func main() {
defer common.LogPanic()
worker_common.Init()
defer util.TimeTrack(time.Now(), "Capturing Archives")
defer glog.Flush()
// Create the task file so that the master knows this worker is still busy.
skutil.LogErr(util.CreateTaskFile(util.ACTIVITY_CAPTURING_ARCHIVES))
defer util.DeleteTaskFile(util.ACTIVITY_CAPTURING_ARCHIVES)
if *chromiumBuild == "" {
glog.Error("Must specify --chromium_build")
return
}
// Reset the local chromium checkout.
if err := util.ResetCheckout(util.ChromiumSrcDir); err != nil {
glog.Errorf("Could not reset %s: %s", util.ChromiumSrcDir, err)
return
}
// Sync the local chromium checkout.
if err := util.SyncDir(util.ChromiumSrcDir); err != nil {
glog.Errorf("Could not gclient sync %s: %s", util.ChromiumSrcDir, err)
return
}
// Delete and remake the local webpage archives directory.
pathToArchives := filepath.Join(util.WebArchivesDir, *pagesetType)
skutil.RemoveAll(pathToArchives)
skutil.MkdirAll(pathToArchives, 0700)
// Instantiate GsUtil object.
gs, err := util.NewGsUtil(nil)
if err != nil {
glog.Error(err)
return
}
// Download the specified chromium build if it does not exist locally.
if err := gs.DownloadChromiumBuild(*chromiumBuild); err != nil {
glog.Error(err)
return
}
// Download pagesets if they do not exist locally.
if err := gs.DownloadWorkerArtifacts(util.PAGESETS_DIR_NAME, *pagesetType, *workerNum); err != nil {
glog.Error(err)
return
}
pathToPagesets := filepath.Join(util.PagesetsDir, *pagesetType)
chromiumBinary := filepath.Join(util.ChromiumBuildsDir, *chromiumBuild, util.BINARY_CHROME)
recordWprBinary := filepath.Join(util.TelemetryBinariesDir, util.BINARY_RECORD_WPR)
timeoutSecs := util.PagesetTypeToInfo[*pagesetType].CaptureArchivesTimeoutSecs
// Loop through all pagesets.
fileInfos, err := ioutil.ReadDir(pathToPagesets)
if err != nil {
glog.Errorf("Unable to read the pagesets dir %s: %s", pathToPagesets, err)
return
}
glog.Infof("The %s fileInfos are: %s", len(fileInfos), fileInfos)
for _, fileInfo := range fileInfos {
pagesetBaseName := filepath.Base(fileInfo.Name())
if pagesetBaseName == util.TIMESTAMP_FILE_NAME || filepath.Ext(pagesetBaseName) == ".pyc" {
// Ignore timestamp files and .pyc files.
continue
}
// Read the pageset.
pagesetPath := filepath.Join(pathToPagesets, fileInfo.Name())
decodedPageset, err := util.ReadPageset(pagesetPath)
if err != nil {
glog.Errorf("Could not read %s: %s", pagesetPath, err)
return
}
glog.Infof("===== Processing %s =====", pagesetPath)
args := []string{
util.CAPTURE_ARCHIVES_DEFAULT_CT_BENCHMARK,
"--extra-browser-args=--disable-setuid-sandbox",
"--browser=exact",
"--browser-executable=" + chromiumBinary,
"--user-agent=" + decodedPageset.UserAgent,
"--urls-list=" + decodedPageset.UrlsList,
"--archive-data-file=" + decodedPageset.ArchiveDataFile,
}
env := []string{
fmt.Sprintf("PYTHONPATH=%s:$PYTHONPATH", pathToPagesets),
"DISPLAY=:0",
}
skutil.LogErr(util.ExecuteCmd(recordWprBinary, args, env, time.Duration(timeoutSecs)*time.Second, nil, nil))
}
// Write timestamp to the webpage archives dir.
skutil.LogErr(util.CreateTimestampFile(pathToArchives))
// Upload webpage archives dir to Google Storage.
