../models/Project").Project
 ProjectRootDocManager = require "../Project/ProjectRootDocManager"
 UserGetter = require "../User/UserGetter"
@@ -31,19 +30,17 @@ module.exports = CompileManager =
 				
 				CompileManager._ensureRootDocumentIsSet project_id, (error) ->
 					return callback(error) if error?
-					DocumentUpdaterHandler.flushProjectToMongo project_id, (error) ->
+					CompileManager.getProjectCompileLimits project_id, (error, limits) ->
 						return callback(error) if error?
-						CompileManager.getProjectCompileLimits project_id, (error, limits) ->
+						for key, value of limits
+							options[key] = value
+						# only pass user_id down to clsi if this is a per-user compile
+						compileAsUser = if Settings.disablePerUserCompiles then undefined else user_id
+						ClsiManager.sendRequest project_id, compileAsUser, options, (error, status, outputFiles, clsiServerId, validationProblems) ->
 							return callback(error) if error?
-							for key, value of limits
-								options[key] = value
-							# only pass user_id down to clsi if this is a per-user compile
-							compileAsUser = if Settings.disablePerUserCompiles then undefined else user_id
-							ClsiManager.sendRequest project_id, compileAsUser, options, (error, status, outputFiles, clsiServerId, validationProblems) ->
-								return callback(error) if error?
-								logger.log files: outputFiles, "output files"
-								callback(null, status, outputFiles, clsiServerId, limits, validationProblems)
-								
+							logger.log files: outputFiles, "output files"
+							callback(null, status, outputFiles, clsiServerId, limits, validationProblems)
+
 
 	stopCompile: (project_id, user_id, callback = (error) ->) ->
 		CompileManager.getProjectCompileLimits project_id, (error, limits) ->

services/web/app/coffee/Features/DocumentUpdater/DocumentUpdaterHandler.coffee

@@ -123,6 +123,36 @@ module.exports = DocumentUpdaterHandler =
 				logger.error project_id:project_id, doc_id:doc_id, url: url, "doc updater returned a non-success status code: #{res.statusCode}"
 				callback new Error("doc updater returned a non-success status code: #{res.statusCode}")
 
+	getProjectDocsIfMatch: (project_id, projectStateHash, callback = (error, docs) ->) ->
+		# If the project state hasn't changed, we can get all the latest
+		# docs from redis via the docupdater. Otherwise we will need to
+		# fall back to getting them from mongo.
+		timer = new metrics.Timer("get-project-docs")
+		url = "#{settings.apis.documentupdater.url}/project/#{project_id}/doc?state=#{projectStateHash}"
+		logger.log project_id:project_id, "getting project docs from document updater"
+		request.get url, (error, res, body)->
+			timer.done()
+			if error?
+				logger.error err:error, url:url, project_id:project_id, "error getting project docs from doc updater"
+				return callback(error)
+			if res.statusCode is 409 # HTTP response code "409 Conflict"
+				# Docupdater has checked the projectStateHash and found that
+				# it has changed. This means that the docs currently in redis
+				# aren't the only change to the project and the full set of
+				# docs/files should be retreived from docstore/filestore
+				# instead.
+				return callback()
+			else if res.statusCode >= 200 and res.statusCode < 300
+				logger.log project_id:project_id, "got project docs from document document updater"
+				try
+					docs = JSON.parse(body)
+				catch error
+					return callback(error)
+				callback null, docs
+			else
+				logger.error project_id:project_id, url: url, "doc updater returned a non-success status code: #{res.statusCode}"
+				callback new Error("doc updater returned a non-success status code: #{res.statusCode}")
+
 	acceptChanges: (project_id, doc_id, change_ids = [], callback = (error) ->) ->
 		timer = new metrics.Timer("accept-changes")
 		reqSettings =

services/web/app/coffee/Features/Project/ProjectEntityHandler.coffee

@@ -20,19 +20,11 @@ CooldownManager = require '../Cooldown/CooldownManager'
 
