Merge pull request #1476 from sharelatex/spd-character-sets

Detect encoding of non-utf8 files and convert during load

GitOrigin-RevId: 1856a42457c52b125793dd001b92a014f358da69

services/web/app/coffee/Features/ThirdPartyDataStore/UpdateMerger.coffee

@@ -24,10 +24,10 @@ module.exports = UpdateMerger =
 				return callback(null, "existing-file")
 			if _.some(docs, (d) -> d.path is path)
 				return callback(null, "existing-doc") 
-			# existing file not found in project, fall back to extension check
-			FileTypeManager.isBinary path, fsPath, (err, isFile)->
+			# existing file not found in project, so check the file type to determine if doc
+			FileTypeManager.getType path, fsPath, (err, isBinary)->
 				return callback(err) if err?
-				if isFile
+				if isBinary
 					callback(null, "new-file") # extension was not text
 				else
 					callback(null, "new-doc")

services/web/app/coffee/Features/Uploads/FileSystemImportManager.coffee

@@ -6,14 +6,14 @@ EditorController = require "../Editor/EditorController"
 logger = require("logger-sharelatex")
 
 module.exports = FileSystemImportManager =
-	addDoc: (user_id, project_id, folder_id, name, path, replace, callback = (error, doc)-> )->
+	addDoc: (user_id, project_id, folder_id, name, path, charset, replace, callback = (error, doc)-> )->
 		FileSystemImportManager._isSafeOnFileSystem path, (err, isSafe)->
 			if !isSafe
 				logger.log user_id:user_id, project_id:project_id, folder_id:folder_id, name:name, path:path, "add doc is from symlink, stopping process"
 				return callback("path is symlink")
-			fs.readFile path, "utf8", (error, content = "") ->
+			fs.readFile path, charset, (error, content) ->
 				return callback(error) if error?
-				content = content.replace(/\r/g, "")
+				content = content.replace(/\r\n?/g, "\n") # convert Windows line endings to unix. very old macs also created \r-separated lines
 				lines = content.split("\n")
 				if replace
 					EditorController.upsertDoc project_id, folder_id, name, lines, "upload", user_id, callback
@@ -72,14 +72,14 @@ module.exports = FileSystemImportManager =
 				if isDirectory
 					FileSystemImportManager.addFolder user_id, project_id, folder_id, name, path, replace, callback
 				else
-					FileTypeManager.isBinary name, path, (error, isBinary) =>
+					FileTypeManager.getType name, path, (error, isBinary, charset) =>
 						return callback(error) if error?
 						if isBinary
 							FileSystemImportManager.addFile user_id, project_id, folder_id, name, path, replace, (err, entity) ->
 								entity?.type = 'file'
 								callback(err, entity)
 						else
-							FileSystemImportManager.addDoc user_id, project_id, folder_id, name, path, replace, (err, entity) ->
+							FileSystemImportManager.addDoc user_id, project_id, folder_id, name, path, charset, replace, (err, entity) ->
 								entity?.type = 'doc'
 								callback(err, entity)
 

services/web/app/coffee/Features/Uploads/FileTypeManager.coffee

@@ -1,9 +1,11 @@
 fs = require "fs"
 Path = require("path")
+isUtf8 = require('is-utf8');
+
 
 module.exports = FileTypeManager =
 	TEXT_EXTENSIONS : [
-		"tex", "latex", "sty", "cls", "bst", "bib", "bibtex", "txt", "tikz", "rtex", "md", "asy", "latexmkrc"
+		"tex", "latex", "sty", "cls", "bst", "bib", "bibtex", "txt", "tikz", "rtex", "md", "asy", "latexmkrc", "lbx", "bbx", "cbx", "m"
 	]
 
 	IGNORE_EXTENSIONS : [
@@ -29,22 +31,29 @@ module.exports = FileTypeManager =
 			return callback(error) if error?
 			callback(null, stats?.isDirectory())
 
