Add basic repository lfs management (#7199)

This PR adds basic repository LFS management UI including the ability to find all possible pointers within the repository. Locks are not managed at present but would be addable through some simple additions.

* Add basic repository lfs management
* add auto-associate function
* Add functionality to find commits with this lfs file
* Add link to find commits on the lfs file view
* Adjust commit view to state the likely branch causing the commit
* Only read Oid from database

Author: zeripath

models/lfs.go

@@ -8,6 +8,8 @@ import (
 	"io"
 
 	"code.gitea.io/gitea/modules/timeutil"
+
+	"xorm.io/builder"
 )
 
 // LFSMetaObject stores metadata for LFS tracked files.
@@ -106,19 +108,91 @@ func (repo *Repository) GetLFSMetaObjectByOid(oid string) (*LFSMetaObject, error
 
 // RemoveLFSMetaObjectByOid removes a LFSMetaObject entry from database by its OID.
 // It may return ErrLFSObjectNotExist or a database error.
-func (repo *Repository) RemoveLFSMetaObjectByOid(oid string) error {
+func (repo *Repository) RemoveLFSMetaObjectByOid(oid string) (int64, error) {
 	if len(oid) == 0 {
-		return ErrLFSObjectNotExist
+		return 0, ErrLFSObjectNotExist
 	}
 
 	sess := x.NewSession()
 	defer sess.Close()
 	if err := sess.Begin(); err != nil {
-		return err
+		return -1, err
 	}
 
 	m := &LFSMetaObject{Oid: oid, RepositoryID: repo.ID}
 	if _, err := sess.Delete(m); err != nil {
+		return -1, err
+	}
+
+	count, err := sess.Count(&LFSMetaObject{Oid: oid})
+	if err != nil {
+		return count, err
+	}
+
+	return count, sess.Commit()
+}
+
+// GetLFSMetaObjects returns all LFSMetaObjects associated with a repository
+func (repo *Repository) GetLFSMetaObjects(page, pageSize int) ([]*LFSMetaObject, error) {
+	sess := x.NewSession()
+	defer sess.Close()
+
+	if page >= 0 && pageSize > 0 {
+		start := 0
+		if page > 0 {
+			start = (page - 1) * pageSize
+		}
+		sess.Limit(pageSize, start)
+	}
+	lfsObjects := make([]*LFSMetaObject, 0, pageSize)
+	return lfsObjects, sess.Find(&lfsObjects, &LFSMetaObject{RepositoryID: repo.ID})
+}
+
+// CountLFSMetaObjects returns a count of all LFSMetaObjects associated with a repository
+func (repo *Repository) CountLFSMetaObjects() (int64, error) {
+	return x.Count(&LFSMetaObject{RepositoryID: repo.ID})
+}
+
+// LFSObjectAccessible checks if a provided Oid is accessible to the user
+func LFSObjectAccessible(user *User, oid string) (bool, error) {
+	if user.IsAdmin {
+		count, err := x.Count(&LFSMetaObject{Oid: oid})
+		return (count > 0), err
+	}
+	cond := accessibleRepositoryCondition(user.ID)
+	count, err := x.Where(cond).Join("INNER", "repository", "`lfs_meta_object`.repository_id = `repository`.id").Count(&LFSMetaObject{Oid: oid})
+	return (count > 0), err
+}
+
+// LFSAutoAssociate auto associates accessible LFSMetaObjects
+func LFSAutoAssociate(metas []*LFSMetaObject, user *User, repoID int64) error {
+	sess := x.NewSession()
+	defer sess.Close()
+	if err := sess.Begin(); err != nil {
+		return err
+	}
+
+	oids := make([]interface{}, len(metas))
+	oidMap := make(map[string]*LFSMetaObject, len(metas))
+	for i, meta := range metas {
+		oids[i] = meta.Oid
+		oidMap[meta.Oid] = meta
+	}
+
+	cond := builder.NewCond()
+	if !user.IsAdmin {
+		cond = builder.In("`lfs_meta_object`.repository_id",
+			builder.Select("`repository`.id").From("repository").Where(accessibleRepositoryCondition(user.ID)))
+	}
+	newMetas := make([]*LFSMetaObject, 0, len(metas))
+	if err := sess.Cols("oid").Where(cond).In("oid", oids...).GroupBy("oid").Find(&newMetas); err != nil {
+		return err
+	}
+	for i := range newMetas {
+		newMetas[i].Size = oidMap[newMetas[i].Oid].Size
+		newMetas[i].RepositoryID = repoID
+	}
+	if _, err := sess.InsertMulti(newMetas); err != nil {
 		return err
 	}
 

