github-linguist 2.9.7 → 2.10.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: b3f06b9230afcfcea464a68df82268ec42c98030
-  data.tar.gz: 67e84ae38ba71cbb8c0dca2763a0d060bf764b84
+  metadata.gz: d32dfca063a58b4fdda048d20fbf9ee735a87a47
+  data.tar.gz: 8603ce6993bb645e7d847f991142436d999c5e96
 SHA512:
-  metadata.gz: 09205a528b8b4903b5f8ecbbebd740fdfaccb71704fad5fba13de2261689094b6a60605cc03db997a9d3830f84c4bdfe0e26033d823c90a4230f7cdfdc0e1422
-  data.tar.gz: f10dd819715aa9295d3e031097aecf68cb3ce66eb3553ec191d91155ec145de009c74245cc996877d77dd0510f45800808b68c5515e4a2a655f8f92009e67eb9
+  metadata.gz: db134aedc386ccc175b578474a84c75391814b235e01da962b6d2483453cfc80dec48670c5a23c65930fb723c92a0ca48df320c24eeead87fedbe5c6a21a448f
+  data.tar.gz: 89195a956fe06ab069fff7c4bf61c2324cb397964d818fcd7926eb8548b56e58ad73bf566b9cedd88275d10def55e3909e715bd686754b45b83e0aaacbb717b3

data/bin/linguist

@@ -12,8 +12,9 @@ path = ARGV[0] || Dir.pwd
 if File.directory?(path)
   repo = Linguist::Repository.from_directory(path)
   repo.languages.sort_by { |_, size| size }.reverse.each do |language, size|
-    percentage = ((size / repo.size.to_f) * 100).round
-    puts "%-4s %s" % ["#{percentage}%", language]
+    percentage = ((size / repo.size.to_f) * 100)
+    percentage = sprintf '%.2f' % percentage
+    puts "%-7s %s" % ["#{percentage}%", language]
   end
 elsif File.file?(path)
   blob = Linguist::FileBlob.new(path, Dir.pwd)

data/lib/linguist/classifier.rb

@@ -15,8 +15,8 @@ module Linguist
     #
     # Returns nothing.
     #
-    # Set LINGUIST_DEBUG=1 or =2 to see probabilities per-token,
-    # per-language.  See also dump_all_tokens, below.
+    # Set LINGUIST_DEBUG=1 or =2 to see probabilities per-token or
+    # per-language.  See also #dump_all_tokens, below.
     def self.train!(db, language, data)
       tokens = Tokenizer.tokenize(data)
 
@@ -151,10 +151,10 @@ module Linguist
         printf "%#{maxlen}s", ""
         puts "    #" + languages.map { |lang| sprintf("%10s", lang) }.join
         
-        tokmap = Hash.new(0)
-        tokens.each { |tok| tokmap[tok] += 1 }
+        token_map = Hash.new(0)
+        tokens.each { |tok| token_map[tok] += 1 }
         
-        tokmap.sort.each { |tok, count|
+        token_map.sort.each { |tok, count|
           arr = languages.map { |lang| [lang, token_probability(tok, lang)] }
           min = arr.map { |a,b| b }.min
           minlog = Math.log(min)

data/lib/linguist/generated.rb

@@ -58,7 +58,9 @@ module Linguist
         generated_parser? ||
         generated_net_docfile? ||
         generated_net_designer_file? ||
-        generated_protocol_buffer?
+        generated_protocol_buffer? ||
+        generated_jni_header? ||
+        node_modules?
     end
 
     # Internal: Is the blob an XCode project file?
@@ -73,14 +75,16 @@ module Linguist
 
     # Internal: Is the blob minified files?
     #
-    # Consider a file minified if it contains more than 5% spaces.
+    # Consider a file minified if the average line length is
+    # greater then 110c.
+    #
     # Currently, only JS and CSS files are detected by this method.
     #
     # Returns true or false.
     def minified_files?
       return unless ['.js', '.css'].include? extname
-      if data && data.length > 200
-        (data.each_char.count{ |c| c <= ' ' } / data.length.to_f) < 0.05
+      if lines.any?
+        (lines.inject(0) { |n, l| n += l.length } / lines.length) > 110
       else
         false
       end
@@ -181,5 +185,24 @@ module Linguist
 
       return lines[0].include?("Generated by the protocol buffer compiler.  DO NOT EDIT!")
     end
+
+    # Internal: Is the blob a C/C++ header generated by the Java JNI tool javah?
+    #
+    # Returns true of false.
+    def generated_jni_header?
+      return false unless extname == '.h'
+      return false unless lines.count > 2
+
+      return lines[0].include?("/* DO NOT EDIT THIS FILE - it is machine generated */") &&
+               lines[1].include?("#include <jni.h>")
+    end
+
+    # node_modules/ can contain large amounts of files, in general not meant
+    # for humans in pull requests.
+    #
+    # Returns true or false.
+    def node_modules?
+      !!name.match(/node_modules\//)
+    end
   end
 end

