Merge pull request #73 from dpaz/issue72

Cli to analyze a single file

cli/enry/main.go

@@ -9,6 +9,7 @@ import (
 	"log"
 	"os"
 	"path/filepath"
+	"strings"
 
 	"gopkg.in/src-d/enry.v1"
 	"gopkg.in/src-d/enry.v1/data"
@@ -48,6 +49,7 @@ func main() {
 		}
 
 		if relativePath == "." {
+			fmt.Print(printFileAnalysis(root))
 			return nil
 		}
 
@@ -154,6 +156,44 @@ func printPercents(out map[string][]string, buff *bytes.Buffer) {
 	}
 }
 
+func printFileAnalysis(file string) string {
+	content, err := ioutil.ReadFile(file)
+	if err != nil {
+		fmt.Println(err)
+	}
+
+	totalLines, nonBlank := getLines(file, string(content))
+	fileType := getFileType(file, content)
+	language := enry.GetLanguage(file, content)
+	mimeType := enry.GetMimeType(file, language)
+
+	return fmt.Sprintf(
+		`%s: %d lines (%d sloc)
+  type:      %s
+  mime_type: %s
+  language:  %s
+`,
+		filepath.Base(file), totalLines, nonBlank, fileType, mimeType, language,
+	)
+}
+
+func getLines(file string, content string) (int, int) {
+	totalLines := strings.Count(content, "\n")
+	nonBlank := totalLines - strings.Count(content, "\n\n")
+	return totalLines, nonBlank
+}
+
+func getFileType(file string, content []byte) string {
+	switch {
+	case enry.IsImage(file):
+		return "Image"
+	case enry.IsBinary(content):
+		return "Binary"
+	default:
+		return "Text"
+	}
+}
+
 func writeStringLn(s string, buff *bytes.Buffer) {
 	buff.WriteString(s)
 	buff.WriteByte('\n')

