| 
 | 1 | +// Copyright 2019 The Gitea Authors. All rights reserved.  | 
 | 2 | +// Use of this source code is governed by a MIT-style  | 
 | 3 | +// license that can be found in the LICENSE file.  | 
 | 4 | + | 
 | 5 | +package code  | 
 | 6 | + | 
 | 7 | +import (  | 
 | 8 | +	"strconv"  | 
 | 9 | +	"strings"  | 
 | 10 | + | 
 | 11 | +	"code.gitea.io/gitea/models"  | 
 | 12 | +	"code.gitea.io/gitea/modules/git"  | 
 | 13 | +	"code.gitea.io/gitea/modules/log"  | 
 | 14 | +	"code.gitea.io/gitea/modules/setting"  | 
 | 15 | +)  | 
 | 16 | + | 
 | 17 | +type fileUpdate struct {  | 
 | 18 | +	Filename string  | 
 | 19 | +	BlobSha  string  | 
 | 20 | +}  | 
 | 21 | + | 
 | 22 | +// repoChanges changes (file additions/updates/removals) to a repo  | 
 | 23 | +type repoChanges struct {  | 
 | 24 | +	Updates          []fileUpdate  | 
 | 25 | +	RemovedFilenames []string  | 
 | 26 | +}  | 
 | 27 | + | 
 | 28 | +func getDefaultBranchSha(repo *models.Repository) (string, error) {  | 
 | 29 | +	stdout, err := git.NewCommand("show-ref", "-s", git.BranchPrefix+repo.DefaultBranch).RunInDir(repo.RepoPath())  | 
 | 30 | +	if err != nil {  | 
 | 31 | +		return "", err  | 
 | 32 | +	}  | 
 | 33 | +	return strings.TrimSpace(stdout), nil  | 
 | 34 | +}  | 
 | 35 | + | 
 | 36 | +// getRepoChanges returns changes to repo since last indexer update  | 
 | 37 | +func getRepoChanges(repo *models.Repository, revision string) (*repoChanges, error) {  | 
 | 38 | +	if err := repo.GetIndexerStatus(); err != nil {  | 
 | 39 | +		return nil, err  | 
 | 40 | +	}  | 
 | 41 | + | 
 | 42 | +	if len(repo.IndexerStatus.CommitSha) == 0 {  | 
 | 43 | +		return genesisChanges(repo, revision)  | 
 | 44 | +	}  | 
 | 45 | +	return nonGenesisChanges(repo, revision)  | 
 | 46 | +}  | 
 | 47 | + | 
 | 48 | +func isIndexable(entry *git.TreeEntry) bool {  | 
 | 49 | +	if !entry.IsRegular() && !entry.IsExecutable() {  | 
 | 50 | +		return false  | 
 | 51 | +	}  | 
 | 52 | +	name := strings.ToLower(entry.Name())  | 
 | 53 | +	for _, g := range setting.Indexer.ExcludePatterns {  | 
 | 54 | +		if g.Match(name) {  | 
 | 55 | +			return false  | 
 | 56 | +		}  | 
 | 57 | +	}  | 
 | 58 | +	for _, g := range setting.Indexer.IncludePatterns {  | 
 | 59 | +		if g.Match(name) {  | 
 | 60 | +			return true  | 
 | 61 | +		}  | 
 | 62 | +	}  | 
 | 63 | +	return len(setting.Indexer.IncludePatterns) == 0  | 
 | 64 | +}  | 
 | 65 | + | 
 | 66 | +// parseGitLsTreeOutput parses the output of a `git ls-tree -r --full-name` command  | 
 | 67 | +func parseGitLsTreeOutput(stdout []byte) ([]fileUpdate, error) {  | 
 | 68 | +	entries, err := git.ParseTreeEntries(stdout)  | 
 | 69 | +	if err != nil {  | 
 | 70 | +		return nil, err  | 
 | 71 | +	}  | 
 | 72 | +	var idxCount = 0  | 
 | 73 | +	updates := make([]fileUpdate, len(entries))  | 
 | 74 | +	for _, entry := range entries {  | 
 | 75 | +		if isIndexable(entry) {  | 
 | 76 | +			updates[idxCount] = fileUpdate{  | 
 | 77 | +				Filename: entry.Name(),  | 
