|
66
|
1 package chroma
|
|
|
2
|
|
|
3 import (
|
|
|
4 "path/filepath"
|
|
|
5 "sort"
|
|
|
6 "strings"
|
|
|
7 )
|
|
|
8
|
|
|
9 var (
|
|
|
10 ignoredSuffixes = [...]string{
|
|
|
11 // Editor backups
|
|
|
12 "~", ".bak", ".old", ".orig",
|
|
|
13 // Debian and derivatives apt/dpkg/ucf backups
|
|
|
14 ".dpkg-dist", ".dpkg-old", ".ucf-dist", ".ucf-new", ".ucf-old",
|
|
|
15 // Red Hat and derivatives rpm backups
|
|
|
16 ".rpmnew", ".rpmorig", ".rpmsave",
|
|
|
17 // Build system input/template files
|
|
|
18 ".in",
|
|
|
19 }
|
|
|
20 )
|
|
|
21
|
|
|
22 // LexerRegistry is a registry of Lexers.
|
|
|
23 type LexerRegistry struct {
|
|
|
24 Lexers Lexers
|
|
|
25 byName map[string]Lexer
|
|
|
26 byAlias map[string]Lexer
|
|
|
27 }
|
|
|
28
|
|
|
29 // NewLexerRegistry creates a new LexerRegistry of Lexers.
|
|
|
30 func NewLexerRegistry() *LexerRegistry {
|
|
|
31 return &LexerRegistry{
|
|
|
32 byName: map[string]Lexer{},
|
|
|
33 byAlias: map[string]Lexer{},
|
|
|
34 }
|
|
|
35 }
|
|
|
36
|
|
|
37 // Names of all lexers, optionally including aliases.
|
|
|
38 func (l *LexerRegistry) Names(withAliases bool) []string {
|
|
|
39 out := []string{}
|
|
|
40 for _, lexer := range l.Lexers {
|
|
|
41 config := lexer.Config()
|
|
|
42 out = append(out, config.Name)
|
|
|
43 if withAliases {
|
|
|
44 out = append(out, config.Aliases...)
|
|
|
45 }
|
|
|
46 }
|
|
|
47 sort.Strings(out)
|
|
|
48 return out
|
|
|
49 }
|
|
|
50
|
|
|
51 // Get a Lexer by name, alias or file extension.
|
|
|
52 func (l *LexerRegistry) Get(name string) Lexer {
|
|
|
53 if lexer := l.byName[name]; lexer != nil {
|
|
|
54 return lexer
|
|
|
55 }
|
|
|
56 if lexer := l.byAlias[name]; lexer != nil {
|
|
|
57 return lexer
|
|
|
58 }
|
|
|
59 if lexer := l.byName[strings.ToLower(name)]; lexer != nil {
|
|
|
60 return lexer
|
|
|
61 }
|
|
|
62 if lexer := l.byAlias[strings.ToLower(name)]; lexer != nil {
|
|
|
63 return lexer
|
|
|
64 }
|
|
|
65
|
|
|
66 candidates := PrioritisedLexers{}
|
|
|
67 // Try file extension.
|
|
|
68 if lexer := l.Match("filename." + name); lexer != nil {
|
|
|
69 candidates = append(candidates, lexer)
|
|
|
70 }
|
|
|
71 // Try exact filename.
|
|
|
72 if lexer := l.Match(name); lexer != nil {
|
|
|
73 candidates = append(candidates, lexer)
|
|
|
74 }
|
|
|
75 if len(candidates) == 0 {
|
|
|
76 return nil
|
|
|
77 }
|
|
|
78 sort.Sort(candidates)
|
|
|
79 return candidates[0]
|
|
|
80 }
|
|
|
81
|
|
|
82 // MatchMimeType attempts to find a lexer for the given MIME type.
|
|
|
83 func (l *LexerRegistry) MatchMimeType(mimeType string) Lexer {
|
|
|
84 matched := PrioritisedLexers{}
|
|
|
85 for _, l := range l.Lexers {
|
|
|
86 for _, lmt := range l.Config().MimeTypes {
|
|
|
87 if mimeType == lmt {
|
|
|
88 matched = append(matched, l)
|
|
|
89 }
|
|
|
90 }
|
|
|
91 }
|
|
|
92 if len(matched) != 0 {
|
|
|
93 sort.Sort(matched)
|
|
|
94 return matched[0]
|
|
|
95 }
|
|
|
96 return nil
|
|
|
97 }
|
|
|
98
|
|
|
99 // Match returns the first lexer matching filename.
