Expose LanguageInfo with all Linguist data

As discussed in https://github.com/go-enry/go-enry/issues/54, this provides an
API for accessing a LanguageInfo struct which is populated with all the data
from the Linguist YAML source file. Functions are provided to access the
LanguageInfo by name or ID.

The other top-level functions like GetLanguageExtensions, GetLanguageGroup, etc.
could in principle be implemented using this structure, which would simplify the
code generation. But that would be a big change so I didn't do any of that.
Perhaps in the next major version something like that would make sense.

common.go

@@ -3,6 +3,7 @@ package enry
 import (
 	"bufio"
 	"bytes"
+	"fmt"
 	"path"
 	"path/filepath"
 	"strings"
@@ -522,11 +523,11 @@ type Type int
 
 // Type's values.
 const (
-	Unknown Type = iota
-	Data
-	Programming
-	Markup
-	Prose
+	Unknown     Type = Type(data.TypeUnknown)
+	Data             = Type(data.TypeData)
+	Programming      = Type(data.TypeProgramming)
+	Markup           = Type(data.TypeMarkup)
+	Prose            = Type(data.TypeProse)
 )
 
 // GetLanguageType returns the type of the given language.
@@ -558,3 +559,21 @@ func GetLanguageGroup(language string) string {
 
 	return ""
 }
+
+// GetLanguageInfo returns the LanguageInfo for a given language name, or an error if not found.
+func GetLanguageInfo(language string) (data.LanguageInfo, error) {
+	if info, ok := data.LanguageInfoByName[language]; ok {
+		return info, nil
+	}
+
+	return data.LanguageInfo{}, fmt.Errorf("language %q not found", language)
+}
+
+// GetLanguageInfo returns the LanguageInfo for a given language name, or an error if not found.
+func GetLanguageInfoByID(id int) (data.LanguageInfo, error) {
+	if info, ok := data.LanguageInfoByID[id]; ok {
+		return info, nil
+	}
+
+	return data.LanguageInfo{}, fmt.Errorf("language %q not found", id)
+}

common_test.go

@@ -642,3 +642,56 @@ func (s *EnryTestSuite) TestGetLanguageID() {
 		assert.Equal(s.T(), test.found, found, fmt.Sprintf("%v: found = %t, expected: %t", test.name, found, test.found))
 	}
 }
+
+func (s *EnryTestSuite) TestGetLanguageInfo() {
+	tests := []struct {
+		name       string
+		language   string
+		expectedID int
+		error      bool
+	}{
+		{name: "TestGetLanguageID_1", language: "1C Enterprise", expectedID: 0},
+		{name: "TestGetLanguageID_2", language: "BestLanguageEver", error: true},
+		{name: "TestGetLanguageID_3", language: "C++", expectedID: 43},
+		{name: "TestGetLanguageID_5", language: "Objective-C", expectedID: 257},
+		{name: "TestGetLanguageID_6", language: "golang", error: true}, // Aliases are not supported
+		{name: "TestGetLanguageID_7", language: "Go", expectedID: 132},
+		{name: "TestGetLanguageID_8", language: "Makefile", expectedID: 220},
+	}
+
+	for _, test := range tests {
+		info, err := GetLanguageInfo(test.language)
+		if test.error {
+			assert.Error(s.T(), err, "%v: expected error for %q", test.name, test.language)
+		} else {
+			assert.NoError(s.T(), err)
+			assert.Equal(s.T(), test.expectedID, info.LanguageID, fmt.Sprintf("%v: id = %v, expected: %v", test.name, info.LanguageID, test.expectedID))
+		}
+	}
+}
+
+func (s *EnryTestSuite) TestGetLanguageInfoByID() {
+	tests := []struct {
+		name         string
+		id           int
+		expectedName string
+		error        bool
+	}{
+		{name: "TestGetLanguageID_1", id: 0, expectedName: "1C Enterprise"},
+		{name: "TestGetLanguageID_2", id: -1, error: true},
+		{name: "TestGetLanguageID_3", id: 43, expectedName: "C++"},
+		{name: "TestGetLanguageID_5", id: 257, expectedName: "Objective-C"},
+		{name: "TestGetLanguageID_7", id: 132, expectedName: "Go"},
+		{name: "TestGetLanguageID_8", id: 220, expectedName: "Makefile"},
+	}
+
+	for _, test := range tests {
+		info, err := GetLanguageInfoByID(test.id)
+		if test.error {
+			assert.Error(s.T(), err, "%v: expected error for %q", test.name, test.id)
+		} else {
+			assert.NoError(s.T(), err)
+			assert.Equal(s.T(), test.expectedName, info.Name, fmt.Sprintf("%v: id = %v, expected: %v", test.name, test.id, test.expectedName))
+		}
+	}
+}

