Code/repo search (#2582)

Indexed search of repository contents (for default branch only)

8 files changed, 367 insertions, 10 deletions

diff --git a/models/fixtures/repo_indexer_status.yml b/models/fixtures/repo_indexer_status.yml
new file mode 100644
index 0000000000..ca780a73aa
--- /dev/null
+++ b/models/fixtures/repo_indexer_status.yml
@@ -0,0 +1 @@
+[] # empty
diff --git a/models/migrations/migrations.go b/models/migrations/migrations.go
index f1cb1a67ed..ba27568fd4 100644
--- a/models/migrations/migrations.go
+++ b/models/migrations/migrations.go
@@ -144,6 +144,8 @@ var migrations = []Migration{
 	NewMigration("remove organization watch repositories", removeOrganizationWatchRepo),
 	// v47 -> v48
 	NewMigration("add deleted branches", addDeletedBranch),
+	// v48 -> v49
+	NewMigration("add repo indexer status", addRepoIndexerStatus),
 }
 
 // Migrate database to current version
diff --git a/models/migrations/v48.go b/models/migrations/v48.go
new file mode 100644
index 0000000000..6cea66b5ac
--- /dev/null
+++ b/models/migrations/v48.go
@@ -0,0 +1,25 @@
+// Copyright 2017 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package migrations
+
+import (
+	"fmt"
+
+	"github.com/go-xorm/xorm"
+)
+
+func addRepoIndexerStatus(x *xorm.Engine) error {
+	// RepoIndexerStatus see models/repo_indexer.go
+	type RepoIndexerStatus struct {
+		ID        int64  `xorm:"pk autoincr"`
+		RepoID    int64  `xorm:"INDEX NOT NULL"`
+		CommitSha string `xorm:"VARCHAR(40)"`
+	}
+
+	if err := x.Sync2(new(RepoIndexerStatus)); err != nil {
+		return fmt.Errorf("Sync2: %v", err)
+	}
+	return nil
+}
diff --git a/models/models.go b/models/models.go
index 66a7d97872..853b9799ed 100644
--- a/models/models.go
+++ b/models/models.go
@@ -13,6 +13,10 @@ import (
 	"path"
 	"strings"
 
+	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/modules/util"
+
 	// Needed for the MySQL driver
 	_ "github.com/go-sql-driver/mysql"
 	"github.com/go-xorm/core"
@@ -23,9 +27,6 @@ import (
 
 	// Needed for the MSSSQL driver
 	_ "github.com/denisenkom/go-mssqldb"
-
-	"code.gitea.io/gitea/modules/log"
-	"code.gitea.io/gitea/modules/setting"
 )
 
 // Engine represents a xorm engine or session.
@@ -115,6 +116,7 @@ func init() {
 		new(Stopwatch),
 		new(TrackedTime),
 		new(DeletedBranch),
+		new(RepoIndexerStatus),
 	)
 
 	gonicNames := []string{"SSL", "UID"}
@@ -150,8 +152,13 @@ func LoadConfigs() {
 	DbCfg.Timeout = sec.Key("SQLITE_TIMEOUT").MustInt(500)
 
 	sec = setting.Cfg.Section("indexer")
-	setting.Indexer.IssuePath = sec.Key("ISSUE_INDEXER_PATH").MustString("indexers/issues.bleve")
+	setting.Indexer.IssuePath = absolutePath(
+		sec.Key("ISSUE_INDEXER_PATH").MustString("indexers/issues.bleve"))
+	setting.Indexer.RepoIndexerEnabled = sec.Key("REPO_INDEXER_ENABLED").MustBool(false)
+	setting.Indexer.RepoPath = absolutePath(
+		sec.Key("REPO_INDEXER_PATH").MustString("indexers/repos.bleve"))
 	setting.Indexer.UpdateQueueLength = sec.Key("UPDATE_BUFFER_LEN").MustInt(20)
+	setting.Indexer.MaxIndexerFileSize = sec.Key("MAX_FILE_SIZE").MustInt64(512 * 1024 * 1024)
 }
 
 // parsePostgreSQLHostPort parses given input in various forms defined in
@@ -336,3 +343,12 @@ func DumpDatabase(filePath string, dbType string) error {
 	}
 	return x.DumpTablesToFile(tbs, filePath)
 }
+
+// absolutePath make path absolute if it is relative
+func absolutePath(path string) string {
+	workDir, err := setting.WorkDir()
+	if err != nil {
+		log.Fatal(4, "Failed to get work directory: %v", err)
+	}
+	return util.EnsureAbsolutePath(path, workDir)
+}
diff --git a/models/repo.go b/models/repo.go
index eca71568ee..8d57ae51a5 100644
--- a/models/repo.go
+++ b/models/repo.go
@@ -205,10 +205,11 @@ type Repository struct {
 	ExternalMetas map[string]string `xorm:"-"`
 	Units         []*RepoUnit       `xorm:"-"`
 
