// This checks the given slice of pointers that don't exist in .git/lfs/objects
// against the server. Anything the server already has does not need to be
// uploaded again.
func (c *uploadContext) checkMissing(missing []*lfs.WrappedPointer, missingSize int64) {
numMissing := len(missing)
if numMissing == 0 {
return
}
checkQueue := lfs.NewDownloadCheckQueue(numMissing, missingSize)
for _, p := range missing {
checkQueue.Add(lfs.NewDownloadable(p))
}
// this channel is filled with oids for which Check() succeeded & Transfer() was called
transferc := checkQueue.Watch()
done := make(chan int)
go func() {
for oid := range transferc {
c.SetUploaded(oid)
}
done <- 1
}()
// Currently this is needed to flush the batch but is not enough to sync transferc completely
checkQueue.Wait()
<-done
}
// Fetch and report completion of each OID to a channel (optional, pass nil to skip)
func fetchAndReportToChan(pointers []*lfs.WrappedPointer, include, exclude []string, out chan<- *lfs.WrappedPointer) {
totalSize := int64(0)
for _, p := range pointers {
totalSize += p.Size
}
q := lfs.NewDownloadQueue(len(pointers), totalSize, false)
for _, p := range pointers {
// Only add to download queue if local file is not the right size already
// This avoids previous case of over-reporting a requirement for files we already have
// which would only be skipped by PointerSmudgeObject later
passFilter := lfs.FilenamePassesIncludeExcludeFilter(p.Name, include, exclude)
if !lfs.ObjectExistsOfSize(p.Oid, p.Size) && passFilter {
q.Add(lfs.NewDownloadable(p))
} else {
// If we already have it, or it won't be fetched
// report it to chan immediately to support pull/checkout
if out != nil {
out <- p
}
}
}
if out != nil {
dlwatch := q.Watch()
go func() {
// fetch only reports single OID, but OID *might* be referenced by multiple
// WrappedPointers if same content is at multiple paths, so map oid->slice
oidToPointers := make(map[string][]*lfs.WrappedPointer, len(pointers))
for _, pointer := range pointers {
plist := oidToPointers[pointer.Oid]
oidToPointers[pointer.Oid] = append(plist, pointer)
}
for oid := range dlwatch {
plist, ok := oidToPointers[oid]
if !ok {
continue
}
for _, p := range plist {
out <- p
}
}
close(out)
}()
}
processQueue := time.Now()
q.Wait()
tracerx.PerformanceSince("process queue", processQueue)
}
func prune(fetchPruneConfig config.FetchPruneConfig, verifyRemote, dryRun, verbose bool) {
localObjects := make([]localstorage.Object, 0, 100)
retainedObjects := tools.NewStringSetWithCapacity(100)
var reachableObjects tools.StringSet
var taskwait sync.WaitGroup
// Add all the base funcs to the waitgroup before starting them, in case
// one completes really fast & hits 0 unexpectedly
// each main process can Add() to the wg itself if it subdivides the task
taskwait.Add(4) // 1..4: localObjects, current & recent refs, unpushed, worktree
if verifyRemote {
taskwait.Add(1) // 5
}
progressChan := make(PruneProgressChan, 100)
// Collect errors
errorChan := make(chan error, 10)
var errorwait sync.WaitGroup
errorwait.Add(1)
var taskErrors []error
go pruneTaskCollectErrors(&taskErrors, errorChan, &errorwait)
// Populate the single list of local objects
go pruneTaskGetLocalObjects(&localObjects, progressChan, &taskwait)
// Now find files to be retained from many sources
retainChan := make(chan string, 100)
go pruneTaskGetRetainedCurrentAndRecentRefs(fetchPruneConfig, retainChan, errorChan, &taskwait)
go pruneTaskGetRetainedUnpushed(fetchPruneConfig, retainChan, errorChan, &taskwait)
go pruneTaskGetRetainedWorktree(retainChan, errorChan, &taskwait)
if verifyRemote {
reachableObjects = tools.NewStringSetWithCapacity(100)
go pruneTaskGetReachableObjects(&reachableObjects, errorChan, &taskwait)
}
// Now collect all the retained objects, on separate wait
var retainwait sync.WaitGroup
retainwait.Add(1)
go pruneTaskCollectRetained(&retainedObjects, retainChan, progressChan, &retainwait)
// Report progress
var progresswait sync.WaitGroup
progresswait.Add(1)
go pruneTaskDisplayProgress(progressChan, &progresswait)
taskwait.Wait() // wait for subtasks
close(retainChan) // triggers retain collector to end now all tasks have
retainwait.Wait() // make sure all retained objects added
close(errorChan) // triggers error collector to end now all tasks have
errorwait.Wait() // make sure all errors have been processed
pruneCheckErrors(taskErrors)
prunableObjects := make([]string, 0, len(localObjects)/2)
// Build list of prunables (also queue for verify at same time if applicable)
var verifyQueue *lfs.TransferQueue
var verifiedObjects tools.StringSet
var totalSize int64
var verboseOutput bytes.Buffer
var verifyc chan string
if verifyRemote {
cfg.CurrentRemote = fetchPruneConfig.PruneRemoteName
// build queue now, no estimates or progress output
verifyQueue = lfs.NewDownloadCheckQueue(0, 0)
verifiedObjects = tools.NewStringSetWithCapacity(len(localObjects) / 2)
// this channel is filled with oids for which Check() succeeded & Transfer() was called
verifyc = verifyQueue.Watch()