data/type.go

@@ -3,6 +3,48 @@
 
 package data
 
+// Type represent language's type. Either data, programming, markup, prose, or unknown.
+type Type int
+
+// Type's values.
+const (
+	TypeUnknown Type = iota
+	TypeData
+	TypeProgramming
+	TypeMarkup
+	TypeProse
+)
+
+func (t Type) String() string {
+	switch t {
+	case TypeData:
+		return "data"
+	case TypeProgramming:
+		return "programming"
+	case TypeMarkup:
+		return "markup"
+	case TypeProse:
+		return "prose"
+	default:
+		return "unknown"
+	}
+}
+
+func TypeForString(s string) Type {
+	switch s {
+	case "data":
+		return TypeData
+	case "programming":
+		return TypeProgramming
+	case "markup":
+		return TypeMarkup
+	case "prose":
+		return TypeProse
+	default:
+		return TypeUnknown
+	}
+}
+
 var LanguagesType = map[string]int{
 	"1C Enterprise":                      2,
 	"4D":                                 2,

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

@@ -0,0 +1,99 @@
+package data
+
+// LanguageInfo exposes the data for a language's Linguist YAML entry as a Go struct.
+// See https://github.com/github/linguist/blob/master/lib/linguist/languages.yml
+type LanguageInfo struct {
+  Name           string
+  FSName         string
+	Type           Type
+	Color          string
+	Group          string
+	Aliases        []string
+	Extensions     []string
+	Interpreters   []string
+	Filenames      []string
+	MimeType       string
+  TMScope        string
+  AceMode        string
+  CodemirrorMode string
+  Wrap           bool
+	LanguageID     int
+}
+
+// LanguageInfoByName allows accessing LanguageInfo by a language's primary name.
+var LanguageInfoByName = map[string]LanguageInfo{
+	{{range $language, $info := . -}}
+	"{{$language}}": LanguageInfo{
+    Name: "{{$language}}",
+    FSName: "{{$info.FSName}}",
+		Type: TypeForString("{{$info.Type}}"),
+		Color: "{{$info.Color}}",
+		Group: "{{$info.Group}}",
+		Aliases: []string{
+			{{range $alias := $info.Aliases -}}
+			"{{$alias}}",
+			{{end -}}
+		},
+		Extensions: []string{
+			{{range $extension := $info.Extensions -}}
+			"{{$extension}}",
+			{{end -}}
+		},
+		Interpreters: []string{
+			{{range $interpreter := $info.Interpreters -}}
+			"{{$interpreter}}",
+			{{end -}}
+		},
+		Filenames: []string{
+			{{range $filename := $info.Filenames -}}
+			"{{$filename}}",
+			{{end -}}
+		},
+		MimeType: "{{$info.MimeType}}",
+    TMScope: "{{$info.TMScope}}",
+    AceMode: "{{$info.AceMode}}",
+    CodemirrorMode: "{{$info.CodemirrorMode}}",
+    Wrap: {{$info.Wrap}},
+		LanguageID: {{$info.LanguageID}},
+	},
+	{{end -}}
+}
+
+// LanguageInfoByID allows accessing LanguageInfo by a language's ID.
+var LanguageInfoByID = map[int]LanguageInfo{
+	{{range $language, $info := . -}}
+	{{$info.LanguageID}}: LanguageInfo{
+    Name: "{{$language}}",
+    FSName: "{{$info.FSName}}",
+		Type: TypeForString("{{$info.Type}}"),
+		Color: "{{$info.Color}}",
+		Group: "{{$info.Group}}",
+		Aliases: []string{
+			{{range $alias := $info.Aliases -}}
+			"{{$alias}}",
+			{{end -}}
+		},
+		Extensions: []string{
+			{{range $extension := $info.Extensions -}}
+			"{{$extension}}",
+			{{end -}}
+		},
+		Interpreters: []string{
+			{{range $interpreter := $info.Interpreters -}}
+			"{{$interpreter}}",
+			{{end -}}
+		},
+		Filenames: []string{
+			{{range $filename := $info.Filenames -}}
+			"{{$filename}}",
+			{{end -}}
+		},
+		MimeType: "{{$info.MimeType}}",
+    TMScope: "{{$info.TMScope}}",
+    AceMode: "{{$info.AceMode}}",
+    CodemirrorMode: "{{$info.CodemirrorMode}}",
+    Wrap: {{$info.Wrap}},
+		LanguageID: {{$info.LanguageID}},
+	},
+	{{end -}}
+}