-	IsFork   bool        `xorm:"INDEX NOT NULL DEFAULT false"`
-	ForkID   int64       `xorm:"INDEX"`
-	BaseRepo *Repository `xorm:"-"`
-	Size     int64       `xorm:"NOT NULL DEFAULT 0"`
+	IsFork        bool               `xorm:"INDEX NOT NULL DEFAULT false"`
+	ForkID        int64              `xorm:"INDEX"`
+	BaseRepo      *Repository        `xorm:"-"`
+	Size          int64              `xorm:"NOT NULL DEFAULT 0"`
+	IndexerStatus *RepoIndexerStatus `xorm:"-"`
 
 	Created     time.Time `xorm:"-"`
 	CreatedUnix int64     `xorm:"INDEX created"`
@@ -782,8 +783,10 @@ func UpdateLocalCopyBranch(repoPath, localPath, branch string) error {
 		if err != nil {
 			return fmt.Errorf("git fetch origin: %v", err)
 		}
-		if err := git.ResetHEAD(localPath, true, "origin/"+branch); err != nil {
-			return fmt.Errorf("git reset --hard origin/%s: %v", branch, err)
+		if len(branch) > 0 {
+			if err := git.ResetHEAD(localPath, true, "origin/"+branch); err != nil {
+				return fmt.Errorf("git reset --hard origin/%s: %v", branch, err)
+			}
 		}
 	}
 	return nil
@@ -989,6 +992,7 @@ func MigrateRepository(doer, u *User, opts MigrateRepoOptions) (*Repository, err
 		if err = SyncReleasesWithTags(repo, gitRepo); err != nil {
 			log.Error(4, "Failed to synchronize tags to releases for repository: %v", err)
 		}
+		UpdateRepoIndexer(repo)
 	}
 
 	if err = repo.UpdateSize(); err != nil {
@@ -1883,6 +1887,7 @@ func DeleteRepository(doer *User, uid, repoID int64) error {
 		go HookQueue.Add(repo.ID)
 	}
 
+	DeleteRepoFromIndexer(repo)
 	return nil
 }
 
diff --git a/models/repo_editor.go b/models/repo_editor.go
index 692fe8c51b..1adaa2c955 100644
--- a/models/repo_editor.go
+++ b/models/repo_editor.go
@@ -178,6 +178,8 @@ func (repo *Repository) UpdateRepoFile(doer *User, opts UpdateRepoFileOptions) (
 	if err != nil {
 		return fmt.Errorf("PushUpdate: %v", err)
 	}
+	UpdateRepoIndexer(repo)
+
 	return nil
 }
 