if err := gs.UploadWorkerArtifacts(util.WEB_ARCHIVES_DIR_NAME, *pagesetType, *workerNum); err != nil {
glog.Error(err)
return
}
}
func main() {
defer common.LogPanic()
worker_common.Init()
defer util.TimeTrack(time.Now(), "Fixing archives")
defer glog.Flush()
// Create the task file so that the master knows this worker is still busy.
skutil.LogErr(util.CreateTaskFile(util.ACTIVITY_FIXING_ARCHIVES))
defer util.DeleteTaskFile(util.ACTIVITY_FIXING_ARCHIVES)
if *pagesetType == "" {
glog.Error("Must specify --pageset_type")
return
}
if *chromiumBuild == "" {
glog.Error("Must specify --chromium_build")
return
}
if *runID == "" {
glog.Error("Must specify --run_id")
return
}
// Reset the local chromium checkout.
if err := util.ResetCheckout(util.ChromiumSrcDir); err != nil {
glog.Errorf("Could not reset %s: %s", util.ChromiumSrcDir, err)
return
}
// Sync the local chromium checkout.
if err := util.SyncDir(util.ChromiumSrcDir); err != nil {
glog.Errorf("Could not gclient sync %s: %s", util.ChromiumSrcDir, err)
return
}
// Instantiate GsUtil object.
gs, err := util.NewGsUtil(nil)
if err != nil {
glog.Error(err)
return
}
// Download the specified chromium build.
if err := gs.DownloadChromiumBuild(*chromiumBuild); err != nil {
glog.Error(err)
return
}
// Delete the chromium build to save space when we are done.
defer skutil.RemoveAll(filepath.Join(util.ChromiumBuildsDir, *chromiumBuild))
chromiumBinary := filepath.Join(util.ChromiumBuildsDir, *chromiumBuild, util.BINARY_CHROME)
// Download pagesets if they do not exist locally.
if err := gs.DownloadWorkerArtifacts(util.PAGESETS_DIR_NAME, *pagesetType, *workerNum); err != nil {
glog.Error(err)
return
}
pathToPagesets := filepath.Join(util.PagesetsDir, *pagesetType)
// Download archives if they do not exist locally.
if err := gs.DownloadWorkerArtifacts(util.WEB_ARCHIVES_DIR_NAME, *pagesetType, *workerNum); err != nil {
glog.Error(err)
return
}
// Establish output paths.
localOutputDir := filepath.Join(util.StorageDir, util.FixArchivesRunsDir, *runID)
skutil.RemoveAll(localOutputDir)
skutil.MkdirAll(localOutputDir, 0700)
defer skutil.RemoveAll(localOutputDir)
// Construct path to the ct_run_benchmark python script.
_, currentFile, _, _ := runtime.Caller(0)
pathToPyFiles := filepath.Join(
filepath.Dir((filepath.Dir(filepath.Dir(filepath.Dir(currentFile))))),
"py")
timeoutSecs := util.PagesetTypeToInfo[*pagesetType].RunChromiumPerfTimeoutSecs
fileInfos, err := ioutil.ReadDir(pathToPagesets)
if err != nil {
glog.Errorf("Unable to read the pagesets dir %s: %s", pathToPagesets, err)
return
}
// Location of the WPR logs.
wprLogs := filepath.Join(util.ChromiumSrcDir, "webpagereplay_logs", "logs.txt")
// Slice that will contain
inconsistentArchives := []string{}
// Loop through all pagesets.
for _, fileInfo := range fileInfos {
benchmarkResults := []float64{}
resourceMissingCounts := []int{}
pagesetBaseName := filepath.Base(fileInfo.Name())
if pagesetBaseName == util.TIMESTAMP_FILE_NAME || filepath.Ext(pagesetBaseName) == ".pyc" {
// Ignore timestamp files and .pyc files.
continue
}
// Convert the filename into a format consumable by the run_benchmarks
// binary.
pagesetName := strings.TrimSuffix(pagesetBaseName, filepath.Ext(pagesetBaseName))
pagesetPath := filepath.Join(pathToPagesets, fileInfo.Name())
glog.Infof("===== Processing %s =====", pagesetPath)
// Repeat runs the specified number of times.
for repeatNum := 1; repeatNum <= *repeatBenchmark; repeatNum++ {
// Delete webpagereplay_logs before every run.