-	isBinary: (name, fsPath, callback = (error, result) ->) ->
+	# returns charset as understood by fs.readFile,
+	getType: (name, fsPath, callback = (error, isBinary, charset) ->) ->
 		parts = name.split(".")
-		extension = parts.slice(-1)[0]
-		if extension?
-			extension = extension.toLowerCase()
-		binaryFile = (@TEXT_EXTENSIONS.indexOf(extension) == -1 or parts.length <= 1) and parts[0] != 'latexmkrc'
-		
-		if binaryFile
-			return callback null, true
+		extension = parts.slice(-1)[0].toLowerCase()
+		isText = (FileTypeManager.TEXT_EXTENSIONS.indexOf(extension) > -1 and parts.length > 1) or parts[0] == 'latexmkrc'
+
+		return callback null, true unless isText
 
 		fs.stat fsPath, (error, stat) ->
 			return callback(error) if error?
 			if stat.size > FileTypeManager.MAX_TEXT_FILE_SIZE
 				return callback null, true # Treat large text file as binary
-			else
-				return callback null, false
+
+			fs.readFile fsPath, (err, bytes) ->
+				return callback(err) if err?
+
+				if isUtf8(bytes)
+					return callback null, false, "utf-8"
+				# check for little-endian unicode bom (nodejs does not support big-endian)
+				if bytes[0] == 0xFF and bytes[1] == 0xFE
+					return callback null, false, "utf-16le"
+
+				callback null, false, "latin1"
 
 	shouldIgnore: (path, callback = (error, result) ->) ->
 		name = Path.basename(path)

services/web/npm-shrinkwrap.json

@@ -5622,9 +5622,8 @@
     },
     "is-utf8": {
       "version": "0.2.1",
-      "from": "is-utf8@>=0.2.0 <0.3.0",
-      "resolved": "https://registry.npmjs.org/is-utf8/-/is-utf8-0.2.1.tgz",
-      "dev": true
+      "from": "is-utf8@>=0.2.1 <0.3.0",
+      "resolved": "https://registry.npmjs.org/is-utf8/-/is-utf8-0.2.1.tgz"
     },
     "is-windows": {
       "version": "1.0.2",

services/web/package.json

@@ -54,6 +54,7 @@
     "heapdump": "^0.3.7",
     "helmet": "^3.8.1",
     "http-proxy": "^1.8.1",
+    "is-utf8": "^0.2.1",
     "jade": "~1.3.1",
     "jquery": "^1.11.1",
     "json2csv": "^4.3.3",

services/web/test/acceptance/coffee/ProjectStructureTests.coffee

@@ -252,6 +252,47 @@ describe "ProjectStructureChanges", ->
 				expect(project.rootFolder[0].folders[0].docs[0].name).to.equal('ao.sty')
 				done()
 
+	describe "uploading a project with files in different encodings", ->
+		before (done) ->
+			MockDocUpdaterApi.clearProjectStructureUpdates()
+
+			zip_file = fs.createReadStream(Path.resolve(__dirname + '/../files/charsets/charsets.zip'))
+
+			@owner.request.post {
+				uri: "project/new/upload",
+				formData:
+					qqfile: zip_file
+			}, (error, res, body) =>
+				throw error if error?
+				if res.statusCode < 200 || res.statusCode >= 300
+					throw new Error("failed to upload project #{res.statusCode}")
+				@uploaded_project_id = JSON.parse(body).project_id
+				done()
+
+		it "should correctly parse windows-1252", ->
+			{docUpdates: updates} = MockDocUpdaterApi.getProjectStructureUpdates(@uploaded_project_id)
+			update = _.find updates, (update) ->
+				update.pathname == '/test-german-windows-1252.tex'
+			expect(update.docLines).to.contain("Der schnelle braune Fuchs sprang träge über den Hund.")
+
+		it "should correctly parse German utf8", ->
+			{docUpdates: updates} = MockDocUpdaterApi.getProjectStructureUpdates(@uploaded_project_id)
+			update = _.find updates, (update) ->
+				update.pathname == '/test-german-utf8x.tex'
+			expect(update.docLines).to.contain("Der schnelle braune Fuchs sprang träge über den Hund.")
+
+		it "should correctly parse little-endian utf16", ->
+			{docUpdates: updates} = MockDocUpdaterApi.getProjectStructureUpdates(@uploaded_project_id)
+			update = _.find updates, (update) ->
+				update.pathname == '/test-greek-utf16-le-bom.tex'
+			expect(update.docLines).to.contain("Η γρήγορη καστανή αλεπού πήδηξε χαλαρά πάνω από το σκυλί.")
+
+		it "should correctly parse Greek utf8", ->
+			{docUpdates: updates} = MockDocUpdaterApi.getProjectStructureUpdates(@uploaded_project_id)
+			update = _.find updates, (update) ->
+				update.pathname == '/test-greek-utf8x.tex'
+			expect(update.docLines).to.contain("Η γρήγορη καστανή αλεπού πήδηξε χαλαρά πάνω από το σκυλί.")
+
 	describe "uploading a file", ->
 		beforeEach (done) ->
 			MockDocUpdaterApi.clearProjectStructureUpdates()
@@ -260,7 +301,7 @@ describe "ProjectStructureChanges", ->
 				@root_folder_id = project.rootFolder[0]._id.toString()
 				@project_0 = project
 				done()
-				
+
 		it "should version a newly uploaded file", (done) ->
 			image_file = fs.createReadStream(Path.resolve(__dirname + '/../files/1pixel.png'))
 