diff --git a/models/repo_indexer.go b/models/repo_indexer.go
new file mode 100644
index 0000000000..a6b049e081
--- /dev/null
+++ b/models/repo_indexer.go
@@ -0,0 +1,302 @@
+// Copyright 2017 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package models
+
+import (
+	"io/ioutil"
+	"os"
+	"path"
+	"strconv"
+	"strings"
+
+	"code.gitea.io/git"
+	"code.gitea.io/gitea/modules/base"
+	"code.gitea.io/gitea/modules/indexer"
+	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/setting"
+
+	"github.com/Unknwon/com"
+)
+
+// RepoIndexerStatus status of a repo's entry in the repo indexer
+// For now, implicitly refers to default branch
+type RepoIndexerStatus struct {
+	ID        int64  `xorm:"pk autoincr"`
+	RepoID    int64  `xorm:"INDEX"`
+	CommitSha string `xorm:"VARCHAR(40)"`
+}
+
+func (repo *Repository) getIndexerStatus() error {
+	if repo.IndexerStatus != nil {
+		return nil
+	}
+	status := &RepoIndexerStatus{RepoID: repo.ID}
+	has, err := x.Get(status)
+	if err != nil {
+		return err
+	} else if !has {
+		status.CommitSha = ""
+	}
+	repo.IndexerStatus = status
+	return nil
+}
+
+func (repo *Repository) updateIndexerStatus(sha string) error {
+	if err := repo.getIndexerStatus(); err != nil {
+		return err
+	}
+	if len(repo.IndexerStatus.CommitSha) == 0 {
+		repo.IndexerStatus.CommitSha = sha
+		_, err := x.Insert(repo.IndexerStatus)
+		return err
+	}
+	repo.IndexerStatus.CommitSha = sha
+	_, err := x.ID(repo.IndexerStatus.ID).Cols("commit_sha").
+		Update(repo.IndexerStatus)
+	return err
+}
+
+type repoIndexerOperation struct {
+	repo    *Repository
+	deleted bool
+}
+
+var repoIndexerOperationQueue chan repoIndexerOperation
+
+// InitRepoIndexer initialize the repo indexer
+func InitRepoIndexer() {
+	if !setting.Indexer.RepoIndexerEnabled {
+		return
+	}
+	indexer.InitRepoIndexer(populateRepoIndexer)
+	repoIndexerOperationQueue = make(chan repoIndexerOperation, setting.Indexer.UpdateQueueLength)
+	go processRepoIndexerOperationQueue()
+}
+
+// populateRepoIndexer populate the repo indexer with data
+func populateRepoIndexer() error {
+	log.Info("Populating repository indexer (this may take a while)")
+	for page := 1; ; page++ {
+		repos, _, err := SearchRepositoryByName(&SearchRepoOptions{
+			Page:     page,
+			PageSize: 10,
+			OrderBy:  SearchOrderByID,
+			Private:  true,
+		})
+		if err != nil {
+			return err
+		} else if len(repos) == 0 {
+			return nil
+		}
+		for _, repo := range repos {
+			if err = updateRepoIndexer(repo); err != nil {
+				// only log error, since this should not prevent
+				// gitea from starting up
+				log.Error(4, "updateRepoIndexer: repoID=%d, %v", repo.ID, err)
+			}
+		}
+	}
+}
+
+type updateBatch struct {
+	updates []indexer.RepoIndexerUpdate
+}
+
+func updateRepoIndexer(repo *Repository) error {
+	changes, err := getRepoChanges(repo)
+	if err != nil {
+		return err
+	} else if changes == nil {
+		return nil
+	}
+
+	batch := indexer.RepoIndexerBatch()
+	for _, filename := range changes.UpdatedFiles {
+		if err := addUpdate(filename, repo, batch); err != nil {
+			return err
+		}
+	}
+	for _, filename := range changes.RemovedFiles {
+		if err := addDelete(filename, repo, batch); err != nil {
+			return err
+		}
+	}
+	if err = batch.Flush(); err != nil {
+		return err
+	}
+	return updateLastIndexSync(repo)
+}
+
+// repoChanges changes (file additions/updates/removals) to a repo
+type repoChanges struct {
+	UpdatedFiles []string
+	RemovedFiles []string
+}
+
+// getRepoChanges returns changes to repo since last indexer update
+func getRepoChanges(repo *Repository) (*repoChanges, error) {
+	repoWorkingPool.CheckIn(com.ToStr(repo.ID))
+	defer repoWorkingPool.CheckOut(com.ToStr(repo.ID))
+
+	if err := repo.UpdateLocalCopyBranch(""); err != nil {
+		return nil, err
+	} else if !git.IsBranchExist(repo.LocalCopyPath(), repo.DefaultBranch) {
+		// repo does not have any commits yet, so nothing to update
+		return nil, nil
+	} else if err = repo.UpdateLocalCopyBranch(repo.DefaultBranch); err != nil {
+		return nil, err
+	} else if err = repo.getIndexerStatus(); err != nil {
+		return nil, err
+	}
+
+	if len(repo.IndexerStatus.CommitSha) == 0 {
+		return genesisChanges(repo)
+	}
+	return nonGenesisChanges(repo)
+}
+
+func addUpdate(filename string, repo *Repository, batch *indexer.Batch) error {
+	filepath := path.Join(repo.LocalCopyPath(), filename)
