Merge pull request #26 from mcarmonaa/modelines

Added Modeline detection strategy

alias.go

@@ -0,0 +1,15 @@
+package slinguist
+
+import "strings"
+
+// GetLanguageByAlias returns the language related to the given alias or Otherlanguage otherwise.
+func GetLanguageByAlias(alias string) (lang string) {
+	a := strings.Split(alias, `,`)[0]
+	a = strings.ToLower(a)
+	lang, ok := languagesByAlias[a]
+	if !ok {
+		lang = OtherLanguage
+	}
+
+	return
+}

alias_test.go

@@ -0,0 +1,26 @@
+package slinguist
+
+import . "gopkg.in/check.v1"
+
+func (s *TSuite) TestGetLanguageByAlias(c *C) {
+	tests := []struct {
+		alias        string
+		expectedLang string
+	}{
+		{alias: "BestLanguageEver", expectedLang: OtherLanguage},
+		{alias: "aspx-vb", expectedLang: "ASP"},
+		{alias: "C++", expectedLang: "C++"},
+		{alias: "c++", expectedLang: "C++"},
+		{alias: "objc", expectedLang: "Objective-C"},
+		{alias: "golang", expectedLang: "Go"},
+		{alias: "GOLANG", expectedLang: "Go"},
+		{alias: "bsdmake", expectedLang: "Makefile"},
+		{alias: "xhTmL", expectedLang: "HTML"},
+		{alias: "python", expectedLang: "Python"},
+	}
+
+	for _, test := range tests {
+		lang := GetLanguageByAlias(test.alias)
+		c.Assert(lang, Equals, test.expectedLang)
+	}
+}

