Merge pull request #20 from mcarmonaa/types

added types.go generation

internal/code-generator/assets/type.go.tmpl

@@ -0,0 +1,26 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH github.com/src-d/simple-linguist/cli/slinguist-generate
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: {{ getCommit }}
+
+type Type int
+
+const (
+	TypeUnknown Type = iota
+	TypeData
+	TypeProgramming
+	TypeMarkup
+	TypeProse
+)
+
+func GetLanguageType(language string) (langType Type) {
+	langType, _ = languagesType[language]
+	return langType
+}
+
+var languagesType = map[string]Type{
+	{{range $language, $type := . -}}
+	"{{ $language }}": {{ $type -}},
+	{{end -}}
+}

internal/code-generator/generator/generator_test.go

@@ -17,6 +17,7 @@ const (
 	formatedContentGold       = "test_files/formated_content.gold"
 	formatedVendorGold        = "test_files/formated_vendor.gold"
 	formatedDocumentationGold = "test_files/formated_documentation.gold"
+	formatedTypesGold         = "test_files/formated_type.gold"
 
 	// Languages test
 	ymlTestFile           = "test_files/languages.test.yml"
@@ -41,6 +42,12 @@ const (
 	documentationGold         = "test_files/documentation.gold"
 	documentationTestTmplPath = "test_files/documentation.test.go.tmpl"
 	documentationTestTmplName = "documentation.test.go.tmpl"
+
+	// Types test
+	typesTestFile     = "test_files/type.test.yml"
+	typesGold         = "test_files/type.gold"
+	typesTestTmplPath = "test_files/type.test.go.tmpl"
+	typesTestTmplName = "type.test.go.tmpl"
 )
 
 func TestFromFile(t *testing.T) {
@@ -56,6 +63,9 @@ func TestFromFile(t *testing.T) {
 	goldDocumentation, err := ioutil.ReadFile(formatedDocumentationGold)
 	assert.NoError(t, err)
 
+	goldTypes, err := ioutil.ReadFile(formatedTypesGold)
+	assert.NoError(t, err)
+
 	outPathLang, err := ioutil.TempFile("/tmp", "generator-test-")
 	assert.NoError(t, err)
 	defer os.Remove(outPathLang.Name())
@@ -72,6 +82,10 @@ func TestFromFile(t *testing.T) {
 	assert.NoError(t, err)
 	defer os.Remove(outPathDocumentation.Name())
 
+	outPathTypes, err := ioutil.TempFile("/tmp", "generator-test-")
+	assert.NoError(t, err)
+	defer os.Remove(outPathTypes.Name())
+
 	tests := []struct {
 		name        string
 		fileToParse string
@@ -122,6 +136,16 @@ func TestFromFile(t *testing.T) {
 			generate:    Documentation,
 			wantOut:     goldDocumentation,
 		},
+		{
+			name:        "tyTestFromFile_Types",
+			fileToParse: typesTestFile,
+			outPath:     outPathTypes.Name(),
+			tmplPath:    typesTestTmplPath,
+			tmplName:    typesTestTmplName,
+			commit:      commitTest,
+			generate:    Types,
+			wantOut:     goldTypes,
+		},
 	}
 
 	for _, tt := range tests {
@@ -270,3 +294,37 @@ func TestDocumentation(t *testing.T) {
 		})
 	}
 }
+
+func TestTypes(t *testing.T) {
+	gold, err := ioutil.ReadFile(typesGold)
+	assert.NoError(t, err)
+
+	input, err := ioutil.ReadFile(typesTestFile)
+	assert.NoError(t, err)
+
+	tests := []struct {
+		name     string
+		input    []byte
+		tmplPath string
+		tmplName string
+		commit   string
+		wantOut  []byte
+	}{
+		{
+			name:     "TestTypes",
+			input:    input,
+			tmplPath: typesTestTmplPath,
+			tmplName: typesTestTmplName,
+			commit:   commitTest,
+			wantOut:  gold,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			out, err := Types(tt.input, tt.tmplPath, tt.tmplName, tt.commit)
+			assert.NoError(t, err)
+			assert.EqualValues(t, tt.wantOut, out, fmt.Sprintf("Types() = %v, want %v", string(out), string(tt.wantOut)))
+		})
+	}
+}

internal/code-generator/generator/langs.go

@@ -2,7 +2,6 @@ package generator
 
 import (
 	"bytes"
-	"errors"
 	"io"
 	"strings"
 	"text/template"
@@ -10,23 +9,28 @@ import (
 	"gopkg.in/yaml.v2"
 )
 
-var (
+type languageInfo struct {
-	// ErrExtensionsNotFound is the error returned if data parsed doesn't contain extensions.
+	Type         string   `yaml:"type,omitempty"`
-	ErrExtensionsNotFound = errors.New("extensions not found")
+	Aliases      []string `yaml:"aliases,omitempty,flow"`
-)
+	Extensions   []string `yaml:"extensions,omitempty,flow"`
+	Interpreters []string `yaml:"interpreters,omitempty,flow"`
+	Group        string   `yaml:"group,omitempty"`
+}
 