data/lib/linguist/language.rb

@@ -1,6 +1,10 @@
 require 'escape_utils'
 require 'pygments'
 require 'yaml'
+begin
+  require 'json'
+rescue LoadError
+end
 
 require 'linguist/classifier'
 require 'linguist/samples'
@@ -27,7 +31,7 @@ module Linguist
     #
     # Returns an array
     def self.detectable_markup
-      ["CSS", "Less", "Sass"]
+      ["CSS", "Less", "Sass", "TeX"]
     end
 
     # Internal: Create a new Language object
@@ -455,7 +459,16 @@ module Linguist
   filenames = Samples::DATA['filenames']
   popular = YAML.load_file(File.expand_path("../popular.yml", __FILE__))
 
-  YAML.load_file(File.expand_path("../languages.yml", __FILE__)).each do |name, options|
+  languages_yml = File.expand_path("../languages.yml", __FILE__)
+  languages_json = File.expand_path("../languages.json", __FILE__)
+
+  if File.exist?(languages_json) && defined?(JSON)
+    languages = JSON.load(File.read(languages_json))
+  else
+    languages = YAML.load_file(languages_yml)
+  end
+
+  languages.each do |name, options|
     options['extensions'] ||= []
     options['filenames'] ||= []
 

data/lib/linguist/languages.yml

@@ -29,6 +29,12 @@ ABAP:
   lexer: ABAP
   primary_extension: .abap
 
+ANTLR:
+  type: programming
+  color: "#9DC3FF"
+  lexer: ANTLR
+  primary_extension: .g4
+
 ASP:
   type: programming
   color: "#6a40fd"
@@ -62,6 +68,10 @@ Ada:
   extensions:
   - .ads
 
+Agda:
+  type: programming
+  primary_extension: .agda
+
 ApacheConf:
   type: markup
   aliases:
@@ -112,11 +122,21 @@ AutoHotkey:
   - ahk
   primary_extension: .ahk
 
+AutoIt:
+  type: programming
+  color: "#36699B"
+  aliases:
+  - au3
+  - AutoIt3
+  - AutoItScript
+  primary_extension: .au3
+
 Awk:
   type: programming
   lexer: Awk
   primary_extension: .awk
   extensions:
+  - .auk
   - .gawk
   - .mawk
   - .nawk
@@ -134,9 +154,23 @@ Batchfile:
 Befunge:
   primary_extension: .befunge
 
+BlitzBasic:
+  type: programming
+  aliases:
+  - blitzplus
+  - blitz3d
+  primary_extension: .bb
+  extensions:
+  - .decls
+
 BlitzMax:
   primary_extension: .bmx
 
+Bluespec:
+  type: programming
+  lexer: verilog
+  primary_extension: .bsv
+
 Boo:
   type: programming
   color: "#d4bec1"
@@ -184,8 +218,10 @@ C++:
   - .H
   - .h++
   - .hh
+  - .hpp
   - .hxx
   - .tcc
+  - .tpp
 
 C-ObjDump:
   type: data
@@ -235,14 +271,26 @@ ChucK:
   lexer: Java
   primary_extension: .ck
 
+Clean:
+  type: programming
+  color: "#3a81ad"
+  lexer: Text only
+  primary_extension: .icl
+  extensions:
+  - .dcl
+
 Clojure:
   type: programming
   ace_mode: clojure
   color: "#db5855"
   primary_extension: .clj
   extensions:
+  - .cl2
+  - .cljc
   - .cljs
+  - .cljscm
   - .cljx
+  - .hic
   filenames:
   - riemann.config
 
@@ -281,6 +329,7 @@ Common Lisp:
   primary_extension: .lisp
   extensions:
   - .asd
+  - .cl
   - .lsp
   - .ny
   - .podsl
@@ -301,6 +350,12 @@ Cucumber:
   lexer: Gherkin
   primary_extension: .feature
 
+Cuda:
+  lexer: CUDA
+  primary_extension: .cu
+  extensions:
+  - .cuh
+
 Cython:
   type: programming
   group: Python
@@ -321,6 +376,14 @@ D-ObjDump:
   lexer: d-objdump
   primary_extension: .d-objdump
 
+DM:
+  type: programming
+  color: "#075ff1"
+  lexer: Text only
+  primary_extension: .dm
+  aliases:
+  - byond
+
 DOT:
   type: programming
   lexer: Text only
@@ -338,6 +401,7 @@ Darcs Patch:
 
