Change language statistics to save size instead of percentage (#11681)

* Change language statistics to save size instead of percentage in database

Co-Authored-By: Cirno the Strongest <1447794+CirnoT@users.noreply.github.com>

* Do not exclude if only language

* Fix edge cases with special langauges

Co-authored-by: Cirno the Strongest <1447794+CirnoT@users.noreply.github.com>

Author: lafriks

models/migrations/migrations.go

@@ -212,6 +212,8 @@ var migrations = []Migration{
 	NewMigration("Add ResolveDoerID to Comment table", addResolveDoerIDCommentColumn),
 	// v139 -> v140
 	NewMigration("prepend refs/heads/ to issue refs", prependRefsHeadsToIssueRefs),
+	// v140 -> v141
+	NewMigration("Save detected language file size to database instead of percent", fixLanguageStatsToSaveSize),
 }
 
 // GetCurrentDBVersion returns the current db version

models/migrations/v140.go

@@ -0,0 +1,56 @@
+// 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/setting"
+
+	"xorm.io/xorm"
+)
+
+func fixLanguageStatsToSaveSize(x *xorm.Engine) error {
+	// LanguageStat see models/repo_language_stats.go
+	type LanguageStat struct {
+		Size int64 `xorm:"NOT NULL DEFAULT 0"`
+	}
+
+	// RepoIndexerType specifies the repository indexer type
+	type RepoIndexerType int
+
+	const (
+		// RepoIndexerTypeCode code indexer
+		RepoIndexerTypeCode RepoIndexerType = iota // 0
+		// RepoIndexerTypeStats repository stats indexer
+		RepoIndexerTypeStats // 1
+	)
+
+	// RepoIndexerStatus see models/repo_indexer.go
+	type RepoIndexerStatus struct {
+		IndexerType RepoIndexerType `xorm:"INDEX(s) NOT NULL DEFAULT 0"`
+	}
+
+	if err := x.Sync2(new(LanguageStat)); err != nil {
+		return fmt.Errorf("Sync2: %v", err)
+	}
+
+	x.Delete(&RepoIndexerStatus{IndexerType: RepoIndexerTypeStats})
+
+	// Delete language stat statuses
+	truncExpr := "TRUNCATE TABLE"
+	if setting.Database.UseSQLite3 {
+		truncExpr = "DELETE FROM"
+	}
+
+	// Delete language stats
+	if _, err := x.Exec(fmt.Sprintf("%s language_stat", truncExpr)); err != nil {
+		return err
+	}
+
+	sess := x.NewSession()
+	defer sess.Close()
+	return dropTableColumns(sess, "language_stat", "percentage")
+}

models/repo_language_stats.go

@@ -20,11 +20,28 @@ type LanguageStat struct {
 	CommitID    string
 	IsPrimary   bool
 	Language    string             `xorm:"VARCHAR(30) UNIQUE(s) INDEX NOT NULL"`
-	Percentage  float32            `xorm:"NUMERIC(5,2) NOT NULL DEFAULT 0"`
+	Percentage  float32            `xorm:"-"`
+	Size        int64              `xorm:"NOT NULL DEFAULT 0"`
 	Color       string             `xorm:"-"`
 	CreatedUnix timeutil.TimeStamp `xorm:"INDEX CREATED"`
 }
 
+// specialLanguages defines list of languages that are excluded from the calculation
+// unless they are the only language present in repository. Only languages which under
+// normal circumstances are not considered to be code should be listed here.
+var specialLanguages = map[string]struct{}{
+	"XML":      {},
+	"JSON":     {},
+	"TOML":     {},
+	"YAML":     {},
+	"INI":      {},
+	"SQL":      {},
+	"SVG":      {},
+	"Text":     {},
+	"Markdown": {},
+	"other":    {},
+}
+
 // LanguageStatList defines a list of language statistics
 type LanguageStatList []*LanguageStat
 