 // Languages reads from buf and builds languages.go file from languagesTmplPath.
 func Languages(data []byte, languagesTmplPath, languagesTmplName, commit string) ([]byte, error) {
-	var yamlSlice yaml.MapSlice
+	languages := make(map[string]*languageInfo)
-	if err := yaml.Unmarshal(data, &yamlSlice); err != nil {
+	if err := yaml.Unmarshal(data, &languages); err != nil {
 		return nil, err
 	}
 
-	languagesByExtension, err := buildExtensionLanguageMap(yamlSlice)
+	orderedKeyList, err := getAlphabeticalOrderedKeys(data)
 	if err != nil {
 		return nil, err
 	}
 
+	languagesByExtension := buildExtensionLanguageMap(languages, orderedKeyList)
+
 	buf := &bytes.Buffer{}
 	if err := executeLanguagesTemplate(buf, languagesByExtension, languagesTmplPath, languagesTmplName, commit); err != nil {
 		return nil, err
@@ -35,46 +39,30 @@ func Languages(data []byte, languagesTmplPath, languagesTmplName, commit string)
 	return buf.Bytes(), nil
 }
 
-func buildExtensionLanguageMap(yamlSlice yaml.MapSlice) (map[string][]string, error) {
+func getAlphabeticalOrderedKeys(data []byte) ([]string, error) {
-	extensionLangsMap := make(map[string][]string)
+	var yamlSlice yaml.MapSlice
-	for _, lang := range yamlSlice {
+	if err := yaml.Unmarshal(data, &yamlSlice); err != nil {
-		extensions, err := findExtensions(lang.Value.(yaml.MapSlice))
-		if err != nil && err != ErrExtensionsNotFound {
 		return nil, err
 	}
 
-		fillMap(extensionLangsMap, lang.Key.(string), extensions)
+	orderedKeyList := make([]string, 0)
+	for _, lang := range yamlSlice {
+		orderedKeyList = append(orderedKeyList, lang.Key.(string))
 	}
 
-	return extensionLangsMap, nil
+	return orderedKeyList, nil
 }
 
-func findExtensions(items yaml.MapSlice) ([]string, error) {
+func buildExtensionLanguageMap(languages map[string]*languageInfo, orderedKeyList []string) map[string][]string {
-	const extField = "extensions"
+	extensionLangsMap := make(map[string][]string)
-	for _, item := range items {
+	for _, lang := range orderedKeyList {
-		if item.Key == extField {
+		langInfo := languages[lang]
-			extensions := toStringSlice(item.Value.([]interface{}))
+		for _, extension := range langInfo.Extensions {
-			return extensions, nil
+			extensionLangsMap[extension] = append(extensionLangsMap[extension], lang)
 		}
 	}
 
-	return nil, ErrExtensionsNotFound
+	return extensionLangsMap
-}
-
-func toStringSlice(slice []interface{}) []string {
-	extensions := make([]string, 0, len(slice))
-	for _, element := range slice {
-		extension := element.(string)
-		extensions = append(extensions, extension)
-	}
-
-	return extensions
-}
-
-func fillMap(extensionLangs map[string][]string, lang string, extensions []string) {
-	for _, extension := range extensions {
-		extensionLangs[extension] = append(extensionLangs[extension], lang)
-	}
 }
 
 func executeLanguagesTemplate(out io.Writer, languagesByExtension map[string][]string, languagesTmplPath, languagesTmpl, commit string) error {

internal/code-generator/generator/test_files/formated_type.gold

@@ -0,0 +1,27 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH github.com/src-d/simple-linguist/cli/slinguist-generate
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: fe8b44ab8a225b1ffa75b983b916ea22fee5b6f7
+
+type Type int
+
+const (
+	TypeUnknown Type = iota
+	TypeData
+	TypeProgramming
+	TypeMarkup
+	TypeProse
+)
+
+func GetLanguageType(language string) (langType Type) {
+	langType, _ = languagesType[language]
+	return langType
+}
+
+var languagesType = map[string]Type{
+	"Scaml":  TypeMarkup,
+	"Scheme": TypeProgramming,
+	"Scilab": TypeProgramming,
+	"Self":   TypeProgramming,
+}

internal/code-generator/generator/test_files/type.gold

@@ -0,0 +1,27 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH github.com/src-d/simple-linguist/cli/slinguist-generate
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: fe8b44ab8a225b1ffa75b983b916ea22fee5b6f7
+
+type Type int
+
+const (
+	TypeUnknown Type = iota
+	TypeData
+	TypeProgramming
+	TypeMarkup
+	TypeProse
+)
+
+func GetLanguageType(language string) (langType Type) {
+	langType, _ = languagesType[language]
+	return langType
+}
+
+var languagesType = map[string]Type{
+	"Scaml": TypeMarkup,
+	"Scheme": TypeProgramming,
+	"Scilab": TypeProgramming,
+	"Self": TypeProgramming,
+	}