 Dart:
   type: programming
+  color: "#98BAD6"
   primary_extension: .dart
 
 DCPU-16 ASM:
@@ -363,7 +427,7 @@ Ecere Projects:
   lexer: JSON
   primary_extension: .epj
 
-Ecl:
+ECL:
   type: programming
   color: "#8a1267"
   primary_extension: .ecl
@@ -397,6 +461,8 @@ Emacs Lisp:
   - elisp
   - emacs
   primary_extension: .el
+  filenames:
+  - .emacs
   extensions:
   - .emacs
 
@@ -512,6 +578,12 @@ Gettext Catalog:
   extensions:
   - .pot
 
+Glyph:
+  type: programming
+  color: "#e4cc98"
+  lexer: Tcl
+  primary_extension: .glf
+
 Go:
   type: programming
   color: "#a89b4d"
@@ -612,7 +684,6 @@ Haskell:
 
 Haxe:
   type: programming
-  lexer: haXe
   ace_mode: haxe
   color: "#346d51"
   primary_extension: .hx
@@ -627,6 +698,17 @@ INI:
   - .properties
   primary_extension: .ini
 
+Idris:
+  type: programming
+  lexer: Text only
+  primary_extension: .idr
+  extensions:
+  - .lidr
+
+Inno Setup:
+  primary_extension: .iss
+  lexer: Text only
+
 IRC log:
   lexer: IRC logs
   search_term: irc
@@ -657,6 +739,21 @@ JSON:
   ace_mode: json
   searchable: false
   primary_extension: .json
+  extensions:
+  - .sublime-keymap
+  - .sublime_metrics
+  - .sublime-mousemap
+  - .sublime-project
+  - .sublime_session
+  - .sublime-settings
+  - .sublime-workspace
+  filenames:
+  - composer.lock
+
+Jade:
+  group: HTML
+  type: markup
+  primary_extension: .jade
 
 Java:
   type: programming
@@ -699,6 +796,12 @@ Julia:
   primary_extension: .jl
   color: "#a270ba"
 
+KRL:
+  lexer: Text only
+  type: programming
+  color: "#f5c800"
+  primary_extension: .krl
+
 Kotlin:
   type: programming
   primary_extension: .kt
@@ -719,7 +822,6 @@ LLVM:
 Lasso:
   type: programming
   lexer: Lasso
-  ace_mode: lasso
   color: "#2584c3"
   primary_extension: .lasso
 
@@ -727,7 +829,6 @@ Less:
   type: markup
   group: CSS
   lexer: CSS
-  ace_mode: less
   primary_extension: .less
 
 LilyPond:
@@ -736,6 +837,13 @@ LilyPond:
   extensions:
   - .ily
 
+Literate Agda:
+  type: programming
+  group: Agda
+  primary_extension: .lagda
+  extensions:
+    - .lagda
+
 Literate CoffeeScript:
   type: programming
   group: CoffeeScript
@@ -836,15 +944,17 @@ Matlab:
 Max:
   type: programming
   color: "#ce279c"
-  lexer: Text only
+  lexer: JSON
   aliases:
   - max/msp
   - maxmsp
   search_term: max/msp
-  primary_extension: .mxt
+  primary_extension: .maxpat
   extensions:
   - .maxhelp
-  - .maxpat
+  - .maxproj
+  - .mxt
+  - .pat
 
 MiniD: # Legacy
   searchable: false
@@ -885,6 +995,12 @@ Nemerle:
   color: "#0d3c6e"
   primary_extension: .n
 
+NetLogo:
+  type: programming
+  lexer: Common Lisp
+  color: "#ff2b2b"
+  primary_extension: .nlogo
+
 Nginx:
   type: markup
   lexer: Nginx configuration file
@@ -975,6 +1091,12 @@ OpenEdge ABL:
   - abl
   primary_extension: .p
 
+Oxygene:
+  type: programming
+  lexer: Text only
+  color: "#5a63a3"
+  primary_extension: .oxygene
+
 PHP:
   type: programming
   ace_mode: php
@@ -1056,6 +1178,9 @@ PowerShell:
   aliases:
   - posh
   primary_extension: .ps1
+  extensions:
+  - .psd1
+  - .psm1
 
 Processing:
   type: programming
@@ -1070,6 +1195,13 @@ Prolog:
   extensions:
   - .pro
 
+Protocol Buffer:
+  type: markup
+  aliases:
+  - protobuf
+  - Protocol Buffers
+  primary_extension: .proto
+
 Puppet:
   type: programming
   color: "#cc5555"
@@ -1106,14 +1238,32 @@ Python traceback:
   searchable: false
   primary_extension: .pytb
 