data/mimeType.go

@@ -0,0 +1,204 @@
+package data
+
+// CODE GENERATED AUTOMATICALLY WITH gopkg.in/src-d/enry.v1/internal/code-generator
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: 37979b26b04e10868017469e5cc56263b0a39c84
+
+var LanguagesMime = map[string]string{
+	"AGS Script":              "text/x-c++src",
+	"APL":                     "text/apl",
+	"ASN.1":                   "text/x-ttcn-asn",
+	"ASP":                     "application/x-aspx",
+	"Alpine Abuild":           "text/x-sh",
+	"Ant Build System":        "application/xml",
+	"Apex":                    "text/x-java",
+	"Arduino":                 "text/x-c++src",
+	"Brainfuck":               "text/x-brainfuck",
+	"C":                       "text/x-csrc",
+	"C#":                      "text/x-csharp",
+	"C++":                     "text/x-c++src",
+	"C2hs Haskell":            "text/x-haskell",
+	"CMake":                   "text/x-cmake",
+	"COBOL":                   "text/x-cobol",
+	"COLLADA":                 "text/xml",
+	"CSON":                    "text/x-coffeescript",
+	"CSS":                     "text/css",
+	"ChucK":                   "text/x-java",
+	"Clojure":                 "text/x-clojure",
+	"Closure Templates":       "text/x-soy",
+	"CoffeeScript":            "text/x-coffeescript",
+	"Common Lisp":             "text/x-common-lisp",
+	"Component Pascal":        "text/x-pascal",
+	"Crystal":                 "text/x-crystal",
+	"Cuda":                    "text/x-c++src",
+	"Cycript":                 "text/javascript",
+	"Cython":                  "text/x-cython",
+	"D":                       "text/x-d",
+	"DTrace":                  "text/x-csrc",
+	"Dart":                    "application/dart",
+	"Diff":                    "text/x-diff",
+	"Dockerfile":              "text/x-dockerfile",
+	"Dylan":                   "text/x-dylan",
+	"EBNF":                    "text/x-ebnf",
+	"ECL":                     "text/x-ecl",
+	"EQ":                      "text/x-csharp",
+	"Eagle":                   "text/xml",
+	"Easybuild":               "text/x-python",
+	"Ecere Projects":          "application/json",
+	"Eiffel":                  "text/x-eiffel",
+	"Elm":                     "text/x-elm",
+	"Emacs Lisp":              "text/x-common-lisp",
+	"EmberScript":             "text/x-coffeescript",
+	"Erlang":                  "text/x-erlang",
+	"F#":                      "text/x-fsharp",
+	"Factor":                  "text/x-factor",
+	"Forth":                   "text/x-forth",
+	"Fortran":                 "text/x-fortran",
+	"GCC Machine Description": "text/x-common-lisp",
+	"GN": "text/x-python",
+	"Game Maker Language": "text/x-c++src",
+	"Genshi":              "text/xml",
+	"Gentoo Ebuild":       "text/x-sh",
+	"Gentoo Eclass":       "text/x-sh",
+	"Glyph":               "text/x-tcl",
+	"Go":                  "text/x-go",
+	"Grammatical Framework": "text/x-haskell",
+	"Groovy":                "text/x-groovy",
+	"Groovy Server Pages":   "application/x-jsp",
+	"HCL":                 "text/x-ruby",
+	"HTML":                "text/html",
+	"HTML+Django":         "text/x-django",
+	"HTML+ECR":            "text/html",
+	"HTML+EEX":            "text/html",
+	"HTML+ERB":            "application/x-erb",
+	"HTML+PHP":            "application/x-httpd-php",
+	"HTTP":                "message/http",
+	"Hack":                "application/x-httpd-php",
+	"Haml":                "text/x-haml",
+	"Haskell":             "text/x-haskell",
+	"Haxe":                "text/x-haxe",
+	"IDL":                 "text/x-idl",
+	"INI":                 "text/x-properties",
+	"IRC log":             "text/mirc",
+	"JSON":                "application/json",
+	"JSON5":               "application/json",
+	"JSONiq":              "application/json",
+	"JSX":                 "text/jsx",
+	"Java":                "text/x-java",
+	"Java Server Pages":   "application/x-jsp",
+	"JavaScript":          "text/javascript",
+	"Julia":               "text/x-julia",
+	"Jupyter Notebook":    "application/json",
+	"Kit":                 "text/html",
+	"Kotlin":              "text/x-kotlin",
+	"LFE":                 "text/x-common-lisp",
+	"LabVIEW":             "text/xml",
+	"Latte":               "text/x-smarty",
+	"Less":                "text/css",
+	"Literate Haskell":    "text/x-literate-haskell",
+	"LiveScript":          "text/x-livescript",
+	"LookML":              "text/x-yaml",
+	"Lua":                 "text/x-lua",
+	"M":                   "text/x-mumps",
+	"MTML":                "text/html",
+	"MUF":                 "text/x-forth",
+	"Makefile":            "text/x-cmake",
+	"Markdown":            "text/x-gfm",
+	"Marko":               "text/html",
+	"Mathematica":         "text/x-mathematica",
+	"Matlab":              "text/x-octave",
+	"Maven POM":           "text/xml",
+	"Max":                 "application/json",
+	"Metal":               "text/x-c++src",
+	"Mirah":               "text/x-ruby",
+	"Modelica":            "text/x-modelica",
+	"NSIS":                "text/x-nsis",
+	"NetLogo":             "text/x-common-lisp",
+	"NewLisp":             "text/x-common-lisp",
+	"Nginx":               "text/x-nginx-conf",
+	"Nu":                  "text/x-scheme",
+	"NumPy":               "text/x-python",
+	"OCaml":               "text/x-ocaml",
+	"Objective-C":         "text/x-objectivec",
+	"Objective-C++":       "text/x-objectivec",
+	"OpenCL":              "text/x-csrc",
+	"OpenRC runscript":    "text/x-sh",
+	"Oz":                  "text/x-oz",
+	"PHP":                 "application/x-httpd-php",
+	"PLSQL":               "text/x-plsql",
+	"PLpgSQL":             "text/x-sql",
+	"Pascal":              "text/x-pascal",
+	"Perl":                "text/x-perl",
+	"Perl 6":              "text/x-perl",
+	"Pic":                 "text/troff",
+	"Pod":                 "text/x-perl",
+	"PowerShell":          "application/x-powershell",
+	"Protocol Buffer":     "text/x-protobuf",
+	"Public Key":          "application/pgp",
+	"Pug":                 "text/x-pug",
+	"Puppet":              "text/x-puppet",
+	"PureScript":          "text/x-haskell",
+	"Python":              "text/x-python",
+	"R":                   "text/x-rsrc",
+	"RAML":                "text/x-yaml",
+	"RHTML":               "application/x-erb",
+	"RMarkdown":           "text/x-gfm",
+	"RPM Spec":            "text/x-rpm-spec",
+	"Reason":              "text/x-rustsrc",
+	"Roff":                "text/troff",
+	"Rouge":               "text/x-clojure",
+	"Ruby":                "text/x-ruby",
+	"Rust":                "text/x-rustsrc",
+	"SAS":                 "text/x-sas",
+	"SCSS":                "text/x-scss",
+	"SPARQL":              "application/sparql-query",
+	"SQL":                 "text/x-sql",
+	"SQLPL":               "text/x-sql",
+	"SRecode Template":    "text/x-common-lisp",
+	"SVG":                 "text/xml",
+	"Sage":                "text/x-python",
+	"SaltStack":           "text/x-yaml",
+	"Sass":                "text/x-sass",
+	"Scala":               "text/x-scala",
+	"Scheme":              "text/x-scheme",
+	"Shell":               "text/x-sh",
+	"ShellSession":        "text/x-sh",
+	"Slim":                "text/x-slim",
+	"Smalltalk":           "text/x-stsrc",
+	"Smarty":              "text/x-smarty",
+	"Squirrel":            "text/x-c++src",
+	"Standard ML":         "text/x-ocaml",
+	"Sublime Text Config": "text/javascript",
+	"Swift":               "text/x-swift",
+	"SystemVerilog":       "text/x-systemverilog",
+	"TOML":                "text/x-toml",
+	"Tcl":                 "text/x-tcl",
+	"Tcsh":                "text/x-sh",
+	"TeX":                 "text/x-stex",
+	"Terra":               "text/x-lua",
+	"Textile":             "text/x-textile",
+	"Turtle":              "text/turtle",
+	"Twig":                "text/x-twig",
+	"TypeScript":          "application/typescript",
+	"Unified Parallel C":  "text/x-csrc",
+	"Unity3D Asset":       "text/x-yaml",
+	"Uno":                 "text/x-csharp",
+	"UnrealScript":        "text/x-java",
+	"VHDL":                "text/x-vhdl",
+	"Verilog":             "text/x-verilog",
+	"Visual Basic":        "text/x-vb",
+	"Volt":                "text/x-d",
+	"WebAssembly":         "text/x-common-lisp",
+	"WebIDL":              "text/x-webidl",
+	"XC":                  "text/x-csrc",
+	"XML":                 "text/xml",
+	"XPages":              "text/xml",
+	"XProc":               "text/xml",
+	"XQuery":              "application/xquery",
+	"XS":                  "text/x-csrc",
+	"XSLT":                "text/xml",
+	"YAML":                "text/x-yaml",
+	"edn":                 "text/x-clojure",
+	"reStructuredText":    "text/x-rst",
+	"wisp":                "text/x-clojure",
+}