internal/code-generator/generator/test_files/type.test.go.tmpl

@@ -0,0 +1,26 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH github.com/src-d/simple-linguist/cli/slinguist-generate
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: {{ getCommit }}
+
+type Type int
+
+const (
+	TypeUnknown Type = iota
+	TypeData
+	TypeProgramming
+	TypeMarkup
+	TypeProse
+)
+
+func GetLanguageType(language string) (langType Type) {
+	langType, _ = languagesType[language]
+	return langType
+}
+
+var languagesType = map[string]Type{
+	{{range $language, $type := . -}}
+	"{{ $language }}": {{ $type -}},
+	{{end -}}
+}

internal/code-generator/generator/test_files/type.test.yml

@@ -0,0 +1,46 @@
+---
+Scaml:
+  group: HTML
+  type: markup
+  extensions:
+  - ".scaml"
+  tm_scope: source.scaml
+  ace_mode: text
+  language_id: 342
+Scheme:
+  type: programming
+  color: "#1e4aec"
+  extensions:
+  - ".scm"
+  - ".sld"
+  - ".sls"
+  - ".sps"
+  - ".ss"
+  interpreters:
+  - guile
+  - bigloo
+  - chicken
+  - csi
+  - gosh
+  - r6rs
+  ace_mode: scheme
+  codemirror_mode: scheme
+  codemirror_mime_type: text/x-scheme
+  language_id: 343
+Scilab:
+  type: programming
+  extensions:
+  - ".sci"
+  - ".sce"
+  - ".tst"
+  ace_mode: text
+  language_id: 344
+Self:
+  type: programming
+  color: "#0579aa"
+  extensions:
+  - ".self"
+  tm_scope: none
+  ace_mode: text
+  language_id: 345
+  

internal/code-generator/generator/types.go

@@ -0,0 +1,55 @@
+package generator
+
+import (
+	"bytes"
+	"io"
+	"text/template"
+
+	yaml "gopkg.in/yaml.v2"
+)
+
+var typeToTypeConst = map[string]string{
+	"data":        "TypeData",
+	"programming": "TypeProgramming",
+	"markup":      "TypeMarkup",
+	"prose":       "TypeProse",
+}
+
+// Types reads from buf and builds type.go file from typeTmplPath.
+func Types(data []byte, typeTmplPath, typeTmplName, commit string) ([]byte, error) {
+	languages := make(map[string]*languageInfo)
+	if err := yaml.Unmarshal(data, &languages); err != nil {
+		return nil, err
+	}
+
+	langTypeMap := buildLanguageTypeMap(languages)
+
+	buf := &bytes.Buffer{}
+	if err := executeTypesTemplate(buf, langTypeMap, typeTmplPath, typeTmplName, commit); err != nil {
+		return nil, err
+	}
+
+	return buf.Bytes(), nil
+}
+
+func buildLanguageTypeMap(languages map[string]*languageInfo) map[string]string {
+	langTypeMap := make(map[string]string)
+	for lang, info := range languages {
+		langTypeMap[lang] = typeToTypeConst[info.Type]
+	}
+
+	return langTypeMap
+}
+
+func executeTypesTemplate(out io.Writer, langTypeMap map[string]string, typeTmplPath, typeTmpl, commit string) error {
+	fmap := template.FuncMap{
+		"getCommit": func() string { return commit },
+	}
+
+	t := template.Must(template.New(typeTmpl).Funcs(fmap).ParseFiles(typeTmplPath))
+	if err := t.Execute(out, langTypeMap); err != nil {
+		return err
+	}
+
+	return nil
+}

internal/code-generator/main.go

@@ -8,49 +8,65 @@ import (
 )
 
 const (
+	// languages.go generation
 	languagesYAML     = ".linguist/lib/linguist/languages.yml"
 	langFile          = "languages.go"
 	languagesTmplPath = "internal/code-generator/assets/languages.go.tmpl"
 	languagesTmpl     = "languages.go.tmpl"
 
+	// content.go generation
 	heuristicsRuby  = ".linguist/lib/linguist/heuristics.rb"
 	contentFile     = "content.go"
 	contentTmplPath = "internal/code-generator/assets/content.go.tmpl"
 	contentTmpl     = "content.go.tmpl"
 
+	// vendor_matchers.go generation
 	vendorYAML     = ".linguist/lib/linguist/vendor.yml"
 	vendorFile     = "vendor_matchers.go"
 	vendorTmplPath = "internal/code-generator/assets/vendor.go.tmpl"
 	vendorTmpl     = "vendor.go.tmpl"
 
+	// documentation_matchers.go generation
 	documentationYAML     = ".linguist/lib/linguist/documentation.yml"
 	documentationFile     = "documentation_matchers.go"
 	documentationTmplPath = "internal/code-generator/assets/documentation.go.tmpl"
 	documentationTmpl     = "documentation.go.tmpl"
 
+	// type.go generation
+	typeFile     = "type.go"
+	typeTmplPath = "internal/code-generator/assets/type.go.tmpl"
+	typeTmpl     = "type.go.tmpl"
+
 	commitPath = ".git/refs/heads/master"
 )
 