+QML:
+  type: markup
+  color: "#44a51c"
+  primary_extension: .qml
+
 R:
   type: programming
   color: "#198ce7"
   lexer: S
   primary_extension: .r
+  extensions:
+  - .R
   filenames:
   - .Rprofile
 
+REALbasic:
+  type: programming
+  lexer: VB.net
+  primary_extension: .rbbas
+  extensions:
+  - .rbfrm
+  - .rbmnu
+  - .rbres
+  - .rbtbar
+  - .rbuistate
+
 RHTML:
   type: markup
   group: HTML
@@ -1152,6 +1302,12 @@ Rebol:
 Redcode:
   primary_extension: .cw
 
+RobotFramework:
+  type: programming
+  primary_extension: .robot
+  # extensions:
+  # - .txt
+
 Rouge:
   type: programming
   lexer: Clojure
@@ -1175,6 +1331,7 @@ Ruby:
   - .gemspec
   - .god
   - .irbrc
+  - .mspec
   - .podspec
   - .rbuild
   - .rbw
@@ -1183,6 +1340,7 @@ Ruby:
   - .thor
   - .watchr
   filenames:
+  - Appraisals
   - Berksfile
   - Gemfile
   - Guardfile
@@ -1224,6 +1382,11 @@ Scala:
   color: "#7dd3b0"
   primary_extension: .scala
 
+Scaml:
+  group: HTML
+  type: markup
+  primary_extension: .scaml
+
 Scheme:
   type: programming
   color: "#1e4aec"
@@ -1302,6 +1465,8 @@ Tcl:
   type: programming
   color: "#e4cc98"
   primary_extension: .tcl
+  extensions:
+  - .adp
 
 Tcsh:
   type: programming
@@ -1322,6 +1487,9 @@ TeX:
   - .dtx
   - .ins
   - .ltx
+  - .mkii
+  - .mkiv
+  - .mkvi
   - .sty
   - .toc
 
@@ -1365,6 +1533,12 @@ Unified Parallel C:
   color: "#755223"
   primary_extension: .upc
 
+UnrealScript:
+  type: programming
+  color: "#a54c4d"
+  lexer: Java
+  primary_extension: .uc
+
 VHDL:
   type: programming
   lexer: vhdl
@@ -1384,9 +1558,7 @@ Verilog:
   color: "#848bf3"
   primary_extension: .v
   extensions:
-  - .sv
-  - .svh
-  - .vh
+  - .veo
 
 VimL:
   type: programming
@@ -1396,6 +1568,7 @@ VimL:
   - vim
   primary_extension: .vim
   filenames:
+  - .vimrc
   - vimrc
   - gvimrc
 
@@ -1406,6 +1579,7 @@ Visual Basic:
   primary_extension: .vb
   extensions:
   - .bas
+  - .frm
   - .frx
   - .vba
   - .vbs
@@ -1432,6 +1606,8 @@ XML:
   extensions:
   - .axml
   - .ccxml
+  - .clixml
+  - .cproject
   - .dita
   - .ditamap
   - .ditaval
@@ -1439,12 +1615,17 @@ XML:
   - .grxml
   - .jelly
   - .kml
+  - .launch
   - .mxml
   - .plist
+  - .pluginspec
+  - .ps1xml
+  - .psc1
   - .pt
   - .rdf
   - .rss
   - .scxml
+  - .srdf
   - .svg
   - .tmCommand
   - .tmLanguage
@@ -1453,12 +1634,14 @@ XML:
   - .tmTheme
   - .tml
   - .ui
+  - .urdf
   - .vxml
   - .wsdl
   - .wxi
   - .wxl
   - .wxs
   - .x3d
+  - .xacro
   - .xaml
   - .xlf
   - .xliff
@@ -1469,6 +1652,7 @@ XML:
   filenames:
   - .classpath
   - .project
+  - phpunit.xml.dist
 
 XProc:
   type: programming
@@ -1483,6 +1667,8 @@ XQuery:
   primary_extension: .xquery
   extensions:
   - .xq
+  - .xql
+  - .xqm
   - .xqy
 
 XS:
@@ -1508,6 +1694,7 @@ YAML:
   primary_extension: .yml
   extensions:
   - .reek
+  - .rviz
   - .yaml
 
 eC:
@@ -1534,6 +1721,11 @@ mupad:
   lexer: MuPAD
   primary_extension: .mu
 
+nesC:
+  type: programming
+  color: "#ffce3b"
+  primary_extension: .nc
+
 ooc:
   type: programming
   lexer: Ooc
@@ -1556,3 +1748,9 @@ wisp:
   ace_mode: clojure
   color: "#7582D1"
   primary_extension: .wisp
+
+xBase:
+  type: programming
+  lexer: Text only
+  color: "#3a4040"
+  primary_extension: .prg