Language statistics bar for repositories (#8037)

* Implementation for calculating language statistics

Impement saving code language statistics to database

Implement rendering langauge stats

Add primary laguage to show in repository list

Implement repository stats indexer queue

Add indexer test

Refactor to use queue module

* Do not timeout for queues

7 files changed, 274 insertions, 32 deletions

diff --git a/models/migrations/migrations.go b/models/migrations/migrations.go
index 2de2a45568..ce3f77ba4e 100644
--- a/models/migrations/migrations.go
+++ b/models/migrations/migrations.go
@@ -186,6 +186,8 @@ var migrations = []Migration{
 	NewMigration("Add some columns on review for migration", addReviewMigrateInfo),
 	// v126 -> v127
 	NewMigration("Fix topic repository count", fixTopicRepositoryCount),
+	// v127 -> v128
+	NewMigration("add repository code language statistics", addLanguageStats),
 }
 
 // Migrate database to current version
diff --git a/models/migrations/v127.go b/models/migrations/v127.go
new file mode 100644
index 0000000000..d8f0de4a6e
--- /dev/null
+++ b/models/migrations/v127.go
@@ -0,0 +1,45 @@
+// Copyright 2020 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"
+
+	"code.gitea.io/gitea/modules/timeutil"
+
+	"xorm.io/xorm"
+)
+
+func addLanguageStats(x *xorm.Engine) error {
+	// LanguageStat see models/repo_language_stats.go
+	type LanguageStat struct {
+		ID          int64 `xorm:"pk autoincr"`
+		RepoID      int64 `xorm:"UNIQUE(s) INDEX NOT NULL"`
+		CommitID    string
+		IsPrimary   bool
+		Language    string             `xorm:"VARCHAR(30) UNIQUE(s) INDEX NOT NULL"`
+		Percentage  float32            `xorm:"NUMERIC(5,2) NOT NULL DEFAULT 0"`
+		Color       string             `xorm:"-"`
+		CreatedUnix timeutil.TimeStamp `xorm:"INDEX CREATED"`
+	}
+
+	type RepoIndexerType int
+
+	// RepoIndexerStatus see models/repo_stats_indexer.go
+	type RepoIndexerStatus struct {
+		ID          int64           `xorm:"pk autoincr"`
+		RepoID      int64           `xorm:"INDEX(s)"`
+		CommitSha   string          `xorm:"VARCHAR(40)"`
+		IndexerType RepoIndexerType `xorm:"INDEX(s) NOT NULL DEFAULT 0"`
+	}
+
+	if err := x.Sync2(new(LanguageStat)); err != nil {
+		return fmt.Errorf("Sync2: %v", err)
+	}
+	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 239a9cf280..b84a179e37 100644
--- a/models/models.go
+++ b/models/models.go
@@ -116,6 +116,7 @@ func init() {
 		new(OAuth2AuthorizationCode),
 		new(OAuth2Grant),
 		new(Task),
+		new(LanguageStat),
 	)
 
 	gonicNames := []string{"SSL", "UID"}
diff --git a/models/repo.go b/models/repo.go
index 2d0f6fccbd..4bb5e3eb01 100644
--- a/models/repo.go
+++ b/models/repo.go
@@ -175,8 +175,9 @@ type Repository struct {
 	*Mirror    `xorm:"-"`
 	Status     RepositoryStatus `xorm:"NOT NULL DEFAULT 0"`
 
-	RenderingMetas map[string]string `xorm:"-"`
-	Units          []*RepoUnit       `xorm:"-"`
+	RenderingMetas  map[string]string `xorm:"-"`
+	Units           []*RepoUnit       `xorm:"-"`
+	PrimaryLanguage *LanguageStat     `xorm:"-"`
 
