github.com/gitbundle/modules@v0.0.0-20231025071548-85b91c5c3b01/lfs/pointer_scanner_nogogit.go (about)

     1  // Copyright 2023 The GitBundle Inc. All rights reserved.
     2  // Copyright 2017 The Gitea Authors. All rights reserved.
     3  // Use of this source code is governed by a MIT-style
     4  // license that can be found in the LICENSE file.
     5  
     6  //go:build !gogit
     7  
     8  package lfs
     9  
    10  import (
    11  	"bufio"
    12  	"context"
    13  	"io"
    14  	"strconv"
    15  	"strings"
    16  	"sync"
    17  
    18  	"github.com/gitbundle/modules/git"
    19  	"github.com/gitbundle/modules/git/pipeline"
    20  )
    21  
    22  // SearchPointerBlobs scans the whole repository for LFS pointer files
    23  func SearchPointerBlobs(ctx context.Context, repo *git.Repository, pointerChan chan<- PointerBlob, errChan chan<- error) {
    24  	basePath := repo.Path
    25  
    26  	catFileCheckReader, catFileCheckWriter := io.Pipe()
    27  	shasToBatchReader, shasToBatchWriter := io.Pipe()
    28  	catFileBatchReader, catFileBatchWriter := io.Pipe()
    29  
    30  	wg := sync.WaitGroup{}
    31  	wg.Add(4)
    32  
    33  	// Create the go-routines in reverse order.
    34  
    35  	// 4. Take the output of cat-file --batch and check if each file in turn
    36  	// to see if they're pointers to files in the LFS store
    37  	go createPointerResultsFromCatFileBatch(ctx, catFileBatchReader, &wg, pointerChan)
    38  
    39  	// 3. Take the shas of the blobs and batch read them
    40  	go pipeline.CatFileBatch(ctx, shasToBatchReader, catFileBatchWriter, &wg, basePath)
    41  
    42  	// 2. From the provided objects restrict to blobs <=1k
    43  	go pipeline.BlobsLessThan1024FromCatFileBatchCheck(catFileCheckReader, shasToBatchWriter, &wg)
    44  
    45  	// 1. Run batch-check on all objects in the repository
    46  	if git.CheckGitVersionAtLeast("2.6.0") != nil {
    47  		revListReader, revListWriter := io.Pipe()
    48  		shasToCheckReader, shasToCheckWriter := io.Pipe()
    49  		wg.Add(2)
    50  		go pipeline.CatFileBatchCheck(ctx, shasToCheckReader, catFileCheckWriter, &wg, basePath)
    51  		go pipeline.BlobsFromRevListObjects(revListReader, shasToCheckWriter, &wg)
    52  		go pipeline.RevListAllObjects(ctx, revListWriter, &wg, basePath, errChan)
    53  	} else {
    54  		go pipeline.CatFileBatchCheckAllObjects(ctx, catFileCheckWriter, &wg, basePath, errChan)
    55  	}
    56  	wg.Wait()
    57  
    58  	close(pointerChan)
    59  	close(errChan)
    60  }
    61  
    62  func createPointerResultsFromCatFileBatch(ctx context.Context, catFileBatchReader *io.PipeReader, wg *sync.WaitGroup, pointerChan chan<- PointerBlob) {
    63  	defer wg.Done()
    64  	defer catFileBatchReader.Close()
    65  
    66  	bufferedReader := bufio.NewReader(catFileBatchReader)
    67  	buf := make([]byte, 1025)
    68  
    69  loop:
    70  	for {
    71  		select {
    72  		case <-ctx.Done():
    73  			break loop
    74  		default:
    75  		}
    76  
    77  		// File descriptor line: sha
    78  		sha, err := bufferedReader.ReadString(' ')
    79  		if err != nil {
    80  			_ = catFileBatchReader.CloseWithError(err)
    81  			break
    82  		}
    83  		sha = strings.TrimSpace(sha)
    84  		// Throw away the blob
    85  		if _, err := bufferedReader.ReadString(' '); err != nil {
    86  			_ = catFileBatchReader.CloseWithError(err)
    87  			break
    88  		}
    89  		sizeStr, err := bufferedReader.ReadString('\n')
    90  		if err != nil {
    91  			_ = catFileBatchReader.CloseWithError(err)
    92  			break
    93  		}
    94  		size, err := strconv.Atoi(sizeStr[:len(sizeStr)-1])
    95  		if err != nil {
    96  			_ = catFileBatchReader.CloseWithError(err)
    97  			break
    98  		}
    99  		pointerBuf := buf[:size+1]
   100  		if _, err := io.ReadFull(bufferedReader, pointerBuf); err != nil {
   101  			_ = catFileBatchReader.CloseWithError(err)
   102  			break
   103  		}
   104  		pointerBuf = pointerBuf[:size]
   105  		// Now we need to check if the pointerBuf is an LFS pointer
   106  		pointer, _ := ReadPointerFromBuffer(pointerBuf)
   107  		if !pointer.IsValid() {
   108  			continue
   109  		}
   110  
   111  		pointerChan <- PointerBlob{Hash: sha, Pointer: pointer}
   112  	}
   113  }