+type generatorArgs struct {
+	fileToParse string
+	outPath     string
+	tmplPath    string
+	tmplName    string
+	commit      string
+	generate    generator.Func
+}
+
 func main() {
 	commit, err := getCommit(commitPath)
 	if err != nil {
 		log.Printf("couldn't find commit: %v", err)
 	}
 
-	if err := generator.FromFile(languagesYAML, langFile, languagesTmplPath, languagesTmpl, commit, generator.Languages); err != nil {
+	argsList := []*generatorArgs{
-		log.Println(err)
+		&generatorArgs{languagesYAML, langFile, languagesTmplPath, languagesTmpl, commit, generator.Languages},
+		&generatorArgs{heuristicsRuby, contentFile, contentTmplPath, contentTmpl, commit, generator.Heuristics},
+		&generatorArgs{vendorYAML, vendorFile, vendorTmplPath, vendorTmpl, commit, generator.Vendor},
+		&generatorArgs{documentationYAML, documentationFile, documentationTmplPath, documentationTmpl, commit, generator.Documentation},
+		&generatorArgs{languagesYAML, typeFile, typeTmplPath, typeTmpl, commit, generator.Types},
 	}
 
-	if err := generator.FromFile(heuristicsRuby, contentFile, contentTmplPath, contentTmpl, commit, generator.Heuristics); err != nil {
+	for _, args := range argsList {
+		if err := generator.FromFile(args.fileToParse, args.outPath, args.tmplPath, args.tmplName, args.commit, args.generate); err != nil {
 			log.Println(err)
 		}
-
-	if err := generator.FromFile(vendorYAML, vendorFile, vendorTmplPath, vendorTmpl, commit, generator.Vendor); err != nil {
-		log.Println(err)
-	}
-
-	if err := generator.FromFile(documentationYAML, documentationFile, documentationTmplPath, documentationTmpl, commit, generator.Documentation); err != nil {
-		log.Println(err)
 	}
 }
 