aliases_map.go

@@ -0,0 +1,637 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH gopkg.in/src-d/simple-linguist.v1/internal/code-generator
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: dae33dc2b20cddc85d1300435c3be7118a7115a9
+
+// languagesByAlias keeps alias for different languages and use the name of the languages as a alias too. All the
+// keys (alias or not) are written in lower case and the whitespaces has been replaced by underscores.
+var languagesByAlias = map[string]string{
+	"1c_enterprise":    "1C Enterprise",
+	"abap":             "ABAP",
+	"abl":              "OpenEdge ABL",
+	"abnf":             "ABNF",
+	"abuild":           "Alpine Abuild",
+	"aconf":            "ApacheConf",
+	"actionscript":     "ActionScript",
+	"actionscript3":    "ActionScript",
+	"actionscript_3":   "ActionScript",
+	"ada":              "Ada",
+	"ada2005":          "Ada",
+	"ada95":            "Ada",
+	"advpl":            "xBase",
+	"afdko":            "OpenType Feature File",
+	"agda":             "Agda",
+	"ags":              "AGS Script",
+	"ags_script":       "AGS Script",
+	"ahk":              "AutoHotkey",
+	"alloy":            "Alloy",
+	"alpine_abuild":    "Alpine Abuild",
+	"ampl":             "AMPL",
+	"ant_build_system": "Ant Build System",
+	"antlr":            "ANTLR",
+	"apache":           "ApacheConf",
+	"apacheconf":       "ApacheConf",
+	"apex":             "Apex",
+	"api_blueprint":    "API Blueprint",
+	"apkbuild":         "Alpine Abuild",
+	"apl":              "APL",
+	"apollo_guidance_computer": "Apollo Guidance Computer",
+	"applescript":              "AppleScript",
+	"arc":                      "Arc",
+	"arduino":                  "Arduino",
+	"arexx":                    "REXX",
+	"as3":                      "ActionScript",
+	"asciidoc":                 "AsciiDoc",
+	"asn.1":                    "ASN.1",
+	"asp":                      "ASP",
+	"aspectj":                  "AspectJ",
+	"aspx":                     "ASP",
+	"aspx-vb":                  "ASP",
+	"assembly":                 "Assembly",
+	"ats":                      "ATS",
+	"ats2":                     "ATS",
+	"au3":                      "AutoIt",
+	"augeas":                   "Augeas",
+	"autoconf":                 "M4Sugar",
+	"autohotkey":               "AutoHotkey",
+	"autoit":                   "AutoIt",
+	"autoit3":                  "AutoIt",
+	"autoitscript":             "AutoIt",
+	"awk":                      "Awk",
+	"b3d":                      "BlitzBasic",
+	"bash":                     "Shell",
+	"bash_session":             "ShellSession",
+	"bat":                      "Batchfile",
+	"batch":                    "Batchfile",
+	"batchfile":                "Batchfile",
+	"befunge":                  "Befunge",
+	"bison":                    "Bison",
+	"bitbake":                  "BitBake",
+	"blade":                    "Blade",
+	"blitz3d":                  "BlitzBasic",
+	"blitzbasic":               "BlitzBasic",
+	"blitzmax":                 "BlitzMax",
+	"blitzplus":                "BlitzBasic",
+	"bluespec":                 "Bluespec",
+	"bmax":                     "BlitzMax",
+	"boo":                      "Boo",
+	"bplus":                    "BlitzBasic",
+	"brainfuck":                "Brainfuck",
+	"brightscript":             "Brightscript",
+	"bro":                      "Bro",
+	"bsdmake":                  "Makefile",
+	"byond":                    "DM",
+	"c":                        "C",
+	"c#":                       "C#",
+	"c++":                      "C++",
+	"c++-objdump":              "Cpp-ObjDump",
+	"c-objdump":                "C-ObjDump",
+	"c2hs":                     "C2hs Haskell",
+	"c2hs_haskell":             "C2hs Haskell",
+	"cap'n_proto":              "Cap'n Proto",
+	"carto":                    "CartoCSS",
+	"cartocss":                 "CartoCSS",
+	"ceylon":                   "Ceylon",
+	"cfc":                      "ColdFusion CFC",
+	"cfm":                      "ColdFusion",
+	"cfml":                     "ColdFusion",
+	"chapel":                   "Chapel",
+	"charity":                  "Charity",
+	"chpl":                     "Chapel",
+	"chuck":                    "ChucK",
+	"cirru":                    "Cirru",
+	"clarion":                  "Clarion",
+	"clean":                    "Clean",
+	"click":                    "Click",
+	"clipper":                  "xBase",
+	"clips":                    "CLIPS",
+	"clojure":                  "Clojure",
+	"cmake":                    "CMake",
+	"cobol":                    "COBOL",
+	"coffee":                   "CoffeeScript",
+	"coffee-script":            "CoffeeScript",
+	"coffeescript":             "CoffeeScript",
+	"coldfusion":               "ColdFusion",
+	"coldfusion_cfc":           "ColdFusion CFC",
+	"coldfusion_html":          "ColdFusion",
+	"collada":                  "COLLADA",
+	"common_lisp":              "Common Lisp",
+	"component_pascal":         "Component Pascal",
+	"console":                  "ShellSession",
+	"cool":                     "Cool",
+	"coq":                      "Coq",
+	"cpp":                      "C++",
+	"cpp-objdump":              "Cpp-ObjDump",
+	"creole":                   "Creole",
+	"crystal":                  "Crystal",
+	"csharp":                   "C#",
+	"cson":                     "CSON",
+	"csound":                   "Csound",
+	"csound-csd":               "Csound Document",
+	"csound-orc":               "Csound",
+	"csound-sco":               "Csound Score",
+	"csound_document":          "Csound Document",
+	"csound_score":             "Csound Score",
+	"css":                      "CSS",
+	"csv":                      "CSV",
+	"cucumber":                 "Gherkin",
+	"cuda":                     "Cuda",
+	"cycript":                  "Cycript",
+	"cython":                   "Cython",
+	"d":                        "D",
+	"d-objdump":                "D-ObjDump",
+	"darcs_patch":              "Darcs Patch",
+	"dart":                     "Dart",
+	"dcl":                      "DIGITAL Command Language",
+	"delphi":                   "Component Pascal",
+	"desktop":                  "desktop",
+	"diff":                     "Diff",
+	"digital_command_language": "DIGITAL Command Language",
+	"django":                   "HTML+Django",
+	"dm":                       "DM",
+	"dns_zone":                 "DNS Zone",
+	"dockerfile":               "Dockerfile",
+	"dogescript":               "Dogescript",
+	"dosbatch":                 "Batchfile",
+	"dosini":                   "INI",
+	"dpatch":                   "Darcs Patch",
+	"dtrace":                   "DTrace",
+	"dtrace-script":            "DTrace",
+	"dylan":                    "Dylan",
+	"e":                        "E",
+	"eagle":                    "Eagle",
+	"ebnf":                     "EBNF",
+	"ec":                       "eC",
+	"ecere_projects":           "Ecere Projects",
+	"ecl":                      "ECL",
+	"eclipse":                  "ECLiPSe",
+	"ecr":                      "HTML+ECR",
+	"edn":                      "edn",
+	"eex":                      "HTML+EEX",
+	"eiffel":                   "Eiffel",
+	"ejs":                      "EJS",
+	"elisp":                    "Emacs Lisp",
+	"elixir":                   "Elixir",
+	"elm":                      "Elm",
+	"emacs":                    "Emacs Lisp",
+	"emacs_lisp":               "Emacs Lisp",
+	"emberscript":              "EmberScript",
+	"eq":                       "EQ",
+	"erb":                      "HTML+ERB",
+	"erlang":                   "Erlang",
+	"f#":                       "F#",
+	"factor":                   "Factor",
+	"fancy":                    "Fancy",
+	"fantom":                   "Fantom",
+	"filebench_wml":            "Filebench WML",
+	"filterscript":             "Filterscript",
+	"fish":                     "fish",
+	"flex":                     "Lex",
+	"flux":                     "FLUX",
+	"formatted":                "Formatted",
+	"forth":                    "Forth",
+	"fortran":                  "Fortran",
+	"foxpro":                   "xBase",
+	"freemarker":               "FreeMarker",
+	"frege":                    "Frege",
+	"fsharp":                   "F#",
+	"ftl":                      "FreeMarker",
+	"fundamental":              "Text",
+	"g-code":                   "G-code",
+	"game_maker_language":      "Game Maker Language",
+	"gams":                     "GAMS",
+	"gap":                      "GAP",
+	"gcc_machine_description": "GCC Machine Description",
+	"gdb":                     "GDB",
+	"gdscript":                "GDScript",
+	"genie":                   "Genie",
+	"genshi":                  "Genshi",
+	"gentoo_ebuild":           "Gentoo Ebuild",
+	"gentoo_eclass":           "Gentoo Eclass",
+	"gettext_catalog":         "Gettext Catalog",
+	"gf":                      "Grammatical Framework",
+	"gherkin":                 "Gherkin",
+	"glsl":                    "GLSL",
+	"glyph":                   "Glyph",
+	"gn":                      "GN",
+	"gnuplot":                 "Gnuplot",
+	"go":                      "Go",
+	"golang":                  "Go",
+	"golo":                    "Golo",
+	"gosu":                    "Gosu",
+	"grace":                   "Grace",
+	"gradle":                  "Gradle",
+	"grammatical_framework":   "Grammatical Framework",
+	"graph_modeling_language": "Graph Modeling Language",
+	"graphql":                 "GraphQL",
+	"graphviz_(dot)":          "Graphviz (DOT)",
+	"groovy":                  "Groovy",
+	"groovy_server_pages":     "Groovy Server Pages",
+	"gsp":                      "Groovy Server Pages",
+	"hack":                     "Hack",
+	"haml":                     "Haml",
+	"handlebars":               "Handlebars",
+	"harbour":                  "Harbour",
+	"haskell":                  "Haskell",
+	"haxe":                     "Haxe",
+	"hbs":                      "Handlebars",
+	"hcl":                      "HCL",
+	"hlsl":                     "HLSL",
+	"html":                     "HTML",
+	"html+django":              "HTML+Django",
+	"html+django/jinja":        "HTML+Django",
+	"html+ecr":                 "HTML+ECR",
+	"html+eex":                 "HTML+EEX",
+	"html+erb":                 "HTML+ERB",
+	"html+jinja":               "HTML+Django",
+	"html+php":                 "HTML+PHP",
+	"html+ruby":                "RHTML",
+	"htmlbars":                 "Handlebars",
+	"htmldjango":               "HTML+Django",
+	"http":                     "HTTP",
+	"hy":                       "Hy",
+	"hylang":                   "Hy",
+	"hyphy":                    "HyPhy",
+	"i7":                       "Inform 7",
+	"idl":                      "IDL",
+	"idris":                    "Idris",
+	"igor":                     "IGOR Pro",
+	"igor_pro":                 "IGOR Pro",
+	"igorpro":                  "IGOR Pro",
+	"inc":                      "PHP",
+	"inform7":                  "Inform 7",
+	"inform_7":                 "Inform 7",
+	"ini":                      "INI",
+	"inno_setup":               "Inno Setup",
+	"io":                       "Io",
+	"ioke":                     "Ioke",
+	"ipython_notebook":         "Jupyter Notebook",
+	"irc":                      "IRC log",
+	"irc_log":                  "IRC log",
+	"irc_logs":                 "IRC log",
+	"isabelle":                 "Isabelle",
+	"isabelle_root":            "Isabelle ROOT",
+	"j":                        "J",
+	"jasmin":                   "Jasmin",
+	"java":                     "Java",
+	"java_server_page":         "Groovy Server Pages",
+	"java_server_pages":        "Java Server Pages",
+	"javascript":               "JavaScript",
+	"jflex":                    "JFlex",
+	"jison":                    "Jison",
+	"jison_lex":                "Jison Lex",
+	"jruby":                    "Ruby",
+	"js":                       "JavaScript",
+	"json":                     "JSON",
+	"json5":                    "JSON5",
+	"jsoniq":                   "JSONiq",
+	"jsonld":                   "JSONLD",
+	"jsp":                      "Java Server Pages",
+	"jsx":                      "JSX",
+	"julia":                    "Julia",
+	"jupyter_notebook":         "Jupyter Notebook",
+	"kicad":                    "KiCad",
+	"kit":                      "Kit",
+	"kotlin":                   "Kotlin",
+	"krl":                      "KRL",
+	"labview":                  "LabVIEW",
+	"lasso":                    "Lasso",
+	"lassoscript":              "Lasso",
+	"latex":                    "TeX",
+	"latte":                    "Latte",
+	"lean":                     "Lean",
+	"less":                     "Less",
+	"lex":                      "Lex",
+	"lfe":                      "LFE",
+	"lhaskell":                 "Literate Haskell",
+	"lhs":                      "Literate Haskell",
