Merge master

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)

lib/linguist/generated.rb

@@ -59,7 +59,8 @@ module Linguist
         generated_net_docfile? ||
         generated_net_designer_file? ||
         generated_protocol_buffer? ||
-        generated_jni_header?
+        generated_jni_header? ||
+        node_modules?
     end
 
     # Internal: Is the blob an XCode project file?
@@ -74,7 +75,7 @@ module Linguist
 
     # Internal: Is the blob minified files?
     #
-    # Consider a file minified if it contains more than 5% spaces.
+    # Consider a file minified if it contains less than 5% spaces.
     # Currently, only JS and CSS files are detected by this method.
     #
     # Returns true or false.
@@ -193,5 +194,13 @@ module Linguist
       return lines[0].include?("/* DO NOT EDIT THIS FILE - it is machine generated */")
       return 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

lib/linguist/language.rb

@@ -27,7 +27,7 @@ module Linguist
     #
     # Returns an array
     def self.detectable_markup
-      ["CSS", "Less", "Sass"]
+      ["CSS", "Less", "Sass", "TeX"]
     end
 
     # Internal: Create a new Language object

lib/linguist/languages.yml

@@ -28,13 +28,13 @@ ABAP:
   type: programming
   lexer: ABAP
   primary_extension: .abap
-  
+
 ANTLR:
   type: programming
   color: "#9DC3FF"
   lexer: ANTLR
   primary_extension: .g4
-  
+
 ASP:
   type: programming
   color: "#6a40fd"
@@ -122,6 +122,15 @@ AutoHotkey:
   - ahk
   primary_extension: .ahk
 
+AutoIt:
+  type: programming
+  color: "#36699B"
+  aliases:
+  - au3
+  - AutoIt3
+  - AutoItScript
+  primary_extension: .au3
+
 Awk:
   type: programming
   lexer: Awk
@@ -145,6 +154,15 @@ Batchfile:
 Befunge:
   primary_extension: .befunge
 
+BlitzBasic:
+  type: programming
+  aliases:
+  - blitzplus
+  - blitz3d
+  primary_extension: .bb
+  extensions:
+  - .decls
+
 BlitzMax:
   primary_extension: .bmx
 
@@ -357,7 +375,7 @@ D-ObjDump:
   type: data
   lexer: d-objdump
   primary_extension: .d-objdump
-  
+
 DM:
   type: programming
   color: "#075ff1"
@@ -383,6 +401,7 @@ Darcs Patch:
 
 Dart:
   type: programming
+  color: "#98BAD6"
   primary_extension: .dart
 
 DCPU-16 ASM:
@@ -393,7 +412,7 @@ DCPU-16 ASM:
   - .dasm
   aliases:
   - dasm16
-  
+
 Diff:
   primary_extension: .diff
 
@@ -408,7 +427,7 @@ Ecere Projects:
   lexer: JSON
   primary_extension: .epj
 
-Ecl:
+ECL:
   type: programming
   color: "#8a1267"
   primary_extension: .ecl
@@ -442,7 +461,7 @@ Emacs Lisp:
   - elisp
   - emacs
   primary_extension: .el
-  filenames:   
+  filenames:
   - .emacs
   extensions:
   - .emacs
@@ -972,6 +991,12 @@ Nemerle:
   color: "#0d3c6e"
   primary_extension: .n
 
+NetLogo:
+  type: programming
+  lexer: Common Lisp
+  color: "#ff2b2b"
+  primary_extension: .nlogo
+
 Nginx:
   type: markup
   lexer: Nginx configuration file
@@ -1311,6 +1336,7 @@ Ruby:
   - .thor
   - .watchr
   filenames:
+  - Appraisals
   - Berksfile
   - Gemfile
   - Guardfile
@@ -1503,6 +1529,12 @@ Unified Parallel C:
   color: "#755223"
   primary_extension: .upc
 
+UnrealScript:
+  type: programming
+  color: "#a54c4d"
+  lexer: Java
+  primary_extension: .uc
+
 VHDL:
   type: programming
   lexer: vhdl
@@ -1581,6 +1613,7 @@ XML:
   - .kml
   - .mxml
   - .plist
+  - .pluginspec
   - .ps1xml
   - .psc1
   - .pt
@@ -1611,6 +1644,7 @@ XML:
   filenames:
   - .classpath
   - .project
+  - phpunit.xml.dist
 
 XProc:
   type: programming

lib/linguist/vendor.yml

@@ -139,7 +139,7 @@
 
 # LICENSE, README, git config files
 - ^COPYING$
-- ^LICENSE$
+- LICENSE$
 - gitattributes$
 - gitignore$
 - gitmodules$