 module.exports = ProjectEntityHandler =
 	getAllFolders: (project_id,  callback) ->
-		logger.log project_id:project_id, "getting all folders for project" 
-		folders = {}
-		processFolder = (basePath, folder) ->
-			folders[basePath] = folder
-			for childFolder in (folder.folders or [])
-				if childFolder.name?
-					processFolder path.join(basePath, childFolder.name), childFolder
-
+		logger.log project_id:project_id, "getting all folders for project"
 		ProjectGetter.getProjectWithoutDocLines project_id, (err, project) ->
 			return callback(err) if err?
 			return callback("no project") if !project?
-			processFolder "/", project.rootFolder[0]
-			callback null, folders
+			ProjectEntityHandler.getAllFoldersFromProject project, callback
 
 	getAllDocs: (project_id, callback) ->
 		logger.log project_id:project_id, "getting all docs for project"
@@ -74,6 +66,43 @@ module.exports = ProjectEntityHandler =
 						files[path.join(folderPath, file.name)] = file
 			callback null, files
 
+	getAllFoldersFromProject: (project, callback) ->
+		folders = {}
+		processFolder = (basePath, folder) ->
+			folders[basePath] = folder
+			for childFolder in (folder.folders or [])
+				if childFolder.name?
+					processFolder path.join(basePath, childFolder.name), childFolder
+
+		processFolder "/", project.rootFolder[0]
+		callback null, folders
+
+	getAllEntitiesFromProject: (project, callback) ->
+		logger.log project:project, "getting all files for project"
+		@getAllFoldersFromProject project, (err, folders = {}) ->
+			return callback(err) if err?
+			docs = []
+			files = []
+			for folderPath, folder of folders
+				for doc in (folder.docs or [])
+					if doc?
+						docs.push({path: path.join(folderPath, doc.name), doc:doc})
+				for file in (folder.fileRefs or [])
+					if file?
+						files.push({path: path.join(folderPath, file.name), file:file})
+			callback null, docs, files
+
+	getAllDocPathsFromProject: (project, callback) ->
+		logger.log project:project, "getting all docs for project"
+		@getAllFoldersFromProject project, (err, folders = {}) ->
+			return callback(err) if err?
+			docPath = {}
+			for folderPath, folder of folders
+				for doc in (folder.docs or [])
+					docPath[doc._id] = path.join(folderPath, doc.name)
+			logger.log count:_.keys(docPath).length, project_id:project._id, "returning docPaths for project"
+			callback null, docPath
+
 	flushProjectToThirdPartyDataStore: (project_id, callback) ->
 		self = @
 		logger.log project_id:project_id, "flushing project to tpds"

services/web/app/views/beta_program/opt_in.pug

@@ -18,7 +18,7 @@ block content
 											| #{translate("beta_program_badge_description")}
 											span.beta-feature-badge
 										p.text-centered
-											strong We're not currently testing anything in beta, but keep checking back!
+											strong We're currently testing lower latency compilation features in beta.
 								.row.text-centered
 									.col-md-12
 										if user.betaProgram

services/web/public/coffee/ide/pdf/controllers/PdfController.coffee

@@ -105,6 +105,7 @@ define [
 				rootDoc_id: options.rootDocOverride_id or null
 				draft: $scope.draft
 				check: checkType
+				incrementalCompilesEnabled: window.user?.betaProgram
 				_csrf: window.csrfToken
 			}, {params: params}
 