+	"lilypond":                 "LilyPond",
+	"limbo":                    "Limbo",
+	"linker_script":            "Linker Script",
+	"linux_kernel_module":      "Linux Kernel Module",
+	"liquid":                   "Liquid",
+	"lisp":                     "Common Lisp",
+	"litcoffee":                "Literate CoffeeScript",
+	"literate_agda":            "Literate Agda",
+	"literate_coffeescript":    "Literate CoffeeScript",
+	"literate_haskell":         "Literate Haskell",
+	"live-script":              "LiveScript",
+	"livescript":               "LiveScript",
+	"llvm":                     "LLVM",
+	"logos":                    "Logos",
+	"logtalk":                  "Logtalk",
+	"lolcode":                  "LOLCODE",
+	"lookml":                   "LookML",
+	"loomscript":               "LoomScript",
+	"ls":                       "LiveScript",
+	"lsl":                      "LSL",
+	"lua":                      "Lua",
+	"m":                        "M",
+	"m4":                       "M4",
+	"m4sugar":                  "M4Sugar",
+	"macruby":                  "Ruby",
+	"make":                     "Makefile",
+	"makefile":                 "Makefile",
+	"mako":                     "Mako",
+	"markdown":                 "Markdown",
+	"marko":                    "Marko",
+	"markojs":                  "Marko",
+	"mask":                     "Mask",
+	"mathematica":              "Mathematica",
+	"matlab":                   "Matlab",
+	"maven_pom":                "Maven POM",
+	"max":                      "Max",
+	"max/msp":                  "Max",
+	"maxmsp":                   "Max",
+	"maxscript":                "MAXScript",
+	"mediawiki":                "MediaWiki",
+	"mercury":                  "Mercury",
+	"meson":                    "Meson",
+	"metal":                    "Metal",
+	"mf":                       "Makefile",
+	"minid":                    "MiniD",
+	"mirah":                    "Mirah",
+	"mma":                      "Mathematica",
+	"modelica":                 "Modelica",
+	"modula-2":                 "Modula-2",
+	"module_management_system": "Module Management System",
+	"monkey":                   "Monkey",
+	"moocode":                  "Moocode",
+	"moonscript":               "MoonScript",
+	"mql4":                     "MQL4",
+	"mql5":                     "MQL5",
+	"mtml":                     "MTML",
+	"muf":                      "MUF",
+	"mumps":                    "M",
+	"mupad":                    "mupad",
+	"myghty":                   "Myghty",
+	"nasm":                     "Assembly",
+	"ncl":                      "NCL",
+	"nemerle":                  "Nemerle",
+	"nesc":                     "nesC",
+	"netlinx":                  "NetLinx",
+	"netlinx+erb":              "NetLinx+ERB",
+	"netlogo":                  "NetLogo",
+	"newlisp":                  "NewLisp",
+	"nginx":                    "Nginx",
+	"nginx_configuration_file": "Nginx",
+	"nim":                   "Nim",
+	"ninja":                 "Ninja",
+	"nit":                   "Nit",
+	"nix":                   "Nix",
+	"nixos":                 "Nix",
+	"njk":                   "HTML+Django",
+	"nl":                    "NL",
+	"node":                  "JavaScript",
+	"nroff":                 "Roff",
+	"nsis":                  "NSIS",
+	"nu":                    "Nu",
+	"numpy":                 "NumPy",
+	"nunjucks":              "HTML+Django",
+	"nush":                  "Nu",
+	"nvim":                  "Vim script",
+	"obj-c":                 "Objective-C",
+	"obj-c++":               "Objective-C++",
+	"obj-j":                 "Objective-J",
+	"objc":                  "Objective-C",
+	"objc++":                "Objective-C++",
+	"objdump":               "ObjDump",
+	"objective-c":           "Objective-C",
+	"objective-c++":         "Objective-C++",
+	"objective-j":           "Objective-J",
+	"objectivec":            "Objective-C",
+	"objectivec++":          "Objective-C++",
+	"objectivej":            "Objective-J",
+	"objectpascal":          "Component Pascal",
+	"objj":                  "Objective-J",
+	"ocaml":                 "OCaml",
+	"octave":                "Matlab",
+	"omgrofl":               "Omgrofl",
+	"ooc":                   "ooc",
+	"opa":                   "Opa",
+	"opal":                  "Opal",
+	"opencl":                "OpenCL",
+	"openedge":              "OpenEdge ABL",
+	"openedge_abl":          "OpenEdge ABL",
+	"openrc":                "OpenRC runscript",
+	"openrc_runscript":      "OpenRC runscript",
+	"openscad":              "OpenSCAD",
+	"opentype_feature_file": "OpenType Feature File",
+	"org":                            "Org",
+	"osascript":                      "AppleScript",
+	"ox":                             "Ox",
+	"oxygene":                        "Oxygene",
+	"oz":                             "Oz",
+	"p4":                             "P4",
+	"pan":                            "Pan",
+	"pandoc":                         "Markdown",
+	"papyrus":                        "Papyrus",
+	"parrot":                         "Parrot",
+	"parrot_assembly":                "Parrot Assembly",
+	"parrot_internal_representation": "Parrot Internal Representation",
+	"pascal":                "Pascal",
+	"pasm":                  "Parrot Assembly",
+	"pawn":                  "PAWN",
+	"perl":                  "Perl",
+	"perl6":                 "Perl6",
+	"php":                   "PHP",
+	"pic":                   "Pic",
+	"pickle":                "Pickle",
+	"picolisp":              "PicoLisp",
+	"piglatin":              "PigLatin",
+	"pike":                  "Pike",
+	"pir":                   "Parrot Internal Representation",
+	"plpgsql":               "PLpgSQL",
+	"plsql":                 "PLSQL",
+	"pod":                   "Pod",
+	"pogoscript":            "PogoScript",
+	"pony":                  "Pony",
+	"posh":                  "PowerShell",
+	"postscr":               "PostScript",
+	"postscript":            "PostScript",
+	"pot":                   "Gettext Catalog",
+	"pov-ray":               "POV-Ray SDL",
+	"pov-ray_sdl":           "POV-Ray SDL",
+	"povray":                "POV-Ray SDL",
+	"powerbuilder":          "PowerBuilder",
+	"powershell":            "PowerShell",
+	"processing":            "Processing",
+	"progress":              "OpenEdge ABL",
+	"prolog":                "Prolog",
+	"propeller_spin":        "Propeller Spin",
+	"protobuf":              "Protocol Buffer",
+	"protocol_buffer":       "Protocol Buffer",
+	"protocol_buffers":      "Protocol Buffer",
+	"public_key":            "Public Key",
+	"pug":                   "Pug",
+	"puppet":                "Puppet",
+	"pure_data":             "Pure Data",
+	"purebasic":             "PureBasic",
+	"purescript":            "PureScript",
+	"pycon":                 "Python console",
+	"pyrex":                 "Cython",
+	"python":                "Python",
+	"python_console":        "Python console",
+	"python_traceback":      "Python traceback",
+	"qmake":                 "QMake",
+	"qml":                   "QML",
+	"r":                     "R",
+	"racket":                "Racket",
+	"ragel":                 "Ragel",
+	"ragel-rb":              "Ragel",
+	"ragel-ruby":            "Ragel",
+	"rake":                  "Ruby",
+	"raml":                  "RAML",
+	"rascal":                "Rascal",
+	"raw":                   "Raw token data",
+	"raw_token_data":        "Raw token data",
+	"rb":                    "Ruby",
+	"rbx":                   "Ruby",
+	"rdoc":                  "RDoc",
+	"realbasic":             "REALbasic",
+	"reason":                "Reason",
+	"rebol":                 "Rebol",
+	"red":                   "Red",
+	"red/system":            "Red",
+	"redcode":               "Redcode",
+	"regex":                 "Regular Expression",
+	"regexp":                "Regular Expression",
+	"regular_expression":    "Regular Expression",
+	"ren'py":                "Ren'Py",
+	"renderscript":          "RenderScript",
+	"renpy":                 "Ren'Py",
+	"restructuredtext":      "reStructuredText",
+	"rexx":                  "REXX",
+	"rhtml":                 "RHTML",
+	"rmarkdown":             "RMarkdown",
+	"robotframework":        "RobotFramework",
+	"roff":                  "Roff",
+	"rouge":                 "Rouge",
+	"rpm_spec":              "RPM Spec",
+	"rscript":               "R",
+	"rss":                   "XML",
+	"rst":                   "reStructuredText",
+	"ruby":                  "Ruby",
+	"runoff":                "RUNOFF",
+	"rust":                  "Rust",
+	"rusthon":               "Python",
+	"sage":                  "Sage",
+	"salt":                  "SaltStack",
+	"saltstack":             "SaltStack",
+	"saltstate":             "SaltStack",
+	"sas":                   "SAS",
+	"sass":                  "Sass",
+	"scala":                 "Scala",
+	"scaml":                 "Scaml",
+	"scheme":                "Scheme",
+	"scilab":                "Scilab",
+	"scss":                  "SCSS",
+	"self":                  "Self",
+	"sh":                    "Shell",
+	"shell":                 "Shell",
+	"shell-script":          "Shell",
+	"shellsession":          "ShellSession",
+	"shen":                  "Shen",
+	"slash":                 "Slash",
+	"slim":                  "Slim",
+	"smali":                 "Smali",
+	"smalltalk":             "Smalltalk",
+	"smarty":                "Smarty",
+	"sml":                   "Standard ML",
+	"smt":                   "SMT",
+	"sourcemod":             "SourcePawn",
+	"sourcepawn":            "SourcePawn",
+	"sparql":                "SPARQL",
+	"specfile":              "RPM Spec",
+	"spline_font_database":  "Spline Font Database",
+	"splus":                 "R",
+	"sqf":                   "SQF",
+	"sql":                   "SQL",
+	"sqlpl":                 "SQLPL",
+	"squeak":                "Smalltalk",
+	"squirrel":              "Squirrel",
+	"srecode_template":      "SRecode Template",
+	"stan":                  "Stan",
+	"standard_ml":           "Standard ML",
+	"stata":                 "Stata",
+	"ston":                  "STON",
+	"stylus":                "Stylus",
+	"sublime_text_config":   "Sublime Text Config",
+	"subrip_text":           "SubRip Text",
+	"supercollider":         "SuperCollider",
+	"svg":                   "SVG",
+	"swift":                 "Swift",
+	"systemverilog":         "SystemVerilog",
+	"tcl":                   "Tcl",
+	"tcsh":                  "Tcsh",
+	"tea":                   "Tea",
+	"terra":                 "Terra",
+	"tex":                   "TeX",
+	"text":                  "Text",
+	"textile":               "Textile",
+	"thrift":                "Thrift",
+	"ti_program":            "TI Program",
+	"tla":                   "TLA",
+	"toml":                  "TOML",
+	"ts":                    "TypeScript",
+	"turing":                "Turing",
+	"turtle":                "Turtle",
+	"twig":                  "Twig",
+	"txl":                   "TXL",
+	"typescript":            "TypeScript",
+	"udiff":                 "Diff",
+	"unified_parallel_c":    "Unified Parallel C",
+	"unity3d_asset":         "Unity3D Asset",
+	"unix_assembly":         "Unix Assembly",
+	"uno":                   "Uno",
+	"unrealscript":          "UnrealScript",
+	"ur":                    "UrWeb",
+	"ur/web":                "UrWeb",
+	"urweb":                 "UrWeb",
+	"vala":                  "Vala",
+	"vb.net":                "Visual Basic",
+	"vbnet":                 "Visual Basic",
+	"vcl":                   "VCL",
+	"verilog":               "Verilog",
+	"vhdl":                  "VHDL",
+	"vim":                   "Vim script",
+	"vim_script":            "Vim script",
+	"viml":                  "Vim script",
+	"visual_basic":          "Visual Basic",
+	"volt":                  "Volt",
+	"vue":                   "Vue",
+	"wavefront_material":    "Wavefront Material",
+	"wavefront_object":      "Wavefront Object",
+	"web_ontology_language": "Web Ontology Language",
+	"webidl":                "WebIDL",
+	"winbatch":              "Batchfile",
+	"wisp":                  "wisp",
+	"world_of_warcraft_addon_data": "World of Warcraft Addon Data",
+	"wsdl":       "XML",
+	"x10":        "X10",
+	"xbase":      "xBase",
+	"xc":         "XC",
+	"xcompose":   "XCompose",
+	"xhtml":      "HTML",
+	"xml":        "XML",
+	"xml+genshi": "Genshi",
+	"xml+kid":    "Genshi",
+	"xojo":       "Xojo",
+	"xpages":     "XPages",
+	"xproc":      "XProc",
+	"xquery":     "XQuery",
+	"xs":         "XS",
+	"xsd":        "XML",
+	"xsl":        "XSLT",
+	"xslt":       "XSLT",
+	"xten":       "X10",
+	"xtend":      "Xtend",
+	"yacc":       "Yacc",
+	"yaml":       "YAML",
+	"yang":       "YANG",
+	"yml":        "YAML",
+	"zephir":     "Zephir",
+	"zimpl":      "Zimpl",
+	"zsh":        "Shell",
+}