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

@@ -0,0 +1,11 @@
+package data
+
+// CODE GENERATED AUTOMATICALLY WITH gopkg.in/src-d/enry.v1/internal/code-generator
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: {{ getCommit }}
+
+var LanguagesMime = map[string]string{
+  {{range $language, $mime := . -}}
+  "{{$language}}": "{{$mime -}}",
+  {{end -}}
+}

internal/code-generator/generator/generator_test.go

@@ -70,6 +70,11 @@ const (
 	commitGold         = "test_files/commit.gold"
 	commitTestTmplPath = "../assets/commit.go.tmpl"
 	commitTestTmplName = "commit.go.tmpl"
+
+	// mime test
+	mimeTypeGold         = "test_files/mimeType.gold"
+	mimeTypeTestTmplPath = "../assets/mimeType.go.tmpl"
+	mimeTypeTestTmplName = "mimeType.go.tmpl"
 )
 
 type GeneratorTestSuite struct {
@@ -218,6 +223,16 @@ func (s *GeneratorTestSuite) TestGenerationFiles() {
 			generate:   Commit,
 			wantOut:    commitGold,
 		},
+		{
+			name:        "MimeType()",
+			fileToParse: filepath.Join(s.tmpLinguist, languagesFile),
+			samplesDir:  "",
+			tmplPath:    mimeTypeTestTmplPath,
+			tmplName:    mimeTypeTestTmplName,
+			commit:      commit,
+			generate:    MimeType,
+			wantOut:     mimeTypeGold,
+		},
 	}
 
 	for _, test := range tests {
@@ -227,7 +242,6 @@ func (s *GeneratorTestSuite) TestGenerationFiles() {
 		outPath, err := ioutil.TempFile("/tmp", "generator-test-")
 		assert.NoError(s.T(), err)
 		defer os.Remove(outPath.Name())
-
 		err = test.generate(test.fileToParse, test.samplesDir, outPath.Name(), test.tmplPath, test.tmplName, test.commit)
 		assert.NoError(s.T(), err)
 		out, err := ioutil.ReadFile(outPath.Name())