|
|
|
100 func (l *LexerRegistry) Match(filename string) Lexer {
|
|
|
101 filename = filepath.Base(filename)
|
|
|
102 matched := PrioritisedLexers{}
|
|
|
103 // First, try primary filename matches.
|
|
|
104 for _, lexer := range l.Lexers {
|
|
|
105 config := lexer.Config()
|
|
|
106 for _, glob := range config.Filenames {
|
|
|
107 ok, err := filepath.Match(glob, filename)
|
|
|
108 if err != nil { // nolint
|
|
|
109 panic(err)
|
|
|
110 } else if ok {
|
|
|
111 matched = append(matched, lexer)
|
|
|
112 } else {
|
|
|
113 for _, suf := range &ignoredSuffixes {
|
|
|
114 ok, err := filepath.Match(glob+suf, filename)
|
|
|
115 if err != nil {
|
|
|
116 panic(err)
|
|
|
117 } else if ok {
|
|
|
118 matched = append(matched, lexer)
|
|
|
119 break
|
|
|
120 }
|
|
|
121 }
|
|
|
122 }
|
|
|
123 }
|
|
|
124 }
|
|
|
125 if len(matched) > 0 {
|
|
|
126 sort.Sort(matched)
|
|
|
127 return matched[0]
|
|
|
128 }
|
|
|
129 matched = nil
|
|
|
130 // Next, try filename aliases.
|
|
|
131 for _, lexer := range l.Lexers {
|
|
|
132 config := lexer.Config()
|
|
|
133 for _, glob := range config.AliasFilenames {
|
|
|
134 ok, err := filepath.Match(glob, filename)
|
|
|
135 if err != nil { // nolint
|
|
|
136 panic(err)
|
|
|
137 } else if ok {
|
|
|
138 matched = append(matched, lexer)
|
|
|
139 } else {
|
|
|
140 for _, suf := range &ignoredSuffixes {
|
|
|
141 ok, err := filepath.Match(glob+suf, filename)
|
|
|
142 if err != nil {
|
|
|
143 panic(err)
|
|
|
144 } else if ok {
|
|
|
145 matched = append(matched, lexer)
|
|
|
146 break
|
|
|
147 }
|
|
|
148 }
|
|
|
149 }
|
|
|
150 }
|
|
|
151 }
|
|
|
152 if len(matched) > 0 {
|
|
|
153 sort.Sort(matched)
|
|
|
154 return matched[0]
|
|
|
155 }
|
|
|
156 return nil
|
|
|
157 }
|
|
|
158
|
|
|
159 // Analyse text content and return the "best" lexer..
|
|
|
160 func (l *LexerRegistry) Analyse(text string) Lexer {
|
|
|
161 var picked Lexer
|
|
|
162 highest := float32(0.0)
|
|
|
163 for _, lexer := range l.Lexers {
|
|
|
164 if analyser, ok := lexer.(Analyser); ok {
|
|
|
165 weight := analyser.AnalyseText(text)
|
|
|
166 if weight > highest {
|
|
|
167 picked = lexer
|
|
|
168 highest = weight
|
|
|
169 }
|
|
|
170 }
|
|
|
171 }
|
|
|
172 return picked
|
|
|
173 }
|
|
|
174
|
|
|
175 // Register a Lexer with the LexerRegistry.
|
|
|
176 func (l *LexerRegistry) Register(lexer Lexer) Lexer {
|
|
|
177 lexer.SetRegistry(l)
|
|
|
178 config := lexer.Config()
|
|
|
179 l.byName[config.Name] = lexer
|
|
|
180 l.byName[strings.ToLower(config.Name)] = lexer
|
|
|
181 for _, alias := range config.Aliases {
|
|
|
182 l.byAlias[alias] = lexer
|
|
|
183 l.byAlias[strings.ToLower(alias)] = lexer
|
|
|
184 }
|
|
|
185 l.Lexers = append(l.Lexers, lexer)
|
|
|
186 return lexer
|
|
|
187 }
|