models/repo_list.go

@@ -176,28 +176,7 @@ func SearchRepository(opts *SearchRepoOptions) (RepositoryList, int64, error) {
 	if opts.Private {
 		if !opts.UserIsAdmin && opts.UserID != 0 && opts.UserID != opts.OwnerID {
 			// OK we're in the context of a User
-			// We should be Either
-			cond = cond.And(builder.Or(
-				// 1. Be able to see all non-private repositories that either:
-				cond.And(
-					builder.Eq{"is_private": false},
-					builder.Or(
-						//   A. Aren't in organisations  __OR__
-						builder.NotIn("owner_id", builder.Select("id").From("`user`").Where(builder.Eq{"type": UserTypeOrganization})),
-						//   B. Isn't a private organisation. (Limited is OK because we're logged in)
-						builder.NotIn("owner_id", builder.Select("id").From("`user`").Where(builder.Eq{"visibility": structs.VisibleTypePrivate}))),
-				),
-				// 2. Be able to see all repositories that we have access to
-				builder.In("id", builder.Select("repo_id").
-					From("`access`").
-					Where(builder.And(
-						builder.Eq{"user_id": opts.UserID},
-						builder.Gt{"mode": int(AccessModeNone)}))),
-				// 3. Be able to see all repositories that we are in a team
-				builder.In("id", builder.Select("`team_repo`.repo_id").
-					From("team_repo").
-					Where(builder.Eq{"`team_user`.uid": opts.UserID}).
-					Join("INNER", "team_user", "`team_user`.team_id = `team_repo`.team_id"))))
+			cond = cond.And(accessibleRepositoryCondition(opts.UserID))
 		}
 	} else {
 		// Not looking at private organisations
@@ -316,6 +295,31 @@ func SearchRepository(opts *SearchRepoOptions) (RepositoryList, int64, error) {
 	return repos, count, nil
 }
 
+// accessibleRepositoryCondition takes a user a returns a condition for checking if a repository is accessible
+func accessibleRepositoryCondition(userID int64) builder.Cond {
+	return builder.Or(
+		// 1. Be able to see all non-private repositories that either:
+		builder.And(
+			builder.Eq{"`repository`.is_private": false},
+			builder.Or(
+				//   A. Aren't in organisations  __OR__
+				builder.NotIn("`repository`.owner_id", builder.Select("id").From("`user`").Where(builder.Eq{"type": UserTypeOrganization})),
+				//   B. Isn't a private organisation. (Limited is OK because we're logged in)
+				builder.NotIn("`repository`.owner_id", builder.Select("id").From("`user`").Where(builder.Eq{"visibility": structs.VisibleTypePrivate}))),
+		),
+		// 2. Be able to see all repositories that we have access to
+		builder.In("`repository`.id", builder.Select("repo_id").
+			From("`access`").
+			Where(builder.And(
+				builder.Eq{"user_id": userID},
+				builder.Gt{"mode": int(AccessModeNone)}))),
+		// 3. Be able to see all repositories that we are in a team
+		builder.In("`repository`.id", builder.Select("`team_repo`.repo_id").
+			From("team_repo").
+			Where(builder.Eq{"`team_user`.uid": userID}).
+			Join("INNER", "team_user", "`team_user`.team_id = `team_repo`.team_id")))
+}
+
 // SearchRepositoryByName takes keyword and part of repository name to search,
 // it returns results in given range and number of total results.
 func SearchRepositoryByName(opts *SearchRepoOptions) (RepositoryList, int64, error) {

