Merge pull request #2696 from github/cut-linguist-4.7.0

Bumping to version 4.7.0

.gitmodules

@@ -646,7 +646,7 @@
 	url = https://github.com/SRI-CSL/SMT.tmbundle.git
 [submodule "vendor/grammars/language-crystal"]
 	path = vendor/grammars/language-crystal
-	url = https://github.com/k2b6s9j/language-crystal
+	url = https://github.com/atom-crystal/language-crystal
 [submodule "vendor/grammars/language-xbase"]
 	path = vendor/grammars/language-xbase
 	url = https://github.com/hernad/atom-language-harbour
@@ -680,3 +680,6 @@
 [submodule "vendor/grammars/Stata.tmbundle"]
 	path = vendor/grammars/Stata.tmbundle
 	url = https://github.com/pschumm/Stata.tmbundle
+[submodule "vendor/grammars/FreeMarker.tmbundle"]
+	path = vendor/grammars/FreeMarker.tmbundle
+	url = https://github.com/freemarker/FreeMarker.tmbundle

CONTRIBUTING.md

@@ -12,7 +12,7 @@ We try only to add new extensions once they have some usage on GitHub. In most c
 
 To add support for a new extension:
 
-0. Add your extension to the language entry in [`languages.yml`][languages].
+0. Add your extension to the language entry in [`languages.yml`][languages], keeping the extensions in alphabetical order.
 0. Add at least one sample for your extension to the [samples directory][samples] in the correct subdirectory.
 0. Open a pull request, linking to a [GitHub search result](https://github.com/search?utf8=%E2%9C%93&q=extension%3Aboot+NOT+nothack&type=Code&ref=searchresults) showing in-the-wild usage.
 

README.md

@@ -17,7 +17,7 @@ The Language stats bar displays languages percentages for the files in the repos
 
 0. Click on the name of the language in the stats bar to see a list of the files that are identified as that language.
 0. If you see files that you didn't write, consider moving the files into one of the [paths for vendored  code](/lib/linguist/vendor.yml), or use the [manual overrides](#overrides) feature to ignore them.
-0. If the files are being misclassified, search for [open issues][issues] to see if anyone else has already reported the issue. Any information you an add, especially links to public repositories, is helpful.
+0. If the files are being misclassified, search for [open issues][issues] to see if anyone else has already reported the issue. Any information you can add, especially links to public repositories, is helpful.
 0. If there are no reported issues of this misclassification, [open an issue][new-issue] and include a link to the repository or a sample of the code that is being misclassified.
 
 ## Overrides

grammars.yml

@@ -42,6 +42,8 @@ vendor/grammars/Docker.tmbundle:
 - source.dockerfile
 vendor/grammars/Elm.tmLanguage:
 - source.elm
+vendor/grammars/FreeMarker.tmbundle:
+- text.html.ftl
 vendor/grammars/G-Code/:
 - source.LS
 - source.MCPOST

lib/linguist/blob.rb

@@ -0,0 +1,73 @@
+require 'linguist/blob_helper'
+
+module Linguist
+  # A Blob is a wrapper around the content of a file to make it quack
+  # like a Grit::Blob. It provides the basic interface: `name`,
+  # `data`, `path` and `size`.
+  class Blob
+    include BlobHelper
+
+    # Public: Initialize a new Blob.
+    #
+    # path    - A path String (does not necessarily exists on the file system).
+    # content - Content of the file.
+    #
+    # Returns a FileBlob.
+    def initialize(path, content)
+      @path = path
+      @content = content
+    end
+
+    # Public: Filename
+    #
+    # Examples
+    #
+    #   Blob.new("/path/to/linguist/lib/linguist.rb", "").path
+    #   # =>  "/path/to/linguist/lib/linguist.rb"
+    #
+    # Returns a String
+    attr_reader :path
+
+    # Public: File name
+    #
+    # Returns a String
+    def name
+      File.basename(@path)
+    end
+
+    # Public: File contents.
+    #
+    # Returns a String.
+    def data
+      @content
+    end
+
+    # Public: Get byte size
+    #
+    # Returns an Integer.
+    def size
+      @content.bytesize
+    end
+
+    # Public: Get file extension.
+    #
+    # Returns a String.
+    def extension
+      extensions.last || ""
+    end
+
+    # Public: Return an array of the file extensions
+    #
+    #     >> Linguist::FileBlob.new("app/views/things/index.html.erb").extensions
+    #     => [".html.erb", ".erb"]
+    #
+    # Returns an Array
+    def extensions
+      basename, *segments = name.downcase.split(".")
+
+      segments.map.with_index do |segment, index|
+        "." + segments[index..-1].join(".")
+      end
+    end
+  end
+end

lib/linguist/file_blob.rb

@@ -1,10 +1,11 @@
 require 'linguist/blob_helper'
+require 'linguist/blob'
 
 module Linguist
   # A FileBlob is a wrapper around a File object to make it quack
   # like a Grit::Blob. It provides the basic interface: `name`,
   # `data`, `path` and `size`.
-  class FileBlob
+  class FileBlob < Blob
     include BlobHelper
 
     # Public: Initialize a new FileBlob from a path
@@ -18,20 +19,6 @@ module Linguist
       @path = base_path ? path.sub("#{base_path}/", '') : path
     end
 
-    # Public: Filename
-    #
-    # Examples
-    #
-    #   FileBlob.new("/path/to/linguist/lib/linguist.rb").path
-    #   # =>  "/path/to/linguist/lib/linguist.rb"
-    #
-    #   FileBlob.new("/path/to/linguist/lib/linguist.rb",
-    #                "/path/to/linguist").path
-    #   # =>  "lib/linguist.rb"
-    #
-    # Returns a String
-    attr_reader :path
-
     # Public: Read file permissions
     #
     # Returns a String like '100644'
@@ -39,13 +26,6 @@ module Linguist
       File.stat(@fullpath).mode.to_s(8)
     end
 
-    # Public: File name
-    #
-    # Returns a String
-    def name
-      File.basename(@fullpath)
-    end
-
     # Public: Read file contents.
     #
     # Returns a String.
@@ -59,26 +39,5 @@ module Linguist
     def size
       File.size(@fullpath)
     end
-
-    # Public: Get file extension.
-    #
-    # Returns a String.
-    def extension
-      extensions.last || ""
-    end
-
-    # Public: Return an array of the file extensions
-    #
-    #     >> Linguist::FileBlob.new("app/views/things/index.html.erb").extensions
-    #     => [".html.erb", ".erb"]
-    #
-    # Returns an Array
-    def extensions
-      basename, *segments = name.downcase.split(".")
-
-      segments.map.with_index do |segment, index|
-        "." + segments[index..-1].join(".")
-      end
-    end
   end
 end

lib/linguist/generated.rb

@@ -71,7 +71,8 @@ module Linguist
       generated_jni_header? ||
       vcr_cassette? ||
       generated_module? ||
-      generated_unity3d_meta?
+      generated_unity3d_meta? ||
+      generated_racc?
     end
 
     # Internal: Is the blob an Xcode file?
@@ -359,5 +360,18 @@ module Linguist
       return false unless lines.count > 1
       return lines[0].include?("fileFormatVersion: ")
     end
+
+    # Internal: Is this a Racc-generated file?
+    #
+    # A Racc-generated file contains:
+    # # This file is automatically generated by Racc x.y.z
+    # on the third line.
+    #
+    # Return true or false
+    def generated_racc?
+      return false unless extname == '.rb'
+      return false unless lines.count > 2
+      return lines[2].start_with?("# This file is automatically generated by Racc")
+    end
   end
 end

lib/linguist/languages.yml

@@ -1029,6 +1029,16 @@ Forth:
   - .fs
   ace_mode: forth
 
+FreeMarker:
+  type: programming
+  color: "#0050b2"
+  aliases:
+  - ftl
+  extensions:
+  - .ftl
+  tm_scope: text.html.ftl
+  ace_mode: ftl
+
 Frege:
   type: programming
   color: "#00cafe"
@@ -1323,6 +1333,16 @@ HTML+Django:
   - htmldjango
   ace_mode: django
 
+HTML+EEX:
+  type: markup
+  tm_scope: text.html.elixir
+  group: HTML
+  aliases:
+  - eex
+  extensions:
+  - .eex
+  ace_mode: text
+
 HTML+ERB:
   type: markup
   tm_scope: text.html.erb
@@ -1332,7 +1352,7 @@ HTML+ERB:
   extensions:
   - .erb
   - .erb.deface
-  ace_mode: html_ruby
+  ace_mode: text
 
 HTML+PHP:
   type: markup
@@ -1681,6 +1701,7 @@ KiCad:
   type: programming
   extensions:
   - .sch
+  - .brd
   - .kicad_pcb
   tm_scope: none
   ace_mode: text
@@ -2063,6 +2084,14 @@ Mercury:
   tm_scope: source.mercury
   ace_mode: prolog
 