+	if stat, err := os.Stat(filepath); err != nil {
+		return err
+	} else if stat.Size() > setting.Indexer.MaxIndexerFileSize {
+		return nil
+	}
+	fileContents, err := ioutil.ReadFile(filepath)
+	if err != nil {
+		return err
+	} else if !base.IsTextFile(fileContents) {
+		return nil
+	}
+	return batch.Add(indexer.RepoIndexerUpdate{
+		Filepath: filename,
+		Op:       indexer.RepoIndexerOpUpdate,
+		Data: &indexer.RepoIndexerData{
+			RepoID:  repo.ID,
+			Content: string(fileContents),
+		},
+	})
+}
+
+func addDelete(filename string, repo *Repository, batch *indexer.Batch) error {
+	return batch.Add(indexer.RepoIndexerUpdate{
+		Filepath: filename,
+		Op:       indexer.RepoIndexerOpDelete,
+		Data: &indexer.RepoIndexerData{
+			RepoID: repo.ID,
+		},
+	})
+}
+
+// genesisChanges get changes to add repo to the indexer for the first time
+func genesisChanges(repo *Repository) (*repoChanges, error) {
+	var changes repoChanges
+	stdout, err := git.NewCommand("ls-files").RunInDir(repo.LocalCopyPath())
+	if err != nil {
+		return nil, err
+	}
+	for _, line := range strings.Split(stdout, "\n") {
+		filename := strings.TrimSpace(line)
+		if len(filename) == 0 {
+			continue
+		} else if filename[0] == '"' {
+			filename, err = strconv.Unquote(filename)
+			if err != nil {
+				return nil, err
+			}
+		}
+		changes.UpdatedFiles = append(changes.UpdatedFiles, filename)
+	}
+	return &changes, nil
+}
+
+// nonGenesisChanges get changes since the previous indexer update
+func nonGenesisChanges(repo *Repository) (*repoChanges, error) {
+	diffCmd := git.NewCommand("diff", "--name-status",
+		repo.IndexerStatus.CommitSha, "HEAD")
+	stdout, err := diffCmd.RunInDir(repo.LocalCopyPath())
+	if err != nil {
+		// previous commit sha may have been removed by a force push, so
+		// try rebuilding from scratch
+		if err = indexer.DeleteRepoFromIndexer(repo.ID); err != nil {
+			return nil, err
+		}
+		return genesisChanges(repo)
+	}
+	var changes repoChanges
+	for _, line := range strings.Split(stdout, "\n") {
+		line = strings.TrimSpace(line)
+		if len(line) == 0 {
+			continue
+		}
+		filename := strings.TrimSpace(line[1:])
+		if len(filename) == 0 {
+			continue
+		} else if filename[0] == '"' {
+			filename, err = strconv.Unquote(filename)
+			if err != nil {
+				return nil, err
+			}
+		}
+
+		switch status := line[0]; status {
+		case 'M', 'A':
+			changes.UpdatedFiles = append(changes.UpdatedFiles, filename)
+		case 'D':
+			changes.RemovedFiles = append(changes.RemovedFiles, filename)
+		default:
+			log.Warn("Unrecognized status: %c (line=%s)", status, line)
+		}
+	}
+	return &changes, nil
+}
+
+func updateLastIndexSync(repo *Repository) error {
+	stdout, err := git.NewCommand("rev-parse", "HEAD").RunInDir(repo.LocalCopyPath())
+	if err != nil {
+		return err
+	}
+	sha := strings.TrimSpace(stdout)
+	return repo.updateIndexerStatus(sha)
+}
+
+func processRepoIndexerOperationQueue() {
+	for {
+		op := <-repoIndexerOperationQueue
+		if op.deleted {
+			if err := indexer.DeleteRepoFromIndexer(op.repo.ID); err != nil {
+				log.Error(4, "DeleteRepoFromIndexer: %v", err)
+			}
+		} else {
+			if err := updateRepoIndexer(op.repo); err != nil {
+				log.Error(4, "updateRepoIndexer: %v", err)
+			}
+		}
+	}
+}
+
+// DeleteRepoFromIndexer remove all of a repository's entries from the indexer
+func DeleteRepoFromIndexer(repo *Repository) {
+	addOperationToQueue(repoIndexerOperation{repo: repo, deleted: true})
+}
+
+// UpdateRepoIndexer update a repository's entries in the indexer
+func UpdateRepoIndexer(repo *Repository) {
+	addOperationToQueue(repoIndexerOperation{repo: repo, deleted: false})
+}
+
+func addOperationToQueue(op repoIndexerOperation) {
+	if !setting.Indexer.RepoIndexerEnabled {
+		return
+	}
+	select {
+	case repoIndexerOperationQueue <- op:
+		break
+	default:
+		go func() {
+			repoIndexerOperationQueue <- op
+		}()
+	}
+}
diff --git a/models/update.go b/models/update.go
index f3bd6cce13..f91559d9ec 100644
--- a/models/update.go
+++ b/models/update.go
@@ -263,6 +263,10 @@ func pushUpdate(opts PushUpdateOptions) (repo *Repository, err error) {
 		commits = ListToPushCommits(l)
 	}
 
+	if opts.RefFullName == git.BranchPrefix+repo.DefaultBranch {
+		UpdateRepoIndexer(repo)
+	}
+
 	if err := CommitRepoAction(CommitRepoActionOptions{
 		PusherName:  opts.PusherName,
 		RepoOwnerID: owner.ID,