common.go

@@ -36,6 +36,10 @@ func GetLanguageExtensions(language string) []string {
 
 // GetLanguage return the Language for a given filename and file content.
 func GetLanguage(filename string, content []byte) string {
+	if lang, safe := GetLanguageByModeline(content); safe {
+		return lang
+	}
+
 	if lang, safe := GetLanguageByFilename(filename); safe {
 		return lang
 	}

content.go

@@ -14,10 +14,11 @@ func GetLanguageByContent(filename string, content []byte) (lang string, safe bo
 	ext := strings.ToLower(filepath.Ext(filename))
 	if fnMatcher, ok := matchers[ext]; ok {
 		lang, safe = fnMatcher(content)
-		return
+	} else {
+		lang = OtherLanguage
 	}
 
-	return GetLanguageByExtension(filename)
+	return
 }
 
 type languageMatcher func([]byte) (string, bool)

extension.go

@@ -7,7 +7,6 @@ import (
 
 func GetLanguageByExtension(filename string) (lang string, safe bool) {
 	ext := strings.ToLower(filepath.Ext(filename))
-
 	lang = OtherLanguage
 	langs, ok := languagesByExtension[ext]
 	if !ok {
@@ -15,9 +14,6 @@ func GetLanguageByExtension(filename string) (lang string, safe bool) {
 	}
 
 	lang = langs[0]
-	if len(langs) == 1 {
-		safe = true
-	}
-
+	safe = len(langs) == 1
 	return
 }

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

@@ -0,0 +1,13 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH gopkg.in/src-d/simple-linguist.v1/internal/code-generator
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: {{ getCommit }}
+
+// languagesByAlias keeps alias for different languages and use the name of the languages as a alias too. All the
+// keys (alias or not) are written in lower case and the whitespaces has been replaced by underscores.
+var languagesByAlias = map[string]string{
+	{{range $alias, $language := . -}}
+	"{{ $alias }}":	{{ printf "%q" $language -}},
+	{{end -}}
+}

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

@@ -14,10 +14,11 @@ func GetLanguageByContent(filename string, content []byte) (lang string, safe bo
 	ext := strings.ToLower(filepath.Ext(filename))
 	if fnMatcher, ok := matchers[ext]; ok {
 		lang, safe = fnMatcher(content)
-		return
+	} else {
+		lang = OtherLanguage
 	}
 
-	return GetLanguageByExtension(filename)
+	return
 }
 
 type languageMatcher func ([]byte) (string, bool)

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

@@ -1,28 +0,0 @@
-package slinguist
-
-// CODE GENERATED AUTOMATICALLY WITH gopkg.in/src-d/simple-linguist.v1/internal/code-generator
-// THIS FILE SHOULD NOT BE EDITED BY HAND
-// Extracted from github/linguist commit: {{ getCommit }}
-
-type Type int
-
-const (
-	// Language's type. Either data, programming, markup, prose, or unknown.
-	Unknown Type = iota
-	Data
-	Programming
-	Markup
-	Prose
-)
-
-// GetLanguageType returns the given language's type.
-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/assets/types.go.tmpl

@@ -4,8 +4,8 @@ package slinguist
 // THIS FILE SHOULD NOT BE EDITED BY HAND
 // Extracted from github/linguist commit: {{ getCommit }}
 