+Metal:
+  type: programming
+  color: "#8f14e9"
+  extensions:
+  - .metal
+  tm_scope: source.c++
+  ace_mode: c_cpp
+
 MiniD: # Legacy
   type: programming
   searchable: false
@@ -2441,6 +2470,7 @@ PHP:
   - .php3
   - .php4
   - .php5
+  - .phps
   - .phpt
   filenames:
   - Phakefile

lib/linguist/version.rb

@@ -1,3 +1,3 @@
 module Linguist
-  VERSION = "4.6.4"
+  VERSION = "4.7.0"
 end

samples/FreeMarker/example.ftl

@@ -0,0 +1,31 @@
+<#import "layout.ftl" as layout>
+
+<#assign results = [
+        {
+            "title": "Example Result",
+            "description": "Lorem ipsum dolor sit amet, pede id pellentesque, sollicitudin turpis sed in sed sed, libero dictum."
+        }
+    ] />
+
+<@layout.page title="FreeMarker Example">
+    <#if results?size == 0>
+        There were no results.
+    <#else>
+        <ul>
+            <#list results as result>
+                <li>
+                    <strong>${result.title}</strong>
+                    <p>${result.description}</p>
+                </li>
+            </#list>
+        </ul>
+    </#if>
+
+    <#-- This is a FreeMarker comment -->
+    <@currentTime />
+</@layout.page>
+
+
+<#macro currentTime>
+    ${.now?string.full}
+</#macro>

samples/FreeMarker/layout.ftl

@@ -0,0 +1,32 @@
+<#ftl strip_text=true />
+
+<#macro page title>
+    <!doctype html>
+    <html lang="${.lang}">
+        <head>
+            <title>${title}</title>
+            <@metaTags />
+        </head>
+        <body>
+            <#nested />
+            <@footer />
+        </body>
+    </html>
+</#macro>
+
+
+<#---
+  Default meta tags
+-->
+<#macro metaTags>
+    <#compress>
+        <meta charset="utf-8">
+        <meta http-equiv="X-UA-Compatible" content="IE=edge">
+        <meta name="viewport" content="width=device-width,initial-scale=1">
+        <meta name="format-detection" content="telephone=no">
+    </#compress>
+</#macro>
+
+<#macro footer>
+    <p>This page is using FreeMarker v${.version}</p>
+</#macro>

samples/HTML+EEX/index.html.eex

@@ -0,0 +1,26 @@
+<h1>Listing Books</h1>
+ 
+<table>
+  <tr>
+    <th>Title</th>
+    <th>Summary</th>
+    <th></th>
+    <th></th>
+    <th></th>
+  </tr>
+
+<%= for book <- @books do %>
+  <tr>
+    <%# comment %>
+    <td><%= book.title %></td>
+    <td><%= book.content %></td>
+    <td><%= link "Show", to: book_path(@conn, :show, book) %></td>
+    <td><%= link "Edit", to: book_path(@conn, :edit, book) %></td>
+    <td><%= link "Delete", to: book_path(@conn, :delete, book), method: :delete, data: [confirm: "Are you sure?"] %></td>
+  </tr>
+<% end %>
+</table>
+ 
+<br />
+ 
+<%= link "New book", to: book_path(@conn, :new) %>

samples/Metal/ITMVisualisationEngine.metal

@@ -0,0 +1,99 @@
+// Copyright 2014 Isis Innovation Limited and the authors of InfiniTAM
+
+#include <metal_stdlib>
+
+#include "../../DeviceAgnostic/ITMSceneReconstructionEngine.h"
+#include "../../DeviceAgnostic/ITMVisualisationEngine.h"
+#include "ITMVisualisationEngine_Metal.h"
+
+using namespace metal;
+
+kernel void genericRaycastVH_device(DEVICEPTR(Vector4f) *pointsRay                                  [[ buffer(0) ]],
+                                    const CONSTPTR(ITMVoxel) *voxelData                             [[ buffer(1) ]],
+                                    const CONSTPTR(typename ITMVoxelIndex::IndexData) *voxelIndex   [[ buffer(2) ]],
+                                    const CONSTPTR(Vector2f) *minmaxdata                            [[ buffer(3) ]],
+                                    const CONSTPTR(CreateICPMaps_Params) *params                    [[ buffer(4) ]],
+                                    uint2 threadIdx                                                 [[ thread_position_in_threadgroup ]],
+                                    uint2 blockIdx                                                  [[ threadgroup_position_in_grid ]],
+                                    uint2 blockDim                                                  [[ threads_per_threadgroup ]])
+{
+    int x = threadIdx.x + blockIdx.x * blockDim.x, y = threadIdx.y + blockIdx.y * blockDim.y;
+    
+    if (x >= params->imgSize.x || y >= params->imgSize.y) return;
+    
+    int locId = x + y * params->imgSize.x;
+    int locId2 = (int)floor((float)x / minmaximg_subsample) + (int)floor((float)y / minmaximg_subsample) * params->imgSize.x;
+    
+    castRay<ITMVoxel, ITMVoxelIndex>(pointsRay[locId], x, y, voxelData, voxelIndex, params->invM, params->invProjParams,
+                                     params->voxelSizes.y, params->lightSource.w, minmaxdata[locId2]);
+}
+
+kernel void genericRaycastVGMissingPoints_device(DEVICEPTR(Vector4f) *forwardProjection                         [[ buffer(0) ]],
+                                                 const CONSTPTR(int) *fwdProjMissingPoints                      [[ buffer(1) ]],
+                                                 const CONSTPTR(ITMVoxel) *voxelData                            [[ buffer(2) ]],
+                                                 const CONSTPTR(typename ITMVoxelIndex::IndexData) *voxelIndex  [[ buffer(3) ]],
+                                                 const CONSTPTR(Vector2f) *minmaxdata                           [[ buffer(4) ]],
+                                                 const CONSTPTR(CreateICPMaps_Params) *params                   [[ buffer(5) ]],
+                                                 uint2 threadIdx                                                [[ thread_position_in_threadgroup ]],
+                                                 uint2 blockIdx                                                 [[ threadgroup_position_in_grid ]],
+                                                 uint2 blockDim                                                 [[ threads_per_threadgroup ]])
+{
+    int pointId = threadIdx.x + blockIdx.x * blockDim.x;
+    
+    if (pointId >= params->imgSize.z) return;
+    
+    int locId = fwdProjMissingPoints[pointId];
+    int y = locId / params->imgSize.x, x = locId - y * params->imgSize.x;
+    int locId2 = (int)floor((float)x / minmaximg_subsample) + (int)floor((float)y / minmaximg_subsample) * params->imgSize.x;
+    
+    castRay<ITMVoxel, ITMVoxelIndex>(forwardProjection[locId], x, y, voxelData, voxelIndex, params->invM, params->invProjParams,
+                                     params->voxelSizes.y, params->lightSource.w, minmaxdata[locId2]);
+}
+
+kernel void renderICP_device(const CONSTPTR(Vector4f) *pointsRay            [[ buffer(0) ]],
+                             DEVICEPTR(Vector4f) *pointsMap                 [[ buffer(1) ]],
+                             DEVICEPTR(Vector4f) *normalsMap                [[ buffer(2) ]],
+                             DEVICEPTR(Vector4u) *outRendering              [[ buffer(3) ]],
+                             const CONSTPTR(CreateICPMaps_Params) *params   [[ buffer(4) ]],
+                             uint2 threadIdx                                [[ thread_position_in_threadgroup ]],
+                             uint2 blockIdx                                 [[ threadgroup_position_in_grid ]],
+                             uint2 blockDim                                 [[ threads_per_threadgroup ]])
+{
+    int x = threadIdx.x + blockIdx.x * blockDim.x, y = threadIdx.y + blockIdx.y * blockDim.y;
+    
+    if (x >= params->imgSize.x || y >= params->imgSize.y) return;
+    
+    processPixelICP<false>(outRendering, pointsMap, normalsMap, pointsRay, params->imgSize.xy, x, y, params->voxelSizes.x, TO_VECTOR3(params->lightSource));
+}
+
+kernel void renderForward_device(DEVICEPTR(Vector4u) *outRendering              [[ buffer(0) ]],
+                                 const CONSTPTR(Vector4f) *pointsRay            [[ buffer(1) ]],
+                                 const CONSTPTR(CreateICPMaps_Params) *params   [[ buffer(2) ]],
+                                 uint2 threadIdx                                [[ thread_position_in_threadgroup ]],
+                                 uint2 blockIdx                                 [[ threadgroup_position_in_grid ]],
+                                 uint2 blockDim                                 [[ threads_per_threadgroup ]])
+{
+    int x = threadIdx.x + blockIdx.x * blockDim.x, y = threadIdx.y + blockIdx.y * blockDim.y;
+    
+    if (x >= params->imgSize.x || y >= params->imgSize.y) return;
+    
+    processPixelForwardRender<false>(outRendering, pointsRay, params->imgSize.xy, x, y, params->voxelSizes.x, TO_VECTOR3(params->lightSource));
+}
+
+kernel void forwardProject_device(DEVICEPTR(Vector4f) *forwardProjection         [[ buffer(0) ]],
+                                  const CONSTPTR(Vector4f) *pointsRay            [[ buffer(1) ]],
+                                  const CONSTPTR(CreateICPMaps_Params) *params   [[ buffer(2) ]],
+                                  uint2 threadIdx                                [[ thread_position_in_threadgroup ]],
+                                  uint2 blockIdx                                 [[ threadgroup_position_in_grid ]],
+                                  uint2 blockDim                                 [[ threads_per_threadgroup ]])
+{
+    int x = (threadIdx.x + blockIdx.x * blockDim.x), y = (threadIdx.y + blockIdx.y * blockDim.y);
+    
+    if (x >= params->imgSize.x || y >= params->imgSize.y) return;
+    
+    int locId = x + y * params->imgSize.x;
+    Vector4f pixel = pointsRay[locId];
+    
+    int locId_new = forwardProjectPixel(pixel * params->voxelSizes.x, params->M, params->projParams, params->imgSize.xy);
+    if (locId_new >= 0) forwardProjection[locId_new] = pixel;
+}

samples/PHP/mail.phps

@@ -0,0 +1,31 @@
+<?php
+/**
+ * This example shows sending a message using PHP's mail() function.
+ */
+
+require '../PHPMailerAutoload.php';
+
+//Create a new PHPMailer instance
+$mail = new PHPMailer;
+//Set who the message is to be sent from
+$mail->setFrom('from@example.com', 'First Last');
+//Set an alternative reply-to address
+$mail->addReplyTo('replyto@example.com', 'First Last');
+//Set who the message is to be sent to
+$mail->addAddress('whoto@example.com', 'John Doe');
+//Set the subject line
+$mail->Subject = 'PHPMailer mail() test';
+//Read an HTML message body from an external file, convert referenced images to embedded,
+//convert HTML into a basic plain-text alternative body
+$mail->msgHTML(file_get_contents('contents.html'), dirname(__FILE__));
+//Replace the plain text body with one created manually
+$mail->AltBody = 'This is a plain-text message body';
+//Attach an image file
+$mail->addAttachment('images/phpmailer_mini.png');
+
+//send the message, check for errors
+if (!$mail->send()) {
+    echo "Mailer Error: " . $mail->ErrorInfo;
+} else {
+    echo "Message sent!";
+}