type.go

@@ -0,0 +1,467 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH github.com/src-d/simple-linguist/cli/slinguist-generate
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: dae33dc2b20cddc85d1300435c3be7118a7115a9
+
+type Type int
+
+const (
+	TypeUnknown Type = iota
+	TypeData
+	TypeProgramming
+	TypeMarkup
+	TypeProse
+)
+
+func GetLanguageType(language string) (langType Type) {
+	langType, _ = languagesType[language]
+	return langType
+}
+
+var languagesType = map[string]Type{
+	"1C Enterprise":    TypeProgramming,
+	"ABAP":             TypeProgramming,
+	"ABNF":             TypeData,
+	"AGS Script":       TypeProgramming,
+	"AMPL":             TypeProgramming,
+	"ANTLR":            TypeProgramming,
+	"API Blueprint":    TypeMarkup,
+	"APL":              TypeProgramming,
+	"ASN.1":            TypeData,
+	"ASP":              TypeProgramming,
+	"ATS":              TypeProgramming,
+	"ActionScript":     TypeProgramming,
+	"Ada":              TypeProgramming,
+	"Agda":             TypeProgramming,
+	"Alloy":            TypeProgramming,
+	"Alpine Abuild":    TypeProgramming,
+	"Ant Build System": TypeData,
+	"ApacheConf":       TypeMarkup,
+	"Apex":             TypeProgramming,
+	"Apollo Guidance Computer": TypeProgramming,
+	"AppleScript":              TypeProgramming,
+	"Arc":                      TypeProgramming,
+	"Arduino":                  TypeProgramming,
+	"AsciiDoc":                 TypeProse,
+	"AspectJ":                  TypeProgramming,
+	"Assembly":                 TypeProgramming,
+	"Augeas":                   TypeProgramming,
+	"AutoHotkey":               TypeProgramming,
+	"AutoIt":                   TypeProgramming,
+	"Awk":                      TypeProgramming,
+	"Batchfile":                TypeProgramming,
+	"Befunge":                  TypeProgramming,
+	"Bison":                    TypeProgramming,
+	"BitBake":                  TypeProgramming,
+	"Blade":                    TypeMarkup,
+	"BlitzBasic":               TypeProgramming,
+	"BlitzMax":                 TypeProgramming,
+	"Bluespec":                 TypeProgramming,
+	"Boo":                      TypeProgramming,
+	"Brainfuck":                TypeProgramming,
+	"Brightscript":             TypeProgramming,
+	"Bro":                      TypeProgramming,
+	"C":                        TypeProgramming,
+	"C#":                       TypeProgramming,
+	"C++":                      TypeProgramming,
+	"C-ObjDump":                TypeData,
+	"C2hs Haskell":             TypeProgramming,
+	"CLIPS":                    TypeProgramming,
+	"CMake":                    TypeProgramming,
+	"COBOL":                    TypeProgramming,
+	"COLLADA":                  TypeData,
+	"CSON":                     TypeData,
+	"CSS":                      TypeMarkup,
+	"CSV":                      TypeData,
+	"Cap'n Proto":              TypeProgramming,
+	"CartoCSS":                 TypeProgramming,
+	"Ceylon":                   TypeProgramming,
+	"Chapel":                   TypeProgramming,
+	"Charity":                  TypeProgramming,
+	"ChucK":                    TypeProgramming,
+	"Cirru":                    TypeProgramming,
+	"Clarion":                  TypeProgramming,
+	"Clean":                    TypeProgramming,
+	"Click":                    TypeProgramming,
+	"Clojure":                  TypeProgramming,
+	"CoffeeScript":             TypeProgramming,
+	"ColdFusion":               TypeProgramming,
+	"ColdFusion CFC":           TypeProgramming,
+	"Common Lisp":              TypeProgramming,
+	"Component Pascal":         TypeProgramming,
+	"Cool":                     TypeProgramming,
+	"Coq":                      TypeProgramming,
+	"Cpp-ObjDump":              TypeData,
+	"Creole":                   TypeProse,
+	"Crystal":                  TypeProgramming,
+	"Csound":                   TypeProgramming,
+	"Csound Document":          TypeProgramming,
+	"Csound Score":             TypeProgramming,
+	"Cuda":                     TypeProgramming,
+	"Cycript":                  TypeProgramming,
+	"Cython":                   TypeProgramming,
+	"D":                        TypeProgramming,
+	"D-ObjDump":                TypeData,
+	"DIGITAL Command Language": TypeProgramming,
+	"DM":             TypeProgramming,
+	"DNS Zone":       TypeData,
+	"DTrace":         TypeProgramming,
+	"Darcs Patch":    TypeData,
+	"Dart":           TypeProgramming,
+	"Diff":           TypeData,
+	"Dockerfile":     TypeData,
+	"Dogescript":     TypeProgramming,
+	"Dylan":          TypeProgramming,
+	"E":              TypeProgramming,
+	"EBNF":           TypeData,
+	"ECL":            TypeProgramming,
+	"ECLiPSe":        TypeProgramming,
+	"EJS":            TypeMarkup,
+	"EQ":             TypeProgramming,
+	"Eagle":          TypeMarkup,
+	"Ecere Projects": TypeData,
+	"Eiffel":         TypeProgramming,
+	"Elixir":         TypeProgramming,
+	"Elm":            TypeProgramming,
+	"Emacs Lisp":     TypeProgramming,
+	"EmberScript":    TypeProgramming,
+	"Erlang":         TypeProgramming,
+	"F#":             TypeProgramming,
+	"FLUX":           TypeProgramming,
+	"Factor":         TypeProgramming,
+	"Fancy":          TypeProgramming,
+	"Fantom":         TypeProgramming,
+	"Filebench WML":  TypeProgramming,
+	"Filterscript":   TypeProgramming,
+	"Formatted":      TypeData,
+	"Forth":          TypeProgramming,
+	"Fortran":        TypeProgramming,
+	"FreeMarker":     TypeProgramming,
+	"Frege":          TypeProgramming,
+	"G-code":         TypeData,
+	"GAMS":           TypeProgramming,
+	"GAP":            TypeProgramming,
+	"GCC Machine Description": TypeProgramming,
+	"GDB":      TypeProgramming,
+	"GDScript": TypeProgramming,
+	"GLSL":     TypeProgramming,
+	"GN":       TypeData,
+	"Game Maker Language":     TypeProgramming,
+	"Genie":                   TypeProgramming,
+	"Genshi":                  TypeProgramming,
+	"Gentoo Ebuild":           TypeProgramming,
+	"Gentoo Eclass":           TypeProgramming,
+	"Gettext Catalog":         TypeProse,
+	"Gherkin":                 TypeProgramming,
+	"Glyph":                   TypeProgramming,
+	"Gnuplot":                 TypeProgramming,
+	"Go":                      TypeProgramming,
+	"Golo":                    TypeProgramming,
+	"Gosu":                    TypeProgramming,
+	"Grace":                   TypeProgramming,
+	"Gradle":                  TypeData,
+	"Grammatical Framework":   TypeProgramming,
+	"Graph Modeling Language": TypeData,
+	"GraphQL":                 TypeData,
+	"Graphviz (DOT)":          TypeData,
+	"Groovy":                  TypeProgramming,
+	"Groovy Server Pages":     TypeProgramming,
+	"HCL":                      TypeProgramming,
+	"HLSL":                     TypeProgramming,
+	"HTML":                     TypeMarkup,
+	"HTML+Django":              TypeMarkup,
+	"HTML+ECR":                 TypeMarkup,