internal/code-generator/generator/langinfo.go

@@ -8,6 +8,7 @@ type languageInfo struct {
 	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"`
 }
 
 func getAlphabeticalOrderedKeys(languages map[string]*languageInfo) []string {

internal/code-generator/generator/mimeType.go

@@ -0,0 +1,55 @@
+package generator
+
+import (
+	"bytes"
+	"io"
+	"io/ioutil"
+	"text/template"
+
+	yaml "gopkg.in/yaml.v2"
+)
+
+func MimeType(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
+	}
+
+	langMimeMap := buildLanguageMimeMap(languages)
+
+	buf := &bytes.Buffer{}
+	if err := executeMimeTemplate(buf, langMimeMap, tmplPath, tmplName, commit); err != nil {
+		return err
+	}
+
+	return formatedWrite(outPath, buf.Bytes())
+}
+
+func buildLanguageMimeMap(languages map[string]*languageInfo) map[string]string {
+	langMimeMap := make(map[string]string)
+	for lang, info := range languages {
+		if len(info.MimeType) != 0 {
+			langMimeMap[lang] = info.MimeType
+		}
+	}
+
+	return langMimeMap
+}
+
+func executeMimeTemplate(out io.Writer, langMimeMap map[string]string, tmplPath, tmplName, commit string) error {
+	fmap := template.FuncMap{
+		"getCommit": func() string { return commit },
+	}
+
+	t := template.Must(template.New(tmplName).Funcs(fmap).ParseFiles(tmplPath))
+	if err := t.Execute(out, langMimeMap); err != nil {
+		return err
+	}
+
+	return nil
+}