@@ -740,3 +781,65 @@ describe "ProjectStructureChanges", ->
 					expect(@project_1.version).to.equal(@project_0.version + 1)
 					done()
 
+
+	describe "uploading a document", ->
+		beforeEach (done) ->
+			MockDocUpdaterApi.clearProjectStructureUpdates()
+			ProjectGetter.getProject example_project_id, (error, project) =>
+				throw error if error?
+				@root_folder_id = project.rootFolder[0]._id.toString()
+				@project_0 = project
+				done()
+
+		describe "with an unusual character set", ->
+			it "should correctly handle utf16-le data", (done) ->
+				document_file = fs.createReadStream(Path.resolve(__dirname + '/../files/charsets/test-greek-utf16-le-bom.tex'))
+
+				req = @owner.request.post {
+					uri: "project/#{example_project_id}/upload",
+					qs:
+						folder_id: @root_folder_id
+					formData:
+						qqfile:
+							value: document_file
+							options:
+								filename: 'test-greek-utf16-le-bom.tex',
+								contentType: 'text/x-tex'
+				}, (error, res, body) =>
+					throw error if error?
+					if res.statusCode < 200 || res.statusCode >= 300
+						throw new Error("failed to upload file #{res.statusCode}")
+
+					example_file_id = JSON.parse(body).entity_id
+
+					{docUpdates:updates} = MockDocUpdaterApi.getProjectStructureUpdates(example_project_id)
+					update = updates[0]
+					expect(update.pathname).to.equal('/test-greek-utf16-le-bom.tex')
+					expect(update.docLines).to.contain("Η γρήγορη καστανή αλεπού πήδηξε χαλαρά πάνω από το σκυλί.")
+					done()
+
+			it "should correctly handle windows1252/iso-8859-1/latin1 data", (done) ->
+				document_file = fs.createReadStream(Path.resolve(__dirname + '/../files/charsets/test-german-windows-1252.tex'))
+
+				req = @owner.request.post {
+					uri: "project/#{example_project_id}/upload",
+					qs:
+						folder_id: @root_folder_id
+					formData:
+						qqfile:
+							value: document_file
+							options:
+								filename: 'test-german-windows-1252.tex',
+								contentType: 'text/x-tex'
+				}, (error, res, body) =>
+					throw error if error?
+					if res.statusCode < 200 || res.statusCode >= 300
+						throw new Error("failed to upload file #{res.statusCode}")
+
+					example_file_id = JSON.parse(body).entity_id
+
+					{docUpdates:updates} = MockDocUpdaterApi.getProjectStructureUpdates(example_project_id)
+					update = updates[0]
+					expect(update.pathname).to.equal('/test-german-windows-1252.tex')
+					expect(update.docLines).to.contain("Der schnelle braune Fuchs sprang träge über den Hund.")
+					done()

services/web/test/acceptance/coffee/RestoringFilesTest.coffee

@@ -1,6 +1,8 @@
 async = require "async"
 expect = require("chai").expect
 _ = require 'underscore'
+fs = require 'fs'
+Path = require 'path'
 