modules/git/pipeline/catfile.go

@@ -0,0 +1,94 @@
+// Copyright 2019 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 pipeline
+
+import (
+	"bufio"
+	"bytes"
+	"fmt"
+	"io"
+	"strconv"
+	"strings"
+	"sync"
+
+	"code.gitea.io/gitea/modules/git"
+	"code.gitea.io/gitea/modules/log"
+)
+
+// CatFileBatchCheck runs cat-file with --batch-check
+func CatFileBatchCheck(shasToCheckReader *io.PipeReader, catFileCheckWriter *io.PipeWriter, wg *sync.WaitGroup, tmpBasePath string) {
+	defer wg.Done()
+	defer shasToCheckReader.Close()
+	defer catFileCheckWriter.Close()
+
+	stderr := new(bytes.Buffer)
+	var errbuf strings.Builder
+	cmd := git.NewCommand("cat-file", "--batch-check")
+	if err := cmd.RunInDirFullPipeline(tmpBasePath, catFileCheckWriter, stderr, shasToCheckReader); err != nil {
+		_ = catFileCheckWriter.CloseWithError(fmt.Errorf("git cat-file --batch-check [%s]: %v - %s", tmpBasePath, err, errbuf.String()))
+	}
+}
+
+// CatFileBatchCheckAllObjects runs cat-file with --batch-check --batch-all
+func CatFileBatchCheckAllObjects(catFileCheckWriter *io.PipeWriter, wg *sync.WaitGroup, tmpBasePath string, errChan chan<- error) {
+	defer wg.Done()
+	defer catFileCheckWriter.Close()
+
+	stderr := new(bytes.Buffer)
+	var errbuf strings.Builder
+	cmd := git.NewCommand("cat-file", "--batch-check", "--batch-all-objects")
+	if err := cmd.RunInDirPipeline(tmpBasePath, catFileCheckWriter, stderr); err != nil {
+		log.Error("git cat-file --batch-check --batch-all-object [%s]: %v - %s", tmpBasePath, err, errbuf.String())
+		err = fmt.Errorf("git cat-file --batch-check --batch-all-object [%s]: %v - %s", tmpBasePath, err, errbuf.String())
+		_ = catFileCheckWriter.CloseWithError(err)
+		errChan <- err
+	}
+}
+
+// CatFileBatch runs cat-file --batch
+func CatFileBatch(shasToBatchReader *io.PipeReader, catFileBatchWriter *io.PipeWriter, wg *sync.WaitGroup, tmpBasePath string) {
+	defer wg.Done()
+	defer shasToBatchReader.Close()
+	defer catFileBatchWriter.Close()
+
+	stderr := new(bytes.Buffer)
+	var errbuf strings.Builder
+	if err := git.NewCommand("cat-file", "--batch").RunInDirFullPipeline(tmpBasePath, catFileBatchWriter, stderr, shasToBatchReader); err != nil {
+		_ = shasToBatchReader.CloseWithError(fmt.Errorf("git rev-list [%s]: %v - %s", tmpBasePath, err, errbuf.String()))
+	}
+}
+
+// BlobsLessThan1024FromCatFileBatchCheck reads a pipeline from cat-file --batch-check and returns the blobs <1024 in size
+func BlobsLessThan1024FromCatFileBatchCheck(catFileCheckReader *io.PipeReader, shasToBatchWriter *io.PipeWriter, wg *sync.WaitGroup) {
+	defer wg.Done()
+	defer catFileCheckReader.Close()
+	scanner := bufio.NewScanner(catFileCheckReader)
+	defer func() {
+		_ = shasToBatchWriter.CloseWithError(scanner.Err())
+	}()
+	for scanner.Scan() {
+		line := scanner.Text()
+		if len(line) == 0 {
+			continue
+		}
+		fields := strings.Split(line, " ")
+		if len(fields) < 3 || fields[1] != "blob" {
+			continue
+		}
+		size, _ := strconv.Atoi(fields[2])
+		if size > 1024 {
+			continue
+		}
+		toWrite := []byte(fields[0] + "\n")
+		for len(toWrite) > 0 {
+			n, err := shasToBatchWriter.Write(toWrite)
+			if err != nil {
+				_ = catFileCheckReader.CloseWithError(err)
+				break
+			}
+			toWrite = toWrite[n:]
+		}
+	}
+}

modules/git/pipeline/namerev.go

@@ -0,0 +1,28 @@
+// Copyright 2019 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 pipeline
+
+import (
+	"bytes"
+	"fmt"
+	"io"
+	"strings"
+	"sync"
+
+	"code.gitea.io/gitea/modules/git"
+)
+
+// NameRevStdin runs name-rev --stdin
+func NameRevStdin(shasToNameReader *io.PipeReader, nameRevStdinWriter *io.PipeWriter, wg *sync.WaitGroup, tmpBasePath string) {
+	defer wg.Done()
+	defer shasToNameReader.Close()
+	defer nameRevStdinWriter.Close()
+
+	stderr := new(bytes.Buffer)
+	var errbuf strings.Builder
+	if err := git.NewCommand("name-rev", "--stdin", "--name-only", "--always").RunInDirFullPipeline(tmpBasePath, nameRevStdinWriter, stderr, shasToNameReader); err != nil {
+		_ = shasToNameReader.CloseWithError(fmt.Errorf("git name-rev [%s]: %v - %s", tmpBasePath, err, errbuf.String()))
+	}
+}