samples/Ruby/racc.rb

@@ -0,0 +1,267 @@
+#
+# DO NOT MODIFY!!!!
+# This file is automatically generated by Racc 1.4.7
+# from Racc grammer file "".
+#
+
+require 'racc/parser.rb'
+module RJSON
+  class Parser < Racc::Parser
+
+
+  require 'rjson/handler'
+
+  attr_reader :handler
+
+  def initialize tokenizer, handler = Handler.new
+    @tokenizer = tokenizer
+    @handler   = handler
+    super()
+  end
+
+  def next_token
+    @tokenizer.next_token
+  end
+
+  def parse
+    do_parse
+    handler
+  end
+##### State transition tables begin ###
+
+racc_action_table = [
+     9,    33,     9,    11,    13,    16,    19,    22,     9,     7,
+    23,     1,     9,    11,    13,    16,    19,    29,    30,     7,
+    21,     1,     9,    11,    13,    16,    19,    31,   nil,     7,
+    21,     1,    23,     7,   nil,     1 ]
+
+racc_action_check = [
+     6,    27,    33,    33,    33,    33,    33,     3,    31,    33,
+     6,    33,    29,    29,    29,    29,    29,    12,    22,    29,
+    12,    29,     2,     2,     2,     2,     2,    25,   nil,     2,
+     2,     2,    25,     0,   nil,     0 ]
+
+racc_action_pointer = [
+    24,   nil,    20,     7,   nil,   nil,    -2,   nil,   nil,   nil,
+   nil,   nil,    10,   nil,   nil,   nil,   nil,   nil,   nil,   nil,
+   nil,   nil,    18,   nil,   nil,    20,   nil,    -7,   nil,    10,
+   nil,     6,   nil,     0,   nil,   nil,   nil ]
+
+racc_action_default = [
+   -27,   -12,   -21,   -27,    -1,    -2,   -27,   -10,   -15,   -26,
+    -8,   -22,   -27,   -23,   -17,   -16,   -24,   -20,   -18,   -25,
+   -19,   -11,   -27,   -13,    -3,   -27,    -6,   -27,    -9,   -21,
+    37,   -27,    -4,   -21,   -14,    -5,    -7 ]
+
+racc_goto_table = [
+     8,    26,    24,    27,    10,     3,    25,     5,     4,    12,
+   nil,   nil,   nil,   nil,    28,   nil,   nil,   nil,   nil,   nil,
+   nil,    32,   nil,   nil,   nil,   nil,    35,    34,    27,   nil,
+   nil,    36 ]
+
+racc_goto_check = [
+     9,     7,     5,     8,    11,     1,     6,     3,     2,    12,
+   nil,   nil,   nil,   nil,    11,   nil,   nil,   nil,   nil,   nil,
+   nil,     5,   nil,   nil,   nil,   nil,     7,     9,     8,   nil,
+   nil,     9 ]
+
+racc_goto_pointer = [
+   nil,     5,     8,     7,   nil,    -4,     0,    -5,    -3,    -2,
+   nil,     2,     7,   nil,   nil ]
+
+racc_goto_default = [
+   nil,   nil,    14,    18,     6,   nil,   nil,   nil,    20,   nil,
+     2,   nil,   nil,    15,    17 ]
+
+racc_reduce_table = [
+  0, 0, :racc_error,
+  1, 14, :_reduce_none,
+  1, 14, :_reduce_none,
+  2, 15, :_reduce_none,
+  3, 15, :_reduce_none,
+  3, 19, :_reduce_none,
+  1, 19, :_reduce_none,
+  3, 20, :_reduce_none,
+  2, 16, :_reduce_none,
+  3, 16, :_reduce_none,
+  1, 23, :_reduce_10,
+  1, 24, :_reduce_11,
+  1, 17, :_reduce_12,
+  1, 18, :_reduce_13,
+  3, 25, :_reduce_none,
+  1, 25, :_reduce_none,
+  1, 22, :_reduce_none,
+  1, 22, :_reduce_none,
+  1, 22, :_reduce_none,
+  1, 26, :_reduce_none,
+  1, 26, :_reduce_20,
+  0, 27, :_reduce_none,
+  1, 27, :_reduce_22,
+  1, 27, :_reduce_23,
+  1, 27, :_reduce_24,
+  1, 27, :_reduce_25,
+  1, 21, :_reduce_26 ]
+
+racc_reduce_n = 27
+
+racc_shift_n = 37
+
+racc_token_table = {
+  false => 0,
+  :error => 1,
+  :STRING => 2,
+  :NUMBER => 3,
+  :TRUE => 4,
+  :FALSE => 5,
+  :NULL => 6,
+  "," => 7,
+  ":" => 8,
+  "[" => 9,
+  "]" => 10,
+  "{" => 11,
+  "}" => 12 }
+
+racc_nt_base = 13
+
+racc_use_result_var = true
+
+Racc_arg = [
+  racc_action_table,
+  racc_action_check,
+  racc_action_default,
+  racc_action_pointer,
+  racc_goto_table,
+  racc_goto_check,
+  racc_goto_default,
+  racc_goto_pointer,
+  racc_nt_base,
+  racc_reduce_table,
+  racc_token_table,
+  racc_shift_n,
+  racc_reduce_n,
+  racc_use_result_var ]
+
+Racc_token_to_s_table = [
+  "$end",
+  "error",
+  "STRING",
+  "NUMBER",
+  "TRUE",
+  "FALSE",
+  "NULL",
+  "\",\"",
+  "\":\"",
+  "\"[\"",
+  "\"]\"",
+  "\"{\"",
+  "\"}\"",
+  "$start",
+  "document",
+  "object",
+  "array",
+  "start_object",
+  "end_object",
+  "pairs",
+  "pair",
+  "string",
+  "value",
+  "start_array",
+  "end_array",
+  "values",
+  "scalar",
+  "literal" ]
+
+Racc_debug_parser = false
+
+##### State transition tables end #####
+
+# reduce 0 omitted
+
+# reduce 1 omitted
+
+# reduce 2 omitted
+
+# reduce 3 omitted
+
+# reduce 4 omitted
+
+# reduce 5 omitted
+
+# reduce 6 omitted
+
+# reduce 7 omitted
+
+# reduce 8 omitted
+
+# reduce 9 omitted
+
+def _reduce_10(val, _values, result)
+ @handler.start_array  
+    result
+end
+
+def _reduce_11(val, _values, result)
+ @handler.end_array    
+    result
+end
+
+def _reduce_12(val, _values, result)
+ @handler.start_object 
+    result
+end
+
+def _reduce_13(val, _values, result)
+ @handler.end_object   
+    result
+end
+
+# reduce 14 omitted
+
+# reduce 15 omitted
+
+# reduce 16 omitted
+
+# reduce 17 omitted
+
+# reduce 18 omitted
+
+# reduce 19 omitted
+
+def _reduce_20(val, _values, result)
+ @handler.scalar val[0] 
+    result
+end
+
+# reduce 21 omitted
+
+def _reduce_22(val, _values, result)
+ n = val[0]; result = n.count('.') > 0 ? n.to_f : n.to_i 
+    result
+end
+
+def _reduce_23(val, _values, result)
+ result = true 
+    result
+end
+
+def _reduce_24(val, _values, result)
+ result = false 
+    result
+end
+
+def _reduce_25(val, _values, result)
+ result = nil 
+    result
+end
+
+def _reduce_26(val, _values, result)
+ @handler.scalar val[0].gsub(/^"|"$/, '') 
+    result
+end
+
+def _reduce_none(val, _values, result)
+  val[0]
+end
+
+  end   # class Parser
+  end   # module RJSON

test/helper.rb

@@ -3,6 +3,7 @@ require "minitest/autorun"
 require "mocha/setup"
 require "linguist"
 require 'color-proximity'
+require "linguist/blob"
 require 'licensee'
 
 def fixtures_path
@@ -10,8 +11,14 @@ def fixtures_path
 end
 
 def fixture_blob(name)
-  name = File.join(fixtures_path, name) unless name =~ /^\//
+  filepath = (name =~ /^\//)? name : File.join(fixtures_path, name)
-  Linguist::FileBlob.new(name, fixtures_path)
+  Linguist::FileBlob.new(filepath, fixtures_path)
+end
+
+def fixture_blob_memory(name)
+  filepath = (name =~ /^\//)? name : File.join(fixtures_path, name)
+  content = File.read(filepath)
+  Linguist::Blob.new(name, content)
 end
 
 def samples_path
@@ -19,6 +26,12 @@ def samples_path
 end
 
 def sample_blob(name)
-  name = File.join(samples_path, name) unless name =~ /^\//
+  filepath = (name =~ /^\//)? name : File.join(samples_path, name)
-  Linguist::FileBlob.new(name, samples_path)
+  Linguist::FileBlob.new(filepath, samples_path)
+end
+
+def sample_blob_memory(name)
+  filepath = (name =~ /^\//)? name : File.join(samples_path, name)
+  content = File.read(filepath)
+  Linguist::Blob.new(name, content)
 end

test/test_blob.rb

@@ -15,50 +15,47 @@ class TestBlob < Minitest::Test
   end
 
   def script_blob(name)
-    blob = sample_blob(name)
+    blob = sample_blob_memory(name)
     blob.instance_variable_set(:@name, 'script')
     blob
   end
 
   def test_name
-    assert_equal "foo.rb", sample_blob("foo.rb").name
+    assert_equal "foo.rb", sample_blob_memory("Ruby/foo.rb").name
   end
 
   def test_mime_type
-    assert_equal "application/postscript", fixture_blob("Binary/octocat.ai").mime_type
+    assert_equal "application/postscript", fixture_blob_memory("Binary/octocat.ai").mime_type
-    assert_equal "application/x-ruby", sample_blob("Ruby/grit.rb").mime_type
+    assert_equal "application/x-ruby", sample_blob_memory("Ruby/grit.rb").mime_type
-    assert_equal "application/x-sh", sample_blob("Shell/script.sh").mime_type
+    assert_equal "application/x-sh", sample_blob_memory("Shell/script.sh").mime_type
-    assert_equal "application/xml", sample_blob("XML/bar.xml").mime_type
+    assert_equal "text/plain", fixture_blob_memory("Data/README").mime_type
-    assert_equal "audio/ogg", fixture_blob("Binary/foo.ogg").mime_type
-    assert_equal "text/plain", fixture_blob("Data/README").mime_type
   end
 
   def test_content_type
-    assert_equal "application/pdf", fixture_blob("Binary/foo.pdf").content_type
+    assert_equal "application/pdf", fixture_blob_memory("Binary/foo.pdf").content_type
-    assert_equal "audio/ogg", fixture_blob("Binary/foo.ogg").content_type
+    assert_equal "image/png", fixture_blob_memory("Binary/foo.png").content_type
-    assert_equal "image/png", fixture_blob("Binary/foo.png").content_type
+    assert_equal "text/plain; charset=iso-8859-2", fixture_blob_memory("Data/README").content_type
-    assert_equal "text/plain; charset=iso-8859-2", fixture_blob("Data/README").content_type
   end
 