skutil.RemoveAll(wprLogs)
skutil.LogErr(os.Chdir(pathToPyFiles))
args := []string{
util.BINARY_RUN_BENCHMARK,
fmt.Sprintf("%s.%s", *benchmarkName, util.BenchmarksToTelemetryName[*benchmarkName]),
"--page-set-name=" + pagesetName,
"--page-set-base-dir=" + pathToPagesets,
"--also-run-disabled-tests",
}
// Add output dir.
outputDirArgValue := filepath.Join(localOutputDir, pagesetName, strconv.Itoa(repeatNum))
args = append(args, "--output-dir="+outputDirArgValue)
// Figure out which browser should be used.
args = append(args, "--browser=exact", "--browser-executable="+chromiumBinary)
// Split benchmark args if not empty and append to args.
if *benchmarkArgs != "" {
for _, benchmarkArg := range strings.Split(*benchmarkArgs, " ") {
args = append(args, benchmarkArg)
}
}
// Add browserArgs if not empty to args.
if *browserArgs != "" {
args = append(args, "--extra-browser-args="+*browserArgs)
}
// Set the PYTHONPATH to the pagesets and the telemetry dirs.
env := []string{
fmt.Sprintf("PYTHONPATH=%s:%s:%s:$PYTHONPATH", pathToPagesets, util.TelemetryBinariesDir, util.TelemetrySrcDir),
"DISPLAY=:0",
}
skutil.LogErr(
util.ExecuteCmd("python", args, env, time.Duration(timeoutSecs)*time.Second, nil, nil))
// Examine the results-pivot-table.csv file and store the mean frame time.
resultsCSV := filepath.Join(outputDirArgValue, "results-pivot-table.csv")
// TODO(rmistry): The format has changed from results.csv to results-pivot-table.csv
headers, values, err := getRowsFromCSV(resultsCSV)
if err != nil {
glog.Errorf("Could not read %s: %s", resultsCSV, err)
continue
}
for i := range headers {
if headers[i] == *benchmarkHeaderToCheck {
value, _ := strconv.ParseFloat(values[i], 64)
benchmarkResults = append(benchmarkResults, value)
break
}
}
// Find how many times "Could not replay" showed up in wprLogs.
content, err := ioutil.ReadFile(wprLogs)
if err != nil {
glog.Errorf("Could not read %s: %s", wprLogs, err)
continue
}
resourceMissingCount := strings.Count(string(content), "Could not replay")
resourceMissingCounts = append(resourceMissingCounts, resourceMissingCount)
}
glog.Infof("Benchmark results for %s are: %v", fileInfo.Name(), benchmarkResults)
percentageChange := getPercentageChange(benchmarkResults)
glog.Infof("Percentage change of results is: %f", percentageChange)
glog.Infof("\"Could not replay\" showed up %v times in %s", resourceMissingCounts, wprLogs)
maxResourceMissingCount := 0
for _, count := range resourceMissingCounts {
if maxResourceMissingCount < count {
maxResourceMissingCount = count
}
}
if percentageChange > *percentageChangeThreshold || maxResourceMissingCount > *resourceMissingCountThreshold {
glog.Infof("The archive for %s is inconsistent!", fileInfo.Name())
inconsistentArchives = append(inconsistentArchives, fmt.Sprintf("%s percentageChange: %f maxResourceMissingCount: %v", fileInfo.Name(), percentageChange, maxResourceMissingCount))
if *deletePageset {
// Delete the pageset.
skutil.RemoveAll(pagesetPath)
}
}
}
if len(inconsistentArchives) > 0 {
glog.Infof("%d archives are inconsistent!", len(inconsistentArchives))
glog.Infof("The list of inconsistentArchives is: %v", inconsistentArchives)
if *deletePageset {
// Write new timestamp to the pagesets dir.
skutil.RemoveAll(filepath.Join(pathToPagesets, util.TIMESTAMP_FILE_NAME))
skutil.LogErr(util.CreateTimestampFile(pathToPagesets))
// Inconsistent pagesets were deleted locally. Upload local pagesets dir
// to Google Storage.