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

@@ -0,0 +1,204 @@
+package data
+
+// CODE GENERATED AUTOMATICALLY WITH gopkg.in/src-d/enry.v1/internal/code-generator
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: d5c8db3fb91963c4b2762ca2ea2ff7cfac109f68
+
+var LanguagesMime = map[string]string{
+	"AGS Script":              "text/x-c++src",
+	"APL":                     "text/apl",
+	"ASN.1":                   "text/x-ttcn-asn",
+	"ASP":                     "application/x-aspx",
+	"Alpine Abuild":           "text/x-sh",
+	"Ant Build System":        "application/xml",
+	"Apex":                    "text/x-java",
+	"Arduino":                 "text/x-c++src",
+	"Brainfuck":               "text/x-brainfuck",
+	"C":                       "text/x-csrc",
+	"C#":                      "text/x-csharp",
+	"C++":                     "text/x-c++src",
+	"C2hs Haskell":            "text/x-haskell",
+	"CMake":                   "text/x-cmake",
+	"COBOL":                   "text/x-cobol",
+	"COLLADA":                 "text/xml",
+	"CSON":                    "text/x-coffeescript",
+	"CSS":                     "text/css",
+	"ChucK":                   "text/x-java",
+	"Clojure":                 "text/x-clojure",
+	"Closure Templates":       "text/x-soy",
+	"CoffeeScript":            "text/x-coffeescript",
+	"Common Lisp":             "text/x-common-lisp",
+	"Component Pascal":        "text/x-pascal",
+	"Crystal":                 "text/x-crystal",
+	"Cuda":                    "text/x-c++src",
+	"Cycript":                 "text/javascript",
+	"Cython":                  "text/x-cython",
+	"D":                       "text/x-d",
+	"DTrace":                  "text/x-csrc",
+	"Dart":                    "application/dart",
+	"Diff":                    "text/x-diff",
+	"Dockerfile":              "text/x-dockerfile",
+	"Dylan":                   "text/x-dylan",
+	"EBNF":                    "text/x-ebnf",
+	"ECL":                     "text/x-ecl",
+	"EQ":                      "text/x-csharp",
+	"Eagle":                   "text/xml",
+	"Easybuild":               "text/x-python",
+	"Ecere Projects":          "application/json",
+	"Eiffel":                  "text/x-eiffel",
+	"Elm":                     "text/x-elm",
+	"Emacs Lisp":              "text/x-common-lisp",
+	"EmberScript":             "text/x-coffeescript",
+	"Erlang":                  "text/x-erlang",
+	"F#":                      "text/x-fsharp",
+	"Factor":                  "text/x-factor",
+	"Forth":                   "text/x-forth",
+	"Fortran":                 "text/x-fortran",
+	"GCC Machine Description": "text/x-common-lisp",
+	"GN": "text/x-python",
+	"Game Maker Language": "text/x-c++src",
+	"Genshi":              "text/xml",
+	"Gentoo Ebuild":       "text/x-sh",
+	"Gentoo Eclass":       "text/x-sh",
+	"Glyph":               "text/x-tcl",
+	"Go":                  "text/x-go",
+	"Grammatical Framework": "text/x-haskell",
+	"Groovy":                "text/x-groovy",
+	"Groovy Server Pages":   "application/x-jsp",
+	"HCL":                 "text/x-ruby",
+	"HTML":                "text/html",
+	"HTML+Django":         "text/x-django",
+	"HTML+ECR":            "text/html",
+	"HTML+EEX":            "text/html",
+	"HTML+ERB":            "application/x-erb",
+	"HTML+PHP":            "application/x-httpd-php",
+	"HTTP":                "message/http",
+	"Hack":                "application/x-httpd-php",
+	"Haml":                "text/x-haml",
+	"Haskell":             "text/x-haskell",
+	"Haxe":                "text/x-haxe",
+	"IDL":                 "text/x-idl",
+	"INI":                 "text/x-properties",
+	"IRC log":             "text/mirc",
+	"JSON":                "application/json",
+	"JSON5":               "application/json",
+	"JSONiq":              "application/json",
+	"JSX":                 "text/jsx",
+	"Java":                "text/x-java",
+	"Java Server Pages":   "application/x-jsp",
+	"JavaScript":          "text/javascript",
+	"Julia":               "text/x-julia",
+	"Jupyter Notebook":    "application/json",
+	"Kit":                 "text/html",
+	"Kotlin":              "text/x-kotlin",
+	"LFE":                 "text/x-common-lisp",
+	"LabVIEW":             "text/xml",
+	"Latte":               "text/x-smarty",
+	"Less":                "text/css",
+	"Literate Haskell":    "text/x-literate-haskell",
+	"LiveScript":          "text/x-livescript",
+	"LookML":              "text/x-yaml",
+	"Lua":                 "text/x-lua",
+	"M":                   "text/x-mumps",
+	"MTML":                "text/html",
+	"MUF":                 "text/x-forth",
+	"Makefile":            "text/x-cmake",
+	"Markdown":            "text/x-gfm",
+	"Marko":               "text/html",
+	"Mathematica":         "text/x-mathematica",
+	"Matlab":              "text/x-octave",
+	"Maven POM":           "text/xml",
+	"Max":                 "application/json",
+	"Metal":               "text/x-c++src",
+	"Mirah":               "text/x-ruby",
+	"Modelica":            "text/x-modelica",
+	"NSIS":                "text/x-nsis",
+	"NetLogo":             "text/x-common-lisp",
+	"NewLisp":             "text/x-common-lisp",
+	"Nginx":               "text/x-nginx-conf",
+	"Nu":                  "text/x-scheme",
+	"NumPy":               "text/x-python",
+	"OCaml":               "text/x-ocaml",
+	"Objective-C":         "text/x-objectivec",
+	"Objective-C++":       "text/x-objectivec",
+	"OpenCL":              "text/x-csrc",
+	"OpenRC runscript":    "text/x-sh",
+	"Oz":                  "text/x-oz",
+	"PHP":                 "application/x-httpd-php",
+	"PLSQL":               "text/x-plsql",
+	"PLpgSQL":             "text/x-sql",
+	"Pascal":              "text/x-pascal",
+	"Perl":                "text/x-perl",
+	"Perl 6":              "text/x-perl",
+	"Pic":                 "text/troff",
+	"Pod":                 "text/x-perl",
+	"PowerShell":          "application/x-powershell",
+	"Protocol Buffer":     "text/x-protobuf",
+	"Public Key":          "application/pgp",
+	"Pug":                 "text/x-pug",
+	"Puppet":              "text/x-puppet",
+	"PureScript":          "text/x-haskell",
+	"Python":              "text/x-python",
+	"R":                   "text/x-rsrc",
+	"RAML":                "text/x-yaml",
+	"RHTML":               "application/x-erb",
+	"RMarkdown":           "text/x-gfm",
+	"RPM Spec":            "text/x-rpm-spec",
+	"Reason":              "text/x-rustsrc",
+	"Roff":                "text/troff",
+	"Rouge":               "text/x-clojure",
+	"Ruby":                "text/x-ruby",
+	"Rust":                "text/x-rustsrc",
+	"SAS":                 "text/x-sas",
+	"SCSS":                "text/x-scss",
+	"SPARQL":              "application/sparql-query",
+	"SQL":                 "text/x-sql",
+	"SQLPL":               "text/x-sql",
+	"SRecode Template":    "text/x-common-lisp",
+	"SVG":                 "text/xml",
+	"Sage":                "text/x-python",
+	"SaltStack":           "text/x-yaml",
+	"Sass":                "text/x-sass",
+	"Scala":               "text/x-scala",
+	"Scheme":              "text/x-scheme",
+	"Shell":               "text/x-sh",
+	"ShellSession":        "text/x-sh",
+	"Slim":                "text/x-slim",
+	"Smalltalk":           "text/x-stsrc",
+	"Smarty":              "text/x-smarty",
+	"Squirrel":            "text/x-c++src",
+	"Standard ML":         "text/x-ocaml",
+	"Sublime Text Config": "text/javascript",
+	"Swift":               "text/x-swift",
+	"SystemVerilog":       "text/x-systemverilog",
+	"TOML":                "text/x-toml",
+	"Tcl":                 "text/x-tcl",
+	"Tcsh":                "text/x-sh",
+	"TeX":                 "text/x-stex",
+	"Terra":               "text/x-lua",
+	"Textile":             "text/x-textile",
+	"Turtle":              "text/turtle",
+	"Twig":                "text/x-twig",
+	"TypeScript":          "application/typescript",
+	"Unified Parallel C":  "text/x-csrc",
+	"Unity3D Asset":       "text/x-yaml",
+	"Uno":                 "text/x-csharp",
+	"UnrealScript":        "text/x-java",
+	"VHDL":                "text/x-vhdl",
+	"Verilog":             "text/x-verilog",
+	"Visual Basic":        "text/x-vb",
+	"Volt":                "text/x-d",
+	"WebAssembly":         "text/x-common-lisp",
+	"WebIDL":              "text/x-webidl",
+	"XC":                  "text/x-csrc",
+	"XML":                 "text/xml",
+	"XPages":              "text/xml",
+	"XProc":               "text/xml",
+	"XQuery":              "application/xquery",
+	"XS":                  "text/x-csrc",
+	"XSLT":                "text/xml",
+	"YAML":                "text/x-yaml",
+	"edn":                 "text/x-clojure",
+	"reStructuredText":    "text/x-rst",
+	"wisp":                "text/x-clojure",
+}