@@ -34,12 +51,53 @@ func (stats LanguageStatList) loadAttributes() {
 	}
 }
 
+func (stats LanguageStatList) getLanguagePercentages() map[string]float32 {
+	langPerc := make(map[string]float32)
+	var otherPerc float32 = 100
+	var total int64
+	// Check that repository has at least one non-special language
+	var skipSpecial bool
+	for _, stat := range stats {
+		if _, ok := specialLanguages[stat.Language]; !ok {
+			skipSpecial = true
+			break
+		}
+	}
+	for _, stat := range stats {
+		// Exclude specific languages from percentage calculation
+		if _, ok := specialLanguages[stat.Language]; ok && skipSpecial {
+			continue
+		}
+		total += stat.Size
+	}
+	if total > 0 {
+		for _, stat := range stats {
+			// Exclude specific languages from percentage calculation
+			if _, ok := specialLanguages[stat.Language]; ok && skipSpecial {
+				continue
+			}
+			perc := float32(math.Round(float64(stat.Size)/float64(total)*1000) / 10)
+			if perc <= 0.1 {
+				continue
+			}
+			otherPerc -= perc
+			langPerc[stat.Language] = perc
+		}
+		otherPerc = float32(math.Round(float64(otherPerc)*10) / 10)
+	} else {
+		otherPerc = 100
+	}
+	if otherPerc > 0 {
+		langPerc["other"] = otherPerc
+	}
+	return langPerc
+}
+
 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 {
+	if err := e.Where("`repo_id` = ?", repo.ID).Desc("`size`").Find(&stats); err != nil {
 		return nil, err
 	}
-	stats.loadAttributes()
 	return stats, nil
 }
 
@@ -54,13 +112,18 @@ func (repo *Repository) GetTopLanguageStats(limit int) (LanguageStatList, error)
 	if err != nil {
 		return nil, err
 	}
+	perc := stats.getLanguagePercentages()
 	topstats := make(LanguageStatList, 0, limit)
 	var other float32
 	for i := range stats {
+		if _, ok := perc[stats[i].Language]; !ok {
+			continue
+		}
 		if stats[i].Language == "other" || len(topstats) >= limit {
-			other += stats[i].Percentage
+			other += perc[stats[i].Language]
 			continue
 		}
+		stats[i].Percentage = perc[stats[i].Language]
 		topstats = append(topstats, stats[i])
 	}
 	if other > 0 {
@@ -71,11 +134,12 @@ func (repo *Repository) GetTopLanguageStats(limit int) (LanguageStatList, error)
 			Percentage: float32(math.Round(float64(other)*10) / 10),
 		})
 	}
+	topstats.loadAttributes()
 	return topstats, nil
 }
 
 // UpdateLanguageStats updates the language statistics for repository
-func (repo *Repository) UpdateLanguageStats(commitID string, stats map[string]float32) error {
+func (repo *Repository) UpdateLanguageStats(commitID string, stats map[string]int64) error {
 	sess := x.NewSession()
 	if err := sess.Begin(); err != nil {
 		return err
@@ -87,24 +151,24 @@ func (repo *Repository) UpdateLanguageStats(commitID string, stats map[string]fl
 		return err
 	}
 	var topLang string
-	var p float32
-	for lang, perc := range stats {
-		if perc > p {
-			p = perc
+	var s int64
+	for lang, size := range stats {
+		if size > s {
+			s = size
 			topLang = strings.ToLower(lang)
 		}
 	}
 