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

@@ -1,5 +1,47 @@
 package data
 
+// Type represent language's type. Either data, programming, markup, prose, or unknown.
+type Type int
+
+// Type's values.
+const (
+	TypeUnknown Type = iota
+	TypeData
+	TypeProgramming
+	TypeMarkup
+	TypeProse
+)
+
+func (t Type) String() string {
+  switch t {
+  case TypeData:
+    return "data"
+  case TypeProgramming:
+    return "programming"
+  case TypeMarkup:
+    return "markup"
+  case TypeProse:
+    return "prose"
+  default:
+    return "unknown"
+  }
+}
+
+func TypeForString(s string) Type {
+  switch s {
+  case "data":
+    return TypeData
+  case "programming":
+    return TypeProgramming
+  case "markup":
+    return TypeMarkup
+  case "prose":
+    return TypeProse
+  default:
+    return TypeUnknown
+  }
+}
+
 var LanguagesType = map[string]int{
 	{{range $language, $type := . -}}
 	"{{ $language }}": {{ $type -}},

internal/code-generator/generator/langinfo.go

@@ -1,17 +1,29 @@
 package generator
 
-import "sort"
+import (
+	"bytes"
+	"io"
+	"io/ioutil"
+	"sort"
+
+	"gopkg.in/yaml.v2"
+)
 
 type languageInfo struct {
-	Type         string   `yaml:"type,omitempty"`
-	Color        string   `yaml:"color,omitempty"`
-	Group        string   `yaml:"group,omitempty"`
-	Aliases      []string `yaml:"aliases,omitempty"`
-	Extensions   []string `yaml:"extensions,omitempty,flow"`
-	Interpreters []string `yaml:"interpreters,omitempty,flow"`
-	Filenames    []string `yaml:"filenames,omitempty,flow"`
-	MimeType     string   `yaml:"codemirror_mime_type,omitempty,flow"`
-	LanguageID   *int     `yaml:"language_id,omitempty"`
+	FSName         string   `yaml:"fs_name"`
+	Type           string   `yaml:"type,omitempty"`
+	Color          string   `yaml:"color,omitempty"`
+	Group          string   `yaml:"group,omitempty"`
+	Aliases        []string `yaml:"aliases,omitempty"`
+	Extensions     []string `yaml:"extensions,omitempty,flow"`
+	Interpreters   []string `yaml:"interpreters,omitempty,flow"`
+	Filenames      []string `yaml:"filenames,omitempty,flow"`
+	MimeType       string   `yaml:"codemirror_mime_type,omitempty,flow"`
+	TMScope        string   `yaml:"tm_scope"`
+	AceMode        string   `yaml:"ace_mode"`
+	CodemirrorMode string   `yaml:"codemirror_mode"`
+	Wrap           bool     `yaml:"wrap"`
+	LanguageID     *int     `yaml:"language_id,omitempty"`
 }
 