if err := gs.UploadWorkerArtifacts(util.PAGESETS_DIR_NAME, *pagesetType, *workerNum); err != nil {
glog.Error(err)
return
}
}
}
}
func main() {
defer common.LogPanic()
worker_common.Init()
if !*worker_common.Local {
defer util.CleanTmpDir()
}
defer util.TimeTrack(time.Now(), "Capturing SKPs")
defer glog.Flush()
// Validate required arguments.
if *chromiumBuild == "" {
glog.Error("Must specify --chromium_build")
return
}
if *runID == "" {
glog.Error("Must specify --run_id")
return
}
if *targetPlatform == util.PLATFORM_ANDROID {
glog.Error("Android is not yet supported for capturing SKPs.")
return
}
// Reset the local chromium checkout.
if err := util.ResetCheckout(util.ChromiumSrcDir); err != nil {
glog.Errorf("Could not reset %s: %s", util.ChromiumSrcDir, err)
return
}
// Sync the local chromium checkout.
if err := util.SyncDir(util.ChromiumSrcDir); err != nil {
glog.Errorf("Could not gclient sync %s: %s", util.ChromiumSrcDir, err)
return
}
// Create the task file so that the master knows this worker is still busy.
skutil.LogErr(util.CreateTaskFile(util.ACTIVITY_CAPTURING_SKPS))
defer util.DeleteTaskFile(util.ACTIVITY_CAPTURING_SKPS)
// Instantiate GsUtil object.
gs, err := util.NewGsUtil(nil)
if err != nil {
glog.Error(err)
return
}
// Download the specified chromium build.
if err := gs.DownloadChromiumBuild(*chromiumBuild); err != nil {
glog.Error(err)
return
}
// Delete the chromium build to save space when we are done.
defer skutil.RemoveAll(filepath.Join(util.ChromiumBuildsDir, *chromiumBuild))
chromiumBinary := filepath.Join(util.ChromiumBuildsDir, *chromiumBuild, util.BINARY_CHROME)
if *targetPlatform == util.PLATFORM_ANDROID {
// Install the APK on the Android device.
if err := util.InstallChromeAPK(*chromiumBuild); err != nil {
glog.Errorf("Could not install the chromium APK: %s", err)
return
}
}
// Download pagesets if they do not exist locally.
if err := gs.DownloadWorkerArtifacts(util.PAGESETS_DIR_NAME, *pagesetType, *workerNum); err != nil {
glog.Error(err)
return
}
pathToPagesets := filepath.Join(util.PagesetsDir, *pagesetType)
// Download archives if they do not exist locally.
if err := gs.DownloadWorkerArtifacts(util.WEB_ARCHIVES_DIR_NAME, *pagesetType, *workerNum); err != nil {
glog.Error(err)
return
}
// Create the dir that SKPs will be stored in.
pathToSkps := filepath.Join(util.SkpsDir, *pagesetType, *chromiumBuild)
// Delete and remake the local SKPs directory.
skutil.RemoveAll(pathToSkps)
skutil.MkdirAll(pathToSkps, 0700)
// Establish output paths.
localOutputDir := filepath.Join(util.StorageDir, util.BenchmarkRunsDir, *runID)
skutil.RemoveAll(localOutputDir)
skutil.MkdirAll(localOutputDir, 0700)
defer skutil.RemoveAll(localOutputDir)
// Construct path to the ct_run_benchmark python script.
_, currentFile, _, _ := runtime.Caller(0)
pathToPyFiles := filepath.Join(
filepath.Dir((filepath.Dir(filepath.Dir(filepath.Dir(currentFile))))),
"py")
timeoutSecs := util.PagesetTypeToInfo[*pagesetType].CaptureSKPsTimeoutSecs
fileInfos, err := ioutil.ReadDir(pathToPagesets)
if err != nil {
glog.Errorf("Unable to read the pagesets dir %s: %s", pathToPagesets, err)
return
}
// Create channel that contains all pageset file names. This channel will
// be consumed by the worker pool.
pagesetRequests := util.GetClosedChannelOfPagesets(fileInfos)
var wg sync.WaitGroup
// Use a RWMutex for the chromeProcessesCleaner goroutine to communicate to
// the workers (acting as "readers") when it wants to be the "writer" and
// kill all zombie chrome processes.