   def test_disposition
-    assert_equal "attachment; filename=foo+bar.jar", fixture_blob("Binary/foo bar.jar").disposition
+    assert_equal "attachment; filename=foo+bar.jar", fixture_blob_memory("Binary/foo bar.jar").disposition
-    assert_equal "attachment; filename=foo.bin", fixture_blob("Binary/foo.bin").disposition
+    assert_equal "attachment; filename=foo.bin", fixture_blob_memory("Binary/foo.bin").disposition
-    assert_equal "attachment; filename=linguist.gem", fixture_blob("Binary/linguist.gem").disposition
+    assert_equal "attachment; filename=linguist.gem", fixture_blob_memory("Binary/linguist.gem").disposition
-    assert_equal "attachment; filename=octocat.ai", fixture_blob("Binary/octocat.ai").disposition
+    assert_equal "attachment; filename=octocat.ai", fixture_blob_memory("Binary/octocat.ai").disposition
-    assert_equal "inline", fixture_blob("Data/README").disposition
+    assert_equal "inline", fixture_blob_memory("Data/README").disposition
-    assert_equal "inline", sample_blob("Text/foo.txt").disposition
+    assert_equal "inline", sample_blob_memory("Text/foo.txt").disposition
-    assert_equal "inline", sample_blob("Ruby/grit.rb").disposition
+    assert_equal "inline", sample_blob_memory("Ruby/grit.rb").disposition
-    assert_equal "inline", fixture_blob("Binary/octocat.png").disposition
+    assert_equal "inline", fixture_blob_memory("Binary/octocat.png").disposition
   end
 
   def test_data
-    assert_equal "module Foo\nend\n", sample_blob("Ruby/foo.rb").data
+    assert_equal "module Foo\nend\n", sample_blob_memory("Ruby/foo.rb").data
   end
 
   def test_lines
-    assert_equal ["module Foo", "end", ""], sample_blob("Ruby/foo.rb").lines
+    assert_equal ["module Foo", "end", ""], sample_blob_memory("Ruby/foo.rb").lines
-    assert_equal ["line 1", "line 2", ""], sample_blob("Text/mac.txt").lines
+    assert_equal ["line 1", "line 2", ""], sample_blob_memory("Text/mac.txt").lines
-    assert_equal 475, sample_blob("Emacs Lisp/ess-julia.el").lines.length
+    assert_equal 475, sample_blob_memory("Emacs Lisp/ess-julia.el").lines.length
   end
 
   def test_lines_maintains_original_encoding
@@ -66,534 +63,173 @@ class TestBlob < Minitest::Test
     # earlier versions of the gem made implicit guarantees that the encoding of
     # each `line` is in the same encoding as the file was originally read (in
     # practice, UTF-8 or ASCII-8BIT)
-    assert_equal Encoding.default_external, fixture_blob("Data/utf16le").lines.first.encoding
+    assert_equal Encoding.default_external, fixture_blob_memory("Data/utf16le").lines.first.encoding
   end
 
   def test_size
-    assert_equal 15, sample_blob("Ruby/foo.rb").size
+    assert_equal 15, sample_blob_memory("Ruby/foo.rb").size
   end
 
   def test_loc
-    assert_equal 3, sample_blob("Ruby/foo.rb").loc
+    assert_equal 3, sample_blob_memory("Ruby/foo.rb").loc
   end
 
   def test_sloc
-    assert_equal 2, sample_blob("Ruby/foo.rb").sloc
+    assert_equal 2, sample_blob_memory("Ruby/foo.rb").sloc
-    assert_equal 3, fixture_blob("Data/utf16le-windows").sloc
+    assert_equal 3, fixture_blob_memory("Data/utf16le-windows").sloc
-    assert_equal 1, fixture_blob("Data/iso8859-8-i").sloc
+    assert_equal 1, fixture_blob_memory("Data/iso8859-8-i").sloc
   end
 
   def test_encoding
-    assert_equal "ISO-8859-2", fixture_blob("Data/README").encoding
+    assert_equal "ISO-8859-2", fixture_blob_memory("Data/README").encoding
-    assert_equal "ISO-8859-2", fixture_blob("Data/README").ruby_encoding
+    assert_equal "ISO-8859-2", fixture_blob_memory("Data/README").ruby_encoding
-    assert_equal "UTF-8", sample_blob("Text/foo.txt").encoding
+    assert_equal "UTF-8", sample_blob_memory("Text/foo.txt").encoding
-    assert_equal "UTF-8", sample_blob("Text/foo.txt").ruby_encoding
+    assert_equal "UTF-8", sample_blob_memory("Text/foo.txt").ruby_encoding
-    assert_equal "UTF-16LE", fixture_blob("Data/utf16le").encoding
+    assert_equal "UTF-16LE", fixture_blob_memory("Data/utf16le").encoding
-    assert_equal "UTF-16LE", fixture_blob("Data/utf16le").ruby_encoding
+    assert_equal "UTF-16LE", fixture_blob_memory("Data/utf16le").ruby_encoding
-    assert_equal "UTF-16LE", fixture_blob("Data/utf16le-windows").encoding
+    assert_equal "UTF-16LE", fixture_blob_memory("Data/utf16le-windows").encoding
-    assert_equal "UTF-16LE", fixture_blob("Data/utf16le-windows").ruby_encoding
+    assert_equal "UTF-16LE", fixture_blob_memory("Data/utf16le-windows").ruby_encoding
-    assert_equal "ISO-2022-KR", sample_blob("Text/ISO-2022-KR.txt").encoding
+    assert_equal "ISO-2022-KR", sample_blob_memory("Text/ISO-2022-KR.txt").encoding
-    assert_equal "binary", sample_blob("Text/ISO-2022-KR.txt").ruby_encoding
+    assert_equal "binary", sample_blob_memory("Text/ISO-2022-KR.txt").ruby_encoding
-    assert_nil fixture_blob("Binary/dog.o").encoding
+    assert_nil fixture_blob_memory("Binary/dog.o").encoding
   end
 
   def test_binary
-    # Large blobs aren't loaded
+    assert fixture_blob_memory("Binary/git.deb").binary?
-    large_blob = sample_blob("git.exe")
+    assert fixture_blob_memory("Binary/hello.pbc").binary?
-    large_blob.instance_eval do
+    assert fixture_blob_memory("Binary/linguist.gem").binary?
-      def data; end
+    assert fixture_blob_memory("Binary/octocat.ai").binary?
-    end
+    assert fixture_blob_memory("Binary/octocat.png").binary?
-    assert large_blob.binary?
+    assert fixture_blob_memory("Binary/zip").binary?
-
+    assert !fixture_blob_memory("Data/README").binary?
-    assert fixture_blob("Binary/git.deb").binary?
+    assert !sample_blob_memory("Ruby/foo.rb").binary?
-    assert fixture_blob("Binary/git.exe").binary?
+    assert !sample_blob_memory("Perl/script.pl").binary?
-    assert fixture_blob("Binary/hello.pbc").binary?
-    assert fixture_blob("Binary/linguist.gem").binary?
-    assert fixture_blob("Binary/octocat.ai").binary?
-    assert fixture_blob("Binary/octocat.png").binary?
-    assert fixture_blob("Binary/zip").binary?
-    assert !fixture_blob("Data/README").binary?
-    assert !sample_blob("Ruby/foo.rb").binary?
-    assert !sample_blob("Perl/script.pl").binary?
   end
 
   def test_all_binary
     Samples.each do |sample|
-      blob = sample_blob(sample[:path])
+      blob = sample_blob_memory(sample[:path])
       assert ! (blob.likely_binary? || blob.binary?), "#{sample[:path]} is a binary file"
     end
   end
 
   def test_text
-    assert fixture_blob("Data/README").text?
+    assert fixture_blob_memory("Data/README").text?
-    assert fixture_blob("Data/md").text?
+    assert fixture_blob_memory("Data/md").text?
-    assert sample_blob("Shell/script.sh").text?
+    assert sample_blob_memory("Shell/script.sh").text?
-    assert fixture_blob("Data/txt").text?
+    assert fixture_blob_memory("Data/txt").text?
   end
 
   def test_image
-    assert fixture_blob("Binary/octocat.gif").image?
+    assert fixture_blob_memory("Binary/octocat.png").image?
-    assert fixture_blob("Binary/octocat.jpeg").image?
+    assert !fixture_blob_memory("Binary/octocat.ai").image?
-    assert fixture_blob("Binary/octocat.jpg").image?
+    assert !fixture_blob_memory("Binary/octocat.psd").image?
-    assert fixture_blob("Binary/octocat.png").image?
-    assert !fixture_blob("Binary/octocat.ai").image?
-    assert !fixture_blob("Binary/octocat.psd").image?
   end
 
   def test_solid
-    assert fixture_blob("Binary/cube.stl").solid?
+    assert fixture_blob_memory("Binary/cube.stl").solid?
-    assert fixture_blob("Data/cube.stl").solid?
+    assert fixture_blob_memory("Data/cube.stl").solid?
   end
 
   def test_csv
-    assert fixture_blob("Data/cars.csv").csv?
+    assert fixture_blob_memory("Data/cars.csv").csv?
   end
 
   def test_pdf
-    assert fixture_blob("Binary/foo.pdf").pdf?
+    assert fixture_blob_memory("Binary/foo.pdf").pdf?
   end
 
   def test_viewable
-    assert fixture_blob("Data/README").viewable?
+    assert fixture_blob_memory("Data/README").viewable?
-    assert sample_blob("Ruby/foo.rb").viewable?
+    assert sample_blob_memory("Ruby/foo.rb").viewable?
-    assert sample_blob("Perl/script.pl").viewable?
+    assert sample_blob_memory("Perl/script.pl").viewable?
-    assert !fixture_blob("Binary/linguist.gem").viewable?
+    assert !fixture_blob_memory("Binary/linguist.gem").viewable?
-    assert !fixture_blob("Binary/octocat.ai").viewable?
+    assert !fixture_blob_memory("Binary/octocat.ai").viewable?
-    assert !fixture_blob("Binary/octocat.png").viewable?
+    assert !fixture_blob_memory("Binary/octocat.png").viewable?
   end
 
   def test_generated
-    assert !fixture_blob("Data/README").generated?
+    assert !fixture_blob_memory("Data/README").generated?
-
-    # Xcode project files
-    assert !sample_blob("XML/MainMenu.xib").generated?
-    assert fixture_blob("Binary/MainMenu.nib").generated?
-    assert !sample_blob("XML/project.pbxproj").generated?
-
-    # Gemfile.lock is NOT generated
-    assert !sample_blob("Gemfile.lock").generated?
 
