language_sniffer 1.0.1 → 1.0.2

data/lib/language_sniffer/blob_helper.rb

@@ -1,3 +1,4 @@
+# encoding: UTF-8
 require 'language_sniffer/language'
 require 'language_sniffer/pathname'
 require 'yaml'
@@ -31,7 +32,11 @@ module LanguageSniffer
     #
     # Returns an Array of lines
     def lines
-      @lines ||= (data ? data.split("\n", -1) : [])
+      @lines ||= begin
+        (data ? data.split("\n", -1) : [])
+      rescue ArgumentError # invalid byte sequence in UTF-8
+        []
+      end
     end
 
     # Public: Get number of lines of code
@@ -225,6 +230,24 @@ module LanguageSniffer
       end
     end
 
+    # Internal: Guess language of .cls files
+    #
+    # Returns a Language.
+    def guess_cls_language
+      if lines.grep(/^(%|\\)/).any?
+        Language['TeX']
+      elsif lines.grep(/^\s*(CLASS|METHOD|INTERFACE).*:\s*/i).any? || lines.grep(/^\s*(USING|DEFINE)/i).any?
+        Language['OpenEdge ABL']
+      elsif lines.grep(/\{$/).any? || lines.grep(/\}$/).any?
+        Language['Apex']
+      elsif lines.grep(/^(\'\*|Attribute|Option|Sub|Private|Protected|Public|Friend)/i).any?
+        Language['Visual Basic']
+      else
+        # The most common language should be the fallback
+        Language['TeX']
+      end
+    end
+
     # Internal: Guess language of header files (.h).
     #
     # Returns a Language.
@@ -297,6 +320,39 @@ module LanguageSniffer
       end
     end
 
+    # Internal: Guess language of .t files.
+    #
+    # Returns a Language.
+    def guess_t_language
+      score = 0
+      score += 1 if lines.grep(/^% /).any?
+      score += data.gsub(/ := /).count
+      score += data.gsub(/proc |procedure |fcn |function /).count
+      score += data.gsub(/var \w+: \w+/).count
+
+      # Tell-tale signs its gotta be Perl
+      if lines.grep(/^(my )?(sub |\$|@|%)\w+/).any?
+        score = 0
+      end
+
+      if score >= 3
+        Language['Turing']
+      else
+        Language['Perl']
+      end
+    end
+
+    # Internal: Guess language of .v files.
+    #
+    # Returns a Language
+    def guess_v_language
+      if lines.grep(/^(\/\*|\/\/|module|parameter|input|output|wire|reg|always|initial|begin|\`)/).any?
+        Language['Verilog']
+      else
+        Language['Coq']
+      end
+    end
+
     # Internal: Guess language of .gsp files.
     #
     # Returns a Language.

data/lib/language_sniffer/languages.yml

@@ -1,8 +1,15 @@
 # Defines all Languages known
 #
+# All languages have an associated lexer for syntax highlighting. It
+# defaults to name.downcase, which covers most cases. Make sure the
+# lexer exists in lexers.yml. This is a list of available in our
+# version of pygments.
+#
 # type              - Either data, programming, markup, or nil
+# lexer             - An explicit lexer String (defaults to name.downcase)
 # aliases           - An Array of additional aliases (implicitly
 #                     includes name.downcase)
+# ace_mode          - A String name of Ace Mode (if available)
 # extension         - An Array of associated extensions
 # primary_extension - A String for the main extension associated with
 #                     the langauge. (defaults to extensions.first)
@@ -10,6 +17,7 @@
 # searchable        - Boolean flag to enable searching (defaults to true)
 # search_term       - Deprecated: Some languages maybe indexed under a
 #                     different alias. Avoid defining new exceptions.
+# color             - CSS hex color to represent the language.
 #
 # Any additions or modifications (even trivial) should have corresponding
 # test change in `test/test_blob.rb`.
@@ -48,7 +56,15 @@ Ada:
   - .adb
   - .ads
 
+Apex:
+  type: programming
+  lexer: Text only
+  extensions:
+  - .cls
+
 AppleScript:
+  aliases:
+  - osascript
   primary_extension: .scpt
   extensions:
   - .applescript
@@ -60,6 +76,12 @@ Arc:
   extensions:
   - .arc
 