}
for _, file := range localObjects {
if !retainedObjects.Contains(file.Oid) {
prunableObjects = append(prunableObjects, file.Oid)
totalSize += file.Size
if verbose {
// Save up verbose output for the end, spinner still going
verboseOutput.WriteString(fmt.Sprintf(" * %v (%v)\n", file.Oid, humanizeBytes(file.Size)))
}
if verifyRemote {
tracerx.Printf("VERIFYING: %v", file.Oid)
pointer := lfs.NewPointer(file.Oid, file.Size, nil)
verifyQueue.Add(lfs.NewDownloadable(&lfs.WrappedPointer{Pointer: pointer}))
}
}
}
if verifyRemote {
var verifywait sync.WaitGroup
verifywait.Add(1)
go func() {
for oid := range verifyc {
verifiedObjects.Add(oid)
tracerx.Printf("VERIFIED: %v", oid)
progressChan <- PruneProgress{PruneProgressTypeVerify, 1}
}
verifywait.Done()
}()
verifyQueue.Wait()
verifywait.Wait()
close(progressChan) // after verify (uses spinner) but before check
progresswait.Wait()
pruneCheckVerified(prunableObjects, reachableObjects, verifiedObjects)
} else {
close(progressChan)
progresswait.Wait()
}
if len(prunableObjects) == 0 {
Print("Nothing to prune")
return
}
if dryRun {
Print("%d files would be pruned (%v)", len(prunableObjects), humanizeBytes(totalSize))
if verbose {
Print(verboseOutput.String())
}
} else {
Print("Pruning %d files, (%v)", len(prunableObjects), humanizeBytes(totalSize))
if verbose {
Print(verboseOutput.String())
}
pruneDeleteFiles(prunableObjects)
}
}
func fetchCommand(cmd *cobra.Command, args []string) {
var ref string
var err error
if len(args) == 1 {
ref = args[0]
} else {
ref, err = git.CurrentRef()
if err != nil {
Panic(err, "Could not fetch")
}
}
pointers, err := lfs.ScanRefs(ref, "")
if err != nil {
Panic(err, "Could not scan for Git LFS files")
}
q := lfs.NewDownloadQueue(lfs.Config.ConcurrentTransfers(), len(pointers))
for _, p := range pointers {
q.Add(lfs.NewDownloadable(p))
}
target, err := git.ResolveRef(ref)
if err != nil {
Panic(err, "Could not resolve git ref")
}
current, err := git.CurrentRef()
if err != nil {
Panic(err, "Could not fetch the current git ref")
}
if target == current {
// We just downloaded the files for the current ref, we can copy them into
// the working directory and update the git index. We're doing this in a
// goroutine so they can be copied as they come in, for efficiency.
watch := q.Watch()
go func() {
files := make(map[string]*lfs.WrappedPointer, len(pointers))
for _, pointer := range pointers {
files[pointer.Oid] = pointer
}
// Fire up the update-index command
cmd := exec.Command("git", "update-index", "-q", "--refresh", "--stdin")
stdin, err := cmd.StdinPipe()
if err != nil {
Panic(err, "Could not update the index")
}
if err := cmd.Start(); err != nil {
Panic(err, "Could not update the index")
}
// As files come in, write them to the wd and update the index
for oid := range watch {
pointer, ok := files[oid]
if !ok {
continue
}
file, err := os.Create(pointer.Name)
if err != nil {
Panic(err, "Could not create working directory file")
}
if err := lfs.PointerSmudge(file, pointer.Pointer, pointer.Name, nil); err != nil {
Panic(err, "Could not write working directory file")
}
file.Close()
stdin.Write([]byte(pointer.Name + "\n"))
}
stdin.Close()
if err := cmd.Wait(); err != nil {
Panic(err, "Error updating the git index")
}
}()
processQueue := time.Now()
q.Process()
tracerx.PerformanceSince("process queue", processQueue)
}
}
// Fetch and report completion of each OID to a channel (optional, pass nil to skip)
// Returns true if all completed with no errors, false if errors were written to stderr/log
func fetchAndReportToChan(allpointers []*lfs.WrappedPointer, include, exclude []string, out chan<- *lfs.WrappedPointer) bool {
// Lazily initialize the current remote.
if len(cfg.CurrentRemote) == 0 {
// Actively find the default remote, don't just assume origin
defaultRemote, err := git.DefaultRemote()
if err != nil {
Exit("No default remote")
}
cfg.CurrentRemote = defaultRemote
}
ready, pointers, totalSize := readyAndMissingPointers(allpointers, include, exclude)
q := lfs.NewDownloadQueue(len(pointers), totalSize, false)
if out != nil {
// If we already have it, or it won't be fetched
// report it to chan immediately to support pull/checkout
for _, p := range ready {
out <- p
}
dlwatch := q.Watch()
go func() {
// fetch only reports single OID, but OID *might* be referenced by multiple
// WrappedPointers if same content is at multiple paths, so map oid->slice
oidToPointers := make(map[string][]*lfs.WrappedPointer, len(pointers))
for _, pointer := range pointers {
plist := oidToPointers[pointer.Oid]
oidToPointers[pointer.Oid] = append(plist, pointer)
}
for oid := range dlwatch {
plist, ok := oidToPointers[oid]
if !ok {
continue
}
for _, p := range plist {
out <- p
}
}
close(out)
}()
}
for _, p := range pointers {
tracerx.Printf("fetch %v [%v]", p.Name, p.Oid)
q.Add(lfs.NewDownloadable(p))
}
processQueue := time.Now()
q.Wait()
tracerx.PerformanceSince("process queue", processQueue)
ok := true
for _, err := range q.Errors() {
ok = false
FullError(err)
}
return ok
}