 	IsFork                          bool               `xorm:"INDEX NOT NULL DEFAULT false"`
 	ForkID                          int64              `xorm:"INDEX"`
@@ -185,7 +186,8 @@ type Repository struct {
 	TemplateID                      int64              `xorm:"INDEX"`
 	TemplateRepo                    *Repository        `xorm:"-"`
 	Size                            int64              `xorm:"NOT NULL DEFAULT 0"`
-	IndexerStatus                   *RepoIndexerStatus `xorm:"-"`
+	CodeIndexerStatus               *RepoIndexerStatus `xorm:"-"`
+	StatsIndexerStatus              *RepoIndexerStatus `xorm:"-"`
 	IsFsckEnabled                   bool               `xorm:"NOT NULL DEFAULT true"`
 	CloseIssuesViaCommitInAnyBranch bool               `xorm:"NOT NULL DEFAULT false"`
 	Topics                          []string           `xorm:"TEXT JSON"`
@@ -1504,6 +1506,7 @@ func DeleteRepository(doer *User, uid, repoID int64) error {
 		&Notification{RepoID: repoID},
 		&CommitStatus{RepoID: repoID},
 		&RepoIndexerStatus{RepoID: repoID},
+		&LanguageStat{RepoID: repoID},
 		&Comment{RefRepoID: repoID},
 		&Task{RepoID: repoID},
 	); err != nil {
diff --git a/models/repo_indexer.go b/models/repo_indexer.go
index a9a516175d..1f5ab928ab 100644
--- a/models/repo_indexer.go
+++ b/models/repo_indexer.go
@@ -10,21 +10,32 @@ import (
 	"xorm.io/builder"
 )
 
+// RepoIndexerType specifies the repository indexer type
+type RepoIndexerType int
+
+const (
+	// RepoIndexerTypeCode code indexer
+	RepoIndexerTypeCode RepoIndexerType = iota // 0
+	// RepoIndexerTypeStats repository stats indexer
+	RepoIndexerTypeStats // 1
+)
+
 // 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)"`
+	ID          int64           `xorm:"pk autoincr"`
+	RepoID      int64           `xorm:"INDEX(s)"`
+	CommitSha   string          `xorm:"VARCHAR(40)"`
+	IndexerType RepoIndexerType `xorm:"INDEX(s) NOT NULL DEFAULT 0"`
 }
 
 // GetUnindexedRepos returns repos which do not have an indexer status
-func GetUnindexedRepos(maxRepoID int64, page, pageSize int) ([]int64, error) {
+func GetUnindexedRepos(indexerType RepoIndexerType, maxRepoID int64, page, pageSize int) ([]int64, error) {
 	ids := make([]int64, 0, 50)
 	cond := builder.Cond(builder.IsNull{
 		"repo_indexer_status.id",
 	})
-	sess := x.Table("repository").Join("LEFT OUTER", "repo_indexer_status", "repository.id = repo_indexer_status.repo_id")
+	sess := x.Table("repository").Join("LEFT OUTER", "repo_indexer_status", "repository.id = repo_indexer_status.repo_id AND repo_indexer_status.indexer_type = ?", indexerType)
 	if maxRepoID > 0 {
 		cond = builder.And(cond, builder.Lte{
 			"repository.id": maxRepoID,
@@ -43,40 +54,64 @@ func GetUnindexedRepos(maxRepoID int64, page, pageSize int) ([]int64, error) {
 	return ids, err
 }
 
-// GetIndexerStatus loads repo codes indxer status
-func (repo *Repository) GetIndexerStatus() error {
-	if repo.IndexerStatus != nil {
-		return nil
+// getIndexerStatus loads repo codes indxer status
+func (repo *Repository) getIndexerStatus(e Engine, indexerType RepoIndexerType) (*RepoIndexerStatus, error) {
+	switch indexerType {
+	case RepoIndexerTypeCode:
+		if repo.CodeIndexerStatus != nil {
+			return repo.CodeIndexerStatus, nil
+		}
+	case RepoIndexerTypeStats:
+		if repo.StatsIndexerStatus != nil {
+			return repo.StatsIndexerStatus, nil
+		}
 	}
-	status := &RepoIndexerStatus{RepoID: repo.ID}
-	has, err := x.Get(status)
+	status := &RepoIndexerStatus{RepoID: repo.ID, IndexerType: indexerType}
+	has, err := e.Get(status)
 	if err != nil {
-		return err
+		return nil, err
 	} else if !has {
 		status.CommitSha = ""
 	}
-	repo.IndexerStatus = status
-	return nil
+	switch indexerType {
+	case RepoIndexerTypeCode:
+		repo.CodeIndexerStatus = status
+	case RepoIndexerTypeStats:
+		repo.StatsIndexerStatus = status
+	}
+	return status, nil
 }
 
