forgejo_backup/modules/lfs/pointer_scanner.go
Gusted b5e608f3e2 feat: bump the minimum required Git version from 2.0.0 to 2.34.1 (#8328)
- Resolves forgejo/discussions#324
- Remove all checks of `CheckGitVersionAtLeast` that checked for a version below 2.34.1
- The version was chosen because Debian stable supports 2.39.5 and Ubuntu 22.04 LTS supports 2.34.1

Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/8328
Reviewed-by: 0ko <0ko@noreply.codeberg.org>
Reviewed-by: Earl Warren <earl-warren@noreply.codeberg.org>
Co-authored-by: Gusted <postmaster@gusted.xyz>
Co-committed-by: Gusted <postmaster@gusted.xyz>
2025-06-29 00:44:18 +02:00

100 lines
2.7 KiB
Go

// Copyright 2021 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package lfs
import (
"bufio"
"context"
"io"
"strconv"
"strings"
"sync"
"forgejo.org/modules/git"
"forgejo.org/modules/git/pipeline"
)
// SearchPointerBlobs scans the whole repository for LFS pointer files
func SearchPointerBlobs(ctx context.Context, repo *git.Repository, pointerChan chan<- PointerBlob, errChan chan<- error) {
basePath := repo.Path
catFileCheckReader, catFileCheckWriter := io.Pipe()
shasToBatchReader, shasToBatchWriter := io.Pipe()
catFileBatchReader, catFileBatchWriter := io.Pipe()
wg := sync.WaitGroup{}
wg.Add(4)
// Create the go-routines in reverse order.
// 4. Take the output of cat-file --batch and check if each file in turn
// to see if they're pointers to files in the LFS store
go createPointerResultsFromCatFileBatch(ctx, catFileBatchReader, &wg, pointerChan)
// 3. Take the shas of the blobs and batch read them
go pipeline.CatFileBatch(ctx, shasToBatchReader, catFileBatchWriter, &wg, basePath)
// 2. From the provided objects restrict to blobs <=1k
go pipeline.BlobsLessThan1024FromCatFileBatchCheck(catFileCheckReader, shasToBatchWriter, &wg)
// 1. Run batch-check on all objects in the repository
go pipeline.CatFileBatchCheckAllObjects(ctx, catFileCheckWriter, &wg, basePath, errChan)
wg.Wait()
close(pointerChan)
close(errChan)
}
func createPointerResultsFromCatFileBatch(ctx context.Context, catFileBatchReader *io.PipeReader, wg *sync.WaitGroup, pointerChan chan<- PointerBlob) {
defer wg.Done()
defer catFileBatchReader.Close()
bufferedReader := bufio.NewReader(catFileBatchReader)
buf := make([]byte, 1025)
loop:
for {
select {
case <-ctx.Done():
break loop
default:
}
// File descriptor line: sha
sha, err := bufferedReader.ReadString(' ')
if err != nil {
_ = catFileBatchReader.CloseWithError(err)
break
}
sha = strings.TrimSpace(sha)
// Throw away the blob
if _, err := bufferedReader.ReadString(' '); err != nil {
_ = catFileBatchReader.CloseWithError(err)
break
}
sizeStr, err := bufferedReader.ReadString('\n')
if err != nil {
_ = catFileBatchReader.CloseWithError(err)
break
}
size, err := strconv.Atoi(sizeStr[:len(sizeStr)-1])
if err != nil {
_ = catFileBatchReader.CloseWithError(err)
break
}
pointerBuf := buf[:size+1]
if _, err := io.ReadFull(bufferedReader, pointerBuf); err != nil {
_ = catFileBatchReader.CloseWithError(err)
break
}
pointerBuf = pointerBuf[:size]
// Now we need to check if the pointerBuf is an LFS pointer
pointer, _ := ReadPointerFromBuffer(pointerBuf)
if !pointer.IsValid() {
continue
}
pointerChan <- PointerBlob{Hash: sha, Pointer: pointer}
}
}