+Arduino:
+  type: programming
+  lexer: C++
+  extensions:
+  - .ino
+
 Assembly:
   type: programming
   lexer: NASM
@@ -69,6 +91,11 @@ Assembly:
   extensions:
   - .asm
 
+Augeas:
+  type: programming
+  extensions:
+  - .aug
+
 AutoHotkey:
   type: programming
   lexer: autohotkey
@@ -106,6 +133,11 @@ Brainfuck:
   - .b
   - .bf
 
+Bro:
+  type: programming
+  extensions:
+  - .bro
+
 C:
   type: programming
   overrides:
@@ -114,9 +146,11 @@ C:
   extensions:
   - .c
   - .h
+  - .w
 
 C#:
   type: programming
+  ace_mode: csharp
   search_term: csharp
   aliases:
   - csharp
@@ -125,6 +159,7 @@ C#:
 
 C++:
   type: programming
+  ace_mode: c_cpp
   search_term: cpp
   aliases:
   - cpp
@@ -148,6 +183,15 @@ C-ObjDump:
   extensions:
   - .c-objdump
 
+C2hs Haskell:
+  type: programming
+  lexer: Haskell
+  group: Haskell
+  aliases:
+  - c2hs
+  extensions:
+  - .chs
+
 CMake:
   extensions:
   - .cmake
@@ -156,6 +200,7 @@ CMake:
   - CMakeLists.txt
 
 CSS:
+  ace_mode: css
   extensions:
   - .css
 
@@ -166,6 +211,7 @@ ChucK:
 
 Clojure:
   type: programming
+  ace_mode: clojure
   primary_extension: .clj
   extensions:
   - .clj
@@ -173,6 +219,7 @@ Clojure:
 
 CoffeeScript:
   type: programming
+  ace_mode: coffee
   aliases:
   - coffee
   extensions:
@@ -183,6 +230,7 @@ CoffeeScript:
 ColdFusion:
   type: programming
   lexer: Coldfusion HTML
+  ace_mode: coldfusion
   search_term: cfm
   aliases:
   - cfm
@@ -198,8 +246,14 @@ Common Lisp:
   primary_extension: .lisp
   extensions:
   - .lisp
+  - .lsp
   - .ny
 
+Coq:
+  type: programming
+  extensions:
+  - .v
+
 Cpp-ObjDump:
   type: data
   lexer: cpp-objdump
@@ -242,13 +296,29 @@ Darcs Patch:
   - .darcspatch
   - .dpatch
 
+Dart:
+  type: programming
+  extensions:
+  - .dart
+
 Delphi:
   type: programming
   primary_extension: .pas
   extensions:
   - .dpr
+  - .lpr
   - .pas
 
+DCPU-16 ASM:
+  type: programming
+  lexer: dasm16
+  primary_extension: .dasm16
+  extensions:
+  - .dasm
+  - .dasm16
+  aliases:
+  - dasm16
+
 Diff:
   extensions:
   - .diff
@@ -259,17 +329,32 @@ Dylan:
   extensions:
   - .dylan
 
+Ecere Projects:
+  type: data
+  group: JavaScript
+  lexer: JSON
+  extensions:
+  - .epj
+
 Eiffel:
   type: programming
   lexer: Text only
   extensions:
   - .e
 
+Elixir:
+  type: programming
+  primary_extension: .ex
+  extensions:
+  - .ex
+  - .exs
+
 Emacs Lisp:
   type: programming
   lexer: Scheme
   aliases:
   - elisp
+  - emacs
   primary_extension: .el
   extensions:
   - .el
@@ -284,7 +369,7 @@ Erlang:
 
 F#:
   type: programming
-  lexer: OCaml
+  lexer: FSharp
   search_term: ocaml
   extensions:
   - .fs
@@ -297,9 +382,21 @@ FORTRAN:
   primary_extension: .f90
   extensions:
   - .F
+  - .F03
+  - .F08
+  - .F77
   - .F90
+  - .F95
+  - .FOR
+  - .FPP
   - .f
+  - .f03
+  - .f08
+  - .f77
   - .f90
+  - .f95
+  - .for
+  - .fpp
 