-var languagesByExtension = map[string][]string{
-	{{range $extension, $languages := . -}}
-	"{{ $extension }}":	{ {{- $languages | formatStringSlice -}} },
+var languagesType = map[string]Type{
+	{{range $language, $type := . -}}
+	"{{ $language }}": {{ $type -}},
 	{{end -}}
 }

internal/code-generator/generator/aliases.go

@@ -0,0 +1,63 @@
+package generator
+
+import (
+	"bytes"
+	"io"
+	"text/template"
+
+	"strings"
+
+	yaml "gopkg.in/yaml.v2"
+)
+
+// Aliases reads from buf and builds aliases_map.go file from aliasesTmplPath.
+func Aliases(data []byte, aliasesTmplPath, aliasesTmplName, commit string) ([]byte, error) {
+	languages := make(map[string]*languageInfo)
+	if err := yaml.Unmarshal(data, &languages); err != nil {
+		return nil, err
+	}
+
+	orderedLangList := getAlphabeticalOrderedKeys(languages)
+	languagesByAlias := buildAliasLanguageMap(languages, orderedLangList)
+
+	buf := &bytes.Buffer{}
+	if err := executeAliasesTemplate(buf, languagesByAlias, aliasesTmplPath, aliasesTmplName, commit); err != nil {
+		return nil, err
+	}
+
+	return buf.Bytes(), nil
+}
+
+func buildAliasLanguageMap(languages map[string]*languageInfo, orderedLangList []string) map[string]string {
+	aliasLangsMap := make(map[string]string)
+	for _, lang := range orderedLangList {
+		langInfo := languages[lang]
+		key := convertToAliasKey(lang)
+		aliasLangsMap[key] = lang
+		for _, alias := range langInfo.Aliases {
+			key := convertToAliasKey(alias)
+			aliasLangsMap[key] = lang
+		}
+	}
+
+	return aliasLangsMap
+}
+
+func convertToAliasKey(s string) (key string) {
+	key = strings.Replace(s, ` `, `_`, -1)
+	key = strings.ToLower(key)
+	return
+}
+
+func executeAliasesTemplate(out io.Writer, languagesByAlias map[string]string, aliasesTmplPath, aliasesTmpl, commit string) error {
+	fmap := template.FuncMap{
+		"getCommit": func() string { return commit },
+	}
+
+	t := template.Must(template.New(aliasesTmpl).Funcs(fmap).ParseFiles(aliasesTmplPath))
+	if err := t.Execute(out, languagesByAlias); err != nil {
+		return err
+	}
+
+	return nil
+}

internal/code-generator/generator/generator_test.go

@@ -37,10 +37,10 @@ const (
 	documentationTestTmplName = "documentation.go.tmpl"
 
 	// Types test
-	typesTestFile     = "test_files/type.test.yml"
-	typesGold         = "test_files/type.gold"
-	typesTestTmplPath = "../assets/type.go.tmpl"
-	typesTestTmplName = "type.go.tmpl"
+	typesTestFile     = "test_files/types.test.yml"
+	typesGold         = "test_files/types.gold"
+	typesTestTmplPath = "../assets/types.go.tmpl"
+	typesTestTmplName = "types.go.tmpl"
 
 	// Interpreters test
 	interpretersTestFile     = "test_files/interpreters.test.yml"
@@ -53,6 +53,12 @@ const (
 	filenamesGold         = "test_files/filenames.gold"
 	filenamesTestTmplPath = "../assets/filenames.go.tmpl"
 	filenamesTestTmplName = "filenames.go.tmpl"
+
+	// Aliases test
+	aliasesTestFile     = "test_files/aliases.test.yml"
+	aliasesGold         = "test_files/aliases.gold"
+	aliasesTestTmplPath = "../assets/aliases.go.tmpl"
+	aliasesTestTmplName = "aliases.go.tmpl"
 )
 
 func TestFromFile(t *testing.T) {
@@ -128,6 +134,15 @@ func TestFromFile(t *testing.T) {
 			generate:    Filenames,
 			wantOut:     filenamesGold,
 		},
+		{
+			name:        "TestFromFile_Aliases",
+			fileToParse: aliasesTestFile,
+			tmplPath:    aliasesTestTmplPath,
+			tmplName:    aliasesTestTmplName,
+			commit:      commitTest,
+			generate:    Aliases,
+			wantOut:     aliasesGold,
+		},
 	}
 
 	for _, tt := range tests {

internal/code-generator/generator/langinfo.go

@@ -4,6 +4,7 @@ import "sort"
 
 type languageInfo struct {
 	Type         string   `yaml:"type,omitempty"`
+	Aliases      []string `yaml:"aliases,omitempty"`
 	Extensions   []string `yaml:"extensions,omitempty,flow"`
 	Interpreters []string `yaml:"interpreters,omitempty,flow"`
 	Filenames    []string `yaml:"filenames,omitempty,flow"`

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

@@ -0,0 +1,42 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH gopkg.in/src-d/simple-linguist.v1/internal/code-generator
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: fe8b44ab8a225b1ffa75b983b916ea22fee5b6f7
+
+// languagesByAlias keeps alias for different languages and use the name of the languages as a alias too. All the
+// keys (alias or not) are written in lower case and the whitespaces has been replaced by underscores.
+var languagesByAlias = map[string]string{
+	"asp":               "ASP",
+	"aspx":              "ASP",
+	"aspx-vb":           "ASP",
+	"au3":               "AutoIt",
+	"autoit":            "AutoIt",
+	"autoit3":           "AutoIt",
+	"autoitscript":      "AutoIt",
+	"bat":               "Batchfile",
+	"batch":             "Batchfile",
+	"batchfile":         "Batchfile",
+	"bsdmake":           "Makefile",
+	"c++":               "C++",
+	"cpp":               "C++",
+	"django":            "HTML+Django",
+	"dosbatch":          "Batchfile",
+	"go":                "Go",
+	"golang":            "Go",
+	"gradle":            "Gradle",
+	"html+django":       "HTML+Django",
+	"html+django/jinja": "HTML+Django",
+	"html+jinja":        "HTML+Django",
+	"htmldjango":        "HTML+Django",
+	"make":              "Makefile",
+	"makefile":          "Makefile",
+	"mf":                "Makefile",
+	"njk":               "HTML+Django",
+	"nunjucks":          "HTML+Django",
+	"obj-c":             "Objective-C",
+	"objc":              "Objective-C",
+	"objective-c":       "Objective-C",
+	"objectivec":        "Objective-C",
+	"winbatch":          "Batchfile",
+}

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

@@ -0,0 +1,42 @@
+---
+ASP:
+  aliases:
+  - aspx
+  - aspx-vb
+AutoIt:
+  aliases:
+  - au3
+  - AutoIt3
+  - AutoItScript
+Batchfile:
+  aliases:
+  - bat
+  - batch
+  - dosbatch
+  - winbatch
+C++:
+  aliases:
+  - cpp
+Go:
+  aliases:
+  - golang
+Gradle:
+  type: data
+HTML+Django:
+  aliases:
+  - django
+  - html+django/jinja
+  - html+jinja
+  - htmldjango
+  - njk
+  - nunjucks
+Makefile:
+  aliases:
+  - bsdmake
+  - make
+  - mf
+Objective-C:
+  aliases:
+  - obj-c
+  - objc
+  - objectivec

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

@@ -14,10 +14,11 @@ func GetLanguageByContent(filename string, content []byte) (lang string, safe bo
 	ext := strings.ToLower(filepath.Ext(filename))
 	if fnMatcher, ok := matchers[ext]; ok {
 		lang, safe = fnMatcher(content)
-		return
+	} else {
+		lang = OtherLanguage
 	}
 
-	return GetLanguageByExtension(filename)
+	return
 }
 
 type languageMatcher func([]byte) (string, bool)

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

@@ -1,28 +0,0 @@
-package slinguist
-
-// CODE GENERATED AUTOMATICALLY WITH gopkg.in/src-d/simple-linguist.v1/internal/code-generator
-// THIS FILE SHOULD NOT BE EDITED BY HAND
-// Extracted from github/linguist commit: {{ getCommit }}
-
-type Type int
-
-const (
-	// Language's type. Either data, programming, markup, prose, or unknown.
-	Unknown Type = iota
-	Data
-	Programming
-	Markup
-	Prose
-)
-
-// GetLanguageType returns the given language's type.
-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/types.gold

@@ -4,23 +4,6 @@ package slinguist
 // THIS FILE SHOULD NOT BE EDITED BY HAND
 // Extracted from github/linguist commit: fe8b44ab8a225b1ffa75b983b916ea22fee5b6f7
 
-type Type int
-
-const (
-	// Language's type. Either data, programming, markup, prose, or unknown.
-	Unknown Type = iota
-	Data
-	Programming
-	Markup
-	Prose
-)
-
-// GetLanguageType returns the given language's type.
-func GetLanguageType(language string) (langType Type) {
-	langType, _ = languagesType[language]
-	return langType
-}
-
 var languagesType = map[string]Type{
 	"Scaml":  Markup,
 	"Scheme": Programming,

internal/code-generator/main.go

@@ -35,9 +35,9 @@ const (
 	documentationTmpl     = "documentation.go.tmpl"
 
 	// type.go generation
-	typeFile     = "type.go"
-	typeTmplPath = "internal/code-generator/assets/type.go.tmpl"
-	typeTmpl     = "type.go.tmpl"
+	typeFile     = "types_map.go"
+	typeTmplPath = "internal/code-generator/assets/types.go.tmpl"
+	typeTmpl     = "types.go.tmpl"
 
 	// interpreters_map.go generation
 	interpretersFile     = "interpreters_map.go"
@@ -49,6 +49,11 @@ const (
 	filenamesTmplPath = "internal/code-generator/assets/filenames.go.tmpl"
 	filenamesTmpl     = "filenames.go.tmpl"
 
+	// aliases_map.go generation
+	aliasesFile     = "aliases_map.go"
+	aliasesTmplPath = "internal/code-generator/assets/aliases.go.tmpl"
+	aliasesTmpl     = "aliases.go.tmpl"
+
 	commitPath = ".git/refs/heads/master"
 )
 