services/web/test/UnitTests/coffee/Compile/ClsiManagerTests.coffee

@@ -12,6 +12,8 @@ describe "ClsiManager", ->
 			getCookieJar: sinon.stub().callsArgWith(1, null, @jar)
 			setServerId: sinon.stub().callsArgWith(2)
 			_getServerId:sinon.stub()
+		@ClsiStateManager =
+			computeHash: sinon.stub().callsArgWith(1, null, "01234567890abcdef")
 		@ClsiFormatChecker =
 			checkRecoursesForProblems:sinon.stub().callsArgWith(1)
 		@ClsiManager = SandboxedModule.require modulePath, requires:
@@ -26,10 +28,18 @@ describe "ClsiManager", ->
 						url: "https://clsipremium.example.com"
 			"../../models/Project": Project: @Project = {}
 			"../Project/ProjectEntityHandler": @ProjectEntityHandler = {}
+			"../Project/ProjectGetter": @ProjectGetter = {}
+			"../DocumentUpdater/DocumentUpdaterHandler": @DocumentUpdaterHandler =
+				getProjectDocsIfMatch: sinon.stub().callsArgWith(2,null,null)
 			"./ClsiCookieManager": @ClsiCookieManager
+			"./ClsiStateManager": @ClsiStateManager
 			"logger-sharelatex": @logger = { log: sinon.stub(), error: sinon.stub(), warn: sinon.stub() }
 			"request": @request = sinon.stub()
 			"./ClsiFormatChecker": @ClsiFormatChecker
+			"metrics-sharelatex": @Metrics =
+				Timer: class Timer
+					done: sinon.stub()
+				inc: sinon.stub()
 		@project_id = "project-id"
 		@user_id = "user-id"
 		@callback = sinon.stub()
@@ -93,6 +103,25 @@ describe "ClsiManager", ->
 			it "should call the callback with a failure statue", ->
 				@callback.calledWith(null, @status).should.equal true
 
+		describe "with a sync conflict", ->
+			beforeEach ->
+				@ClsiManager.sendRequestOnce = sinon.stub()
+				@ClsiManager.sendRequestOnce.withArgs(@project_id, @user_id, {syncType:"full"}).callsArgWith(3, null,	@status = "success")
+				@ClsiManager.sendRequestOnce.withArgs(@project_id, @user_id, {}).callsArgWith(3, null, "conflict")
+				@ClsiManager.sendRequest @project_id, @user_id, {}, @callback
+
+			it "should call the sendRequestOnce method twice", ->
+				@ClsiManager.sendRequestOnce.calledTwice.should.equal true
+
+			it "should call the sendRequestOnce method with syncType:full", ->
+				@ClsiManager.sendRequestOnce.calledWith(@project_id, @user_id, {syncType:"full"}).should.equal true
+
+			it "should call the sendRequestOnce method without syncType:full", ->
+				@ClsiManager.sendRequestOnce.calledWith(@project_id, @user_id, {}).should.equal true
+
+			it "should call the callback with a success status", ->
+				@callback.calledWith(null, @status, ).should.equal true
+
 	describe "deleteAuxFiles", ->
 		beforeEach ->
 			@ClsiManager._makeRequest = sinon.stub().callsArg(2)
@@ -144,6 +173,8 @@ describe "ClsiManager", ->
 			@Project.findById = sinon.stub().callsArgWith(2, null, @project)
 			@ProjectEntityHandler.getAllDocs = sinon.stub().callsArgWith(1, null, @docs)
 			@ProjectEntityHandler.getAllFiles = sinon.stub().callsArgWith(1, null, @files)
+			@ProjectGetter.getProject = sinon.stub().callsArgWith(2, null, @project)
+			@DocumentUpdaterHandler.flushProjectToMongo = sinon.stub().callsArgWith(1, null)
 
 		describe "with a valid project", ->
 			beforeEach (done) ->
@@ -152,8 +183,13 @@ describe "ClsiManager", ->
 					done()
 
 			it "should get the project with the required fields", ->
-				@Project.findById
-					.calledWith(@project_id, {compiler:1, rootDoc_id: 1, imageName: 1})
+				@ProjectGetter.getProject
+					.calledWith(@project_id, {compiler:1, rootDoc_id: 1, imageName: 1, rootFolder: 1})
+					.should.equal true
+
+			it "should flush the project to the database", ->
+				@DocumentUpdaterHandler.flushProjectToMongo
+					.calledWith(@project_id)
 					.should.equal true
 
 			it "should get all the docs", ->
@@ -175,6 +211,8 @@ describe "ClsiManager", ->
 							imageName: @image
 							draft: false
 							check: undefined
+							syncType: undefined # "full"
+							syncState: undefined # "01234567890abcdef"
 						rootResourcePath: "main.tex"
 						resources: [{
 							path:    "main.tex"
@@ -189,6 +227,51 @@ describe "ClsiManager", ->
 						}]
 				)
 