     # Generated .NET Docfiles
-    assert sample_blob("XML/net_docfile.xml").generated?
+    assert sample_blob_memory("XML/net_docfile.xml").generated?
 
     # Long line
-    assert !sample_blob("JavaScript/uglify.js").generated?
+    assert !sample_blob_memory("JavaScript/uglify.js").generated?
 
     # Inlined JS, but mostly code
-    assert !sample_blob("JavaScript/json2_backbone.js").generated?
+    assert !sample_blob_memory("JavaScript/json2_backbone.js").generated?
 
     # Minified JS
-    assert !sample_blob("JavaScript/jquery-1.6.1.js").generated?
+    assert !sample_blob_memory("JavaScript/jquery-1.6.1.js").generated?
-    assert sample_blob("JavaScript/jquery-1.6.1.min.js").generated?
+    assert sample_blob_memory("JavaScript/jquery-1.6.1.min.js").generated?
-    assert sample_blob("JavaScript/jquery-1.4.2.min.js").generated?
+    assert sample_blob_memory("JavaScript/jquery-1.4.2.min.js").generated?
-
-    # CoffeeScript-generated JS
-    # TODO
-
-    # TypeScript-generated JS
-    # TODO
 
     # Composer generated composer.lock file
-    assert sample_blob("JSON/composer.lock").generated?
+    assert sample_blob_memory("JSON/composer.lock").generated?
 
     # PEG.js-generated parsers
-    assert sample_blob("JavaScript/parser.js").generated?
+    assert sample_blob_memory("JavaScript/parser.js").generated?
 
     # Generated PostScript
-    assert !sample_blob("PostScript/sierpinski.ps").generated?
+    assert !sample_blob_memory("PostScript/sierpinski.ps").generated?
 
     # These examples are too basic to tell
-    assert !sample_blob("JavaScript/hello.js").generated?
+    assert !sample_blob_memory("JavaScript/hello.js").generated?
 
-    assert sample_blob("JavaScript/intro-old.js").generated?
+    assert sample_blob_memory("JavaScript/intro-old.js").generated?
-    assert sample_blob("JavaScript/classes-old.js").generated?
+    assert sample_blob_memory("JavaScript/classes-old.js").generated?
 
-    assert sample_blob("JavaScript/intro.js").generated?
+    assert sample_blob_memory("JavaScript/intro.js").generated?
-    assert sample_blob("JavaScript/classes.js").generated?
+    assert sample_blob_memory("JavaScript/classes.js").generated?
 
     # Protocol Buffer generated code
-    assert sample_blob("C++/protocol-buffer.pb.h").generated?
+    assert sample_blob_memory("C++/protocol-buffer.pb.h").generated?
-    assert sample_blob("C++/protocol-buffer.pb.cc").generated?
+    assert sample_blob_memory("C++/protocol-buffer.pb.cc").generated?
-    assert sample_blob("Java/ProtocolBuffer.java").generated?
+    assert sample_blob_memory("Java/ProtocolBuffer.java").generated?
-    assert sample_blob("Python/protocol_buffer_pb2.py").generated?
+    assert sample_blob_memory("Python/protocol_buffer_pb2.py").generated?
-    assert sample_blob("Go/api.pb.go").generated?
+    assert sample_blob_memory("Go/api.pb.go").generated?
-    assert sample_blob("Go/embedded.go").generated?
+    assert sample_blob_memory("Go/embedded.go").generated?
 
     # Apache Thrift generated code
-    assert sample_blob("Python/gen-py-linguist-thrift.py").generated?
+    assert sample_blob_memory("Python/gen-py-linguist-thrift.py").generated?
-    assert sample_blob("Go/gen-go-linguist-thrift.go").generated?
+    assert sample_blob_memory("Go/gen-go-linguist-thrift.go").generated?
-    assert sample_blob("Java/gen-java-linguist-thrift.java").generated?
+    assert sample_blob_memory("Java/gen-java-linguist-thrift.java").generated?
-    assert sample_blob("JavaScript/gen-js-linguist-thrift.js").generated?
+    assert sample_blob_memory("JavaScript/gen-js-linguist-thrift.js").generated?
-    assert sample_blob("Ruby/gen-rb-linguist-thrift.rb").generated?
+    assert sample_blob_memory("Ruby/gen-rb-linguist-thrift.rb").generated?
-    assert sample_blob("Objective-C/gen-cocoa-linguist-thrift.m").generated?
+    assert sample_blob_memory("Objective-C/gen-cocoa-linguist-thrift.m").generated?
 
     # Generated JNI
-    assert sample_blob("C/jni_layer.h").generated?
+    assert sample_blob_memory("C/jni_layer.h").generated?
 
     # Minified CSS
-    assert !sample_blob("CSS/bootstrap.css").generated?
+    assert !sample_blob_memory("CSS/bootstrap.css").generated?
-    assert sample_blob("CSS/bootstrap.min.css").generated?
+    assert sample_blob_memory("CSS/bootstrap.min.css").generated?
 
     # Generated VCR
-    assert sample_blob("YAML/vcr_cassette.yml").generated?
+    assert sample_blob_memory("YAML/vcr_cassette.yml").generated?
 
     # Generated by Zephir
-    assert sample_blob("Zephir/filenames/exception.zep.c").generated?
+    assert !sample_blob_memory("Zephir/Router.zep").generated?
-    assert sample_blob("Zephir/filenames/exception.zep.h").generated?
-    assert sample_blob("Zephir/filenames/exception.zep.php").generated?
-    assert !sample_blob("Zephir/Router.zep").generated?
-
-    assert sample_blob("node_modules/grunt/lib/grunt.js").generated?
-
-    # Godep saved dependencies
-    assert sample_blob("Godeps/Godeps.json").generated?
-    assert sample_blob("Godeps/_workspace/src/github.com/kr/s3/sign.go").generated?
 
     # Cython-generated C/C++
-    assert sample_blob("C/sgd_fast.c").generated?
+    assert sample_blob_memory("C/sgd_fast.c").generated?
-    assert sample_blob("C++/wrapper_inner.cpp").generated?
+    assert sample_blob_memory("C++/wrapper_inner.cpp").generated?
 
     # Unity3D-generated metadata
-    assert sample_blob("Unity3D Asset/Tiles.meta").generated?
+    assert sample_blob_memory("Unity3D Asset/Tiles.meta").generated?
+
+    # Racc-generated Ruby
+    assert sample_blob_memory("Ruby/racc.rb").generated?
   end
 