 func getAlphabeticalOrderedKeys(languages map[string]*languageInfo) []string {
@@ -23,3 +35,28 @@ func getAlphabeticalOrderedKeys(languages map[string]*languageInfo) []string {
 	sort.Strings(keyList)
 	return keyList
 }
+
+// LanguageInfo generates maps in Go with language name -> LanguageInfo and language ID -> LanguageInfo.
+// It is of generator.File type.
+func LanguageInfo(fileToParse, samplesDir, outPath, tmplPath, tmplName, commit string) error {
+	data, err := ioutil.ReadFile(fileToParse)
+	if err != nil {
+		return err
+	}
+
+	languages := make(map[string]*languageInfo)
+	if err := yaml.Unmarshal(data, &languages); err != nil {
+		return err
+	}
+
+	buf := &bytes.Buffer{}
+	if err := executeLanguageInfoTemplate(buf, languages, tmplPath, tmplName, commit); err != nil {
+		return err
+	}
+
+	return formatedWrite(outPath, buf.Bytes())
+}
+
+func executeLanguageInfoTemplate(out io.Writer, languages map[string]*languageInfo, tmplPath, tmplName, commit string) error {
+	return executeTemplate(out, tmplName, tmplPath, commit, nil, languages) // TODO: use function map to put language by id?
+}

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

@@ -3,6 +3,48 @@
 
 package data
 
+// Type represent language's type. Either data, programming, markup, prose, or unknown.
+type Type int
+
+// Type's values.
+const (
+	TypeUnknown Type = iota
+	TypeData
+	TypeProgramming
+	TypeMarkup
+	TypeProse
+)
+
+func (t Type) String() string {
+	switch t {
+	case TypeData:
+		return "data"
+	case TypeProgramming:
+		return "programming"
+	case TypeMarkup:
+		return "markup"
+	case TypeProse:
+		return "prose"
+	default:
+		return "unknown"
+	}
+}
+
+func TypeForString(s string) Type {
+	switch s {
+	case "data":
+		return TypeData
+	case "programming":
+		return TypeProgramming
+	case "markup":
+		return TypeMarkup
+	case "prose":
+		return TypeProse
+	default:
+		return TypeUnknown
+	}
+}
+
 var LanguagesType = map[string]int{
 	"1C Enterprise":                      2,
 	"4D":                                 2,

internal/code-generator/generator/types.go

@@ -2,9 +2,10 @@ package generator
 
 import (
 	"bytes"
-	"gopkg.in/yaml.v2"
 	"io"
 	"io/ioutil"
+
+	"gopkg.in/yaml.v2"
 )
 
 var typeToTypeConst = map[string]int{

internal/code-generator/main.go

@@ -87,9 +87,14 @@ var (
 
 	// id.go generation
 	idFile     = "data/id.go"
-	idTmplPath = "internal/code-generator/assets/id.go.tmpl"
+	idTmplPath = filepath.Join(assetsDir, "id.go.tmpl")
 	idTmpl     = "id.go.tmpl"
 
+	// languageInfo.go generation
+	languageInfoFile     = filepath.Join("data", "languageInfo.go")
+	langaugeInfoTmplPath = filepath.Join(assetsDir, "languageInfo.go.tmpl")
+	langaugeInfoTmpl     = "languageInfo.go.tmpl"
+
 	commitPath = filepath.Join(".linguist", ".git", "HEAD")
 )
 
@@ -124,11 +129,12 @@ func main() {
 		{generator.Colors, languagesYAML, "", colorsFile, colorsTmplPath, colorsTmpl, commit},
 		{generator.Groups, languagesYAML, "", groupsFile, groupsTmplPath, groupsTmpl, commit},
 		{generator.ID, languagesYAML, "", idFile, idTmplPath, idTmpl, commit},
+		{generator.LanguageInfo, languagesYAML, "", languageInfoFile, langaugeInfoTmplPath, langaugeInfoTmpl, commit},
 	}
 
 	for _, file := range fileList {
 		if err := file.generate(file.fileToParse, file.samplesDir, file.outPath, file.tmplPath, file.tmplName, file.commit); err != nil {
-			log.Println(err)
+			log.Printf("Error generating template %q to %q: %+v", file.tmplPath, file.outPath, err)
 		}
 	}
 }