@@ -75,6 +80,7 @@ func main() {
 		&generatorArgs{languagesYAML, typeFile, typeTmplPath, typeTmpl, commit, generator.Types},
 		&generatorArgs{languagesYAML, interpretersFile, interpretersTmplPath, interpretersTmpl, commit, generator.Interpreters},
 		&generatorArgs{languagesYAML, filenamesFile, filenamesTmplPath, filenamesTmpl, commit, generator.Filenames},
+		&generatorArgs{languagesYAML, aliasesFile, aliasesTmplPath, aliasesTmpl, commit, generator.Aliases},
 	}
 
 	for _, args := range argsList {

modeline.go

@@ -0,0 +1,105 @@
+package slinguist
+
+import (
+	"bytes"
+	"regexp"
+)
+
+// GetLanguageByModeline returns the language of the given content looking for the modeline,
+// and safe to indicate the sureness of returned language.
+func GetLanguageByModeline(content []byte) (lang string, safe bool) {
+	headFoot := getHeaderAndFooter(content)
+	for _, getLang := range modelinesFunc {
+		lang = getLang(headFoot)
+		safe = lang != OtherLanguage
+		if safe {
+			return
+		}
+	}
+
+	return
+}
+
+func getHeaderAndFooter(content []byte) []byte {
+	const (
+		searchScope = 5
+		eol         = `\n`
+	)
+
+	if bytes.Count(content, []byte(eol)) < 2*searchScope {
+		return content
+	}
+
+	splitted := bytes.Split(content, []byte(eol))
+	header := splitted[:searchScope]
+	footer := splitted[len(splitted)-searchScope:]
+	headerAndFooter := append(header, footer...)
+	return bytes.Join(headerAndFooter, []byte(eol))
+}
+
+var modelinesFunc = []func(content []byte) string{
+	GetLanguageByEmacsModeline,
+	GetLanguageByVimModeline,
+}
+
+var (
+	reEmacsModeline = regexp.MustCompile(`.*-\*-\s*(.+?)\s*-\*-.*(?m:$)`)
+	reEmacsLang     = regexp.MustCompile(`.*(?i:mode)\s*:\s*([^\s;]+)\s*;*.*`)
+	reVimModeline   = regexp.MustCompile(`(?:(?m:\s|^)vi(?:m[<=>]?\d+|m)?|[\t\x20]*ex)\s*[:]\s*(.*)(?m:$)`)
+	reVimLang       = regexp.MustCompile(`(?i:filetype|ft|syntax)\s*=(\w+)(?:\s|:|$)`)
+)
+
+// GetLanguageByEmacsModeline detecs if the content has a emacs modeline and try to get a
+// language basing on alias. If couldn't retrieve a valid language, it returns OtherLanguage.
+func GetLanguageByEmacsModeline(content []byte) (lang string) {
+	matched := reEmacsModeline.FindAllSubmatch(content, -1)
+	if matched == nil {
+		return OtherLanguage
+	}
+
+	// only take the last matched line, discard previous lines
+	lastLineMatched := matched[len(matched)-1][1]
+	matchedAlias := reEmacsLang.FindSubmatch(lastLineMatched)
+	var alias string
+	if matchedAlias != nil {
+		alias = string(matchedAlias[1])
+	} else {
+		alias = string(lastLineMatched)
+	}
+
+	lang = GetLanguageByAlias(alias)
+	return
+}
+
+// GetLanguageByVimModeline detecs if the content has a vim modeline and try to get a
+// language basing on alias. If couldn't retrieve a valid language, it returns OtherLanguage.
+func GetLanguageByVimModeline(content []byte) (lang string) {
+	matched := reVimModeline.FindAllSubmatch(content, -1)
+	if matched == nil {
+		return OtherLanguage
+	}
+
+	// only take the last matched line, discard previous lines
+	lastLineMatched := matched[len(matched)-1][1]
+	matchedAlias := reVimLang.FindAllSubmatch(lastLineMatched, -1)
+	if matchedAlias == nil {
+		return OtherLanguage
+	}
+
+	alias := string(matchedAlias[0][1])
+	if len(matchedAlias) > 1 {
+		// cases:
+		// matchedAlias = [["syntax=ruby " "ruby"] ["ft=python " "python"] ["filetype=perl " "perl"]] returns OtherLanguage;
+		// matchedAlias = [["syntax=python " "python"] ["ft=python " "python"] ["filetype=python " "python"]] returns "Python";
+		for _, match := range matchedAlias {
+			otherAlias := string(match[1])
+			if otherAlias != alias {
+				alias = OtherLanguage
+				break
+			}
+		}
+	}
+
+	lang = GetLanguageByAlias(alias)
+	return
+}

modeline_test.go

@@ -0,0 +1,83 @@
+package slinguist
+
+import (
+	"io/ioutil"
+	"path/filepath"
+
+	. "gopkg.in/check.v1"
+)
+
+const (
+	modelinesDir = ".linguist/test/fixtures/Data/Modelines"
+)
+
+func (s *TSuite) TestGetLanguageByModeline(c *C) {
+	linguistTests := []struct {
+		filename     string
+		expectedLang string
+		expectedSafe bool
+	}{
+		// Emacs
+		{filename: "example_smalltalk.md", expectedLang: "Smalltalk", expectedSafe: true},
+		{filename: "fundamentalEmacs.c", expectedLang: "Text", expectedSafe: true},
+		{filename: "iamphp.inc", expectedLang: "PHP", expectedSafe: true},
+		{filename: "seeplusplusEmacs1", expectedLang: "C++", expectedSafe: true},
+		{filename: "seeplusplusEmacs2", expectedLang: "C++", expectedSafe: true},
+		{filename: "seeplusplusEmacs3", expectedLang: "C++", expectedSafe: true},
+		{filename: "seeplusplusEmacs4", expectedLang: "C++", expectedSafe: true},
+		{filename: "seeplusplusEmacs5", expectedLang: "C++", expectedSafe: true},
+		{filename: "seeplusplusEmacs6", expectedLang: "C++", expectedSafe: true},
+		{filename: "seeplusplusEmacs7", expectedLang: "C++", expectedSafe: true},
+		{filename: "seeplusplusEmacs9", expectedLang: "C++", expectedSafe: true},
+		{filename: "seeplusplusEmacs10", expectedLang: "C++", expectedSafe: true},
+		{filename: "seeplusplusEmacs11", expectedLang: "C++", expectedSafe: true},
+		{filename: "seeplusplusEmacs12", expectedLang: "C++", expectedSafe: true},
+
+		// Vim
+		{filename: "seeplusplus", expectedLang: "C++", expectedSafe: true},
+		{filename: "iamjs.pl", expectedLang: "JavaScript", expectedSafe: true},
+		{filename: "iamjs2.pl", expectedLang: "JavaScript", expectedSafe: true},
+		{filename: "not_perl.pl", expectedLang: "Prolog", expectedSafe: true},
+		{filename: "ruby", expectedLang: "Ruby", expectedSafe: true},
+		{filename: "ruby2", expectedLang: "Ruby", expectedSafe: true},
+		{filename: "ruby3", expectedLang: "Ruby", expectedSafe: true},
+		{filename: "ruby4", expectedLang: "Ruby", expectedSafe: true},
+		{filename: "ruby5", expectedLang: "Ruby", expectedSafe: true},
+		{filename: "ruby6", expectedLang: "Ruby", expectedSafe: true},
+		{filename: "ruby7", expectedLang: "Ruby", expectedSafe: true},
+		{filename: "ruby8", expectedLang: "Ruby", expectedSafe: true},
+		{filename: "ruby9", expectedLang: "Ruby", expectedSafe: true},
+		{filename: "ruby10", expectedLang: "Ruby", expectedSafe: true},
+		{filename: "ruby11", expectedLang: "Ruby", expectedSafe: true},
+		{filename: "ruby12", expectedLang: "Ruby", expectedSafe: true},
+	}
+
+	for _, test := range linguistTests {
+		content, err := ioutil.ReadFile(filepath.Join(modelinesDir, test.filename))
+		c.Assert(err, Equals, nil)
+
+		lang, safe := GetLanguageByModeline(content)
+		c.Assert(lang, Equals, test.expectedLang)
+		c.Assert(safe, Equals, test.expectedSafe)
+	}
+
+	const (
+		wrongVim = `# vim: set syntax=ruby ft  =python filetype=perl :`
+		rightVim = `/* vim: set syntax=python ft   =python filetype=python */`
+	)
+
+	tests := []struct {
+		content      []byte
+		expectedLang string
+		expectedSafe bool
+	}{
+		{content: []byte(wrongVim), expectedLang: OtherLanguage, expectedSafe: false},
+		{content: []byte(rightVim), expectedLang: "Python", expectedSafe: true},
+	}
+
+	for _, test := range tests {
+		lang, safe := GetLanguageByModeline(test.content)
+		c.Assert(lang, Equals, test.expectedLang)
+		c.Assert(safe, Equals, test.expectedSafe)
+	}
+}

type.go

@@ -1,9 +1,5 @@
 package slinguist
 
-// CODE GENERATED AUTOMATICALLY WITH gopkg.in/src-d/simple-linguist.v1/internal/code-generator
-// THIS FILE SHOULD NOT BE EDITED BY HAND
-// Extracted from github/linguist commit: dae33dc2b20cddc85d1300435c3be7118a7115a9
-
 type Type int
 
