Merge branch 'master' into benchmarking

Conflicts: lib/linguist/heuristics.rb lib/linguist/languages.yml
2025-10-29 17:50:22 +00:00 · 2014-09-10 17:06:48 -05:00
parent 8cdb8ed48d 4a017d9033
commit 0c47f2af75
49 changed files with 10868 additions and 438 deletions
--- a/lib/linguist/generated.rb
+++ b/lib/linguist/generated.rb
@@ -235,7 +235,7 @@ module Linguist
    #
    # Returns true or false.
    def composer_lock?
-      !!name.match(/composer.lock/)
+      !!name.match(/composer\.lock/)
    end

    # Internal: Is the blob a generated by Zephir
--- a/lib/linguist/heuristics.rb
+++ b/lib/linguist/heuristics.rb
@@ -12,28 +12,15 @@ module Linguist
    #
    # Returns an array of Languages or []
    def self.find_by_heuristics(data, languages)
-      determined = nil
      if active?
-        if languages.all? { |l| ["Objective-C", "C++", "C"].include?(l) }
-          determined = disambiguate_c(data, languages)
-        end
        if languages.all? { |l| ["Perl", "Prolog"].include?(l) }
-          determined = disambiguate_pl(data, languages)
+          result = disambiguate_pl(data, languages)
        end
        if languages.all? { |l| ["ECL", "Prolog"].include?(l) }
-          determined = disambiguate_ecl(data, languages)
-        end
-        if languages.all? { |l| ["TypeScript", "XML"].include?(l) }
-          determined = disambiguate_ts(data, languages)
-        end
-        if languages.all? { |l| ["Common Lisp", "OpenCL"].include?(l) }
-          determined = disambiguate_cl(data, languages)
-        end
-        if languages.all? { |l| ["Rebol", "R"].include?(l) }
-          determined = disambiguate_r(data, languages)
+          result = disambiguate_ecl(data, languages)
        end
+        return result
      end
-      determined
    end

    # .h extensions are ambigious between C, C++, and Objective-C.
@@ -42,7 +29,7 @@ module Linguist
    # Returns an array of Languages or []
    def self.disambiguate_c(data, languages)
      matches = []
-      matches << Language["Objective-C"] if data.include?("i")
+      matches << Language["Objective-C"] if data.include?("@interface")
      matches << Language["C++"] if data.include?("#include <cstdint>")
      matches
    end
--- a/lib/linguist/language.rb
+++ b/lib/linguist/language.rb
@@ -136,7 +136,7 @@ module Linguist
        elsif (determined = Heuristics.find_by_heuristics(data, possible_language_names)) && !determined.empty?
          determined.first
        # Lastly, fall back to the probablistic classifier.
-        elsif classified = Classifier.classify(Samples::DATA, data, possible_language_names ).first
+        elsif classified = Classifier.classify(Samples::DATA, data, possible_language_names).first
          # Return the actual Language object based of the string language name (i.e., first element of `#classify`)
          Language[classified[0]]
        end
@@ -532,6 +532,7 @@ module Linguist
    if extnames = extensions[name]
      extnames.each do |extname|
        if !options['extensions'].include?(extname)
+          warn "#{name} has a sample with extension (#{extname}) that isn't explicitly defined in languages.yml" unless extname == '.script!'
          options['extensions'] << extname
        end
      end
--- a/lib/linguist/languages.yml
+++ b/lib/linguist/languages.yml
@@ -28,6 +28,16 @@ ABAP:
  extensions:
  - .abap

+AGS Script:
+  type: programming
+  lexer: C++
+  color: "#B9D9FF"
+  aliases:
+  - ags
+  extensions:
+  - .asc
+  - .ash
+
 ANTLR:
  type: programming
  color: "#9DC3FF"
@@ -35,6 +45,12 @@ ANTLR:
  extensions:
  - .g4

+APL:
+  type: programming
+  color: "#8a0707"
+  extensions:
+  - .apl
+
 ASP:
  type: programming
  color: "#6a40fd"
@@ -260,6 +276,7 @@ C:
  extensions:
  - .c
  - .cats
+  - .h
  - .w

 C#:
@@ -282,7 +299,21 @@ C++:
  aliases:
  - cpp
  extensions:
-  - .a
+  - .cpp
+  - .C
+  - .c++
+  - .cc
+  - .cxx
+  - .H
+  - .h
+  - .h++
+  - .hh
+  - .hpp
+  - .hxx
+  - .inl
+  - .tcc
+  - .tpp
+  - .ipp