 ProjectGetter = require "../../../app/js/Features/Project/ProjectGetter.js"
 
@@ -83,7 +85,8 @@ describe "RestoringFiles", ->
 
 		describe "restoring a binary file", ->
 			beforeEach (done) ->
-				MockProjectHistoryApi.addOldFile(@project_id, 42, "image.png", "Mock image.png content")
+				@pngData = fs.readFileSync(Path.resolve(__dirname, '../files/1pixel.png'), 'binary')
+				MockProjectHistoryApi.addOldFile(@project_id, 42, "image.png", @pngData)
 				@owner.request {
 					method: "POST",
 					url: "/project/#{@project_id}/restore_file",
@@ -101,7 +104,7 @@ describe "RestoringFiles", ->
 					file = _.find project.rootFolder[0].fileRefs, (file) ->
 						file.name == 'image.png'
 					file = MockFileStoreApi.files[@project_id][file._id]
-					expect(file.content).to.equal "Mock image.png content"
+					expect(file.content).to.equal @pngData
 					done()
 
 		describe "restoring to a directory that exists", ->

services/web/test/acceptance/files/charsets/test-german-utf8x.tex

@@ -0,0 +1,8 @@
+\documentclass[12pt]{article}
+\usepackage[german,english]{babel}
+\usepackage[utf8x]{inputenc}
+\begin{document}
+    \title{Untitled}
+    \selectlanguage{german}
+    Der schnelle braune Fuchs sprang träge über den Hund.
+\end{document}

services/web/test/acceptance/files/charsets/test-german-windows-1252.tex

@@ -0,0 +1,8 @@
+\documentclass[12pt]{article}
+\usepackage[german,english]{babel}
+\usepackage[cp1252]{inputenc}
+\begin{document}
+    \title{Untitled}
+    \selectlanguage{german}
+    Der schnelle braune Fuchs sprang träge über den Hund.
+\end{document}

services/web/test/acceptance/files/charsets/test-greek-utf8x.tex

@@ -0,0 +1,8 @@
+\documentclass[12pt]{article}
+\usepackage[greek,english]{babel}
+\usepackage[utf8x]{inputenc}
+\begin{document}
+    \title{Untitled}
+    \selectlanguage{greek}
+    Η γρήγορη καστανή αλεπού πήδηξε χαλαρά πάνω από το σκυλί.
+\end{document}

services/web/test/unit/coffee/ThirdPartyDataStore/UpdateMergerTests.coffee

@@ -50,12 +50,12 @@ describe 'UpdateMerger :', ->
 		describe "doc updates for a new doc", ->
 
 			beforeEach ->
-				@FileTypeManager.isBinary = sinon.stub().yields(null, false)
+				@FileTypeManager.getType = sinon.stub().yields(null, false)
 				@updateMerger.p.processDoc = sinon.stub().yields()
 				@updateMerger.mergeUpdate @user_id, @project_id, @docPath, @updateRequest, @source, @callback
 
 			it 'should look at the file contents', ->
-				@FileTypeManager.isBinary.called.should.equal true
+				@FileTypeManager.getType.called.should.equal true
 
 			it 'should process update as doc', ->
 				@updateMerger.p.processDoc
@@ -67,12 +67,12 @@ describe 'UpdateMerger :', ->
 
 		describe "file updates for a new file ", ->
 			beforeEach ->
-				@FileTypeManager.isBinary = sinon.stub().yields(null, true)
+				@FileTypeManager.getType = sinon.stub().yields(null, true)
 				@updateMerger.p.processFile = sinon.stub().yields()
 				@updateMerger.mergeUpdate @user_id, @project_id, @filePath, @updateRequest, @source, @callback
 