-	for lang, perc := range stats {
+	for lang, size := 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 {
+				s.Size = size
+				if _, err := sess.ID(s.ID).Cols("`commit_id`", "`size`", "`is_primary`").Update(s); err != nil {
 					return err
 				}
 				upd = true
@@ -114,11 +178,11 @@ func (repo *Repository) UpdateLanguageStats(commitID string, stats map[string]fl
 		// Insert new language
 		if !upd {
 			if _, err := sess.Insert(&LanguageStat{
-				RepoID:     repo.ID,
-				CommitID:   commitID,
-				IsPrimary:  llang == topLang,
-				Language:   lang,
-				Percentage: perc,
+				RepoID:    repo.ID,
+				CommitID:  commitID,
+				IsPrimary: llang == topLang,
+				Language:  lang,
+				Size:      size,
 			}); err != nil {
 				return err
 			}
@@ -153,7 +217,7 @@ func CopyLanguageStat(originalRepo, destRepo *Repository) error {
 		return err
 	}
 	RepoLang := make(LanguageStatList, 0, 6)
-	if err := sess.Where("`repo_id` = ?", originalRepo.ID).Desc("`percentage`").Find(&RepoLang); err != nil {
+	if err := sess.Where("`repo_id` = ?", originalRepo.ID).Desc("`size`").Find(&RepoLang); err != nil {
 		return err
 	}
 	if len(RepoLang) > 0 {

modules/git/repo_language_stats.go

@@ -8,7 +8,6 @@ import (
 	"bytes"
 	"io"
 	"io/ioutil"
-	"math"
 
 	"code.gitea.io/gitea/modules/analyze"
 
@@ -21,7 +20,7 @@ import (
 const fileSizeLimit int64 = 16 * 1024 * 1024
 
 // GetLanguageStats calculates language stats for git repository at specified commit
-func (repo *Repository) GetLanguageStats(commitID string) (map[string]float32, error) {
+func (repo *Repository) GetLanguageStats(commitID string) (map[string]int64, error) {
 	r, err := git.PlainOpen(repo.Path)
 	if err != nil {
 		return nil, err
@@ -43,7 +42,6 @@ func (repo *Repository) GetLanguageStats(commitID string) (map[string]float32, e
 	}
 
 	sizes := make(map[string]int64)
-	var total int64
 	err = tree.Files().ForEach(func(f *object.File) error {
 		if enry.IsVendor(f.Name) || enry.IsDotFile(f.Name) ||
 			enry.IsDocumentation(f.Name) || enry.IsConfiguration(f.Name) {
@@ -60,33 +58,22 @@ func (repo *Repository) GetLanguageStats(commitID string) (map[string]float32, e
 
 		language := analyze.GetCodeLanguage(f.Name, content)
 		if language == enry.OtherLanguage || language == "" {
-			return nil
+			language = "other"
 		}
 
 		sizes[language] += f.Size
-		total += f.Size
 
 		return nil
 	})
 	if err != nil {
 		return nil, err
 	}
 
-	stats := make(map[string]float32)
-	var otherPerc float32 = 100
-	for language, size := range sizes {
-		perc := float32(math.Round(float64(size)/float64(total)*1000) / 10)
-		if perc <= 0.1 {
-			continue
-		}
-		otherPerc -= perc
-		stats[language] = perc
+	if len(sizes) == 0 {
+		sizes["other"] = 0
 	}
-	otherPerc = float32(math.Round(float64(otherPerc)*10) / 10)
-	if otherPerc > 0 {
-		stats["other"] = otherPerc
-	}
-	return stats, nil
+
+	return sizes, nil
 }
 
 func readFile(f *object.File, limit int64) ([]byte, error) {

modules/indexer/stats/indexer_test.go

@@ -34,6 +34,9 @@ func TestRepoStatsIndex(t *testing.T) {
 
 	repo, err := models.GetRepositoryByID(1)
 	assert.NoError(t, err)
+	status, err := repo.GetIndexerStatus(models.RepoIndexerTypeStats)
+	assert.NoError(t, err)
+	assert.Equal(t, "65f1bf27bc3bf70f64657658635e66094edbcb4d", status.CommitSha)
 	langs, err := repo.GetTopLanguageStats(5)
 	assert.NoError(t, err)
 	assert.Len(t, langs, 1)