 Factor:
   type: programming
@@ -313,6 +410,11 @@ Fancy:
   - .fancypack
   - .fy
 
+Fantom:
+  type: programming
+  extensions:
+  - .fan
+
 GAS:
   type: programming
   group: Assembly
@@ -375,7 +477,7 @@ Groff:
 
 Groovy:
   type: programming
-  lexer: Java
+  ace_mode: groovy
   primary_extension: .groovy
   extensions:
   - .gradle
@@ -393,6 +495,7 @@ Groovy Server Pages:
 
 HTML:
   type: markup
+  ace_mode: html
   primary_extension: .html
   extensions:
   - .htm
@@ -425,6 +528,7 @@ HTML+PHP:
 HaXe:
   type: programming
   lexer: haXe
+  ace_mode: haxe
   extensions:
   - .hx
   - .hxml
@@ -473,13 +577,14 @@ Ioke:
 JSON:
   type: data
   group: JavaScript
-  lexer: JavaScript
+  ace_mode: json
   searchable: false
   extensions:
   - .json
 
 Java:
   type: programming
+  ace_mode: java
   extensions:
   - .java
   - .pde
@@ -495,13 +600,16 @@ Java Server Pages:
 
 JavaScript:
   type: programming
+  ace_mode: javascript
   aliases:
   - js
   - node
   primary_extension: .js
   extensions:
+  - .bones
   - .jake
   - .js
+  - .jsfl
   - .jsm
   - .jss
   - .jsx
@@ -511,6 +619,18 @@ JavaScript:
   filenames:
   - Jakefile
 
+Julia:
+  type: programming
+  extensions:
+  - .jl
+
+Kotlin:
+  type: programming
+  extensions:
+  - .kt
+  - .ktm
+  - .kts
+
 LLVM:
   extensions:
   - .ll
@@ -531,9 +651,14 @@ Literate Haskell:
   extensions:
   - .lhs
 
+Logtalk:
+  type: programming
+  extensions:
+  - .lgt
+
 Lua:
   type: programming
-  primary_extension: .lua
+  ace_mode: lua
   extensions:
   - .lua
   - .nse
@@ -541,16 +666,21 @@ Lua:
 Makefile:
   extensions:
   - .mak
+  - .mk
   filenames:
+  - makefile
   - Makefile
+  - GNUmakefile
 
 Mako:
   extensions:
+  - .mako
   - .mao
 
 Markdown:
   type: markup
   lexer: Text only
+  ace_mode: markdown
   primary_extension: .md
   extensions:
   - .markdown
@@ -600,8 +730,10 @@ Nemerle:
   - .n
 
 Nimrod:
+  type: programming
   extensions:
   - .nim
+  - .nimrod
 
 Nu:
   type: programming
@@ -623,6 +755,7 @@ NumPy:
 
 OCaml:
   type: programming
+  ace_mode: ocaml
   primary_extension: .ml
   extensions:
   - .ml
@@ -652,6 +785,11 @@ Objective-J:
   - .j
   - .sj
 
+Opa:
+  type: programming
+  extensions:
+  - .opa
+
 OpenCL:
   type: programming
   group: C
@@ -659,8 +797,20 @@ OpenCL:
   extensions:
   - .cl
 
+OpenEdge ABL:
+  type: programming
+  aliases:
+  - progress
+  - openedge
+  - abl
+  primary_extension: .p
+  extensions:
+  - .cls
+  - .p
+
 PHP:
   type: programming
+  ace_mode: php
   extensions:
   - .aw
   - .ctp
@@ -669,6 +819,8 @@ PHP:
   - .php4
   - .php5
   - .phpt
+  filenames:
+  - Phakefile
 
 Parrot:
   type: programming
@@ -695,19 +847,31 @@ Parrot Assembly:
 
 Perl:
   type: programming
+  ace_mode: perl
   overrides:
   - .pl
+  - .t
   primary_extension: .pl
   extensions:
   - .PL
   - .perl
   - .ph
   - .pl
+  - .plx
   - .pm
   - .pod
   - .psgi
   - .t
 