   def test_vendored
-    assert !fixture_blob("Data/README").vendored?
+    assert !fixture_blob_memory("Data/README").vendored?
-    assert !sample_blob("ext/extconf.rb").vendored?
-
-    # Dependencies
-    assert sample_blob("dependencies/windows/headers/GL/glext.h").vendored?
-
-    # Node dependencies
-    assert sample_blob("node_modules/coffee-script/lib/coffee-script.js").vendored?
-
-    # Bower Components
-    assert sample_blob("bower_components/custom/custom.js").vendored?
-    assert sample_blob("app/bower_components/custom/custom.js").vendored?
-    assert sample_blob("vendor/assets/bower_components/custom/custom.js").vendored?
-
-    # Go dependencies
-    assert !sample_blob("Godeps/Godeps.json").vendored?
-    assert sample_blob("Godeps/_workspace/src/github.com/kr/s3/sign.go").vendored?
-
-    # Rails vendor/
-    assert sample_blob("vendor/plugins/will_paginate/lib/will_paginate.rb").vendored?
-
-    # Vendor/
-    assert sample_blob("Vendor/my_great_file.h").vendored?
-
-    # 'thirdparty' directory
-    assert sample_blob("thirdparty/lib/main.c").vendored?
-
-    # 'extern(al)' directory
-    assert sample_blob("extern/util/__init__.py").vendored?
-    assert sample_blob("external/jquery.min.js").vendored?
-
-    # C deps
-    assert sample_blob("deps/http_parser/http_parser.c").vendored?
-    assert sample_blob("deps/v8/src/v8.h").vendored?
-
-    assert sample_blob("tools/something/else.c").vendored?
-
-    # Chart.js
-    assert sample_blob("some/vendored/path/Chart.js").vendored?
-    assert !sample_blob("some/vendored/path/chart.js").vendored?
-
-    # Codemirror deps
-    assert sample_blob("codemirror/mode/blah.js").vendored?
-    assert sample_blob("codemirror/5.0/mode/blah.js").vendored?
-
-    # Debian packaging
-    assert sample_blob("debian/cron.d").vendored?
-
-    # Erlang
-    assert sample_blob("rebar").vendored?
-
-    # git config files
-
-    assert_predicate fixture_blob("some/path/.gitattributes"), :vendored?
-    assert_predicate fixture_blob(".gitignore"), :vendored?
-    assert_predicate fixture_blob("special/path/.gitmodules"), :vendored?
-
-    # Minified JavaScript and CSS
-    assert sample_blob("foo.min.js").vendored?
-    assert sample_blob("foo.min.css").vendored?
-    assert sample_blob("foo-min.js").vendored?
-    assert sample_blob("foo-min.css").vendored?
-    assert !sample_blob("foomin.css").vendored?
-    assert !sample_blob("foo.min.txt").vendored?
-
-    #.osx
-    assert sample_blob(".osx").vendored?
-
-    # Prototype
-    assert !sample_blob("public/javascripts/application.js").vendored?
-    assert sample_blob("public/javascripts/prototype.js").vendored?
-    assert sample_blob("public/javascripts/effects.js").vendored?
-    assert sample_blob("public/javascripts/controls.js").vendored?
-    assert sample_blob("public/javascripts/dragdrop.js").vendored?
-
-    # jQuery
-    assert sample_blob("jquery.js").vendored?
-    assert sample_blob("public/javascripts/jquery.js").vendored?
-    assert sample_blob("public/javascripts/jquery.min.js").vendored?
-    assert sample_blob("public/javascripts/jquery-1.7.js").vendored?
-    assert sample_blob("public/javascripts/jquery-1.7.min.js").vendored?
-    assert sample_blob("public/javascripts/jquery-1.5.2.js").vendored?
-    assert sample_blob("public/javascripts/jquery-1.6.1.js").vendored?
-    assert sample_blob("public/javascripts/jquery-1.6.1.min.js").vendored?
-    assert sample_blob("public/javascripts/jquery-1.10.1.js").vendored?
-    assert sample_blob("public/javascripts/jquery-1.10.1.min.js").vendored?
-    assert !sample_blob("public/javascripts/jquery.github.menu.js").vendored?
-
-    # jQuery UI
-    assert sample_blob("themes/ui-lightness/jquery-ui.css").vendored?
-    assert sample_blob("themes/ui-lightness/jquery-ui-1.8.22.custom.css").vendored?
-    assert sample_blob("themes/ui-lightness/jquery.ui.accordion.css").vendored?
-    assert sample_blob("ui/i18n/jquery.ui.datepicker-ar.js").vendored?
-    assert sample_blob("ui/i18n/jquery-ui-i18n.js").vendored?
-    assert sample_blob("ui/jquery.effects.blind.js").vendored?
-    assert sample_blob("ui/jquery-ui-1.8.22.custom.js").vendored?
-    assert sample_blob("ui/jquery-ui-1.8.22.custom.min.js").vendored?
-    assert sample_blob("ui/jquery-ui-1.8.22.js").vendored?
-    assert sample_blob("ui/jquery-ui-1.8.js").vendored?
-    assert sample_blob("ui/jquery-ui.min.js").vendored?
-    assert sample_blob("ui/jquery.ui.accordion.js").vendored?
-    assert sample_blob("ui/minified/jquery.effects.blind.min.js").vendored?
-    assert sample_blob("ui/minified/jquery.ui.accordion.min.js").vendored?
-
-    # jQuery Gantt
-    assert sample_blob("web-app/jquery-gantt/js/jquery.fn.gantt.js").vendored?
-
-    # jQuery fancyBox
-    assert sample_blob("web-app/fancybox/jquery.fancybox.js").vendored?
-
-    # Fuel UX
-    assert sample_blob("web-app/fuelux/js/fuelux.js").vendored?
-
-    # jQuery File Upload
-    assert sample_blob("fileupload-9.0.0/jquery.fileupload-process.js").vendored?
-
-    # Slick
-    assert sample_blob("web-app/slickgrid/controls/slick.columnpicker.js").vendored?
-
-    # Leaflet plugins
-    assert sample_blob("leaflet-plugins/Leaflet.Coordinates-0.5.0.src.js").vendored?
-    assert sample_blob("leaflet-plugins/leaflet.draw-src.js").vendored?
-    assert sample_blob("leaflet-plugins/leaflet.spin.js").vendored?
-
-    # MooTools
-    assert sample_blob("public/javascripts/mootools-core-1.3.2-full-compat.js").vendored?
-    assert sample_blob("public/javascripts/mootools-core-1.3.2-full-compat-yc.js").vendored?
-
-    # Dojo
-    assert sample_blob("public/javascripts/dojo.js").vendored?
-
-    # MochiKit
-    assert sample_blob("public/javascripts/MochiKit.js").vendored?
-
-    # YUI
-    assert sample_blob("public/javascripts/yahoo-dom-event.js").vendored?
-    assert sample_blob("public/javascripts/yahoo-min.js").vendored?
-    assert sample_blob("public/javascripts/yuiloader-dom-event.js").vendored?
-
-    # WYS editors
-    assert sample_blob("public/javascripts/ckeditor.js").vendored?
-    assert sample_blob("public/javascripts/tiny_mce.js").vendored?
-    assert sample_blob("public/javascripts/tiny_mce_popup.js").vendored?
-    assert sample_blob("public/javascripts/tiny_mce_src.js").vendored?
-
-    # AngularJS
-    assert sample_blob("public/javascripts/angular.js").vendored?
-    assert sample_blob("public/javascripts/angular.min.js").vendored?
-
-    # D3.js
-    assert sample_blob("public/javascripts/d3.v3.js").vendored?
-    assert sample_blob("public/javascripts/d3.v3.min.js").vendored?
-
-    # Modernizr
-    assert sample_blob("public/javascripts/modernizr-2.7.1.js").vendored?
-    assert sample_blob("public/javascripts/modernizr.custom.01009.js").vendored?
-
-    # Fabric
-    assert sample_blob("fabfile.py").vendored?
-
-    # WAF
-    assert sample_blob("waf").vendored?
-
-    # Visual Studio IntelliSense
-    assert sample_blob("Scripts/jquery-1.7-vsdoc.js").vendored?
-
-    # Microsoft Ajax
-    assert sample_blob("Scripts/MicrosoftAjax.debug.js").vendored?
-    assert sample_blob("Scripts/MicrosoftAjax.js").vendored?
-    assert sample_blob("Scripts/MicrosoftMvcAjax.debug.js").vendored?
-    assert sample_blob("Scripts/MicrosoftMvcAjax.js").vendored?
-    assert sample_blob("Scripts/MicrosoftMvcValidation.debug.js").vendored?
-    assert sample_blob("Scripts/MicrosoftMvcValidation.js").vendored?
-
-    # jQuery validation plugin (MS bundles this with asp.net mvc)
-    assert sample_blob("Scripts/jquery.validate.js").vendored?
-    assert sample_blob("Scripts/jquery.validate.min.js").vendored?
-    assert sample_blob("Scripts/jquery.validate.unobtrusive.js").vendored?
-    assert sample_blob("Scripts/jquery.validate.unobtrusive.min.js").vendored?
-    assert sample_blob("Scripts/jquery.unobtrusive-ajax.js").vendored?
-    assert sample_blob("Scripts/jquery.unobtrusive-ajax.min.js").vendored?
-
-    # NuGet Packages
-    assert sample_blob("packages/Modernizr.2.0.6/Content/Scripts/modernizr-2.0.6-development-only.js").vendored?
-
-    # Font Awesome
-    assert sample_blob("some/asset/path/font-awesome.min.css").vendored?
-    assert sample_blob("some/asset/path/font-awesome.css").vendored?
-
-    # Normalize
-    assert sample_blob("some/asset/path/normalize.css").vendored?
-
-    # Carthage
-    assert sample_blob('Carthage/blah').vendored?
-
-    # Cocoapods
-    assert sample_blob('Pods/blah').vendored?
-
-    # Html5shiv
-    assert sample_blob("Scripts/html5shiv.js").vendored?
-    assert sample_blob("Scripts/html5shiv.min.js").vendored?
-
-    # Test fixtures
-    assert sample_blob("test/fixtures/random.rkt").vendored?
-    assert sample_blob("Test/fixtures/random.rkt").vendored?
-    assert sample_blob("tests/fixtures/random.rkt").vendored?
-
-    # Cordova/PhoneGap
-    assert sample_blob("cordova.js").vendored?
-    assert sample_blob("cordova.min.js").vendored?
-    assert sample_blob("cordova-2.1.0.js").vendored?
-    assert sample_blob("cordova-2.1.0.min.js").vendored?
-
-    # Foundation js
-    assert sample_blob("foundation.js").vendored?
-    assert sample_blob("foundation.min.js").vendored?
-    assert sample_blob("foundation.abide.js").vendored?
-
-    # Vagrant
-    assert sample_blob("Vagrantfile").vendored?
-
-    # Gradle
-    assert sample_blob("gradlew").vendored?
-    assert sample_blob("gradlew.bat").vendored?
-    assert sample_blob("gradle/wrapper/gradle-wrapper.properties").vendored?
-    assert sample_blob("subproject/gradlew").vendored?
-    assert sample_blob("subproject/gradlew.bat").vendored?
-    assert sample_blob("subproject/gradle/wrapper/gradle-wrapper.properties").vendored?
-
-    # Octicons
-    assert sample_blob("octicons.css").vendored?
-    assert sample_blob("public/octicons.min.css").vendored?
-    assert sample_blob("public/octicons/sprockets-octicons.scss").vendored?
-
-    # Typesafe Activator
-    assert sample_blob("activator").vendored?
-    assert sample_blob("activator.bat").vendored?
-    assert sample_blob("subproject/activator").vendored?
-    assert sample_blob("subproject/activator.bat").vendored?
-
-    assert_predicate fixture_blob(".google_apis/bar.jar"), :vendored?
-    assert_predicate fixture_blob("foo/.google_apis/bar.jar"), :vendored?
-
-    # Sphinx docs
-    assert sample_blob("docs/_build/asset.doc").vendored?
-    assert sample_blob("docs/theme/file.css").vendored?
-
-    # Vagrant
-    assert sample_blob("puphpet/file.pp").vendored?
-
-    # Fabric.io
-    assert sample_blob("Fabric.framework/Fabric.h").vendored?
-
-    # Crashlytics
-    assert sample_blob("Crashlytics.framework/Crashlytics.h").vendored?
-    
-    # Xcode
-    assert sample_blob("myapp/My Template.xctemplate/___FILEBASENAME___.h").vendored?
-    assert sample_blob("myapp/My Images.xcassets/some/stuff.imageset/Contents.json").vendored?
-    assert !sample_blob("myapp/MyData.json").vendored?
-  end
-
-  def test_documentation
-    assert_predicate fixture_blob("doc/foo.html"), :documentation?
-    assert_predicate fixture_blob("docs/foo.html"), :documentation?
-    refute_predicate fixture_blob("project/doc/foo.html"), :documentation?
-    refute_predicate fixture_blob("project/docs/foo.html"), :documentation?
-
-    assert_predicate fixture_blob("Documentation/foo.md"), :documentation?
-    assert_predicate fixture_blob("documentation/foo.md"), :documentation?
-    assert_predicate fixture_blob("project/Documentation/foo.md"), :documentation?
-    assert_predicate fixture_blob("project/documentation/foo.md"), :documentation?
-
-    assert_predicate fixture_blob("javadoc/foo.html"), :documentation?
-    assert_predicate fixture_blob("project/javadoc/foo.html"), :documentation?
-
-    assert_predicate fixture_blob("man/foo.html"), :documentation?
-    refute_predicate fixture_blob("project/man/foo.html"), :documentation?
-
-    assert_predicate fixture_blob("README"), :documentation?
-    assert_predicate fixture_blob("README.md"), :documentation?
-    assert_predicate fixture_blob("README.txt"), :documentation?
-    assert_predicate fixture_blob("Readme"), :documentation?
-    assert_predicate fixture_blob("readme"), :documentation?
-    assert_predicate fixture_blob("foo/README"), :documentation?
-
-    assert_predicate fixture_blob("CHANGE"), :documentation?
-    assert_predicate fixture_blob("CHANGE.md"), :documentation?
-    assert_predicate fixture_blob("CHANGE.txt"), :documentation?
-    assert_predicate fixture_blob("foo/CHANGE"), :documentation?
-
-    assert_predicate fixture_blob("CHANGELOG"), :documentation?
-    assert_predicate fixture_blob("CHANGELOG.md"), :documentation?
-    assert_predicate fixture_blob("CHANGELOG.txt"), :documentation?
-    assert_predicate fixture_blob("foo/CHANGELOG"), :documentation?
-
-    assert_predicate fixture_blob("CHANGES"), :documentation?
-    assert_predicate fixture_blob("CHANGES.md"), :documentation?
-    assert_predicate fixture_blob("CHANGES.txt"), :documentation?
-    assert_predicate fixture_blob("foo/CHANGES"), :documentation?
-
-    assert_predicate fixture_blob("CONTRIBUTING"), :documentation?
-    assert_predicate fixture_blob("CONTRIBUTING.md"), :documentation?
-    assert_predicate fixture_blob("CONTRIBUTING.txt"), :documentation?
-    assert_predicate fixture_blob("foo/CONTRIBUTING"), :documentation?
-
-    assert_predicate fixture_blob("examples/some-file.pl"), :documentation?
-    assert_predicate fixture_blob("Examples/some-example-file.rb"), :documentation?
-
-    assert_predicate fixture_blob("LICENSE"), :documentation?
-    assert_predicate fixture_blob("LICENCE.md"), :documentation?
-    assert_predicate fixture_blob("License.txt"), :documentation?
-    assert_predicate fixture_blob("LICENSE.txt"), :documentation?
-    assert_predicate fixture_blob("foo/LICENSE"), :documentation?
-
-    assert_predicate fixture_blob("COPYING"), :documentation?
-    assert_predicate fixture_blob("COPYING.md"), :documentation?
-    assert_predicate fixture_blob("COPYING.txt"), :documentation?
-    assert_predicate fixture_blob("foo/COPYING"), :documentation?
-
-    assert_predicate fixture_blob("INSTALL"), :documentation?
-    assert_predicate fixture_blob("INSTALL.md"), :documentation?
-    assert_predicate fixture_blob("INSTALL.txt"), :documentation?
-    assert_predicate fixture_blob("foo/INSTALL"), :documentation?
-
-    refute_predicate fixture_blob("foo.md"), :documentation?
-
-    # Samples
-    assert sample_blob("Samples/Ruby/foo.rb").documentation?
-
-    assert_predicate fixture_blob("INSTALL.txt"), :documentation?
   end
 