+		describe "with the incremental compile option", ->
+			beforeEach (done) ->
+				@ClsiStateManager.computeHash = sinon.stub().callsArgWith(1, null, @project_state_hash = "01234567890abcdef")
+				@DocumentUpdaterHandler.getProjectDocsIfMatch = sinon.stub().callsArgWith(2, null, [{_id:@doc_1._id, lines:  @doc_1.lines, v: 123}])
+				@ProjectEntityHandler.getAllDocPathsFromProject = sinon.stub().callsArgWith(1, null, {"mock-doc-id-1":"main.tex"})
+				@ClsiManager._buildRequest @project_id, {timeout:100, incrementalCompilesEnabled:true}, (error, request) =>
+					@request = request
+					done()
+
+			it "should get the project with the required fields", ->
+				@ProjectGetter.getProject
+					.calledWith(@project_id, {compiler:1, rootDoc_id: 1, imageName: 1, rootFolder: 1})
+					.should.equal true
+
+			it "should flush the project to the database", ->
+				@DocumentUpdaterHandler.flushProjectToMongo
+					.calledWith(@project_id)
+					.should.equal true
+
+			it "should get only the live docs from the docupdater", ->
+				@DocumentUpdaterHandler.getProjectDocsIfMatch
+					.calledWith(@project_id)
+					.should.equal true
+
+			it "should not get any of the files", ->
+				@ProjectEntityHandler.getAllFiles
+					.called.should.equal false
+
+			it "should build up the CLSI request", ->
+				expect(@request).to.deep.equal(
+					compile:
+						options:
+							compiler: @compiler
+							timeout : 100
+							imageName: @image
+							draft: false
+							check: undefined
+							syncType: "incremental"
+							syncState: "01234567890abcdef"
+						rootResourcePath: "main.tex"
+						resources: [{
+							path:    "main.tex"
+							content: @doc_1.lines.join("\n")
+						}]
+				)
 