 			it 'should look at the file contents', ->
-				@FileTypeManager.isBinary.called.should.equal true
+				@FileTypeManager.getType.called.should.equal true
 
 			it 'should process update as file', ->
 				@updateMerger.p.processFile
@@ -84,12 +84,12 @@ describe 'UpdateMerger :', ->
 
 		describe "doc updates for an existing doc", ->
 			beforeEach ->
-				@FileTypeManager.isBinary = sinon.stub()
+				@FileTypeManager.getType = sinon.stub()
 				@updateMerger.p.processDoc = sinon.stub().yields()
 				@updateMerger.mergeUpdate @user_id, @project_id, @existingDocPath, @updateRequest, @source, @callback
 
 			it 'should not look at the file contents', ->
-				@FileTypeManager.isBinary.called.should.equal false
+				@FileTypeManager.getType.called.should.equal false
 
 			it 'should process update as doc', ->
 				@updateMerger.p.processDoc
@@ -101,12 +101,12 @@ describe 'UpdateMerger :', ->
 
 		describe "file updates for an existing file", ->
 			beforeEach ->
-				@FileTypeManager.isBinary = sinon.stub()
+				@FileTypeManager.getType = sinon.stub()
 				@updateMerger.p.processFile = sinon.stub().yields()
 				@updateMerger.mergeUpdate @user_id, @project_id, @existingFilePath, @updateRequest, @source, @callback
 
 			it 'should not look at the file contents', ->
-				@FileTypeManager.isBinary.called.should.equal false
+				@FileTypeManager.getType.called.should.equal false
 
 			it 'should process update as file', ->
 				@updateMerger.p.processFile

services/web/test/unit/coffee/Uploads/FileSystemImportManagerTests.coffee

@@ -13,11 +13,15 @@ describe "FileSystemImportManager", ->
 		@replace = "replace-boolean-flag-mock"
 		@user_id = "mock-user-123"
 		@callback = sinon.stub()
+		@encoding = "latin1"
+		@DocumentHelper =
+			convertTexEncodingsToUtf8: sinon.stub().returnsArg(0)
 		@FileSystemImportManager = SandboxedModule.require modulePath, requires:
 			"fs" : @fs = {}
 			"../Editor/EditorController": @EditorController = {}
 			"./FileTypeManager": @FileTypeManager = {}
 			"../Project/ProjectLocator": @ProjectLocator = {}
+			"../Documents/DocumentHelper": @DocumentHelper
 			"logger-sharelatex":
 				log:->
 				err:->
@@ -34,7 +38,7 @@ describe "FileSystemImportManager", ->
 			beforeEach ->
 				@FileSystemImportManager._isSafeOnFileSystem = sinon.stub().callsArgWith(1, null, false)
 				@EditorController.addDoc = sinon.stub()
-				@FileSystemImportManager.addDoc @user_id, @project_id, @folder_id, @name, @path_on_disk, false, @callback
+				@FileSystemImportManager.addDoc @user_id, @project_id, @folder_id, @name, @path_on_disk, @encoding, false, @callback
 
 			it "should not read the file from disk", ->
 				@fs.readFile.called.should.equal false
@@ -45,10 +49,10 @@ describe "FileSystemImportManager", ->
 		describe "with replace set to false", ->
 			beforeEach ->
 				@EditorController.addDoc = sinon.stub().callsArg(6)
-				@FileSystemImportManager.addDoc @user_id, @project_id, @folder_id, @name, @path_on_disk, false, @callback
+				@FileSystemImportManager.addDoc @user_id, @project_id, @folder_id, @name, @path_on_disk, @encoding, false, @callback
 
 			it "should read the file from disk", ->
-				@fs.readFile.calledWith(@path_on_disk, "utf8").should.equal true
+				@fs.readFile.calledWith(@path_on_disk).should.equal true
 
