1
// Copyright 2020 The Gitea Authors. All rights reserved.
2
// SPDX-License-Identifier: MIT
12
"code.gitea.io/gitea/modules/analyze"
13
"code.gitea.io/gitea/modules/optional"
15
"github.com/go-enry/go-enry/v2"
16
"github.com/go-git/go-git/v5"
17
"github.com/go-git/go-git/v5/plumbing"
18
"github.com/go-git/go-git/v5/plumbing/object"
21
// GetLanguageStats calculates language stats for git repository at specified commit
22
func (repo *Repository) GetLanguageStats(commitID string) (map[string]int64, error) {
23
r, err := git.PlainOpen(repo.Path)
28
rev, err := r.ResolveRevision(plumbing.Revision(commitID))
33
commit, err := r.CommitObject(*rev)
38
tree, err := commit.Tree()
43
checker, deferable := repo.CheckAttributeReader(commitID)
46
// sizes contains the current calculated size of all files by language
47
sizes := make(map[string]int64)
48
// by default we will only count the sizes of programming languages or markup languages
49
// unless they are explicitly set using linguist-language
50
includedLanguage := map[string]bool{}
51
// or if there's only one language in the repository
52
firstExcludedLanguage := ""
53
firstExcludedLanguageSize := int64(0)
55
err = tree.Files().ForEach(func(f *object.File) error {
60
isVendored := optional.None[bool]()
61
isGenerated := optional.None[bool]()
62
isDocumentation := optional.None[bool]()
63
isDetectable := optional.None[bool]()
66
attrs, err := checker.CheckPath(f.Name)
68
isVendored = AttributeToBool(attrs, AttributeLinguistVendored)
69
if isVendored.ValueOrDefault(false) {
73
isGenerated = AttributeToBool(attrs, AttributeLinguistGenerated)
74
if isGenerated.ValueOrDefault(false) {
78
isDocumentation = AttributeToBool(attrs, AttributeLinguistDocumentation)
79
if isDocumentation.ValueOrDefault(false) {
83
isDetectable = AttributeToBool(attrs, AttributeLinguistDetectable)
84
if !isDetectable.ValueOrDefault(true) {
88
hasLanguage := TryReadLanguageAttribute(attrs)
89
if hasLanguage.Value() != "" {
90
language := hasLanguage.Value()
92
// group languages, such as Pug -> HTML; SCSS -> CSS
93
group := enry.GetLanguageGroup(language)
98
// this language will always be added to the size
99
sizes[language] += f.Size
105
if (!isVendored.Has() && analyze.IsVendor(f.Name)) ||
106
enry.IsDotFile(f.Name) ||
107
(!isDocumentation.Has() && enry.IsDocumentation(f.Name)) ||
108
enry.IsConfiguration(f.Name) {
112
// If content can not be read or file is too big just do detection by filename
114
if f.Size <= bigFileSize {
115
content, _ = readFile(f, fileSizeLimit)
117
if !isGenerated.Has() && enry.IsGenerated(f.Name, content) {
121
language := analyze.GetCodeLanguage(f.Name, content)
122
if language == enry.OtherLanguage || language == "" {
126
// group languages, such as Pug -> HTML; SCSS -> CSS
127
group := enry.GetLanguageGroup(language)
132
included, checked := includedLanguage[language]
134
langtype := enry.GetLanguageType(language)
135
included = langtype == enry.Programming || langtype == enry.Markup
136
includedLanguage[language] = included
138
if included || isDetectable.ValueOrDefault(false) {
139
sizes[language] += f.Size
140
} else if len(sizes) == 0 && (firstExcludedLanguage == "" || firstExcludedLanguage == language) {
141
firstExcludedLanguage = language
142
firstExcludedLanguageSize += f.Size
151
// If there are no included languages add the first excluded language
152
if len(sizes) == 0 && firstExcludedLanguage != "" {
153
sizes[firstExcludedLanguage] = firstExcludedLanguageSize
156
return mergeLanguageStats(sizes), nil
159
func readFile(f *object.File, limit int64) ([]byte, error) {
171
if limit > 0 && size > limit {
174
buf := bytes.NewBuffer(nil)
176
_, err = io.Copy(buf, io.LimitReader(r, limit))
177
return buf.Bytes(), err