var mutex sync.RWMutex
// Loop through workers in the worker pool.
for i := 0; i < WORKER_POOL_SIZE; i++ {
// Increment the WaitGroup counter.
wg.Add(1)
// Create and run a goroutine closure that captures SKPs.
go func() {
// Decrement the WaitGroup counter when the goroutine completes.
defer wg.Done()
for pagesetName := range pagesetRequests {
mutex.RLock()
// Read the pageset.
pagesetPath := filepath.Join(pathToPagesets, pagesetName)
decodedPageset, err := util.ReadPageset(pagesetPath)
if err != nil {
glog.Errorf("Could not read %s: %s", pagesetPath, err)
continue
}
glog.Infof("===== Processing %s =====", pagesetPath)
skutil.LogErr(os.Chdir(pathToPyFiles))
args := []string{
filepath.Join(util.TelemetryBinariesDir, util.BINARY_RUN_BENCHMARK),
util.BenchmarksToTelemetryName[util.BENCHMARK_SKPICTURE_PRINTER],
"--also-run-disabled-tests",
"--page-repeat=1", // Only need one run for SKPs.
"--skp-outdir=" + pathToSkps,
"--extra-browser-args=" + util.DEFAULT_BROWSER_ARGS,
"--user-agent=" + decodedPageset.UserAgent,
"--urls-list=" + decodedPageset.UrlsList,
"--archive-data-file=" + decodedPageset.ArchiveDataFile,
}
// Figure out which browser should be used.
if *targetPlatform == util.PLATFORM_ANDROID {
args = append(args, "--browser=android-chromium")
} else {
args = append(args, "--browser=exact", "--browser-executable="+chromiumBinary)
}
// Set the PYTHONPATH to the pagesets and the telemetry dirs.
env := []string{
fmt.Sprintf("PYTHONPATH=%s:%s:%s:$PYTHONPATH", pathToPagesets, util.TelemetryBinariesDir, util.TelemetrySrcDir),
"DISPLAY=:0",
}
skutil.LogErr(
util.ExecuteCmd("python", args, env, time.Duration(timeoutSecs)*time.Second, nil, nil))
mutex.RUnlock()
}
}()
}
if !*worker_common.Local {
// Start the cleaner.
go util.ChromeProcessesCleaner(&mutex, *chromeCleanerTimer)
}
// Wait for all spawned goroutines to complete.
wg.Wait()
// Move, validate and upload all SKP files.
// List all directories in pathToSkps and copy out the skps.
skpFileInfos, err := ioutil.ReadDir(pathToSkps)
if err != nil {
glog.Errorf("Unable to read %s: %s", pathToSkps, err)
return
}
for _, fileInfo := range skpFileInfos {
if !fileInfo.IsDir() {
// We are only interested in directories.
continue
}
skpName := fileInfo.Name()
// Find the largest layer in this directory.
layerInfos, err := ioutil.ReadDir(filepath.Join(pathToSkps, skpName))
if err != nil {
glog.Errorf("Unable to read %s: %s", filepath.Join(pathToSkps, skpName), err)
}
if len(layerInfos) > 0 {
largestLayerInfo := layerInfos[0]
for _, layerInfo := range layerInfos {
fmt.Println(layerInfo.Size())
if layerInfo.Size() > largestLayerInfo.Size() {
largestLayerInfo = layerInfo
}
}
// Only save SKPs greater than 6000 bytes. Less than that are probably
// malformed.
if largestLayerInfo.Size() > 6000 {
layerPath := filepath.Join(pathToSkps, skpName, largestLayerInfo.Name())
skutil.Rename(layerPath, filepath.Join(pathToSkps, skpName+".skp"))
} else {
glog.Warningf("Skipping %s because size was less than 5000 bytes", skpName)
}
}
// We extracted what we needed from the directory, now delete it.
skutil.RemoveAll(filepath.Join(pathToSkps, skpName))
}
glog.Info("Calling remove_invalid_skps.py")
// Sync Skia tree.
skutil.LogErr(util.SyncDir(util.SkiaTreeDir))
// Build tools.