-// UpdateIndexerStatus updates indexer status
-func (repo *Repository) UpdateIndexerStatus(sha string) error {
-	if err := repo.GetIndexerStatus(); err != nil {
+// GetIndexerStatus loads repo codes indxer status
+func (repo *Repository) GetIndexerStatus(indexerType RepoIndexerType) (*RepoIndexerStatus, error) {
+	return repo.getIndexerStatus(x, indexerType)
+}
+
+// updateIndexerStatus updates indexer status
+func (repo *Repository) updateIndexerStatus(e Engine, indexerType RepoIndexerType, sha string) error {
+	status, err := repo.getIndexerStatus(e, indexerType)
+	if err != nil {
 		return fmt.Errorf("UpdateIndexerStatus: Unable to getIndexerStatus for repo: %s Error: %v", repo.FullName(), err)
 	}
-	if len(repo.IndexerStatus.CommitSha) == 0 {
-		repo.IndexerStatus.CommitSha = sha
-		_, err := x.Insert(repo.IndexerStatus)
+
+	if len(status.CommitSha) == 0 {
+		status.CommitSha = sha
+		_, err := e.Insert(status)
 		if err != nil {
 			return fmt.Errorf("UpdateIndexerStatus: Unable to insert repoIndexerStatus for repo: %s Sha: %s Error: %v", repo.FullName(), sha, err)
 		}
 		return nil
 	}
-	repo.IndexerStatus.CommitSha = sha
-	_, err := x.ID(repo.IndexerStatus.ID).Cols("commit_sha").
-		Update(repo.IndexerStatus)
+	status.CommitSha = sha
+	_, err = e.ID(status.ID).Cols("commit_sha").
+		Update(status)
 	if err != nil {
 		return fmt.Errorf("UpdateIndexerStatus: Unable to update repoIndexerStatus for repo: %s Sha: %s Error: %v", repo.FullName(), sha, err)
 	}
 	return nil
 }
+
+// UpdateIndexerStatus updates indexer status
+func (repo *Repository) UpdateIndexerStatus(indexerType RepoIndexerType, sha string) error {
+	return repo.updateIndexerStatus(x, indexerType, sha)
+}
diff --git a/models/repo_language_stats.go b/models/repo_language_stats.go
new file mode 100644
index 0000000000..4c3171e292
--- /dev/null
+++ b/models/repo_language_stats.go
@@ -0,0 +1,137 @@
+// Copyright 2020 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 (
+	"math"
+	"strings"
+
+	"code.gitea.io/gitea/modules/timeutil"
+
+	"github.com/src-d/enry/v2"
+)
+
+// LanguageStat describes language statistics of a repository
+type LanguageStat struct {
+	ID          int64 `xorm:"pk autoincr"`
+	RepoID      int64 `xorm:"UNIQUE(s) INDEX NOT NULL"`
+	CommitID    string
+	IsPrimary   bool
+	Language    string             `xorm:"VARCHAR(30) UNIQUE(s) INDEX NOT NULL"`
+	Percentage  float32            `xorm:"NUMERIC(5,2) NOT NULL DEFAULT 0"`
+	Color       string             `xorm:"-"`
+	CreatedUnix timeutil.TimeStamp `xorm:"INDEX CREATED"`
+}
+
+// LanguageStatList defines a list of language statistics
+type LanguageStatList []*LanguageStat
+
+func (stats LanguageStatList) loadAttributes() {
+	for i := range stats {
+		stats[i].Color = enry.GetColor(stats[i].Language)
+	}
+}
+
+func (repo *Repository) getLanguageStats(e Engine) (LanguageStatList, error) {
+	stats := make(LanguageStatList, 0, 6)
+	if err := e.Where("`repo_id` = ?", repo.ID).Desc("`percentage`").Find(&stats); err != nil {
+		return nil, err
+	}
+	stats.loadAttributes()
+	return stats, nil
+}
+
+// GetLanguageStats returns the language statistics for a repository
+func (repo *Repository) GetLanguageStats() (LanguageStatList, error) {
+	return repo.getLanguageStats(x)
+}
+
+// GetTopLanguageStats returns the top language statistics for a repository
+func (repo *Repository) GetTopLanguageStats(limit int) (LanguageStatList, error) {
+	stats, err := repo.getLanguageStats(x)
+	if err != nil {
+		return nil, err
+	}
+	topstats := make(LanguageStatList, 0, limit)
+	var other float32
+	for i := range stats {
+		if stats[i].Language == "other" || len(topstats) >= limit {
+			other += stats[i].Percentage
+			continue
+		}
+		topstats = append(topstats, stats[i])
+	}
+	if other > 0 {
+		topstats = append(topstats, &LanguageStat{
+			RepoID:     repo.ID,
+			Language:   "other",
+			Color:      "#cccccc",
+			Percentage: float32(math.Round(float64(other)*10) / 10),
+		})
+	}
+	return topstats, nil
+}
+
+// UpdateLanguageStats updates the language statistics for repository
+func (repo *Repository) UpdateLanguageStats(commitID string, stats map[string]float32) error {
+	sess := x.NewSession()
+	if err := sess.Begin(); err != nil {
+		return err
+	}
+	defer sess.Close()
+
+	oldstats, err := repo.getLanguageStats(sess)
+	if err != nil {
+		return err
+	}
+	var topLang string
+	var p float32
+	for lang, perc := range stats {
+		if perc > p {
+			p = perc
+			topLang = strings.ToLower(lang)
+		}
+	}
+
+	for lang, perc := range stats {
+		upd := false
+		llang := strings.ToLower(lang)
+		for _, s := range oldstats {
+			// Update already existing language
+			if strings.ToLower(s.Language) == llang {
+				s.CommitID = commitID
+				s.IsPrimary = llang == topLang
+				s.Percentage = perc
+				if _, err := sess.ID(s.ID).Cols("`commit_id`", "`percentage`", "`is_primary`").Update(s); err != nil {
+					return err
+				}
+				upd = true
+				break
+			}
+		}
+		// Insert new language
+		if !upd {
+			if _, err := sess.Insert(&LanguageStat{
+				RepoID:     repo.ID,
+				CommitID:   commitID,
+				IsPrimary:  llang == topLang,
+				Language:   lang,
+				Percentage: perc,
+			}); err != nil {
+				return err
+			}
+		}
+	}
+	// Delete old languages
+	if _, err := sess.Where("`id` IN (SELECT `id` FROM `language_stat` WHERE `repo_id` = ? AND `commit_id` != ?)", repo.ID, commitID).Delete(&LanguageStat{}); err != nil {
+		return err
+	}
+
+	if err = repo.updateIndexerStatus(sess, RepoIndexerTypeStats, commitID); err != nil {
+		return err
+	}
+
+	return sess.Commit()
+}
diff --git a/models/repo_list.go b/models/repo_list.go
index d3a113d26c..6385de4b32 100644
--- a/models/repo_list.go
+++ b/models/repo_list.go
@@ -46,11 +46,14 @@ func (repos RepositoryList) loadAttributes(e Engine) error {
 		return nil
 	}
 