 		describe "when root doc override is valid", ->
 			beforeEach (done) ->

services/web/test/UnitTests/coffee/Compile/ClsiStateManagerTests.coffee

@@ -0,0 +1,148 @@
+sinon = require('sinon')
+chai = require('chai')
+should = chai.should()
+expect = chai.expect
+modulePath = "../../../../app/js/Features/Compile/ClsiStateManager.js"
+SandboxedModule = require('sandboxed-module')
+
+describe "ClsiStateManager", ->
+	beforeEach ->
+		@ClsiStateManager = SandboxedModule.require modulePath, requires:
+			"settings-sharelatex": @settings = {}
+			"../Project/ProjectEntityHandler": @ProjectEntityHandler = {}
+			"logger-sharelatex": @logger = { log: sinon.stub(), error: sinon.stub(), warn: sinon.stub() }
+		@project = "project"
+		@callback = sinon.stub()
+
+	describe "computeHash", ->
+		beforeEach (done) ->
+			@docs = [
+				{path: "/main.tex", doc: {_id: "doc-id-1"}}
+				{path: "/folder/sub.tex", doc: {_id: "doc-id-2"}}
+			]
+			@files = [
+				{path: "/figure.pdf", file: {_id: "file-id-1", rev: 123, created: "aaaaaa"}}
+				{path: "/folder/fig2.pdf", file: {_id: "file-id-2", rev: 456, created: "bbbbbb"}}
+			]
+			@ProjectEntityHandler.getAllEntitiesFromProject = sinon.stub().callsArgWith(1, null, @docs, @files)
+			@ClsiStateManager.computeHash @project, (err, hash) =>
+				@hash0 = hash
+				done()
+
+		describe "with a sample project", ->
+			beforeEach ->
+				@ClsiStateManager.computeHash @project, @callback
+
+			it "should call the callback with a hash value", ->
+				@callback
+					.calledWith(null, "9c2c2428e4147db63cacabf6f357af483af6551d")
+					.should.equal true
+
+		describe "when the files and docs are in a different order", ->
+			beforeEach ->
+				[@docs[0], @docs[1]] = [@docs[1], @docs[0]]
+				[@files[0], @files[1]] = [@files[1], @files[0]]
+				@ClsiStateManager.computeHash @project, @callback
+
+			it "should call the callback with the same hash value", ->
+				@callback
+					.calledWith(null, @hash0)
+					.should.equal true
+
+		describe "when a doc is renamed", ->
+			beforeEach (done) ->
+				@docs[0].path = "/new.tex"
+				@ClsiStateManager.computeHash @project, (err, hash) =>
+					@hash1 = hash
+					done()
+
+			it "should call the callback with a different hash value", ->
+				@callback
+					.neverCalledWith(null, @hash0)
+					.should.equal true
+
+		describe "when a file is renamed", ->
+			beforeEach (done) ->
+				@files[0].path = "/newfigure.pdf"
+				@ClsiStateManager.computeHash @project, (err, hash) =>
+					@hash1 = hash
+					done()
+
+			it "should call the callback with a different hash value", ->
+				@callback
+					.neverCalledWith(null, @hash0)
+					.should.equal true
+
+		describe "when a doc is added", ->
+			beforeEach (done) ->
+				@docs.push { path: "/newdoc.tex", doc: {_id: "newdoc-id"}}
+				@ClsiStateManager.computeHash @project, (err, hash) =>
+					@hash1 = hash
+					done()
+
+			it "should call the callback with a different hash value", ->
+				@callback
+					.neverCalledWith(null, @hash0)
+					.should.equal true
+
+		describe "when a file is added", ->
+			beforeEach (done) ->
+				@files.push { path: "/newfile.tex", file: {_id: "newfile-id", rev: 123}}
+				@ClsiStateManager.computeHash @project, (err, hash) =>
+					@hash1 = hash
+					done()
+
+			it "should call the callback with a different hash value", ->
+				@callback
+					.neverCalledWith(null, @hash0)
+					.should.equal true
+
+		describe "when a doc is removed", ->
+			beforeEach (done) ->
+				@docs.pop()
+				@ClsiStateManager.computeHash @project, (err, hash) =>
+					@hash1 = hash
+					done()
+
+			it "should call the callback with a different hash value", ->
+				@callback
+					.neverCalledWith(null, @hash0)
+					.should.equal true
+
+		describe "when a file is removed", ->
+			beforeEach (done) ->
+				@files.pop()
+				@ClsiStateManager.computeHash @project, (err, hash) =>
+					@hash1 = hash
+					done()
+
+			it "should call the callback with a different hash value", ->
+				@callback
+					.neverCalledWith(null, @hash0)
+					.should.equal true
+
+		describe "when a file's revision is updated", ->
+			beforeEach (done) ->
+				@files[0].file.rev++
+				@ClsiStateManager.computeHash @project, (err, hash) =>
+					@hash1 = hash
+					done()
+
+			it "should call the callback with a different hash value", ->
+				@callback
+					.neverCalledWith(null, @hash0)
+					.should.equal true
+
+
+		describe "when a file's date is updated", ->
+			beforeEach (done) ->
+				@files[0].file.created = "zzzzzz"
+				@ClsiStateManager.computeHash @project, (err, hash) =>
+					@hash1 = hash
+					done()
+
+			it "should call the callback with a different hash value", ->
+				@callback
+					.neverCalledWith(null, @hash0)
+					.should.equal true
+

services/web/test/UnitTests/coffee/Compile/CompileManagerTests.coffee

@@ -17,7 +17,6 @@ describe "CompileManager", ->
 				redis: web: {host: "localhost", port: 42}
 			"../../infrastructure/RedisWrapper":
 				client: () => @rclient = { auth: () -> }
-			"../DocumentUpdater/DocumentUpdaterHandler": @DocumentUpdaterHandler = {}
 			"../Project/ProjectRootDocManager": @ProjectRootDocManager = {}
 			"../../models/Project": Project: @Project = {}
 			"../User/UserGetter": @UserGetter = {}
@@ -40,7 +39,6 @@ describe "CompileManager", ->
 		beforeEach ->
 			@CompileManager._checkIfRecentlyCompiled = sinon.stub().callsArgWith(2, null, false)
 			@CompileManager._ensureRootDocumentIsSet = sinon.stub().callsArgWith(1, null)
-			@DocumentUpdaterHandler.flushProjectToMongo = sinon.stub().callsArgWith(1, null)
 			@CompileManager.getProjectCompileLimits = sinon.stub().callsArgWith(1, null, @limits)
 			@ClsiManager.sendRequest = sinon.stub().callsArgWith(3, null, @status = "mock-status", @outputFiles = "mock output files", @output = "mock output")
 