skutil.LogErr(util.BuildSkiaTools())
// Run remove_invalid_skps.py
pathToRemoveSKPs := filepath.Join(pathToPyFiles, "remove_invalid_skps.py")
pathToSKPInfo := filepath.Join(util.SkiaTreeDir, "out", "Release", "skpinfo")
args := []string{
pathToRemoveSKPs,
"--skp_dir=" + pathToSkps,
"--path_to_skpinfo=" + pathToSKPInfo,
}
skutil.LogErr(util.ExecuteCmd("python", args, []string{}, util.REMOVE_INVALID_SKPS_TIMEOUT,
nil, nil))
// Write timestamp to the SKPs dir.
skutil.LogErr(util.CreateTimestampFile(pathToSkps))
// Upload SKPs dir to Google Storage.
if err := gs.UploadWorkerArtifacts(util.SKPS_DIR_NAME, filepath.Join(*pagesetType, *chromiumBuild), *workerNum); err != nil {
glog.Error(err)
return
}
}
func main() {
defer common.LogPanic()
common.Init()
defer util.TimeTrack(time.Now(), "Creating Pagesets")
defer glog.Flush()
// Create the task file so that the master knows this worker is still busy.
skutil.LogErr(util.CreateTaskFile(util.ACTIVITY_CREATING_PAGESETS))
defer util.DeleteTaskFile(util.ACTIVITY_CREATING_PAGESETS)
// Delete and remake the local pagesets directory.
pathToPagesets := filepath.Join(util.PagesetsDir, *pagesetType)
skutil.RemoveAll(pathToPagesets)
skutil.MkdirAll(pathToPagesets, 0700)
// Get info about the specified pageset type.
pagesetTypeInfo := util.PagesetTypeToInfo[*pagesetType]
csvSource := pagesetTypeInfo.CSVSource
numPages := pagesetTypeInfo.NumPages
userAgent := pagesetTypeInfo.UserAgent
// Download the CSV file from Google Storage to a tmp location.
gs, err := util.NewGsUtil(nil)
if err != nil {
glog.Error(err)
return
}
respBody, err := gs.GetRemoteFileContents(csvSource)
if err != nil {
glog.Error(err)
return
}
defer skutil.Close(respBody)
csvFile := filepath.Join(os.TempDir(), filepath.Base(csvSource))
out, err := os.Create(csvFile)
if err != nil {
glog.Errorf("Unable to create file %s: %s", csvFile, err)
return
}
defer skutil.Close(out)
defer skutil.Remove(csvFile)
if _, err = io.Copy(out, respBody); err != nil {
glog.Error(err)
return
}
// Figure out which pagesets this worker should generate.
numPagesPerSlave := numPages / util.NUM_WORKERS
startNum := (*workerNum-1)*numPagesPerSlave + 1
endNum := *workerNum * numPagesPerSlave
// Construct path to the create_page_set.py python script.
_, currentFile, _, _ := runtime.Caller(0)
createPageSetScript := filepath.Join(
filepath.Dir((filepath.Dir(filepath.Dir(filepath.Dir(currentFile))))),
"py", "create_page_set.py")
// Execute the create_page_set.py python script.
timeoutSecs := util.PagesetTypeToInfo[*pagesetType].CreatePagesetsTimeoutSecs
for currNum := startNum; currNum <= endNum; currNum++ {
args := []string{
createPageSetScript,
"-s", strconv.Itoa(currNum),
"-e", strconv.Itoa(currNum),
"-c", csvFile,
"-p", *pagesetType,
"-u", userAgent,
"-o", pathToPagesets,
}
if err := util.ExecuteCmd("python", args, []string{}, time.Duration(timeoutSecs)*time.Second, nil, nil); err != nil {
glog.Error(err)
return
}
}
// Write timestamp to the pagesets dir.
skutil.LogErr(util.CreateTimestampFile(pathToPagesets))
// Upload pagesets dir to Google Storage.
if err := gs.UploadWorkerArtifacts(util.PAGESETS_DIR_NAME, *pagesetType, *workerNum); err != nil {
glog.Error(err)
return
}
}
func main() {
common.Init()
defer util.TimeTrack(time.Now(), "Capturing Archives")
defer glog.Flush()
// Create the task file so that the master knows this worker is still busy.