+PowerShell:
+  type: programming
+  ace_mode: powershell
+  aliases:
+  - posh
+  extensions:
+  - .ps1
+  - .psm1
+
 Prolog:
   type: programming
   extensions:
@@ -715,6 +879,13 @@ Prolog:
   - .pro
   - .prolog
 
+Puppet:
+  type: programming
+  extensions:
+  - .pp
+  filenames:
+  - Modulefile
+
 Pure Data:
   type: programming
   lexer: Text only
@@ -723,11 +894,13 @@ Pure Data:
 
 Python:
   type: programming
+  ace_mode: python
   primary_extension: .py
   extensions:
   - .py
   - .pyw
   - .wsgi
+  - .xpy
 
 Python traceback:
   type: data
@@ -785,6 +958,7 @@ Redcode:
 
 Ruby:
   type: programming
+  ace_mode: ruby
   aliases:
   - jruby
   - macruby
@@ -797,6 +971,7 @@ Ruby:
   - .gemspec
   - .god
   - .irbrc
+  - .podspec
   - .rake
   - .rb
   - .rbuild
@@ -807,32 +982,40 @@ Ruby:
   - .watchr
   filenames:
   - Capfile
-  - Rakefile
-  - Thorfile
   - Gemfile
   - Guardfile
+  - Podfile
+  - Rakefile
+  - Thorfile
   - Vagrantfile
 
 Rust:
   type: programming
   lexer: Text only
-  primary_extension: .rs
   extensions:
-  - .rc
   - .rs
 
 SCSS:
   type: markup
   group: CSS
+  ace_mode: scss
   extensions:
   - .scss
 
 SQL:
   type: data
+  ace_mode: sql
   searchable: false
   extensions:
   - .sql
 
+Sage:
+  type: programming
+  lexer: Python
+  group: Python
+  extensions:
+  - .sage
+
 Sass:
   type: markup
   group: CSS
@@ -841,6 +1024,7 @@ Sass:
 
 Scala:
   type: programming
+  ace_mode: scala
   primary_extension: .scala
   extensions:
   - .sbt
@@ -855,6 +1039,13 @@ Scheme:
   - .sps
   - .ss
 
+Scilab:
+  type: programming
+  primary_extension: .sci
+  extensions:
+  - .sce
+  - .tst
+
 Self:
   type: programming
   lexer: Text only
@@ -881,6 +1072,8 @@ Shell:
   - .zlogin
   - .zsh
   - .zshrc
+  - bashrc
+  - zshrc
 
 Smalltalk:
   type: programming
@@ -893,7 +1086,6 @@ Smarty:
 
 Standard ML:
   type: programming
-  lexer: Standard ML
   aliases:
   - sml
   primary_extension: .sml
@@ -922,26 +1114,46 @@ Tcsh:
 
 TeX:
   type: markup
+  ace_mode: latex
   primary_extension: .tex
+  overrides:
+  - .cls
   extensions:
   - .aux
   - .cls
+  - .dtx
+  - .ins
+  - .ltx
   - .sty
   - .tex
   - .toc
 
+Tea:
+  type: markup
+  extensions:
+  - .tea
+
 Text:
   type: data
   lexer: Text only
+  ace_mode: text
   extensions:
   - .txt
 
 Textile:
   type: markup
   lexer: Text only
+  ace_mode: textile
   extensions:
   - .textile
 
+Turing:
+  type: programming
+  lexer: Text only
+  extensions:
+  - .t
+  - .tu
+
 Twig:
   type: markup
   group: PHP
@@ -951,8 +1163,7 @@ Twig:
 
 VHDL:
   type: programming
-  lexer: Text only
-  primary_extension: .vhd
+  lexer: vhdl
   extensions:
   - .vhd
   - .vhdl
@@ -961,10 +1172,13 @@ Vala:
   type: programming
   extensions:
   - .vala
+  - .vapi
 
 Verilog:
   type: programming
-  lexer: verilog
+  lexer: Verilog
+  overrides:
+  - .v
   extensions:
   - .v
 