   def test_language
     Samples.each do |sample|
-      blob = sample_blob(sample[:path])
+      blob = sample_blob_memory(sample[:path])
       assert blob.language, "No language for #{sample[:path]}"
       assert_equal sample[:language], blob.language.name, blob.name
     end
@@ -612,7 +248,7 @@ class TestBlob < Minitest::Test
         filepath = File.join(dirname, filename)
         next unless File.file?(filepath)
 
-        blob = fixture_blob(filepath)
+        blob = fixture_blob_memory(filepath)
         if language == 'Data'
           assert blob.language.nil?, "A language was found for #{filepath}"
         elsif language == 'Generated'
@@ -626,7 +262,7 @@ class TestBlob < Minitest::Test
   end
 
   def test_minified_files_not_safe_to_highlight
-    assert !sample_blob("JavaScript/jquery-1.6.1.min.js").safe_to_colorize?
+    assert !sample_blob_memory("JavaScript/jquery-1.6.1.min.js").safe_to_colorize?
   end
 
   def test_empty
@@ -639,27 +275,19 @@ class TestBlob < Minitest::Test
   end
 
   def test_include_in_language_stats
-    vendored = sample_blob("bower_components/custom/custom.js")
+    generated = sample_blob_memory("CSS/bootstrap.min.css")
-    assert_predicate vendored, :vendored?
-    refute_predicate vendored, :include_in_language_stats?
-
-    documentation = fixture_blob("README")
-    assert_predicate documentation, :documentation?
-    refute_predicate documentation, :include_in_language_stats?
-
-    generated = sample_blob("CSS/bootstrap.min.css")
     assert_predicate generated, :generated?
     refute_predicate generated, :include_in_language_stats?
 
-    data = sample_blob("Ant Build System/filenames/ant.xml")
+    data = sample_blob_memory("Ant Build System/filenames/ant.xml")
     assert_equal :data, data.language.type
     refute_predicate data, :include_in_language_stats?
 
-    prose = sample_blob("Markdown/tender.md")
+    prose = sample_blob_memory("Markdown/tender.md")
     assert_equal :prose, prose.language.type
     refute_predicate prose, :include_in_language_stats?
 
-    included = sample_blob("HTML/pages.html")
+    included = sample_blob_memory("HTML/pages.html")
     assert_predicate included, :include_in_language_stats?
   end
 end

test/test_file_blob.rb

@@ -1,9 +1,669 @@
 require_relative "./helper"
 