 C-ObjDump:
  type: data
@@ -308,7 +339,7 @@ CLIPS:
 CMake:
  extensions:
  - .cmake
-  - .cmake.in
+  - .in
  filenames:
  - CMakeLists.txt

@@ -333,6 +364,14 @@ Ceylon:
  extensions:
  - .ceylon

+Chapel:
+  type: programming
+  color: "#8dc63f"
+  aliases:
+  - chpl
+  extensions:
+  - .chpl
+
 ChucK:
  lexer: Java
  extensions:
@@ -341,9 +380,8 @@ ChucK:
 Cirru:
  type: programming
  color: "#aaaaff"
-  # ace_mode: cirru
-  # lexer: Cirru
-  lexer: Text only
+  ace_mode: cirru
+  lexer: Cirru
  extensions:
  - .cirru

@@ -367,7 +405,7 @@ Clojure:
  - .cljscm
  - .cljx
  - .hic
-  - .cljs.hl
+  - .hl
  filenames:
  - riemann.config

@@ -390,14 +428,27 @@ CoffeeScript:

 ColdFusion:
  type: programming
+  group: ColdFusion
  lexer: Coldfusion HTML
  ace_mode: coldfusion
  color: "#ed2cd6"
  search_term: cfm
  aliases:
  - cfm
+  - cfml
  extensions:
  - .cfm
+
+ColdFusion CFC:
+  type: programming
+  group: ColdFusion
+  lexer: Coldfusion CFC
+  ace_mode: coldfusion
+  color: "#ed2cd6"
+  search_term: cfc
+  aliases:
+  - cfc
+  extensions:
  - .cfc

 Common Lisp:
@@ -431,6 +482,7 @@ Coq:
  type: programming
  extensions:
  - .coq
+  - .v

 Cpp-ObjDump:
  type: data
@@ -466,6 +518,12 @@ Cuda:
  - .cu
  - .cuh

+Cycript:
+  type: programming
+  lexer: JavaScript
+  extensions:
+  - .cy
+
 Cython:
  type: programming
  group: Python
@@ -520,6 +578,7 @@ Dart:
 Diff:
  extensions:
  - .diff
+  - .patch

 Dogescript:
  type: programming
@@ -568,7 +627,7 @@ Eagle:

 Eiffel:
  type: programming
-  lexer: Text only
+  lexer: Eiffel
  color: "#946d57"
  extensions:
  - .e
@@ -588,7 +647,7 @@ Elm:

 Emacs Lisp:
  type: programming
-  lexer: Scheme
+  lexer: Common Lisp
  color: "#c065db"
  aliases:
  - elisp
@@ -599,11 +658,20 @@ Emacs Lisp:
  - .el
  - .emacs

+EmberScript:
+  type: programming
+  color: "#f64e3e"
+  lexer: CoffeeScript
+  extensions:
+  - .em
+  - .emberscript
+
 Erlang:
  type: programming
  color: "#0faf8d"
  extensions:
  - .erl
+  - .escript
  - .hrl

 F#:
@@ -679,6 +747,7 @@ Forth:
  extensions:
  - .fth
  - .4th
+  - .forth

 Frege:
  type: programming
@@ -787,6 +856,9 @@ Gosu:
  color: "#82937f"
  extensions:
  - .gs
+  - .gst
+  - .gsx
+  - .vark

 Grace:
  type: programming
@@ -822,6 +894,7 @@ Groovy:
  color: "#e69f56"
  extensions:
  - .groovy
+  - .gradle
  - .grt
  - .gtpl
  - .gvy
@@ -844,7 +917,6 @@ HTML:
  extensions:
  - .html
  - .htm
-  - .html.hl
  - .st
  - .xhtml

@@ -864,9 +936,7 @@ HTML+ERB:
  - erb
  extensions:
  - .erb
-  - .erb.deface
-  - .html.erb
-  - .html.erb.deface
+  - .deface

 HTML+PHP:
  type: markup
@@ -884,17 +954,14 @@ Haml:
  type: markup
  extensions:
  - .haml
-  - .haml.deface
-  - .html.haml.deface
+  - .deface

 Handlebars:
  type: markup
-  lexer: Text only
+  lexer: Handlebars
  extensions:
  - .handlebars
  - .hbs
-  - .html.handlebars
-  - .html.hbs

 Harbour:
  type: programming
@@ -920,7 +987,7 @@ Haxe:

 Hy:
  type: programming
-  lexer: Clojure
+  lexer: Hy
  ace_mode: clojure
  color: "#7891b1"
  extensions:
@@ -932,6 +999,13 @@ IDL:
  color: "#e3592c"
  extensions:
  - .pro
+  - .dlm
+
+IGOR Pro:
+  type: programming
+  lexer: Igor
+  extensions:
+  - .ipf