@@ -978,6 +1192,8 @@ VimL:
   filenames:
   - .gvimrc
   - .vimrc
+  - vimrc
+  - gvimrc
 
 Visual Basic:
   type: programming
@@ -992,6 +1208,7 @@ Visual Basic:
 
 XML:
   type: markup
+  ace_mode: xml
   primary_extension: .xml
   extensions:
   - .glade
@@ -1006,6 +1223,8 @@ XML:
   - .wxl
   - .wxs
   - .xaml
+  - .xlf
+  - .xliff
   - .xml
   - .xsd
   - .xsl
@@ -1036,6 +1255,14 @@ YAML:
   filenames:
   - .gemrc
 
+eC:
+  type: programming
+  search_term: ec
+  primary_extension: .ec
+  extensions:
+  - .ec
+  - .eh
+
 mupad:
   lexer: MuPAD
   extensions:

data/lib/language_sniffer/version.rb

@@ -2,7 +2,7 @@ module LanguageSniffer
   module Version
     MAJOR = 1
     MINOR = 0
-    PATCH = 1
+    PATCH = 2
     STRING = "#{MAJOR}.#{MINOR}.#{PATCH}"
   end
 end

metadata

@@ -1,88 +1,76 @@
---- !ruby/object:Gem::Specification 
+--- !ruby/object:Gem::Specification
 name: language_sniffer
-version: !ruby/object:Gem::Version 
-  hash: 21
+version: !ruby/object:Gem::Version
+  version: 1.0.2
   prerelease: 
-  segments: 
-  - 1
-  - 0
-  - 1
-  version: 1.0.1
 platform: ruby
-authors: 
+authors:
 - GitHub, Inc.
 - Michael Grosser
 autorequire: 
 bindir: bin
 cert_chain: []
-
-date: 2011-10-01 00:00:00 +02:00
-default_executable: 
-dependencies: 
-- !ruby/object:Gem::Dependency 
-  requirement: &id001 !ruby/object:Gem::Requirement 
+date: 2012-05-05 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: rake
+  requirement: !ruby/object:Gem::Requirement
     none: false
-    requirements: 
-    - - ">="
-      - !ruby/object:Gem::Version 
-        hash: 3
-        segments: 
-        - 0
-        version: "0"
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
   type: :development
-  name: rake
-  version_requirements: *id001
   prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
 description: 
 email: grosser.michael@gmail.com
-executables: 
+executables:
 - language_sniffer
 extensions: []
-
 extra_rdoc_files: []
-
-files: 
-- lib/language_sniffer.rb
-- lib/language_sniffer/pathname.rb
-- lib/language_sniffer/language.rb
-- lib/language_sniffer/version.rb
+files:
 - lib/language_sniffer/blob_helper.rb
-- lib/language_sniffer/languages.yml
 - lib/language_sniffer/file_blob.rb
+- lib/language_sniffer/language.rb
+- lib/language_sniffer/languages.yml
+- lib/language_sniffer/pathname.rb
+- lib/language_sniffer/version.rb
+- lib/language_sniffer.rb
 - bin/language_sniffer
-has_rdoc: true
 homepage: http://github.com/grosser/language_sniffer
 licenses: []
-
 post_install_message: 
 rdoc_options: []
-
-require_paths: 
+require_paths:
 - lib
-required_ruby_version: !ruby/object:Gem::Requirement 
+required_ruby_version: !ruby/object:Gem::Requirement
   none: false
-  requirements: 
-  - - ">="
-    - !ruby/object:Gem::Version 
-      hash: 3
-      segments: 
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+      segments:
       - 0
-      version: "0"
-required_rubygems_version: !ruby/object:Gem::Requirement 
+      hash: 322465844304122196
+required_rubygems_version: !ruby/object:Gem::Requirement
   none: false
-  requirements: 
-  - - ">="
-    - !ruby/object:Gem::Version 
-      hash: 3
-      segments: 
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+      segments:
       - 0
-      version: "0"
+      hash: 322465844304122196
 requirements: []
-
 rubyforge_project: 
-rubygems_version: 1.6.2
+rubygems_version: 1.8.24
 signing_key: 
 specification_version: 3
 summary: Language detection
 test_files: []
-