 			it "should insert the doc", ->
 				@EditorController.addDoc.calledWith(@project_id, @folder_id, @name, @docLines, "upload", @user_id)
@@ -60,22 +64,37 @@ describe "FileSystemImportManager", ->
 				@docLines = ["one", "two", "three"]
 				@fs.readFile = sinon.stub().callsArgWith(2, null, @docContent)
 				@EditorController.addDoc = sinon.stub().callsArg(6)
-				@FileSystemImportManager.addDoc @user_id, @project_id, @folder_id, @name, @path_on_disk, false, @callback
+				@FileSystemImportManager.addDoc @user_id, @project_id, @folder_id, @name, @path_on_disk, @encoding, false, @callback
 
 			it "should strip the \\r characters before adding", ->
 				@EditorController.addDoc.calledWith(@project_id, @folder_id, @name, @docLines, "upload", @user_id)
 					.should.equal true
+
+		describe "with \r line endings", ->
+			beforeEach ->
+				@docContent = "one\rtwo\rthree"
+				@docLines = ["one", "two", "three"]
+				@fs.readFile = sinon.stub().callsArgWith(2, null, @docContent)
+				@EditorController.addDoc = sinon.stub().callsArg(6)
+				@FileSystemImportManager.addDoc @user_id, @project_id, @folder_id, @name, @path_on_disk, @encoding, false, @callback
+
+			it "should treat the \\r characters as newlines", ->
+				@EditorController.addDoc.calledWith(@project_id, @folder_id, @name, @docLines, "upload", @user_id)
+					.should.equal true
 		
 		describe "with replace set to true", ->
 			beforeEach ->
 				@EditorController.upsertDoc = sinon.stub().yields()
-				@FileSystemImportManager.addDoc @user_id, @project_id, @folder_id, @name, @path_on_disk, true, @callback
+				@FileSystemImportManager.addDoc @user_id, @project_id, @folder_id, @name, @path_on_disk, @encoding, true, @callback
 
 			it "should upsert the doc", ->
 				@EditorController.upsertDoc
 					.calledWith(@project_id, @folder_id, @name, @docLines, "upload", @user_id)
 					.should.equal true
 
+			it "should read the file with the correct encoding", ->
+				sinon.assert.calledWith(@fs.readFile, @path_on_disk, @encoding)
+
 	describe "addFile with replace set to false", ->
 		beforeEach ->
 			@EditorController.addFile = sinon.stub().yields()
@@ -176,7 +195,7 @@ describe "FileSystemImportManager", ->
 		describe "with binary file", ->
 			beforeEach ->
 				@FileTypeManager.isDirectory = sinon.stub().callsArgWith(1, null, false)
-				@FileTypeManager.isBinary = sinon.stub().callsArgWith(2, null, true)
+				@FileTypeManager.getType = sinon.stub().callsArgWith(2, null, true)
 				@FileSystemImportManager._isSafeOnFileSystem = sinon.stub().callsArgWith(1, null, true)
 				@FileSystemImportManager.addFile = sinon.stub().callsArg(6)
 				@FileSystemImportManager.addEntity @user_id, @project_id, @folder_id, @name, @path_on_disk, @replace, @callback
@@ -188,13 +207,10 @@ describe "FileSystemImportManager", ->
 		describe "with text file", ->
 			beforeEach ->
 				@FileTypeManager.isDirectory = sinon.stub().callsArgWith(1, null, false)
-				@FileTypeManager.isBinary = sinon.stub().callsArgWith(2, null, false)
-				@FileSystemImportManager.addDoc = sinon.stub().callsArg(6)
+				@FileTypeManager.getType = sinon.stub().callsArgWith(2, null, false, 'latin1')
+				@FileSystemImportManager.addDoc = sinon.stub().callsArg(7)
 				@FileSystemImportManager._isSafeOnFileSystem = sinon.stub().callsArgWith(1, null, true)
 				@FileSystemImportManager.addEntity @user_id, @project_id, @folder_id, @name, @path_on_disk, @replace, @callback
 
 			it "should call addFile", ->
-				@FileSystemImportManager.addDoc.calledWith(@user_id, @project_id, @folder_id, @name, @path_on_disk, @replace)
-					.should.equal true
-
-
+				sinon.assert.calledWith(@FileSystemImportManager.addDoc, @user_id, @project_id, @folder_id, @name, @path_on_disk, "latin1", @replace)