modules/git/pipeline/revlist.go

@@ -0,0 +1,75 @@
+// Copyright 2019 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 pipeline
+
+import (
+	"bufio"
+	"bytes"
+	"fmt"
+	"io"
+	"strings"
+	"sync"
+
+	"code.gitea.io/gitea/modules/git"
+	"code.gitea.io/gitea/modules/log"
+)
+
+// RevListAllObjects runs rev-list --objects --all and writes to a pipewriter
+func RevListAllObjects(revListWriter *io.PipeWriter, wg *sync.WaitGroup, basePath string, errChan chan<- error) {
+	defer wg.Done()
+	defer revListWriter.Close()
+
+	stderr := new(bytes.Buffer)
+	var errbuf strings.Builder
+	cmd := git.NewCommand("rev-list", "--objects", "--all")
+	if err := cmd.RunInDirPipeline(basePath, revListWriter, stderr); err != nil {
+		log.Error("git rev-list --objects --all [%s]: %v - %s", basePath, err, errbuf.String())
+		err = fmt.Errorf("git rev-list --objects --all [%s]: %v - %s", basePath, err, errbuf.String())
+		_ = revListWriter.CloseWithError(err)
+		errChan <- err
+	}
+}
+
+// RevListObjects run rev-list --objects from headSHA to baseSHA
+func RevListObjects(revListWriter *io.PipeWriter, wg *sync.WaitGroup, tmpBasePath, headSHA, baseSHA string, errChan chan<- error) {
+	defer wg.Done()
+	defer revListWriter.Close()
+	stderr := new(bytes.Buffer)
+	var errbuf strings.Builder
+	cmd := git.NewCommand("rev-list", "--objects", headSHA, "--not", baseSHA)
+	if err := cmd.RunInDirPipeline(tmpBasePath, revListWriter, stderr); err != nil {
+		log.Error("git rev-list [%s]: %v - %s", tmpBasePath, err, errbuf.String())
+		errChan <- fmt.Errorf("git rev-list [%s]: %v - %s", tmpBasePath, err, errbuf.String())
+	}
+}
+
+// BlobsFromRevListObjects reads a RevListAllObjects and only selects blobs
+func BlobsFromRevListObjects(revListReader *io.PipeReader, shasToCheckWriter *io.PipeWriter, wg *sync.WaitGroup) {
+	defer wg.Done()
+	defer revListReader.Close()
+	scanner := bufio.NewScanner(revListReader)
+	defer func() {
+		_ = shasToCheckWriter.CloseWithError(scanner.Err())
+	}()
+	for scanner.Scan() {
+		line := scanner.Text()
+		if len(line) == 0 {
+			continue
+		}
+		fields := strings.Split(line, " ")
+		if len(fields) < 2 || len(fields[1]) == 0 {
+			continue
+		}
+		toWrite := []byte(fields[0] + "\n")
+		for len(toWrite) > 0 {
+			n, err := shasToCheckWriter.Write(toWrite)
+			if err != nil {
+				_ = revListReader.CloseWithError(err)
+				break
+			}
+			toWrite = toWrite[n:]
+		}
+	}
+}

modules/git/repo.go

@@ -117,6 +117,11 @@ func OpenRepository(repoPath string) (*Repository, error) {
 	}, nil
 }
 
+// GoGitRepo gets the go-git repo representation
+func (repo *Repository) GoGitRepo() *gogit.Repository {
+	return repo.gogitRepo
+}
+
 // IsEmpty Check if repository is empty.
 func (repo *Repository) IsEmpty() (bool, error) {
 	var errbuf strings.Builder

modules/lfs/server.go

@@ -332,7 +332,7 @@ func PutHandler(ctx *context.Context) {
 	if err := contentStore.Put(meta, bodyReader); err != nil {
 		ctx.Resp.WriteHeader(500)
 		fmt.Fprintf(ctx.Resp, `{"message":"%s"}`, err)
-		if err = repository.RemoveLFSMetaObjectByOid(rv.Oid); err != nil {
+		if _, err = repository.RemoveLFSMetaObjectByOid(rv.Oid); err != nil {
 			log.Error("RemoveLFSMetaObjectByOid: %v", err)
 		}
 		return

modules/repofiles/update.go

@@ -385,7 +385,7 @@ func CreateOrUpdateRepoFile(repo *models.Repository, doer *models.User, opts *Up
 		contentStore := &lfs.ContentStore{BasePath: setting.LFS.ContentPath}
 		if !contentStore.Exists(lfsMetaObject) {
 			if err := contentStore.Put(lfsMetaObject, strings.NewReader(opts.Content)); err != nil {
-				if err2 := repo.RemoveLFSMetaObjectByOid(lfsMetaObject.Oid); err2 != nil {
+				if _, err2 := repo.RemoveLFSMetaObjectByOid(lfsMetaObject.Oid); err2 != nil {
 					return nil, fmt.Errorf("Error whilst removing failed inserted LFS object %s: %v (Prev Error: %v)", lfsMetaObject.Oid, err2, err)
 				}
 				return nil, err