 const (
@@ -17,453 +13,10 @@ const (
 
 // GetLanguageType returns the given language's type.
 func GetLanguageType(language string) (langType Type) {
-	langType, _ = languagesType[language]
+	langType, ok := languagesType[language]
+	if !ok {
+		langType = Unknown
+	}
+
 	return langType
 }
-
-var languagesType = map[string]Type{
-	"1C Enterprise":    Programming,
-	"ABAP":             Programming,
-	"ABNF":             Data,
-	"AGS Script":       Programming,
-	"AMPL":             Programming,
-	"ANTLR":            Programming,
-	"API Blueprint":    Markup,
-	"APL":              Programming,
-	"ASN.1":            Data,
-	"ASP":              Programming,
-	"ATS":              Programming,
-	"ActionScript":     Programming,
-	"Ada":              Programming,
-	"Agda":             Programming,
-	"Alloy":            Programming,
-	"Alpine Abuild":    Programming,
-	"Ant Build System": Data,
-	"ApacheConf":       Markup,
-	"Apex":             Programming,
-	"Apollo Guidance Computer": Programming,
-	"AppleScript":              Programming,
-	"Arc":                      Programming,
-	"Arduino":                  Programming,
-	"AsciiDoc":                 Prose,
-	"AspectJ":                  Programming,
-	"Assembly":                 Programming,
-	"Augeas":                   Programming,
-	"AutoHotkey":               Programming,
-	"AutoIt":                   Programming,
-	"Awk":                      Programming,
-	"Batchfile":                Programming,
-	"Befunge":                  Programming,
-	"Bison":                    Programming,
-	"BitBake":                  Programming,
-	"Blade":                    Markup,
-	"BlitzBasic":               Programming,
-	"BlitzMax":                 Programming,
-	"Bluespec":                 Programming,
-	"Boo":                      Programming,
-	"Brainfuck":                Programming,
-	"Brightscript":             Programming,
-	"Bro":                      Programming,
-	"C":                        Programming,
-	"C#":                       Programming,
-	"C++":                      Programming,
-	"C-ObjDump":                Data,
-	"C2hs Haskell":             Programming,
-	"CLIPS":                    Programming,
-	"CMake":                    Programming,
-	"COBOL":                    Programming,
-	"COLLADA":                  Data,
-	"CSON":                     Data,
-	"CSS":                      Markup,
-	"CSV":                      Data,
-	"Cap'n Proto":              Programming,
-	"CartoCSS":                 Programming,
-	"Ceylon":                   Programming,
-	"Chapel":                   Programming,
-	"Charity":                  Programming,
-	"ChucK":                    Programming,
-	"Cirru":                    Programming,
-	"Clarion":                  Programming,
-	"Clean":                    Programming,
-	"Click":                    Programming,
-	"Clojure":                  Programming,
-	"CoffeeScript":             Programming,
-	"ColdFusion":               Programming,
-	"ColdFusion CFC":           Programming,
-	"Common Lisp":              Programming,
-	"Component Pascal":         Programming,
-	"Cool":                     Programming,
-	"Coq":                      Programming,
-	"Cpp-ObjDump":              Data,
-	"Creole":                   Prose,
-	"Crystal":                  Programming,
-	"Csound":                   Programming,
-	"Csound Document":          Programming,
-	"Csound Score":             Programming,
-	"Cuda":                     Programming,
-	"Cycript":                  Programming,
-	"Cython":                   Programming,
-	"D":                        Programming,
-	"D-ObjDump":                Data,
-	"DIGITAL Command Language": Programming,
-	"DM":             Programming,
-	"DNS Zone":       Data,
-	"DTrace":         Programming,
-	"Darcs Patch":    Data,
-	"Dart":           Programming,
-	"Diff":           Data,
-	"Dockerfile":     Data,
-	"Dogescript":     Programming,
-	"Dylan":          Programming,
-	"E":              Programming,
-	"EBNF":           Data,
-	"ECL":            Programming,
-	"ECLiPSe":        Programming,
-	"EJS":            Markup,
-	"EQ":             Programming,
-	"Eagle":          Markup,
-	"Ecere Projects": Data,
-	"Eiffel":         Programming,
-	"Elixir":         Programming,
-	"Elm":            Programming,
-	"Emacs Lisp":     Programming,
-	"EmberScript":    Programming,
-	"Erlang":         Programming,
-	"F#":             Programming,
-	"FLUX":           Programming,
-	"Factor":         Programming,
-	"Fancy":          Programming,
-	"Fantom":         Programming,
-	"Filebench WML":  Programming,
-	"Filterscript":   Programming,
-	"Formatted":      Data,
-	"Forth":          Programming,
-	"Fortran":        Programming,
-	"FreeMarker":     Programming,
-	"Frege":          Programming,
-	"G-code":         Data,
-	"GAMS":           Programming,
-	"GAP":            Programming,
-	"GCC Machine Description": Programming,
-	"GDB":      Programming,
-	"GDScript": Programming,
-	"GLSL":     Programming,
-	"GN":       Data,
-	"Game Maker Language":     Programming,
-	"Genie":                   Programming,
-	"Genshi":                  Programming,
-	"Gentoo Ebuild":           Programming,
-	"Gentoo Eclass":           Programming,
-	"Gettext Catalog":         Prose,
-	"Gherkin":                 Programming,
-	"Glyph":                   Programming,
-	"Gnuplot":                 Programming,
-	"Go":                      Programming,
-	"Golo":                    Programming,
-	"Gosu":                    Programming,
-	"Grace":                   Programming,
-	"Gradle":                  Data,
-	"Grammatical Framework":   Programming,
-	"Graph Modeling Language": Data,
-	"GraphQL":                 Data,
-	"Graphviz (DOT)":          Data,
-	"Groovy":                  Programming,
-	"Groovy Server Pages":     Programming,
-	"HCL":                      Programming,
-	"HLSL":                     Programming,
-	"HTML":                     Markup,
-	"HTML+Django":              Markup,
-	"HTML+ECR":                 Markup,
-	"HTML+EEX":                 Markup,
-	"HTML+ERB":                 Markup,
-	"HTML+PHP":                 Markup,
-	"HTTP":                     Data,
-	"Hack":                     Programming,
-	"Haml":                     Markup,
-	"Handlebars":               Markup,
-	"Harbour":                  Programming,
-	"Haskell":                  Programming,
-	"Haxe":                     Programming,
-	"Hy":                       Programming,
-	"HyPhy":                    Programming,
-	"IDL":                      Programming,
-	"IGOR Pro":                 Programming,
-	"INI":                      Data,
-	"IRC log":                  Data,
-	"Idris":                    Programming,
-	"Inform 7":                 Programming,
-	"Inno Setup":               Programming,
-	"Io":                       Programming,
-	"Ioke":                     Programming,
-	"Isabelle":                 Programming,
-	"Isabelle ROOT":            Programming,
-	"J":                        Programming,
-	"JFlex":                    Programming,
-	"JSON":                     Data,
-	"JSON5":                    Data,
-	"JSONLD":                   Data,
-	"JSONiq":                   Programming,
-	"JSX":                      Programming,
-	"Jasmin":                   Programming,
-	"Java":                     Programming,
-	"Java Server Pages":        Programming,
-	"JavaScript":               Programming,
-	"Jison":                    Programming,
-	"Jison Lex":                Programming,
-	"Julia":                    Programming,
-	"Jupyter Notebook":         Markup,
-	"KRL":                      Programming,
-	"KiCad":                    Programming,
-	"Kit":                      Markup,
-	"Kotlin":                   Programming,
-	"LFE":                      Programming,
-	"LLVM":                     Programming,
-	"LOLCODE":                  Programming,
-	"LSL":                      Programming,
-	"LabVIEW":                  Programming,
-	"Lasso":                    Programming,
-	"Latte":                    Markup,
-	"Lean":                     Programming,
-	"Less":                     Markup,
-	"Lex":                      Programming,
-	"LilyPond":                 Programming,
-	"Limbo":                    Programming,
-	"Linker Script":            Data,
-	"Linux Kernel Module":      Data,
-	"Liquid":                   Markup,
-	"Literate Agda":            Programming,
-	"Literate CoffeeScript":    Programming,
-	"Literate Haskell":         Programming,
-	"LiveScript":               Programming,
-	"Logos":                    Programming,
-	"Logtalk":                  Programming,
-	"LookML":                   Programming,
-	"LoomScript":               Programming,
-	"Lua":                      Programming,
-	"M":                        Programming,
-	"M4":                       Programming,
-	"M4Sugar":                  Programming,
-	"MAXScript":                Programming,
-	"MQL4":                     Programming,
-	"MQL5":                     Programming,
-	"MTML":                     Markup,
-	"MUF":                      Programming,
-	"Makefile":                 Programming,
-	"Mako":                     Programming,
-	"Markdown":                 Prose,
-	"Marko":                    Markup,
-	"Mask":                     Markup,
-	"Mathematica":              Programming,
-	"Matlab":                   Programming,
-	"Maven POM":                Data,
-	"Max":                      Programming,
-	"MediaWiki":                Prose,
-	"Mercury":                  Programming,
-	"Meson":                    Programming,
-	"Metal":                    Programming,
-	"MiniD":                    Programming,
-	"Mirah":                    Programming,
-	"Modelica":                 Programming,
-	"Modula-2":                 Programming,
-	"Module Management System": Programming,
-	"Monkey":                   Programming,
-	"Moocode":                  Programming,
-	"MoonScript":               Programming,
-	"Myghty":                   Programming,
-	"NCL":                      Programming,
-	"NL":                       Data,
-	"NSIS":                     Programming,
-	"Nemerle":                  Programming,
-	"NetLinx":                  Programming,
-	"NetLinx+ERB":              Programming,
-	"NetLogo":                  Programming,
-	"NewLisp":                  Programming,
-	"Nginx":                    Markup,
-	"Nim":                      Programming,
-	"Ninja":                    Data,
-	"Nit":                      Programming,
-	"Nix":                      Programming,
-	"Nu":                       Programming,
-	"NumPy":                    Programming,
-	"OCaml":                    Programming,
-	"ObjDump":                  Data,
-	"Objective-C":              Programming,
-	"Objective-C++":            Programming,
-	"Objective-J":              Programming,
-	"Omgrofl":                  Programming,
-	"Opa":                      Programming,
-	"Opal":                     Programming,
-	"OpenCL":                   Programming,
-	"OpenEdge ABL":             Programming,
-	"OpenRC runscript":         Programming,
-	"OpenSCAD":                 Programming,
-	"OpenType Feature File":    Data,
-	"Org":                            Prose,
-	"Ox":                             Programming,
-	"Oxygene":                        Programming,
-	"Oz":                             Programming,
-	"P4":                             Programming,
-	"PAWN":                           Programming,
-	"PHP":                            Programming,
-	"PLSQL":                          Programming,
-	"PLpgSQL":                        Programming,
-	"POV-Ray SDL":                    Programming,
-	"Pan":                            Programming,
-	"Papyrus":                        Programming,
-	"Parrot":                         Programming,
-	"Parrot Assembly":                Programming,
-	"Parrot Internal Representation": Programming,
-	"Pascal":                       Programming,
-	"Perl":                         Programming,
-	"Perl6":                        Programming,
-	"Pic":                          Markup,
-	"Pickle":                       Data,
-	"PicoLisp":                     Programming,
-	"PigLatin":                     Programming,
-	"Pike":                         Programming,
-	"Pod":                          Prose,
-	"PogoScript":                   Programming,
-	"Pony":                         Programming,
-	"PostScript":                   Markup,
-	"PowerBuilder":                 Programming,
-	"PowerShell":                   Programming,
-	"Processing":                   Programming,
-	"Prolog":                       Programming,
-	"Propeller Spin":               Programming,
-	"Protocol Buffer":              Markup,
-	"Public Key":                   Data,
-	"Pug":                          Markup,
-	"Puppet":                       Programming,
-	"Pure Data":                    Programming,
-	"PureBasic":                    Programming,
-	"PureScript":                   Programming,
-	"Python":                       Programming,
-	"Python console":               Programming,
-	"Python traceback":             Data,
-	"QML":                          Programming,
-	"QMake":                        Programming,
-	"R":                            Programming,
-	"RAML":                         Markup,
-	"RDoc":                         Prose,
-	"REALbasic":                    Programming,
-	"REXX":                         Programming,
-	"RHTML":                        Markup,
-	"RMarkdown":                    Prose,
-	"RPM Spec":                     Data,
-	"RUNOFF":                       Markup,
-	"Racket":                       Programming,
-	"Ragel":                        Programming,
-	"Rascal":                       Programming,
-	"Raw token data":               Data,
-	"Reason":                       Programming,
-	"Rebol":                        Programming,
-	"Red":                          Programming,
-	"Redcode":                      Programming,
-	"Regular Expression":           Data,
-	"Ren'Py":                       Programming,
-	"RenderScript":                 Programming,
-	"RobotFramework":               Programming,
-	"Roff":                         Markup,
-	"Rouge":                        Programming,
-	"Ruby":                         Programming,
-	"Rust":                         Programming,
-	"SAS":                          Programming,
-	"SCSS":                         Markup,
-	"SMT":                          Programming,
-	"SPARQL":                       Data,
-	"SQF":                          Programming,
-	"SQL":                          Data,
-	"SQLPL":                        Programming,
-	"SRecode Template":             Markup,
-	"STON":                         Data,
-	"SVG":                          Data,
-	"Sage":                         Programming,
-	"SaltStack":                    Programming,
-	"Sass":                         Markup,
-	"Scala":                        Programming,
-	"Scaml":                        Markup,
-	"Scheme":                       Programming,
-	"Scilab":                       Programming,
-	"Self":                         Programming,
-	"Shell":                        Programming,
-	"ShellSession":                 Programming,
-	"Shen":                         Programming,
-	"Slash":                        Programming,
-	"Slim":                         Markup,
-	"Smali":                        Programming,
-	"Smalltalk":                    Programming,
-	"Smarty":                       Programming,
-	"SourcePawn":                   Programming,
-	"Spline Font Database":         Data,
-	"Squirrel":                     Programming,
-	"Stan":                         Programming,
-	"Standard ML":                  Programming,
-	"Stata":                        Programming,
-	"Stylus":                       Markup,
-	"SubRip Text":                  Data,
-	"Sublime Text Config":          Data,
-	"SuperCollider":                Programming,
-	"Swift":                        Programming,
-	"SystemVerilog":                Programming,
-	"TI Program":                   Programming,
-	"TLA":                          Programming,
-	"TOML":                         Data,
-	"TXL":                          Programming,
-	"Tcl":                          Programming,
-	"Tcsh":                         Programming,
-	"TeX":                          Markup,
-	"Tea":                          Markup,
-	"Terra":                        Programming,
-	"Text":                         Prose,
-	"Textile":                      Prose,
-	"Thrift":                       Programming,
-	"Turing":                       Programming,
-	"Turtle":                       Data,
-	"Twig":                         Markup,
-	"TypeScript":                   Programming,
-	"Unified Parallel C":           Programming,
-	"Unity3D Asset":                Data,
-	"Unix Assembly":                Programming,
-	"Uno":                          Programming,
-	"UnrealScript":                 Programming,
-	"UrWeb":                        Programming,
-	"VCL":                          Programming,
-	"VHDL":                         Programming,
-	"Vala":                         Programming,
-	"Verilog":                      Programming,
-	"Vim script":                   Programming,
-	"Visual Basic":                 Programming,
-	"Volt":                         Programming,
-	"Vue":                          Markup,
-	"Wavefront Material":           Data,
-	"Wavefront Object":             Data,
-	"Web Ontology Language":        Markup,
-	"WebIDL":                       Programming,
-	"World of Warcraft Addon Data": Data,
-	"X10":              Programming,
-	"XC":               Programming,
-	"XCompose":         Data,
-	"XML":              Data,
-	"XPages":           Programming,
-	"XProc":            Programming,
-	"XQuery":           Programming,
-	"XS":               Programming,
-	"XSLT":             Programming,
-	"Xojo":             Programming,
-	"Xtend":            Programming,
-	"YAML":             Data,
-	"YANG":             Data,
-	"Yacc":             Programming,
-	"Zephir":           Programming,
-	"Zimpl":            Programming,
-	"desktop":          Data,
-	"eC":               Programming,
-	"edn":              Data,
-	"fish":             Programming,
-	"mupad":            Programming,
-	"nesC":             Programming,
-	"ooc":              Programming,
-	"reStructuredText": Prose,
-	"wisp":             Programming,
-	"xBase":            Programming,
-}