 INI:
  type: data
@@ -954,7 +1028,7 @@ Idris:

 Inform 7:
  type: programming
-  lexer: Text only
+  lexer: Inform 7
  wrap: true
  extensions:
  - .ni
@@ -1006,6 +1080,7 @@ JSON:
  searchable: false
  extensions:
  - .json
+  - .lock
  - .sublime-keymap
  - .sublime-mousemap
  - .sublime-project
@@ -1128,12 +1203,31 @@ LLVM:
  extensions:
  - .ll

+LSL:
+  type: programming
+  lexer: LSL
+  ace_mode: lsl
+  extensions:
+  - .lsl
+  interpreters:
+  - lsl
+  color: '#3d9970'
+
+LabVIEW:
+  type: programming
+  lexer: Text only
+  extensions:
+  - .lvproj
+
 Lasso:
  type: programming
  lexer: Lasso
  color: "#2584c3"
  extensions:
  - .lasso
+  - .las
+  - .lasso9
+  - .ldml

 Latte:
  type: markup
@@ -1212,6 +1306,14 @@ Logtalk:
  - .lgt
  - .logtalk

+LookML:
+  type: programming
+  lexer: YAML
+  ace_mode: yaml
+  color: "#652B81"
+  extensions:
+  - .lookml
+
 Lua:
  type: programming
  ace_mode: lua
@@ -1219,6 +1321,7 @@ Lua:
  extensions:
  - .lua
  - .nse
+  - .pd_lua
  - .rbxs
  interpreters:
  - lua
@@ -1284,7 +1387,7 @@ Mathematica:
  - .mathematica
  - .m
  - .nb
-  lexer: Text only
+  lexer: Mathematica

 Matlab:
  type: programming
@@ -1364,6 +1467,7 @@ Myghty:
 NSIS:
  extensions:
  - .nsi
+  - .nsh

 Nemerle:
  type: programming
@@ -1428,6 +1532,7 @@ OCaml:
  color: "#3be133"
  extensions:
  - .ml
+  - .eliom
  - .eliomi
  - .ml4
  - .mli
@@ -1448,6 +1553,7 @@ Objective-C:
  - objc
  extensions:
  - .m
+  - .h

 Objective-C++:
  type: programming
@@ -1479,6 +1585,13 @@ Opa:
  extensions:
  - .opa

+Opal:
+  type: programming
+  color: "#f7ede0"
+  lexer: Text only
+  extensions:
+  - .opal
+
 OpenCL:
  type: programming
  group: C
@@ -1495,6 +1608,13 @@ OpenEdge ABL:
  - abl
  extensions:
  - .p
+  - .cls
+
+OpenSCAD:
+  type: programming
+  lexer: Text only
+  extensions:
+  - .scad

 Org:
  type: prose
@@ -1533,6 +1653,7 @@ PHP:
  - .php
  - .aw
  - .ctp
+  - .module
  - .php3
  - .php4
  - .php5
@@ -1544,7 +1665,7 @@ PHP:

 Pan:
  type: programming
-  lexer: Text only
+  lexer: Pan
  color: '#cc0000'
  extensions:
  - .pan
@@ -1581,6 +1702,7 @@ Pascal:
  extensions:
  - .pas
  - .dfm
+  - .dpr
  - .lpr

 Perl:
@@ -1590,12 +1712,15 @@ Perl:
  extensions:
  - .pl
  - .PL
+  - .cgi
+  - .fcgi
  - .perl
  - .ph
  - .plx
  - .pm
  - .pod
  - .psgi
+  - .t
  interpreters:
  - perl

@@ -1612,6 +1737,13 @@ Perl6:
  - .pl6
  - .pm6

+PigLatin:
+  type: programming
+  color: "#fcd7de"
+  lexer: Text only
+  extensions:
+  - .pig
+
 Pike:
  type: programming
  color: "#066ab2"
@@ -1662,9 +1794,9 @@ Prolog:
  type: programming
  color: "#74283c"
  extensions:
-  - .prolog
-  - .ecl
  - .pl
+  - .ecl
+  - .prolog

 Propeller Spin:
  type: programming
@@ -1709,6 +1841,7 @@ Python:
  color: "#3581ba"
  extensions:
  - .py
+  - .cgi
  - .gyp
  - .lmi
  - .pyde
@@ -1764,7 +1897,7 @@ R:

 RDoc:
  type: prose
-  lexer: Text only
+  lexer: Rd
  ace_mode: rdoc
  wrap: true
  extensions:
@@ -1804,6 +1937,7 @@ Racket:
  - .rkt
  - .rktd
  - .rktl
+  - .scrbl

 Ragel in Ruby Host:
  type: programming
@@ -1873,7 +2007,10 @@ Ruby:
  - .god
  - .irbrc
  - .mspec
+  - .pluginspec
  - .podspec
+  - .rabl
+  - .rake
  - .rbuild
  - .rbw
  - .rbx
@@ -1918,6 +2055,14 @@ SCSS:
  extensions:
  - .scss

+SQF:
+  type: programming
+  color: "#FFCB1F"
+  lexer: C++
+  extensions:
+  - .sqf
+  - .hqf
+
 SQL:
  type: data
  ace_mode: sql
@@ -1947,6 +2092,7 @@ Sass:
  group: CSS
  extensions:
  - .sass
+  - .scss

 Scala:
  type: programming
@@ -1954,6 +2100,7 @@ Scala:
  color: "#7dd3b0"
  extensions:
  - .scala
+  - .sbt
  - .sc

 Scaml:
@@ -1969,6 +2116,7 @@ Scheme:
  - .scm
  - .sld
  - .sls
+  - .sps
  - .ss
  interpreters:
  - guile
@@ -1980,6 +2128,8 @@ Scilab:
  type: programming
  extensions:
  - .sci
+  - .sce
+  - .tst

 Self:
  type: programming
@@ -1999,8 +2149,11 @@ Shell:
  - zsh
  extensions:
  - .sh
+  - .bash
  - .bats
+  - .cgi
  - .tmux
+  - .zsh
  interpreters:
  - bash
  - sh
@@ -2067,6 +2220,7 @@ Standard ML:
  extensions:
  - .ML
  - .fun
+  - .sig
  - .sml

 Stata:
@@ -2147,10 +2301,13 @@ TeX:
  extensions:
  - .tex
  - .aux
+  - .bbx
  - .bib
+  - .cbx
  - .cls
  - .dtx
  - .ins
+  - .lbx
  - .ltx
  - .mkii
  - .mkiv
@@ -2267,6 +2424,7 @@ Visual Basic:
  extensions:
  - .vb
  - .bas
+  - .cls
  - .frm
  - .frx
  - .vba
@@ -2295,6 +2453,7 @@ XML:
  - wsdl
  extensions:
  - .xml
+  - .ant
  - .axml
  - .ccxml
  - .clixml
@@ -2308,6 +2467,7 @@ XML:
  - .fsproj
  - .glade
  - .grxml
+  - .ivy
  - .jelly
  - .kml
  - .launch
--- a/lib/linguist/samples.json
+++ b/lib/linguist/samples.json
--- a/lib/linguist/vendor.yml
+++ b/lib/linguist/vendor.yml
@@ -36,6 +36,10 @@
 # Bootstrap minified css and js
 - (^|/)bootstrap([^.]*)(\.min)?\.(js|css)$

+# Font Awesome
+- font-awesome.min.css
+- font-awesome.css
+
 # Foundation css
 - foundation.min.css
 - foundation.css
@@ -43,8 +47,17 @@
 # Normalize.css
 - normalize.css

+# Bourbon SCSS
+- (^|/)[Bb]ourbon/.*\.css$
+- (^|/)[Bb]ourbon/.*\.scss$
+
+# Animate.css
+- animate.css
+- animate.min.css
+
 # Vendored dependencies
- thirdparty/
+- third[-_]?party/
+- 3rd[-_]?party/
 - vendors?/
 - extern(al)?/

@@ -112,6 +125,10 @@
 - (^|/)modernizr\-\d\.\d+(\.\d+)?(\.min)?\.js$
 - (^|/)modernizr\.custom\.\d+\.js$

+# Knockout
+- (^|/)knockout-(\d+\.){3}(debug\.)?js$
+- knockout-min.js
+
 ## Python ##

 # django
@@ -198,6 +215,9 @@
 - (^|/)cordova([^.]*)(\.min)?\.js$
 - (^|/)cordova\-\d\.\d(\.\d)?(\.min)?\.js$

+# Foundation js
+- foundation(\..*)?\.js$
+
 # Vagrant
 - ^Vagrantfile$

--- a/lib/linguist/version.rb
+++ b/lib/linguist/version.rb
@@ -1,3 +1,3 @@
 module Linguist
-  VERSION = "3.1.0"
+  VERSION = "3.1.5"
 end