skutil.LogErr(util.CreateTaskFile(util.ACTIVITY_CAPTURING_ARCHIVES))
defer util.DeleteTaskFile(util.ACTIVITY_CAPTURING_ARCHIVES)
if *chromiumBuild == "" {
glog.Error("Must specify --chromium_build")
return
}
// Reset the local chromium checkout.
if err := util.ResetCheckout(util.ChromiumSrcDir); err != nil {
glog.Errorf("Could not reset %s: %s", util.ChromiumSrcDir, err)
return
}
// Sync the local chromium checkout.
if err := util.SyncDir(util.ChromiumSrcDir); err != nil {
glog.Errorf("Could not gclient sync %s: %s", util.ChromiumSrcDir, err)
return
}
// Delete and remake the local webpage archives directory.
pathToArchives := filepath.Join(util.WebArchivesDir, *pagesetType)
skutil.RemoveAll(pathToArchives)
skutil.MkdirAll(pathToArchives, 0700)
// Instantiate GsUtil object.
gs, err := util.NewGsUtil(nil)
if err != nil {
glog.Error(err)
return
}
// Download the specified chromium build if it does not exist locally.
if err := gs.DownloadChromiumBuild(*chromiumBuild); err != nil {
glog.Error(err)
return
}
// Download pagesets if they do not exist locally.
if err := gs.DownloadWorkerArtifacts(util.PAGESETS_DIR_NAME, *pagesetType, *workerNum); err != nil {
glog.Error(err)
return
}
pathToPagesets := filepath.Join(util.PagesetsDir, *pagesetType)
chromiumBinary := filepath.Join(util.ChromiumBuildsDir, *chromiumBuild, util.BINARY_CHROME)
recordWprBinary := filepath.Join(util.TelemetryBinariesDir, util.BINARY_RECORD_WPR)
timeoutSecs := util.PagesetTypeToInfo[*pagesetType].CaptureArchivesTimeoutSecs
// Loop through all pagesets.
fileInfos, err := ioutil.ReadDir(pathToPagesets)
if err != nil {
glog.Errorf("Unable to read the pagesets dir %s: %s", pathToPagesets, err)
return
}
// TODO(rmistry): Remove this hack once the 1M webpage archives have been captured.
glog.Infof("The length of fileInfos is: %s", len(fileInfos))
fileInfos = fileInfos[18500:20000]
glog.Infof("The fileInfos are: %s", fileInfos)
for _, fileInfo := range fileInfos {
pagesetBaseName := filepath.Base(fileInfo.Name())
if pagesetBaseName == util.TIMESTAMP_FILE_NAME || filepath.Ext(pagesetBaseName) == ".pyc" {
// Ignore timestamp files and .pyc files.
continue
}
// Convert the filename into a format consumable by the record_wpr binary.
pagesetArchiveName := strings.TrimSuffix(pagesetBaseName, filepath.Ext(pagesetBaseName))
pagesetPath := filepath.Join(pathToPagesets, fileInfo.Name())
glog.Infof("===== Processing %s =====", pagesetPath)
args := []string{
"--extra-browser-args=--disable-setuid-sandbox",
"--browser=exact",
"--browser-executable=" + chromiumBinary,
fmt.Sprintf("%s_page_set", pagesetArchiveName),
"--page-set-base-dir=" + pathToPagesets,
}
env := []string{
fmt.Sprintf("PYTHONPATH=%s:$PYTHONPATH", pathToPagesets),
"DISPLAY=:0",
}
skutil.LogErr(util.ExecuteCmd(recordWprBinary, args, env, time.Duration(timeoutSecs)*time.Second, nil, nil))
}
// Write timestamp to the webpage archives dir.
skutil.LogErr(util.CreateTimestampFile(pathToArchives))
// Upload webpage archives dir to Google Storage.
if err := gs.UploadWorkerArtifacts(util.WEB_ARCHIVES_DIR_NAME, *pagesetType, *workerNum); err != nil {
glog.Error(err)
return
}
}