type_test.go

@@ -29,4 +29,7 @@ func (s *TSuite) TestGetLanguageType(c *C) {
 
 	langType = GetLanguageType("Textile")
 	c.Assert(langType, Equals, Prose)
+
+	langType = GetLanguageType("Whatever")
+	c.Assert(langType, Equals, Unknown)
 }

types_map.go

@@ -0,0 +1,452 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH gopkg.in/src-d/simple-linguist.v1/internal/code-generator
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: dae33dc2b20cddc85d1300435c3be7118a7115a9
+
+var languagesType = map[string]Type{
+	"1C Enterprise":    Programming,
+	"ABAP":             Programming,
+	"ABNF":             Data,
+	"AGS Script":       Programming,
+	"AMPL":             Programming,
+	"ANTLR":            Programming,
+	"API Blueprint":    Markup,
+	"APL":              Programming,
+	"ASN.1":            Data,
+	"ASP":              Programming,
+	"ATS":              Programming,
+	"ActionScript":     Programming,
+	"Ada":              Programming,
+	"Agda":             Programming,
+	"Alloy":            Programming,
+	"Alpine Abuild":    Programming,
+	"Ant Build System": Data,
+	"ApacheConf":       Markup,
+	"Apex":             Programming,
+	"Apollo Guidance Computer": Programming,
+	"AppleScript":              Programming,
+	"Arc":                      Programming,
+	"Arduino":                  Programming,
+	"AsciiDoc":                 Prose,
+	"AspectJ":                  Programming,
+	"Assembly":                 Programming,
+	"Augeas":                   Programming,
+	"AutoHotkey":               Programming,
+	"AutoIt":                   Programming,
+	"Awk":                      Programming,
+	"Batchfile":                Programming,
+	"Befunge":                  Programming,
+	"Bison":                    Programming,
+	"BitBake":                  Programming,
+	"Blade":                    Markup,
+	"BlitzBasic":               Programming,
+	"BlitzMax":                 Programming,
+	"Bluespec":                 Programming,
+	"Boo":                      Programming,
+	"Brainfuck":                Programming,
+	"Brightscript":             Programming,
+	"Bro":                      Programming,
+	"C":                        Programming,
+	"C#":                       Programming,
+	"C++":                      Programming,
+	"C-ObjDump":                Data,
+	"C2hs Haskell":             Programming,
+	"CLIPS":                    Programming,
+	"CMake":                    Programming,
+	"COBOL":                    Programming,
+	"COLLADA":                  Data,
+	"CSON":                     Data,
+	"CSS":                      Markup,
+	"CSV":                      Data,
+	"Cap'n Proto":              Programming,
+	"CartoCSS":                 Programming,
+	"Ceylon":                   Programming,
+	"Chapel":                   Programming,
+	"Charity":                  Programming,
+	"ChucK":                    Programming,
+	"Cirru":                    Programming,
+	"Clarion":                  Programming,
+	"Clean":                    Programming,
+	"Click":                    Programming,
+	"Clojure":                  Programming,
+	"CoffeeScript":             Programming,
+	"ColdFusion":               Programming,
+	"ColdFusion CFC":           Programming,
+	"Common Lisp":              Programming,
+	"Component Pascal":         Programming,
+	"Cool":                     Programming,
+	"Coq":                      Programming,
+	"Cpp-ObjDump":              Data,
+	"Creole":                   Prose,
+	"Crystal":                  Programming,
+	"Csound":                   Programming,
+	"Csound Document":          Programming,
+	"Csound Score":             Programming,
+	"Cuda":                     Programming,
+	"Cycript":                  Programming,
+	"Cython":                   Programming,
+	"D":                        Programming,
+	"D-ObjDump":                Data,
+	"DIGITAL Command Language": Programming,
+	"DM":             Programming,
+	"DNS Zone":       Data,
+	"DTrace":         Programming,
+	"Darcs Patch":    Data,
+	"Dart":           Programming,
+	"Diff":           Data,
+	"Dockerfile":     Data,
+	"Dogescript":     Programming,
+	"Dylan":          Programming,
+	"E":              Programming,
+	"EBNF":           Data,
+	"ECL":            Programming,
+	"ECLiPSe":        Programming,
+	"EJS":            Markup,
+	"EQ":             Programming,
+	"Eagle":          Markup,
+	"Ecere Projects": Data,
+	"Eiffel":         Programming,
+	"Elixir":         Programming,
+	"Elm":            Programming,
+	"Emacs Lisp":     Programming,
+	"EmberScript":    Programming,
+	"Erlang":         Programming,
+	"F#":             Programming,
+	"FLUX":           Programming,
+	"Factor":         Programming,
+	"Fancy":          Programming,
+	"Fantom":         Programming,
+	"Filebench WML":  Programming,
+	"Filterscript":   Programming,
+	"Formatted":      Data,
+	"Forth":          Programming,
+	"Fortran":        Programming,
+	"FreeMarker":     Programming,
+	"Frege":          Programming,
+	"G-code":         Data,
+	"GAMS":           Programming,
+	"GAP":            Programming,
+	"GCC Machine Description": Programming,
+	"GDB":      Programming,
+	"GDScript": Programming,
+	"GLSL":     Programming,
+	"GN":       Data,
+	"Game Maker Language":     Programming,
+	"Genie":                   Programming,
+	"Genshi":                  Programming,
+	"Gentoo Ebuild":           Programming,
+	"Gentoo Eclass":           Programming,
+	"Gettext Catalog":         Prose,
+	"Gherkin":                 Programming,
+	"Glyph":                   Programming,
+	"Gnuplot":                 Programming,
+	"Go":                      Programming,
+	"Golo":                    Programming,
+	"Gosu":                    Programming,
+	"Grace":                   Programming,
+	"Gradle":                  Data,
+	"Grammatical Framework":   Programming,
+	"Graph Modeling Language": Data,
+	"GraphQL":                 Data,
+	"Graphviz (DOT)":          Data,
+	"Groovy":                  Programming,
+	"Groovy Server Pages":     Programming,
+	"HCL":                      Programming,
+	"HLSL":                     Programming,
+	"HTML":                     Markup,
+	"HTML+Django":              Markup,
+	"HTML+ECR":                 Markup,
+	"HTML+EEX":                 Markup,
+	"HTML+ERB":                 Markup,
+	"HTML+PHP":                 Markup,
+	"HTTP":                     Data,
+	"Hack":                     Programming,
+	"Haml":                     Markup,
+	"Handlebars":               Markup,
+	"Harbour":                  Programming,
+	"Haskell":                  Programming,
+	"Haxe":                     Programming,
+	"Hy":                       Programming,
+	"HyPhy":                    Programming,
+	"IDL":                      Programming,
+	"IGOR Pro":                 Programming,
+	"INI":                      Data,
+	"IRC log":                  Data,
+	"Idris":                    Programming,
+	"Inform 7":                 Programming,
+	"Inno Setup":               Programming,
+	"Io":                       Programming,
+	"Ioke":                     Programming,
+	"Isabelle":                 Programming,
+	"Isabelle ROOT":            Programming,
+	"J":                        Programming,
+	"JFlex":                    Programming,
+	"JSON":                     Data,
+	"JSON5":                    Data,
+	"JSONLD":                   Data,
+	"JSONiq":                   Programming,
+	"JSX":                      Programming,
+	"Jasmin":                   Programming,
+	"Java":                     Programming,
+	"Java Server Pages":        Programming,
+	"JavaScript":               Programming,
+	"Jison":                    Programming,
+	"Jison Lex":                Programming,
+	"Julia":                    Programming,
+	"Jupyter Notebook":         Markup,
+	"KRL":                      Programming,
+	"KiCad":                    Programming,
+	"Kit":                      Markup,
+	"Kotlin":                   Programming,
+	"LFE":                      Programming,
+	"LLVM":                     Programming,
+	"LOLCODE":                  Programming,
+	"LSL":                      Programming,
+	"LabVIEW":                  Programming,
+	"Lasso":                    Programming,
+	"Latte":                    Markup,
+	"Lean":                     Programming,
+	"Less":                     Markup,
+	"Lex":                      Programming,
+	"LilyPond":                 Programming,
+	"Limbo":                    Programming,
+	"Linker Script":            Data,
+	"Linux Kernel Module":      Data,
+	"Liquid":                   Markup,
+	"Literate Agda":            Programming,
+	"Literate CoffeeScript":    Programming,
+	"Literate Haskell":         Programming,
+	"LiveScript":               Programming,
+	"Logos":                    Programming,
+	"Logtalk":                  Programming,
+	"LookML":                   Programming,
+	"LoomScript":               Programming,
+	"Lua":                      Programming,
+	"M":                        Programming,
+	"M4":                       Programming,
+	"M4Sugar":                  Programming,
+	"MAXScript":                Programming,
+	"MQL4":                     Programming,
+	"MQL5":                     Programming,
+	"MTML":                     Markup,
+	"MUF":                      Programming,
+	"Makefile":                 Programming,
+	"Mako":                     Programming,
+	"Markdown":                 Prose,
+	"Marko":                    Markup,
+	"Mask":                     Markup,
+	"Mathematica":              Programming,
+	"Matlab":                   Programming,
+	"Maven POM":                Data,
+	"Max":                      Programming,
+	"MediaWiki":                Prose,
+	"Mercury":                  Programming,
+	"Meson":                    Programming,
+	"Metal":                    Programming,
+	"MiniD":                    Programming,
+	"Mirah":                    Programming,
+	"Modelica":                 Programming,
+	"Modula-2":                 Programming,
+	"Module Management System": Programming,
+	"Monkey":                   Programming,
+	"Moocode":                  Programming,
+	"MoonScript":               Programming,
+	"Myghty":                   Programming,
+	"NCL":                      Programming,
+	"NL":                       Data,
+	"NSIS":                     Programming,
+	"Nemerle":                  Programming,
+	"NetLinx":                  Programming,
+	"NetLinx+ERB":              Programming,
+	"NetLogo":                  Programming,
+	"NewLisp":                  Programming,
+	"Nginx":                    Markup,
+	"Nim":                      Programming,
+	"Ninja":                    Data,
+	"Nit":                      Programming,
+	"Nix":                      Programming,
+	"Nu":                       Programming,
+	"NumPy":                    Programming,
+	"OCaml":                    Programming,
+	"ObjDump":                  Data,
+	"Objective-C":              Programming,
+	"Objective-C++":            Programming,
+	"Objective-J":              Programming,
+	"Omgrofl":                  Programming,
+	"Opa":                      Programming,
+	"Opal":                     Programming,
+	"OpenCL":                   Programming,
+	"OpenEdge ABL":             Programming,
+	"OpenRC runscript":         Programming,
+	"OpenSCAD":                 Programming,
+	"OpenType Feature File":    Data,
+	"Org":                            Prose,
+	"Ox":                             Programming,
+	"Oxygene":                        Programming,
+	"Oz":                             Programming,
+	"P4":                             Programming,
+	"PAWN":                           Programming,
+	"PHP":                            Programming,
+	"PLSQL":                          Programming,
+	"PLpgSQL":                        Programming,
+	"POV-Ray SDL":                    Programming,
+	"Pan":                            Programming,
+	"Papyrus":                        Programming,
+	"Parrot":                         Programming,
+	"Parrot Assembly":                Programming,
+	"Parrot Internal Representation": Programming,
+	"Pascal":                       Programming,
+	"Perl":                         Programming,
+	"Perl6":                        Programming,
+	"Pic":                          Markup,
+	"Pickle":                       Data,
+	"PicoLisp":                     Programming,
+	"PigLatin":                     Programming,
+	"Pike":                         Programming,
+	"Pod":                          Prose,
+	"PogoScript":                   Programming,
+	"Pony":                         Programming,
+	"PostScript":                   Markup,
+	"PowerBuilder":                 Programming,
+	"PowerShell":                   Programming,
+	"Processing":                   Programming,
+	"Prolog":                       Programming,
+	"Propeller Spin":               Programming,
+	"Protocol Buffer":              Markup,
+	"Public Key":                   Data,
+	"Pug":                          Markup,
+	"Puppet":                       Programming,
+	"Pure Data":                    Programming,
+	"PureBasic":                    Programming,
+	"PureScript":                   Programming,
+	"Python":                       Programming,
+	"Python console":               Programming,
+	"Python traceback":             Data,
+	"QML":                          Programming,
+	"QMake":                        Programming,
+	"R":                            Programming,
+	"RAML":                         Markup,
+	"RDoc":                         Prose,
+	"REALbasic":                    Programming,
+	"REXX":                         Programming,
+	"RHTML":                        Markup,
+	"RMarkdown":                    Prose,
+	"RPM Spec":                     Data,
+	"RUNOFF":                       Markup,
+	"Racket":                       Programming,
+	"Ragel":                        Programming,
+	"Rascal":                       Programming,
+	"Raw token data":               Data,
+	"Reason":                       Programming,
+	"Rebol":                        Programming,
+	"Red":                          Programming,
+	"Redcode":                      Programming,
+	"Regular Expression":           Data,
+	"Ren'Py":                       Programming,
+	"RenderScript":                 Programming,
+	"RobotFramework":               Programming,
+	"Roff":                         Markup,
+	"Rouge":                        Programming,
+	"Ruby":                         Programming,
+	"Rust":                         Programming,
+	"SAS":                          Programming,
+	"SCSS":                         Markup,
+	"SMT":                          Programming,
+	"SPARQL":                       Data,
+	"SQF":                          Programming,
+	"SQL":                          Data,
+	"SQLPL":                        Programming,
+	"SRecode Template":             Markup,
+	"STON":                         Data,
+	"SVG":                          Data,
+	"Sage":                         Programming,
+	"SaltStack":                    Programming,
+	"Sass":                         Markup,
+	"Scala":                        Programming,
+	"Scaml":                        Markup,
+	"Scheme":                       Programming,
+	"Scilab":                       Programming,
+	"Self":                         Programming,
+	"Shell":                        Programming,
+	"ShellSession":                 Programming,
+	"Shen":                         Programming,
+	"Slash":                        Programming,
+	"Slim":                         Markup,
+	"Smali":                        Programming,
+	"Smalltalk":                    Programming,
+	"Smarty":                       Programming,
+	"SourcePawn":                   Programming,
+	"Spline Font Database":         Data,
+	"Squirrel":                     Programming,
+	"Stan":                         Programming,
+	"Standard ML":                  Programming,
+	"Stata":                        Programming,
+	"Stylus":                       Markup,
+	"SubRip Text":                  Data,
+	"Sublime Text Config":          Data,
+	"SuperCollider":                Programming,
+	"Swift":                        Programming,
+	"SystemVerilog":                Programming,
+	"TI Program":                   Programming,
+	"TLA":                          Programming,
+	"TOML":                         Data,
+	"TXL":                          Programming,
+	"Tcl":                          Programming,
+	"Tcsh":                         Programming,
+	"TeX":                          Markup,
+	"Tea":                          Markup,
+	"Terra":                        Programming,
+	"Text":                         Prose,
+	"Textile":                      Prose,
+	"Thrift":                       Programming,
+	"Turing":                       Programming,
+	"Turtle":                       Data,
+	"Twig":                         Markup,
+	"TypeScript":                   Programming,
+	"Unified Parallel C":           Programming,
+	"Unity3D Asset":                Data,
+	"Unix Assembly":                Programming,
+	"Uno":                          Programming,
+	"UnrealScript":                 Programming,
+	"UrWeb":                        Programming,
+	"VCL":                          Programming,
+	"VHDL":                         Programming,
+	"Vala":                         Programming,
+	"Verilog":                      Programming,
+	"Vim script":                   Programming,
+	"Visual Basic":                 Programming,
+	"Volt":                         Programming,
+	"Vue":                          Markup,
+	"Wavefront Material":           Data,
+	"Wavefront Object":             Data,
+	"Web Ontology Language":        Markup,
+	"WebIDL":                       Programming,
+	"World of Warcraft Addon Data": Data,
+	"X10":              Programming,
+	"XC":               Programming,
+	"XCompose":         Data,
+	"XML":              Data,
+	"XPages":           Programming,
+	"XProc":            Programming,
+	"XQuery":           Programming,
+	"XS":               Programming,
+	"XSLT":             Programming,
+	"Xojo":             Programming,
+	"Xtend":            Programming,
+	"YAML":             Data,
+	"YANG":             Data,
+	"Yacc":             Programming,
+	"Zephir":           Programming,
+	"Zimpl":            Programming,
+	"desktop":          Data,
+	"eC":               Programming,
+	"edn":              Data,
+	"fish":             Programming,
+	"mupad":            Programming,
+	"nesC":             Programming,
+	"ooc":              Programming,
+	"reStructuredText": Prose,
+	"wisp":             Programming,
+	"xBase":            Programming,
+}