@@ -54,11 +52,6 @@ describe "CompileManager", ->
 					.calledWith(@project_id, @user_id)
 					.should.equal true
 
-			it "should flush the project to the database", ->
-				@DocumentUpdaterHandler.flushProjectToMongo
-					.calledWith(@project_id)
-					.should.equal true
-
 			it "should ensure that the root document is set", ->
 				@CompileManager._ensureRootDocumentIsSet
 					.calledWith(@project_id)

services/web/test/UnitTests/coffee/DocumentUpdater/DocumentUpdaterHandlerTests.coffee

@@ -252,6 +252,47 @@ describe 'DocumentUpdaterHandler', ->
 					.calledWith(new Error("doc updater returned failure status code: 500"))
 					.should.equal true
 
+	describe "getProjectDocsIfMatch", ->
+		beforeEach ->
+			@callback = sinon.stub()
+			@project_state_hash = "1234567890abcdef"
+
+		describe "successfully", ->
+			beforeEach ->
+				@doc0 =
+					_id: @doc_id
+					lines: @lines
+					v: @version
+				@docs = [ @doc0, @doc0, @doc0 ]
+				@body = JSON.stringify @docs
+				@request.get = sinon.stub().callsArgWith(1, null, {statusCode: 200}, @body)
+				@handler.getProjectDocsIfMatch @project_id, @project_state_hash, @callback
+
+			it 'should get the documenst from the document updater', ->
+				url = "#{@settings.apis.documentupdater.url}/project/#{@project_id}/doc?state=#{@project_state_hash}"
+				@request.get.calledWith(url).should.equal true
+
+			it "should call the callback with the documents", ->
+				@callback.calledWithExactly(null, @docs).should.equal true
+
+		describe "when the document updater API returns an error", ->
+			beforeEach ->
+				@request.get = sinon.stub().callsArgWith(1, @error = new Error("something went wrong"), null, null)
+				@handler.getProjectDocsIfMatch @project_id, @project_state_hash, @callback
+
+			it "should return an error to the callback", ->
+				@callback.calledWith(@error).should.equal true
+
+		describe "when the document updater returns a conflict error code", ->
+			beforeEach ->
+				@request.get = sinon.stub().callsArgWith(1, null, { statusCode: 409 }, "Conflict")
+				@handler.getProjectDocsIfMatch @project_id, @project_state_hash, @callback
+
+			it "should return the callback with no documents", ->
+				@callback
+					.alwaysCalledWithExactly()
+					.should.equal true
+
 	describe "acceptChanges", ->
 		beforeEach ->
 			@change_id = "mock-change-id-1"

services/web/test/UnitTests/coffee/Project/ProjectEntityHandlerTests.coffee

@@ -781,6 +781,41 @@ describe 'ProjectEntityHandler', ->
 					})
 					.should.equal true
 
+		describe "getAllFoldersFromProject", ->
+			beforeEach ->
+				@callback = sinon.stub()
+				@ProjectEntityHandler.getAllFoldersFromProject @project, @callback
+
+			it "should call the callback with the folders", ->
+				@callback
+					.calledWith(null, {
+						"/": @project.rootFolder[0]
+						"/folder1": @folder1
+					})
+					.should.equal true
+
+		describe "getAllDocPathsFromProject", ->
+			beforeEach ->
+				@docs = [{
+					_id:   @doc1._id
+					lines: @lines1 = ["one"]
+					rev:   @rev1 = 1
+				}, {
+					_id:   @doc2._id
+					lines: @lines2 = ["two"]
+					rev:   @rev2 = 2
+				}]
+				@callback = sinon.stub()
+				@ProjectEntityHandler.getAllDocPathsFromProject @project, @callback
+
+			it "should call the callback with the path for each doc_id", ->
+				@expected = {}
+				@expected[@doc1._id] = "/#{@doc1.name}"
+				@expected[@doc2._id] = "/folder1/#{@doc2.name}"
+				@callback
+					.calledWith(null, @expected)
+					.should.equal true
+
 	describe "flushProjectToThirdPartyDataStore", ->
 		beforeEach (done) ->
 			@project = {