+	"HTML+EEX":                 TypeMarkup,
+	"HTML+ERB":                 TypeMarkup,
+	"HTML+PHP":                 TypeMarkup,
+	"HTTP":                     TypeData,
+	"Hack":                     TypeProgramming,
+	"Haml":                     TypeMarkup,
+	"Handlebars":               TypeMarkup,
+	"Harbour":                  TypeProgramming,
+	"Haskell":                  TypeProgramming,
+	"Haxe":                     TypeProgramming,
+	"Hy":                       TypeProgramming,
+	"HyPhy":                    TypeProgramming,
+	"IDL":                      TypeProgramming,
+	"IGOR Pro":                 TypeProgramming,
+	"INI":                      TypeData,
+	"IRC log":                  TypeData,
+	"Idris":                    TypeProgramming,
+	"Inform 7":                 TypeProgramming,
+	"Inno Setup":               TypeProgramming,
+	"Io":                       TypeProgramming,
+	"Ioke":                     TypeProgramming,
+	"Isabelle":                 TypeProgramming,
+	"Isabelle ROOT":            TypeProgramming,
+	"J":                        TypeProgramming,
+	"JFlex":                    TypeProgramming,
+	"JSON":                     TypeData,
+	"JSON5":                    TypeData,
+	"JSONLD":                   TypeData,
+	"JSONiq":                   TypeProgramming,
+	"JSX":                      TypeProgramming,
+	"Jasmin":                   TypeProgramming,
+	"Java":                     TypeProgramming,
+	"Java Server Pages":        TypeProgramming,
+	"JavaScript":               TypeProgramming,
+	"Jison":                    TypeProgramming,
+	"Jison Lex":                TypeProgramming,
+	"Julia":                    TypeProgramming,
+	"Jupyter Notebook":         TypeMarkup,
+	"KRL":                      TypeProgramming,
+	"KiCad":                    TypeProgramming,
+	"Kit":                      TypeMarkup,
+	"Kotlin":                   TypeProgramming,
+	"LFE":                      TypeProgramming,
+	"LLVM":                     TypeProgramming,
+	"LOLCODE":                  TypeProgramming,
+	"LSL":                      TypeProgramming,
+	"LabVIEW":                  TypeProgramming,
+	"Lasso":                    TypeProgramming,
+	"Latte":                    TypeMarkup,
+	"Lean":                     TypeProgramming,
+	"Less":                     TypeMarkup,
+	"Lex":                      TypeProgramming,
+	"LilyPond":                 TypeProgramming,
+	"Limbo":                    TypeProgramming,
+	"Linker Script":            TypeData,
+	"Linux Kernel Module":      TypeData,
+	"Liquid":                   TypeMarkup,
+	"Literate Agda":            TypeProgramming,
+	"Literate CoffeeScript":    TypeProgramming,
+	"Literate Haskell":         TypeProgramming,
+	"LiveScript":               TypeProgramming,
+	"Logos":                    TypeProgramming,
+	"Logtalk":                  TypeProgramming,
+	"LookML":                   TypeProgramming,
+	"LoomScript":               TypeProgramming,
+	"Lua":                      TypeProgramming,
+	"M":                        TypeProgramming,
+	"M4":                       TypeProgramming,
+	"M4Sugar":                  TypeProgramming,
+	"MAXScript":                TypeProgramming,
+	"MQL4":                     TypeProgramming,
+	"MQL5":                     TypeProgramming,
+	"MTML":                     TypeMarkup,
+	"MUF":                      TypeProgramming,
+	"Makefile":                 TypeProgramming,
+	"Mako":                     TypeProgramming,
+	"Markdown":                 TypeProse,
+	"Marko":                    TypeMarkup,
+	"Mask":                     TypeMarkup,
+	"Mathematica":              TypeProgramming,
+	"Matlab":                   TypeProgramming,
+	"Maven POM":                TypeData,
+	"Max":                      TypeProgramming,
+	"MediaWiki":                TypeProse,
+	"Mercury":                  TypeProgramming,
+	"Meson":                    TypeProgramming,
+	"Metal":                    TypeProgramming,
+	"MiniD":                    TypeProgramming,
+	"Mirah":                    TypeProgramming,
+	"Modelica":                 TypeProgramming,
+	"Modula-2":                 TypeProgramming,
+	"Module Management System": TypeProgramming,
+	"Monkey":                   TypeProgramming,
+	"Moocode":                  TypeProgramming,
+	"MoonScript":               TypeProgramming,
+	"Myghty":                   TypeProgramming,
+	"NCL":                      TypeProgramming,
+	"NL":                       TypeData,
+	"NSIS":                     TypeProgramming,
+	"Nemerle":                  TypeProgramming,
+	"NetLinx":                  TypeProgramming,
+	"NetLinx+ERB":              TypeProgramming,
+	"NetLogo":                  TypeProgramming,
+	"NewLisp":                  TypeProgramming,
+	"Nginx":                    TypeMarkup,
+	"Nim":                      TypeProgramming,
+	"Ninja":                    TypeData,
+	"Nit":                      TypeProgramming,
+	"Nix":                      TypeProgramming,
+	"Nu":                       TypeProgramming,
+	"NumPy":                    TypeProgramming,
+	"OCaml":                    TypeProgramming,
+	"ObjDump":                  TypeData,
+	"Objective-C":              TypeProgramming,
+	"Objective-C++":            TypeProgramming,
+	"Objective-J":              TypeProgramming,
+	"Omgrofl":                  TypeProgramming,
+	"Opa":                      TypeProgramming,
+	"Opal":                     TypeProgramming,
+	"OpenCL":                   TypeProgramming,
+	"OpenEdge ABL":             TypeProgramming,
+	"OpenRC runscript":         TypeProgramming,
+	"OpenSCAD":                 TypeProgramming,
+	"OpenType Feature File":    TypeData,
+	"Org":                            TypeProse,
+	"Ox":                             TypeProgramming,
+	"Oxygene":                        TypeProgramming,
+	"Oz":                             TypeProgramming,
+	"P4":                             TypeProgramming,
+	"PAWN":                           TypeProgramming,
+	"PHP":                            TypeProgramming,
+	"PLSQL":                          TypeProgramming,
+	"PLpgSQL":                        TypeProgramming,
+	"POV-Ray SDL":                    TypeProgramming,
+	"Pan":                            TypeProgramming,
+	"Papyrus":                        TypeProgramming,
+	"Parrot":                         TypeProgramming,
+	"Parrot Assembly":                TypeProgramming,
+	"Parrot Internal Representation": TypeProgramming,
+	"Pascal":                       TypeProgramming,
+	"Perl":                         TypeProgramming,
+	"Perl6":                        TypeProgramming,
+	"Pic":                          TypeMarkup,
+	"Pickle":                       TypeData,
+	"PicoLisp":                     TypeProgramming,
+	"PigLatin":                     TypeProgramming,
+	"Pike":                         TypeProgramming,