-	// Load owners.
 	set := make(map[int64]struct{})
+	repoIDs := make([]int64, len(repos))
 	for i := range repos {
 		set[repos[i].OwnerID] = struct{}{}
+		repoIDs[i] = repos[i].ID
 	}
+
+	// Load owners.
 	users := make(map[int64]*User, len(set))
 	if err := e.
 		Where("id > 0").
@@ -61,6 +64,25 @@ func (repos RepositoryList) loadAttributes(e Engine) error {
 	for i := range repos {
 		repos[i].Owner = users[repos[i].OwnerID]
 	}
+
+	// Load primary language.
+	stats := make(LanguageStatList, 0, len(repos))
+	if err := e.
+		Where("`is_primary` = ? AND `language` != ?", true, "other").
+		In("`repo_id`", repoIDs).
+		Find(&stats); err != nil {
+		return fmt.Errorf("find primary languages: %v", err)
+	}
+	stats.loadAttributes()
+	for i := range repos {
+		for _, st := range stats {
+			if st.RepoID == repos[i].ID {
+				repos[i].PrimaryLanguage = st
+				break
+			}
+		}
+	}
+
 	return nil
 }
 
@@ -119,7 +141,6 @@ type SearchRepoOptions struct {
 	OrderBy         SearchOrderBy
 	Private         bool // Include private repositories in results
 	StarredByID     int64
-	IsProfile       bool
 	AllPublic       bool // Include also all public repositories of users and public organisations
 	AllLimited      bool // Include also all public repositories of limited organisations
 	// None -> include collaborative AND non-collaborative
@@ -306,10 +327,8 @@ func SearchRepository(opts *SearchRepoOptions) (RepositoryList, int64, error) {
 		return nil, 0, fmt.Errorf("Repo: %v", err)
 	}
 
-	if !opts.IsProfile {
-		if err = repos.loadAttributes(sess); err != nil {
-			return nil, 0, fmt.Errorf("LoadAttributes: %v", err)
-		}
+	if err = repos.loadAttributes(sess); err != nil {
+		return nil, 0, fmt.Errorf("LoadAttributes: %v", err)
 	}
 
 	return repos, count, nil