 | 78 | +				BlobSha:  entry.ID.String(),  | 
 | 79 | +			}  | 
 | 80 | +			idxCount++  | 
 | 81 | +		}  | 
 | 82 | +	}  | 
 | 83 | +	return updates[:idxCount], nil  | 
 | 84 | +}  | 
 | 85 | + | 
 | 86 | +// genesisChanges get changes to add repo to the indexer for the first time  | 
 | 87 | +func genesisChanges(repo *models.Repository, revision string) (*repoChanges, error) {  | 
 | 88 | +	var changes repoChanges  | 
 | 89 | +	stdout, err := git.NewCommand("ls-tree", "--full-tree", "-r", revision).  | 
 | 90 | +		RunInDirBytes(repo.RepoPath())  | 
 | 91 | +	if err != nil {  | 
 | 92 | +		return nil, err  | 
 | 93 | +	}  | 
 | 94 | +	changes.Updates, err = parseGitLsTreeOutput(stdout)  | 
 | 95 | +	return &changes, err  | 
 | 96 | +}  | 
 | 97 | + | 
 | 98 | +// nonGenesisChanges get changes since the previous indexer update  | 
 | 99 | +func nonGenesisChanges(repo *models.Repository, revision string) (*repoChanges, error) {  | 
 | 100 | +	diffCmd := git.NewCommand("diff", "--name-status",  | 
 | 101 | +		repo.IndexerStatus.CommitSha, revision)  | 
 | 102 | +	stdout, err := diffCmd.RunInDir(repo.RepoPath())  | 
 | 103 | +	if err != nil {  | 
 | 104 | +		// previous commit sha may have been removed by a force push, so  | 
 | 105 | +		// try rebuilding from scratch  | 
 | 106 | +		log.Warn("git diff: %v", err)  | 
 | 107 | +		if err = indexer.Delete(repo.ID); err != nil {  | 
 | 108 | +			return nil, err  | 
 | 109 | +		}  | 
 | 110 | +		return genesisChanges(repo, revision)  | 
 | 111 | +	}  | 
 | 112 | +	var changes repoChanges  | 
 | 113 | +	updatedFilenames := make([]string, 0, 10)  | 
 | 114 | +	for _, line := range strings.Split(stdout, "\n") {  | 
 | 115 | +		line = strings.TrimSpace(line)  | 
 | 116 | +		if len(line) == 0 {  | 
 | 117 | +			continue  | 
 | 118 | +		}  | 
 | 119 | +		filename := strings.TrimSpace(line[1:])  | 
 | 120 | +		if len(filename) == 0 {  | 
 | 121 | +			continue  | 
 | 122 | +		} else if filename[0] == '"' {  | 
 | 123 | +			filename, err = strconv.Unquote(filename)  | 
 | 124 | +			if err != nil {  | 
 | 125 | +				return nil, err  | 
 | 126 | +			}  | 
 | 127 | +		}  | 
 | 128 | + | 
 | 129 | +		switch status := line[0]; status {  | 
 | 130 | +		case 'M', 'A':  | 
 | 131 | +			updatedFilenames = append(updatedFilenames, filename)  | 
 | 132 | +		case 'D':  | 
 | 133 | +			changes.RemovedFilenames = append(changes.RemovedFilenames, filename)  | 
 | 134 | +		default:  | 
 | 135 | +			log.Warn("Unrecognized status: %c (line=%s)", status, line)  | 
 | 136 | +		}  | 
 | 137 | +	}  | 
 | 138 | + | 
 | 139 | +	cmd := git.NewCommand("ls-tree", "--full-tree", revision, "--")  | 
 | 140 | +	cmd.AddArguments(updatedFilenames...)  | 
 | 141 | +	lsTreeStdout, err := cmd.RunInDirBytes(repo.RepoPath())  | 
 | 142 | +	if err != nil {  | 
 | 143 | +		return nil, err  | 
 | 144 | +	}  | 
 | 145 | +	changes.Updates, err = parseGitLsTreeOutput(lsTreeStdout)  | 
 | 146 | +	return &changes, err  | 
 | 147 | +}  | 
0 commit comments