+	"Pod":                          TypeProse,
+	"PogoScript":                   TypeProgramming,
+	"Pony":                         TypeProgramming,
+	"PostScript":                   TypeMarkup,
+	"PowerBuilder":                 TypeProgramming,
+	"PowerShell":                   TypeProgramming,
+	"Processing":                   TypeProgramming,
+	"Prolog":                       TypeProgramming,
+	"Propeller Spin":               TypeProgramming,
+	"Protocol Buffer":              TypeMarkup,
+	"Public Key":                   TypeData,
+	"Pug":                          TypeMarkup,
+	"Puppet":                       TypeProgramming,
+	"Pure Data":                    TypeProgramming,
+	"PureBasic":                    TypeProgramming,
+	"PureScript":                   TypeProgramming,
+	"Python":                       TypeProgramming,
+	"Python console":               TypeProgramming,
+	"Python traceback":             TypeData,
+	"QML":                          TypeProgramming,
+	"QMake":                        TypeProgramming,
+	"R":                            TypeProgramming,
+	"RAML":                         TypeMarkup,
+	"RDoc":                         TypeProse,
+	"REALbasic":                    TypeProgramming,
+	"REXX":                         TypeProgramming,
+	"RHTML":                        TypeMarkup,
+	"RMarkdown":                    TypeProse,
+	"RPM Spec":                     TypeData,
+	"RUNOFF":                       TypeMarkup,
+	"Racket":                       TypeProgramming,
+	"Ragel":                        TypeProgramming,
+	"Rascal":                       TypeProgramming,
+	"Raw token data":               TypeData,
+	"Reason":                       TypeProgramming,
+	"Rebol":                        TypeProgramming,
+	"Red":                          TypeProgramming,
+	"Redcode":                      TypeProgramming,
+	"Regular Expression":           TypeData,
+	"Ren'Py":                       TypeProgramming,
+	"RenderScript":                 TypeProgramming,
+	"RobotFramework":               TypeProgramming,
+	"Roff":                         TypeMarkup,
+	"Rouge":                        TypeProgramming,
+	"Ruby":                         TypeProgramming,
+	"Rust":                         TypeProgramming,
+	"SAS":                          TypeProgramming,
+	"SCSS":                         TypeMarkup,
+	"SMT":                          TypeProgramming,
+	"SPARQL":                       TypeData,
+	"SQF":                          TypeProgramming,
+	"SQL":                          TypeData,
+	"SQLPL":                        TypeProgramming,
+	"SRecode Template":             TypeMarkup,
+	"STON":                         TypeData,
+	"SVG":                          TypeData,
+	"Sage":                         TypeProgramming,
+	"SaltStack":                    TypeProgramming,
+	"Sass":                         TypeMarkup,
+	"Scala":                        TypeProgramming,
+	"Scaml":                        TypeMarkup,
+	"Scheme":                       TypeProgramming,
+	"Scilab":                       TypeProgramming,
+	"Self":                         TypeProgramming,
+	"Shell":                        TypeProgramming,
+	"ShellSession":                 TypeProgramming,
+	"Shen":                         TypeProgramming,
+	"Slash":                        TypeProgramming,
+	"Slim":                         TypeMarkup,
+	"Smali":                        TypeProgramming,
+	"Smalltalk":                    TypeProgramming,
+	"Smarty":                       TypeProgramming,
+	"SourcePawn":                   TypeProgramming,
+	"Spline Font Database":         TypeData,
+	"Squirrel":                     TypeProgramming,
+	"Stan":                         TypeProgramming,
+	"Standard ML":                  TypeProgramming,
+	"Stata":                        TypeProgramming,
+	"Stylus":                       TypeMarkup,
+	"SubRip Text":                  TypeData,
+	"Sublime Text Config":          TypeData,
+	"SuperCollider":                TypeProgramming,
+	"Swift":                        TypeProgramming,
+	"SystemVerilog":                TypeProgramming,
+	"TI Program":                   TypeProgramming,
+	"TLA":                          TypeProgramming,
+	"TOML":                         TypeData,
+	"TXL":                          TypeProgramming,
+	"Tcl":                          TypeProgramming,
+	"Tcsh":                         TypeProgramming,
+	"TeX":                          TypeMarkup,
+	"Tea":                          TypeMarkup,
+	"Terra":                        TypeProgramming,
+	"Text":                         TypeProse,
+	"Textile":                      TypeProse,
+	"Thrift":                       TypeProgramming,
+	"Turing":                       TypeProgramming,
+	"Turtle":                       TypeData,
+	"Twig":                         TypeMarkup,
+	"TypeScript":                   TypeProgramming,
+	"Unified Parallel C":           TypeProgramming,
+	"Unity3D Asset":                TypeData,
+	"Unix Assembly":                TypeProgramming,
+	"Uno":                          TypeProgramming,
+	"UnrealScript":                 TypeProgramming,
+	"UrWeb":                        TypeProgramming,
+	"VCL":                          TypeProgramming,
+	"VHDL":                         TypeProgramming,
+	"Vala":                         TypeProgramming,
+	"Verilog":                      TypeProgramming,
+	"Vim script":                   TypeProgramming,
+	"Visual Basic":                 TypeProgramming,
+	"Volt":                         TypeProgramming,
+	"Vue":                          TypeMarkup,
+	"Wavefront Material":           TypeData,
+	"Wavefront Object":             TypeData,
+	"Web Ontology Language":        TypeMarkup,
+	"WebIDL":                       TypeProgramming,
+	"World of Warcraft Addon Data": TypeData,
+	"X10":              TypeProgramming,
+	"XC":               TypeProgramming,
+	"XCompose":         TypeData,
+	"XML":              TypeData,
+	"XPages":           TypeProgramming,
+	"XProc":            TypeProgramming,
+	"XQuery":           TypeProgramming,
+	"XS":               TypeProgramming,
+	"XSLT":             TypeProgramming,
+	"Xojo":             TypeProgramming,
+	"Xtend":            TypeProgramming,
+	"YAML":             TypeData,
+	"YANG":             TypeData,
+	"Yacc":             TypeProgramming,
+	"Zephir":           TypeProgramming,
+	"Zimpl":            TypeProgramming,
+	"desktop":          TypeData,
+	"eC":               TypeProgramming,
+	"edn":              TypeData,
+	"fish":             TypeProgramming,
+	"mupad":            TypeProgramming,
+	"nesC":             TypeProgramming,
+	"ooc":              TypeProgramming,
+	"reStructuredText": TypeProse,
+	"wisp":             TypeProgramming,
+	"xBase":            TypeProgramming,
+}