services/web/test/unit/coffee/Uploads/FileTypeManagerTests.coffee

@@ -3,16 +3,20 @@ chai = require('chai')
 should = chai.should()
 modulePath = "../../../../app/js/Features/Uploads/FileTypeManager.js"
 SandboxedModule = require('sandboxed-module')
+isUtf8 = require('is-utf8')
 
 describe "FileTypeManager", ->
 	beforeEach ->
+		@isUtf8 = sinon.spy(isUtf8)
 		@fs = {}
-		@Magic = {}
 		@path = "/path/to/test"
 		@callback = sinon.stub()
+		@ced = sinon.stub()
+		@DocumentHelper =
+			getEncodingFromTexContent: sinon.stub()
 		@FileTypeManager = SandboxedModule.require modulePath, requires:
 			"fs": @fs
-			"mmmagic" : Magic: (options) => @Magic
+			"is-utf8": @isUtf8
 
 	describe "isDirectory", ->
 		beforeEach ->
@@ -35,67 +39,117 @@ describe "FileTypeManager", ->
 			it "should return false", ->
 				@callback.calledWith(null, false).should.equal true
 
-	describe "isBinary", ->
+	describe "getType", ->
 		beforeEach ->
 			@stat = { size: 100 }
+			@contents = "Ich bin eine kleine Teekanne, kurz und kräftig."
 			@fs.stat = sinon.stub().callsArgWith(1, null, @stat)
+			@fs.readFile = sinon.stub().callsArgWith(1, null, Buffer.from(@contents, "utf-8"))
+			@fs.readFile.withArgs("/path/on/disk/utf16.tex").callsArgWith(1, null, Buffer.from("\uFEFF" + @contents, "utf-16le"))
+			@fs.readFile.withArgs("/path/on/disk/latin1.tex").callsArgWith(1, null, Buffer.from(@contents, "latin1"))
+			@encoding = "ASCII"
 
-		it "should return .tex files as not binary", ->
-			@FileTypeManager.isBinary "file.tex", "/path/on/disk", (error, binary) ->
-				binary.should.equal false
+		describe "when the file extension is text", ->
+			it "should return .tex files as not binary", ->
+				@FileTypeManager.getType "file.tex", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return .bib files as not binary", ->
-			@FileTypeManager.isBinary "file.bib", "/path/on/disk", (error, binary) ->
-				binary.should.equal false
+			it "should return .bib files as not binary", ->
+				@FileTypeManager.getType "file.bib", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return .bibtex files as not binary", ->
-			@FileTypeManager.isBinary "file.bibtex", "/path/on/disk", (error, binary) ->
-				binary.should.equal false
+			it "should return .bibtex files as not binary", ->
+				@FileTypeManager.getType "file.bibtex", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return .cls files as not binary", ->
-			@FileTypeManager.isBinary "file.cls", "/path/on/disk", (error, binary) ->
-				binary.should.equal false
+			it "should return .cls files as not binary", ->
+				@FileTypeManager.getType "file.cls", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return .sty files as not binary", ->
-			@FileTypeManager.isBinary "file.sty", "/path/on/disk", (error, binary) ->
-				binary.should.equal false
+			it "should return .sty files as not binary", ->
+				@FileTypeManager.getType "file.sty", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return .bst files as not binary", ->
-			@FileTypeManager.isBinary "file.bst", "/path/on/disk", (error, binary) ->
-				binary.should.equal false
+			it "should return .bst files as not binary", ->
+				@FileTypeManager.getType "file.bst", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return .eps files as binary", ->
-			@FileTypeManager.isBinary "file.eps", "/path/on/disk", (error, binary) ->
-				binary.should.equal true
+			it "should return .latexmkrc file as not binary", ->
+				@FileTypeManager.getType ".latexmkrc", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return .dvi files as binary", ->
-			@FileTypeManager.isBinary "file.dvi", "/path/on/disk", (error, binary) ->
-				binary.should.equal true
+			it "should return latexmkrc file as not binary", ->
+				@FileTypeManager.getType "latexmkrc", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return .png files as binary", ->
-			@FileTypeManager.isBinary "file.png", "/path/on/disk", (error, binary) ->
-				binary.should.equal true
+			it "should return lbx file as not binary", ->
+				@FileTypeManager.getType "file.lbx", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return files without extensions as binary", ->
-			@FileTypeManager.isBinary "tex", "/path/on/disk", (error, binary) ->
-				binary.should.equal true
+			it "should return bbx file as not binary", ->
+				@FileTypeManager.getType "file.bbx", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return .latexmkrc file as not binary", ->
-			@FileTypeManager.isBinary ".latexmkrc", "/path/on/disk", (error, binary) ->
-				binary.should.equal false
+			it "should return cbx file as not binary", ->
+				@FileTypeManager.getType "file.cbx", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return latexmkrc file as not binary", ->
-			@FileTypeManager.isBinary "latexmkrc", "/path/on/disk", (error, binary) ->
-				binary.should.equal false
+			it "should return m file as not binary", ->
+				@FileTypeManager.getType "file.m", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should ignore the case of an extension", ->
-			@FileTypeManager.isBinary "file.TEX", "/path/on/disk", (error, binary) ->
-				binary.should.equal false
+			it "should ignore the case of an extension", ->
+				@FileTypeManager.getType "file.TEX", "/path/on/disk", (error, binary) ->
+					binary.should.equal false
 