internal/code-generator/main.go

@@ -67,6 +67,11 @@ const (
 	commitTmplPath = "internal/code-generator/assets/commit.go.tmpl"
 	commitTmpl     = "commit.go.tmpl"
 
+	// mimeType.go generation
+	mimeTypeFile     = "data/mimeType.go"
+	mimeTypeTmplPath = "internal/code-generator/assets/mimeType.go.tmpl"
+	mimeTypeTmpl     = "mimeType.go.tmpl"
+
 	commitPath = ".linguist/.git/HEAD"
 )
 
@@ -97,6 +102,7 @@ func main() {
 		&generatorFiles{generator.Aliases, languagesYAML, "", aliasesFile, aliasesTmplPath, aliasesTmpl, commit},
 		&generatorFiles{generator.Frequencies, "", samplesDir, frequenciesFile, frequenciesTmplPath, frequenciesTmpl, commit},
 		&generatorFiles{generator.Commit, "", "", commitFile, commitTmplPath, commitTmpl, commit},
+		&generatorFiles{generator.MimeType, languagesYAML, "", mimeTypeFile, mimeTypeTmplPath, mimeTypeTmpl, commit},
 	}
 
 	for _, file := range fileList {

utils.go

@@ -53,6 +53,27 @@ func IsDocumentation(path string) bool {
 	return data.DocumentationMatchers.Match(path)
 }
 
+func IsImage(path string) bool {
+	extension := filepath.Ext(path)
+	if extension == ".png" || extension == ".jpg" || extension == ".jpeg" || extension == ".gif" {
+		return true
+	}
+
+	return false
+}
+
+func GetMimeType(path string, language string) string {
+	if mime, ok := data.LanguagesMime[language]; ok {
+		return mime
+	}
+
+	if IsImage(path) {
+		return "image/" + filepath.Ext(path)[1:]
+	}
+
+	return "text/plain"
+}
+
 const sniffLen = 8000
 
 // IsBinary detects if data is a binary value based on: