commit 7e96933cf2269f0792824d3a09d41a4c88bd0095 Author: James Allen Date: Thu Aug 8 16:42:14 2013 +0100 Initial commit of script for compressing one docs history diff --git a/services/track-changes/.gitignore b/services/track-changes/.gitignore new file mode 100644 index 0000000000..40d016e39d --- /dev/null +++ b/services/track-changes/.gitignore @@ -0,0 +1,4 @@ +**.swp +node_modules/ +app/js +test/unit/js diff --git a/services/track-changes/app/coffee/ConcatManager.coffee b/services/track-changes/app/coffee/ConcatManager.coffee new file mode 100644 index 0000000000..0dd2790fe8 --- /dev/null +++ b/services/track-changes/app/coffee/ConcatManager.coffee @@ -0,0 +1,121 @@ +strInject = (s1, pos, s2) -> s1[...pos] + s2 + s1[pos..] +strRemove = (s1, pos, length) -> s1[...pos] + s1[(pos + length)..] + +module.exports = ConcatManager = + normalizeUpdate: (update) -> + updates = [] + for op in update.op + updates.push + op: [op] + meta: + start_ts: update.meta.ts + end_ts: update.meta.ts + user_id: update.meta.user_id + return updates + + MAX_TIME_BETWEEN_UPDATES: oneMinute = 60 * 1000 + + concatTwoUpdates: (firstUpdate, secondUpdate) -> + firstUpdate = + op: firstUpdate.op + meta: + user_id: firstUpdate.meta.user_id + start_ts: firstUpdate.meta.start_ts or firstUpdate.meta.ts + end_ts: firstUpdate.meta.end_ts or firstUpdate.meta.ts + secondUpdate = + op: secondUpdate.op + meta: + user_id: secondUpdate.meta.user_id + start_ts: secondUpdate.meta.start_ts or secondUpdate.meta.ts + end_ts: secondUpdate.meta.end_ts or secondUpdate.meta.ts + + if firstUpdate.meta.user_id != secondUpdate.meta.user_id + return [firstUpdate, secondUpdate] + + if secondUpdate.meta.start_ts - firstUpdate.meta.end_ts > ConcatManager.MAX_TIME_BETWEEN_UPDATES + return [firstUpdate, secondUpdate] + + firstOp = firstUpdate.op[0] + secondOp = secondUpdate.op[0] + # Two inserts + if firstOp.i? and secondOp.i? and firstOp.p <= secondOp.p <= (firstOp.p + firstOp.i.length) + return [ + meta: + start_ts: firstUpdate.meta.start_ts + end_ts: secondUpdate.meta.end_ts + user_id: firstUpdate.meta.user_id + op: [ + p: firstOp.p + i: strInject(firstOp.i, secondOp.p - firstOp.p, secondOp.i) + ] + ] + # Two deletes + else if firstOp.d? and secondOp.d? and secondOp.p <= firstOp.p <= (secondOp.p + secondOp.d.length) + return [ + meta: + start_ts: firstUpdate.meta.start_ts + end_ts: secondUpdate.meta.end_ts + user_id: firstUpdate.meta.user_id + op: [ + p: secondOp.p + d: strInject(secondOp.d, firstOp.p - secondOp.p, firstOp.d) + ] + ] + # An insert and then a delete + else if firstOp.i? and secondOp.d? and firstOp.p <= secondOp.p <= (firstOp.p + firstOp.i.length) + offset = secondOp.p - firstOp.p + insertedText = firstOp.i.slice(offset, offset + secondOp.d.length) + if insertedText == secondOp.d + insert = strRemove(firstOp.i, offset, secondOp.d.length) + return [] if insert == "" + return [ + meta: + start_ts: firstUpdate.meta.start_ts + end_ts: secondUpdate.meta.end_ts + user_id: firstUpdate.meta.user_id + op: [ + p: firstOp.p + i: insert + ] + ] + else + # This shouldn't be possible! + return [firstUpdate, secondUpdate] + else if firstOp.d? and secondOp.i? and firstOp.p == secondOp.p + offset = firstOp.d.indexOf(secondOp.i) + if offset == -1 + return [firstUpdate, secondUpdate] + headD = firstOp.d.slice(0, offset) + inserted = firstOp.d.slice(offset, secondOp.i.length) + tailD = firstOp.d.slice(offset + secondOp.i.length) + headP = firstOp.p + tailP = firstOp.p + secondOp.i.length + updates = [] + if headD != "" + updates.push + meta: + start_ts: firstUpdate.meta.start_ts + end_ts: secondUpdate.meta.end_ts + user_id: firstUpdate.meta.user_id + op: [ + p: headP + d: headD + ] + if tailD != "" + updates.push + meta: + start_ts: firstUpdate.meta.start_ts + end_ts: secondUpdate.meta.end_ts + user_id: firstUpdate.meta.user_id + op: [ + p: tailP + d: tailD + ] + if updates.length == 2 + updates[0].meta.start_ts = updates[0].meta.end_ts = firstUpdate.meta.start_ts + updates[1].meta.start_ts = updates[1].meta.end_ts = secondUpdate.meta.end_ts + return updates + + else + return [firstUpdate, secondUpdate] + diff --git a/services/track-changes/app/coffee/mongojs.coffee b/services/track-changes/app/coffee/mongojs.coffee new file mode 100644 index 0000000000..667f03fcd5 --- /dev/null +++ b/services/track-changes/app/coffee/mongojs.coffee @@ -0,0 +1,8 @@ +Settings = require "settings-sharelatex" +mongojs = require "mongojs" +db = mongojs.connect(Settings.mongo.url, ["docHistory", "docOps"]) +module.exports = + db: db + ObjectId: mongojs.ObjectId + + diff --git a/services/track-changes/compressHistory.coffee b/services/track-changes/compressHistory.coffee new file mode 100644 index 0000000000..ae2cbde219 --- /dev/null +++ b/services/track-changes/compressHistory.coffee @@ -0,0 +1,74 @@ +{db, ObjectId} = require "./app/js/mongojs" +ConcatManager = require "./app/js/ConcatManager" + +doc_id = process.argv.pop() +console.log "DOC ID", doc_id + +OPS_TO_LEAVE = 10 + +removeLatestCompressedUpdate = (doc_id, callback = (error) ->) -> + db.docHistory.update { doc_id: ObjectId(doc_id) }, { $pop: { docOps: 1 } }, callback + +getLatestCompressedUpdate = (doc_id, callback = (error) ->) -> + db.docHistory.find { doc_id: ObjectId(doc_id) }, { docOps: { $slice: -1 } }, (error, history) -> + return callback(error) if error? + history = history[0] or { docOps: [] } + callback null, history.docOps.slice(-1)[0] + +insertCompressedUpdates = (doc_id, updates, callback = (error) ->) -> + db.docHistory.update { doc_id: ObjectId(doc_id) }, { $push: { docOps: { $each: updates } } }, { upsert: true }, callback + +trimLastRawUpdate = (doc_id, tailVersion, callback = (error) ->) -> + db.docOps.update { doc_id: ObjectId(doc_id) }, { $pop: { docOps: -1 }, $set: { tailVersion: tailVersion + 1 } }, callback + +done = () -> + console.log "DONE! Here's the history:" + db.docHistory.find { doc_id: ObjectId(doc_id) }, (error, docs) -> + throw error if error? + doc = docs[0] + for update in doc.docOps + op = update.op[0] + if op.i? + console.log update.meta.start_ts, update.meta.end_ts, update.meta.user_id, "INSERT", op.p, op.i + else if op.d? + console.log update.meta.start_ts, update.meta.end_ts, update.meta.user_id, "DELETE", op.p, op.d + process.exit() + +do next = () -> + db.docOps.find { doc_id: ObjectId(doc_id) }, { version: true, tailVersion: true, docOps: { $slice: 1 } }, (error, docs) -> + throw error if error? + throw "doc not found" if docs.length < 1 + doc = docs[0] + tailVersion = doc.tailVersion or 0 + version = doc.version + + rawUpdate = doc.docOps[0] + rawUpdates = ConcatManager.normalizeUpdate(rawUpdate) + + if version - tailVersion > OPS_TO_LEAVE + getLatestCompressedUpdate doc_id, (error, lastCompressedUpdate) -> + throw error if error? + if lastCompressedUpdate? + compressedUpdates = [lastCompressedUpdate] + for rawUpdate in rawUpdates + lastCompressedUpdate = compressedUpdates.pop() + compressedUpdates = compressedUpdates.concat ConcatManager.concatTwoUpdates lastCompressedUpdate, rawUpdate + removeLatestCompressedUpdate doc_id, (error) -> + throw error if error? + insertCompressedUpdates doc_id, compressedUpdates, (error) -> + throw error if error? + trimLastRawUpdate doc_id, tailVersion, (error) -> + throw error if error? + console.log "Pushed compressed op" + next() + else + insertCompressedUpdates doc_id, rawUpdates, (error) -> + trimLastRawUpdate doc_id, tailVersion, (error) -> + throw error if error? + console.log "Pushed first op" + next() + else + console.log "Up to date" + done() + + diff --git a/services/track-changes/config/settings.testing.coffee b/services/track-changes/config/settings.testing.coffee new file mode 100755 index 0000000000..528c6a7114 --- /dev/null +++ b/services/track-changes/config/settings.testing.coffee @@ -0,0 +1,3 @@ +module.exports = + mongo: + url: 'mongodb://127.0.0.1/sharelatexTesting' diff --git a/services/track-changes/package.json b/services/track-changes/package.json new file mode 100644 index 0000000000..1a6be78455 --- /dev/null +++ b/services/track-changes/package.json @@ -0,0 +1,11 @@ +{ + "name": "history-sharelatex", + "version": "0.0.1", + "dependencies": { + "chai": "", + "sandboxed-module": "", + "sinon": "", + "mongojs": "0.7.2", + "settings": "git+ssh://git@bitbucket.org:sharelatex/settings-sharelatex.git#master" + } +} diff --git a/services/track-changes/rakefile.rb b/services/track-changes/rakefile.rb new file mode 100644 index 0000000000..6baa462558 --- /dev/null +++ b/services/track-changes/rakefile.rb @@ -0,0 +1,112 @@ +require 'fileutils' + +siteurl = "https://www.sharelatex.com" + +desc "Compile JavaScirpt into CoffeeScript" +namespace 'setup' do + + desc "installes npm packages json and global stuff like less" + task :installDependencys do + sh %{npm install} + sh %{npm install -g coffee-script} + sh %{git submodule init} + sh %{git submodule update} + end +end + + +namespace 'run' do + desc "compiles and runs the javascirpt version of the app" + task :app => ["compile:app"] do + sh %{node app.js | bunyan} do |ok, res| + if ! ok + raise "error compiling app folder tests : #{res}" + end + puts 'finished app compile' + end + end +end + +namespace 'compile' do + desc "compiles main app folder" + task :app do + puts "Compiling app folder to JS" + FileUtils.rm_rf "app/js" + sh %{coffee -c -o app/js/ app/coffee/} do |ok, res| + if ! ok + raise "error compiling app folder tests : #{res}" + end + puts 'finished app compile' + end + end + + desc "compiles unit tests" + task :unittests => ["compile:app"] do + puts "Compiling Unit Tests to JS" + `coffee -c -o test/unit/js/ test/unit/coffee/` + end + + desc "compiles acceptance tests" + task :acceptancetests => ["compile:app"] do + puts "Compiling Acceptance Tests to JS" + sh %{coffee -c -o test/acceptance/js/ test/acceptance/coffee/} do |ok, res| + if ! ok + raise "error compiling acceptance tests: #{res}" + end + end + end +end + +namespace 'test' do + + desc "runs all test" + task :all => ["test:unit", "test:acceptance"] do + puts "testing everything" + end + + desc "Run Acceptance Tests" + task :acceptance => ["compile:acceptancetests"]do + puts "Running Acceptance Tests" + feature = ENV['feature'] + if feature.nil? + featureFlags = "" + else + featureFlags = "-g \"#{feature}\"" + end + sh %{mocha -R spec #{featureFlags} test/acceptance/js/*} do |ok, res| + if ! ok + raise "error running acceptance tests: #{res}" + end + end + end + + desc "run unit tests" + task :unit => ["compile:unittests"]do + puts "Running Unit Tests" + featurePath = ENV['feature'] + puts featurePath + if featurePath.nil? + featurePath = '' + elsif featurePath.include? '/' + elsif !featurePath.include? '/' + featurePath +='/' + else + featurePath = '' + end + + sh %{mocha -R spec test/unit/js/#{featurePath}* --ignore-leaks} do |ok, res| + if ! ok + raise "error running unit tests : #{res}" + end + end + end +end + + +namespace 'deploy' do + desc "safley deploys app" + task :live do + sh %{git push origin} + sh %{cap live deploy} + end +end diff --git a/services/track-changes/test/unit/coffee/ConcatManager/ConcatManagerTests.coffee b/services/track-changes/test/unit/coffee/ConcatManager/ConcatManagerTests.coffee new file mode 100644 index 0000000000..620813c462 --- /dev/null +++ b/services/track-changes/test/unit/coffee/ConcatManager/ConcatManagerTests.coffee @@ -0,0 +1,214 @@ +sinon = require('sinon') +chai = require('chai') +should = chai.should() +expect = chai.expect +modulePath = "../../../../app/js/ConcatManager.js" +SandboxedModule = require('sandboxed-module') + +describe "ConcatManager", -> + beforeEach -> + @ConcatManager = SandboxedModule.require modulePath + @user_id = "user-id-1" + @other_user_id = "user-id-2" + @ts1 = Date.now() + @ts2 = Date.now() + 1000 + + describe "concatTwoUpdates", -> + describe "insert - insert", -> + it "should append one insert to the other", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, i: "foo" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 6, i: "bar" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 3, i: "foobar" ] + meta: start_ts: @ts1, end_ts: @ts2, user_id: @user_id + }] + + it "should insert one insert inside the other", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, i: "foo" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 5, i: "bar" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 3, i: "fobaro" ] + meta: start_ts: @ts1, end_ts: @ts2, user_id: @user_id + }] + + it "should not append separated inserts", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, i: "foo" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 9, i: "bar" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 3, i: "foo" ] + meta: start_ts: @ts1, end_ts: @ts1, user_id: @user_id + }, { + op: [ p: 9, i: "bar" ] + meta: start_ts: @ts2, end_ts: @ts2, user_id: @user_id + }] + + describe "delete - delete", -> + it "should append one delete to the other", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, d: "foo" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 3, d: "bar" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 3, d: "foobar" ] + meta: start_ts: @ts1, end_ts: @ts2, user_id: @user_id + }] + + it "should insert one delete inside the other", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, d: "foo" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 1, d: "bar" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 1, d: "bafoor" ] + meta: start_ts: @ts1, end_ts: @ts2, user_id: @user_id + }] + + it "should not append separated deletes", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, d: "foo" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 9, d: "bar" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 3, d: "foo" ] + meta: start_ts: @ts1, end_ts: @ts1, user_id: @user_id + }, { + op: [ p: 9, d: "bar" ] + meta: start_ts: @ts2, end_ts: @ts2, user_id: @user_id + }] + + describe "insert - delete", -> + it "should undo a previous insert", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, i: "foo" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 5, d: "o" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 3, i: "fo" ] + meta: start_ts: @ts1, end_ts: @ts2, user_id: @user_id + }] + + it "should remove part of an insert from the middle", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, i: "fobaro" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 5, d: "bar" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 3, i: "foo" ] + meta: start_ts: @ts1, end_ts: @ts2, user_id: @user_id + }] + + it "should cancel out two opposite updates", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, i: "foo" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 3, d: "foo" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [] + + it "should not combine separated updates", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, i: "foo" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 9, d: "bar" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 3, i: "foo" ] + meta: start_ts: @ts1, end_ts: @ts1, user_id: @user_id + }, { + op: [ p: 9, d: "bar" ] + meta: start_ts: @ts2, end_ts: @ts2, user_id: @user_id + }] + + describe "delete - insert", -> + it "should redo a previous delete at the beginning", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, d: "foo" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 3, i: "f" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 4, d: "oo" ] + meta: start_ts: @ts1, end_ts: @ts2, user_id: @user_id + }] + + it "should redo a previous delete from halfway through", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, d: "foobar" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 3, i: "oo" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 3, d: "f" ] + meta: start_ts: @ts1, end_ts: @ts1, user_id: @user_id + }, { + op: [ p: 5, d: "bar" ] + meta: start_ts: @ts2, end_ts: @ts2, user_id: @user_id + }] + + it "should not combine the ops if the insert text does not match the delete text", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, d: "foobar" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 3, i: "xy" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [{ + op: [ p: 3, d: "foobar" ] + meta: start_ts: @ts1, end_ts: @ts1, user_id: @user_id + }, { + op: [ p: 3, i: "xy" ] + meta: start_ts: @ts2, end_ts: @ts2, user_id: @user_id + }] + + it "should cancel two equal updates", -> + expect(@ConcatManager.concatTwoUpdates({ + op: [ p: 3, d: "foo" ] + meta: ts: @ts1, user_id: @user_id + }, { + op: [ p: 3, i: "foo" ] + meta: ts: @ts2, user_id: @user_id + })) + .to.deep.equal [] + + + +