-		it "should return large text files as binary", ->
-			@stat.size = 2 * 1024 * 1024 # 2Mb
-			@FileTypeManager.isBinary "file.tex", "/path/on/disk", (error, binary) ->
-				binary.should.equal true
+			it "should return large text files as binary", ->
+				@stat.size = 2 * 1024 * 1024 # 2Mb
+				@FileTypeManager.getType "file.tex", "/path/on/disk", (error, binary) ->
+					binary.should.equal true
+
+			it "should return try to determine the encoding of large files", ->
+				@stat.size = 2 * 1024 * 1024 # 2Mb
+				@FileTypeManager.getType "file.tex", "/path/on/disk", =>
+					sinon.assert.notCalled(@isUtf8)
+
+			it "should detect the file as utf8", ->
+				@FileTypeManager.getType "file.tex", "/path/on/disk", (error, binary, encoding) =>
+					sinon.assert.calledOnce(@isUtf8)
+					@isUtf8.returned(true).should.equal true
+					encoding.should.equal "utf-8"
+
+			it "should return 'latin1' for non-unicode encodings", ->
+				@FileTypeManager.getType "file.tex", "/path/on/disk/latin1.tex", (error, binary, encoding) =>
+					sinon.assert.calledOnce(@isUtf8)
+					@isUtf8.returned(false).should.equal true
+					encoding.should.equal "latin1"
+
+			it "should detect utf16 with BOM as utf-16", ->
+				@FileTypeManager.getType "file.tex", "/path/on/disk/utf16.tex", (error, binary, encoding) =>
+					sinon.assert.calledOnce(@isUtf8)
+					@isUtf8.returned(false).should.equal true
+					encoding.should.equal "utf-16le"
+
+		describe "when the file extension is non-text", ->
+			it "should return .eps files as binary", ->
+				@FileTypeManager.getType "file.eps", "/path/on/disk", (error, binary) ->
+					binary.should.equal true
+
+			it "should return .dvi files as binary", ->
+				@FileTypeManager.getType "file.dvi", "/path/on/disk", (error, binary) ->
+					binary.should.equal true
+
+			it "should return .png files as binary", ->
+				@FileTypeManager.getType "file.png", "/path/on/disk", (error, binary) ->
+					binary.should.equal true
+
+			it "should return files without extensions as binary", ->
+				@FileTypeManager.getType "tex", "/path/on/disk", (error, binary) ->
+					binary.should.equal true
+
+			it "should not try to get the character encoding", ->
+				@FileTypeManager.getType "file.png", "/path/on/disk", =>
+					sinon.assert.notCalled(@isUtf8)
 
 	describe "shouldIgnore", ->
 		it "should ignore tex auxiliary files", ->