type_test.go

@@ -0,0 +1,32 @@
+package slinguist
+
+import . "gopkg.in/check.v1"
+
+func (s *TSuite) TestGetLanguageType(c *C) {
+	langType := GetLanguageType("BestLanguageEver")
+	c.Assert(langType, Equals, TypeUnknown)
+
+	langType = GetLanguageType("JSON")
+	c.Assert(langType, Equals, TypeData)
+
+	langType = GetLanguageType("COLLADA")
+	c.Assert(langType, Equals, TypeData)
+
+	langType = GetLanguageType("Go")
+	c.Assert(langType, Equals, TypeProgramming)
+
+	langType = GetLanguageType("Brainfuck")
+	c.Assert(langType, Equals, TypeProgramming)
+
+	langType = GetLanguageType("HTML")
+	c.Assert(langType, Equals, TypeMarkup)
+
+	langType = GetLanguageType("Sass")
+	c.Assert(langType, Equals, TypeMarkup)
+
+	langType = GetLanguageType("AsciiDoc")
+	c.Assert(langType, Equals, TypeProse)
+
+	langType = GetLanguageType("Textile")
+	c.Assert(langType, Equals, TypeProse)
+}

utils_test.go

@@ -21,13 +21,6 @@ func (s *TSuite) TestIsVendor(c *C) {
 func (s *TSuite) TestIsDocumentation(c *C) {
 	c.Assert(IsDocumentation("foo"), Equals, false)
 	c.Assert(IsDocumentation("README"), Equals, true)
-	c.Assert(IsDocumentation("samples/something"), Equals, true)
-	c.Assert(IsDocumentation("Docs/whatever"), Equals, true)
-	c.Assert(IsDocumentation("/javadoc/*"), Equals, true)
-	c.Assert(IsDocumentation("License"), Equals, true)
-	c.Assert(IsDocumentation("CONTRIBUTING.bar"), Equals, true)
-	c.Assert(IsDocumentation("/CHANGES"), Equals, true)
-	c.Assert(IsDocumentation("INSTALL"), Equals, true)
 }
 
 func (s *TSuite) TestIsConfiguration(c *C) {