-class TestFileBlob < Minitest::Test
+class TestBlob < Minitest::Test
+  include Linguist
+
+  def setup
+    # git blobs are normally loaded as ASCII-8BIT since they may contain data
+    # with arbitrary encoding not known ahead of time
+    @original_external = Encoding.default_external
+    Encoding.default_external = Encoding.find("ASCII-8BIT")
+  end
+
+  def teardown
+    Encoding.default_external = @original_external
+  end
+
+  def script_blob(name)
+    blob = sample_blob(name)
+    blob.instance_variable_set(:@name, 'script')
+    blob
+  end
+  
   def test_extensions
     assert_equal [".gitignore"], Linguist::FileBlob.new(".gitignore").extensions
     assert_equal [".xml"],  Linguist::FileBlob.new("build.xml").extensions
     assert_equal [".html.erb", ".erb"],  Linguist::FileBlob.new("dotted.dir/index.html.erb").extensions
   end
+
+  def test_name
+    assert_equal "foo.rb", sample_blob("foo.rb").name
+  end
+
+  def test_mime_type
+    assert_equal "application/postscript", fixture_blob("Binary/octocat.ai").mime_type
+    assert_equal "application/x-ruby", sample_blob("Ruby/grit.rb").mime_type
+    assert_equal "application/x-sh", sample_blob("Shell/script.sh").mime_type
+    assert_equal "application/xml", sample_blob("XML/bar.xml").mime_type
+    assert_equal "audio/ogg", fixture_blob("Binary/foo.ogg").mime_type
+    assert_equal "text/plain", fixture_blob("Data/README").mime_type
+  end
+
+  def test_content_type
+    assert_equal "application/pdf", fixture_blob("Binary/foo.pdf").content_type
+    assert_equal "audio/ogg", fixture_blob("Binary/foo.ogg").content_type
+    assert_equal "image/png", fixture_blob("Binary/foo.png").content_type
+    assert_equal "text/plain; charset=iso-8859-2", fixture_blob("Data/README").content_type
+  end
+
+  def test_disposition
+    assert_equal "attachment; filename=foo+bar.jar", fixture_blob("Binary/foo bar.jar").disposition
+    assert_equal "attachment; filename=foo.bin", fixture_blob("Binary/foo.bin").disposition
+    assert_equal "attachment; filename=linguist.gem", fixture_blob("Binary/linguist.gem").disposition
+    assert_equal "attachment; filename=octocat.ai", fixture_blob("Binary/octocat.ai").disposition
+    assert_equal "inline", fixture_blob("Data/README").disposition
+    assert_equal "inline", sample_blob("Text/foo.txt").disposition
+    assert_equal "inline", sample_blob("Ruby/grit.rb").disposition
+    assert_equal "inline", fixture_blob("Binary/octocat.png").disposition
+  end
+
+  def test_data
+    assert_equal "module Foo\nend\n", sample_blob("Ruby/foo.rb").data
+  end
+
+  def test_lines
+    assert_equal ["module Foo", "end", ""], sample_blob("Ruby/foo.rb").lines
+    assert_equal ["line 1", "line 2", ""], sample_blob("Text/mac.txt").lines
+    assert_equal 475, sample_blob("Emacs Lisp/ess-julia.el").lines.length
+  end
+
+  def test_lines_maintains_original_encoding
+    # Even if the file's encoding is detected as something like UTF-16LE,
+    # earlier versions of the gem made implicit guarantees that the encoding of
+    # each `line` is in the same encoding as the file was originally read (in
+    # practice, UTF-8 or ASCII-8BIT)
+    assert_equal Encoding.default_external, fixture_blob("Data/utf16le").lines.first.encoding
+  end
+
+  def test_size
+    assert_equal 15, sample_blob("Ruby/foo.rb").size
+  end
+
+  def test_loc
+    assert_equal 3, sample_blob("Ruby/foo.rb").loc
+  end
+
+  def test_sloc
+    assert_equal 2, sample_blob("Ruby/foo.rb").sloc
+    assert_equal 3, fixture_blob("Data/utf16le-windows").sloc
+    assert_equal 1, fixture_blob("Data/iso8859-8-i").sloc
+  end
+
+  def test_encoding
+    assert_equal "ISO-8859-2", fixture_blob("Data/README").encoding
+    assert_equal "ISO-8859-2", fixture_blob("Data/README").ruby_encoding
+    assert_equal "UTF-8", sample_blob("Text/foo.txt").encoding
+    assert_equal "UTF-8", sample_blob("Text/foo.txt").ruby_encoding
+    assert_equal "UTF-16LE", fixture_blob("Data/utf16le").encoding
+    assert_equal "UTF-16LE", fixture_blob("Data/utf16le").ruby_encoding
+    assert_equal "UTF-16LE", fixture_blob("Data/utf16le-windows").encoding
+    assert_equal "UTF-16LE", fixture_blob("Data/utf16le-windows").ruby_encoding
+    assert_equal "ISO-2022-KR", sample_blob("Text/ISO-2022-KR.txt").encoding
+    assert_equal "binary", sample_blob("Text/ISO-2022-KR.txt").ruby_encoding
+    assert_nil fixture_blob("Binary/dog.o").encoding
+  end
+
+  def test_binary
+    # Large blobs aren't loaded
+    large_blob = sample_blob("git.exe")
+    large_blob.instance_eval do
+      def data; end
+    end
+    assert large_blob.binary?
+
+    assert fixture_blob("Binary/git.deb").binary?
+    assert fixture_blob("Binary/git.exe").binary?
+    assert fixture_blob("Binary/hello.pbc").binary?
+    assert fixture_blob("Binary/linguist.gem").binary?
+    assert fixture_blob("Binary/octocat.ai").binary?
+    assert fixture_blob("Binary/octocat.png").binary?
+    assert fixture_blob("Binary/zip").binary?
+    assert !fixture_blob("Data/README").binary?
+    assert !sample_blob("Ruby/foo.rb").binary?
+    assert !sample_blob("Perl/script.pl").binary?
+  end
+
+  def test_all_binary
+    Samples.each do |sample|
+      blob = sample_blob(sample[:path])
+      assert ! (blob.likely_binary? || blob.binary?), "#{sample[:path]} is a binary file"
+    end
+  end
+
+  def test_text
+    assert fixture_blob("Data/README").text?
+    assert fixture_blob("Data/md").text?
+    assert sample_blob("Shell/script.sh").text?
+    assert fixture_blob("Data/txt").text?
+  end
+
+  def test_image
+    assert fixture_blob("Binary/octocat.gif").image?
+    assert fixture_blob("Binary/octocat.jpeg").image?
+    assert fixture_blob("Binary/octocat.jpg").image?
+    assert fixture_blob("Binary/octocat.png").image?
+    assert !fixture_blob("Binary/octocat.ai").image?
+    assert !fixture_blob("Binary/octocat.psd").image?
+  end
+
+  def test_solid
+    assert fixture_blob("Binary/cube.stl").solid?
+    assert fixture_blob("Data/cube.stl").solid?
+  end
+
+  def test_csv
+    assert fixture_blob("Data/cars.csv").csv?
+  end
+
+  def test_pdf
+    assert fixture_blob("Binary/foo.pdf").pdf?
+  end
+
+  def test_viewable
+    assert fixture_blob("Data/README").viewable?
+    assert sample_blob("Ruby/foo.rb").viewable?
+    assert sample_blob("Perl/script.pl").viewable?
+    assert !fixture_blob("Binary/linguist.gem").viewable?
+    assert !fixture_blob("Binary/octocat.ai").viewable?
+    assert !fixture_blob("Binary/octocat.png").viewable?
+  end
+
+  def test_generated
+    assert !fixture_blob("Data/README").generated?
+
+    # Xcode project files
+    assert !sample_blob("XML/MainMenu.xib").generated?
+    assert fixture_blob("Binary/MainMenu.nib").generated?
+    assert !sample_blob("XML/project.pbxproj").generated?
+
+    # Gemfile.lock is NOT generated
+    assert !sample_blob("Gemfile.lock").generated?
+
+    # Generated .NET Docfiles
+    assert sample_blob("XML/net_docfile.xml").generated?
+
+    # Long line
+    assert !sample_blob("JavaScript/uglify.js").generated?
+
+    # Inlined JS, but mostly code
+    assert !sample_blob("JavaScript/json2_backbone.js").generated?
+
+    # Minified JS
+    assert !sample_blob("JavaScript/jquery-1.6.1.js").generated?
+    assert sample_blob("JavaScript/jquery-1.6.1.min.js").generated?
+    assert sample_blob("JavaScript/jquery-1.4.2.min.js").generated?
+
+    # CoffeeScript-generated JS
+    # TODO
+
+    # TypeScript-generated JS
+    # TODO
+
+    # Composer generated composer.lock file
+    assert sample_blob("JSON/composer.lock").generated?
+
+    # PEG.js-generated parsers
+    assert sample_blob("JavaScript/parser.js").generated?
+
+    # Generated PostScript
+    assert !sample_blob("PostScript/sierpinski.ps").generated?
+
+    # These examples are too basic to tell
+    assert !sample_blob("JavaScript/hello.js").generated?
+
+    assert sample_blob("JavaScript/intro-old.js").generated?
+    assert sample_blob("JavaScript/classes-old.js").generated?
+
+    assert sample_blob("JavaScript/intro.js").generated?
+    assert sample_blob("JavaScript/classes.js").generated?
+
+    # Protocol Buffer generated code
+    assert sample_blob("C++/protocol-buffer.pb.h").generated?
+    assert sample_blob("C++/protocol-buffer.pb.cc").generated?
+    assert sample_blob("Java/ProtocolBuffer.java").generated?
+    assert sample_blob("Python/protocol_buffer_pb2.py").generated?
+    assert sample_blob("Go/api.pb.go").generated?
+    assert sample_blob("Go/embedded.go").generated?
+
+    # Apache Thrift generated code
+    assert sample_blob("Python/gen-py-linguist-thrift.py").generated?
+    assert sample_blob("Go/gen-go-linguist-thrift.go").generated?
+    assert sample_blob("Java/gen-java-linguist-thrift.java").generated?
+    assert sample_blob("JavaScript/gen-js-linguist-thrift.js").generated?
+    assert sample_blob("Ruby/gen-rb-linguist-thrift.rb").generated?
+    assert sample_blob("Objective-C/gen-cocoa-linguist-thrift.m").generated?
+
+    # Generated JNI
+    assert sample_blob("C/jni_layer.h").generated?
+
+    # Minified CSS
+    assert !sample_blob("CSS/bootstrap.css").generated?
+    assert sample_blob("CSS/bootstrap.min.css").generated?
+
+    # Generated VCR
+    assert sample_blob("YAML/vcr_cassette.yml").generated?
+
+    # Generated by Zephir
+    assert sample_blob("Zephir/filenames/exception.zep.c").generated?
+    assert sample_blob("Zephir/filenames/exception.zep.h").generated?
+    assert sample_blob("Zephir/filenames/exception.zep.php").generated?
+    assert !sample_blob("Zephir/Router.zep").generated?
+
+    assert sample_blob("node_modules/grunt/lib/grunt.js").generated?
+
+    # Godep saved dependencies
+    assert sample_blob("Godeps/Godeps.json").generated?
+    assert sample_blob("Godeps/_workspace/src/github.com/kr/s3/sign.go").generated?
+
+    # Cython-generated C/C++
+    assert sample_blob("C/sgd_fast.c").generated?
+    assert sample_blob("C++/wrapper_inner.cpp").generated?
+
+    # Unity3D-generated metadata
+    assert sample_blob("Unity3D Asset/Tiles.meta").generated?
+  end
+
+  def test_vendored
+    assert !fixture_blob("Data/README").vendored?
+    assert !sample_blob("ext/extconf.rb").vendored?
+
+    # Dependencies
+    assert sample_blob("dependencies/windows/headers/GL/glext.h").vendored?
+
+    # Node dependencies
+    assert sample_blob("node_modules/coffee-script/lib/coffee-script.js").vendored?
+
+    # Bower Components
+    assert sample_blob("bower_components/custom/custom.js").vendored?
+    assert sample_blob("app/bower_components/custom/custom.js").vendored?
+    assert sample_blob("vendor/assets/bower_components/custom/custom.js").vendored?
+
+    # Go dependencies
+    assert !sample_blob("Godeps/Godeps.json").vendored?
+    assert sample_blob("Godeps/_workspace/src/github.com/kr/s3/sign.go").vendored?
+
+    # Rails vendor/
+    assert sample_blob("vendor/plugins/will_paginate/lib/will_paginate.rb").vendored?
+
+    # Vendor/
+    assert sample_blob("Vendor/my_great_file.h").vendored?
+
+    # 'thirdparty' directory
+    assert sample_blob("thirdparty/lib/main.c").vendored?
+
+    # 'extern(al)' directory
+    assert sample_blob("extern/util/__init__.py").vendored?
+    assert sample_blob("external/jquery.min.js").vendored?
+
+    # C deps
+    assert sample_blob("deps/http_parser/http_parser.c").vendored?
+    assert sample_blob("deps/v8/src/v8.h").vendored?
+
+    assert sample_blob("tools/something/else.c").vendored?
+
+    # Chart.js
+    assert sample_blob("some/vendored/path/Chart.js").vendored?
+    assert !sample_blob("some/vendored/path/chart.js").vendored?
+
+    # Codemirror deps
+    assert sample_blob("codemirror/mode/blah.js").vendored?
+    assert sample_blob("codemirror/5.0/mode/blah.js").vendored?
+
+    # Debian packaging
+    assert sample_blob("debian/cron.d").vendored?
+
+    # Erlang
+    assert sample_blob("rebar").vendored?
+
+    # git config files
+
+    assert_predicate fixture_blob("some/path/.gitattributes"), :vendored?
+    assert_predicate fixture_blob(".gitignore"), :vendored?
+    assert_predicate fixture_blob("special/path/.gitmodules"), :vendored?
+
+    # Minified JavaScript and CSS
+    assert sample_blob("foo.min.js").vendored?
+    assert sample_blob("foo.min.css").vendored?
+    assert sample_blob("foo-min.js").vendored?
+    assert sample_blob("foo-min.css").vendored?
+    assert !sample_blob("foomin.css").vendored?
+    assert !sample_blob("foo.min.txt").vendored?
+
+    #.osx
+    assert sample_blob(".osx").vendored?
+
+    # Prototype
+    assert !sample_blob("public/javascripts/application.js").vendored?
+    assert sample_blob("public/javascripts/prototype.js").vendored?
+    assert sample_blob("public/javascripts/effects.js").vendored?
+    assert sample_blob("public/javascripts/controls.js").vendored?
+    assert sample_blob("public/javascripts/dragdrop.js").vendored?
+
+    # jQuery
+    assert sample_blob("jquery.js").vendored?
+    assert sample_blob("public/javascripts/jquery.js").vendored?
+    assert sample_blob("public/javascripts/jquery.min.js").vendored?
+    assert sample_blob("public/javascripts/jquery-1.7.js").vendored?
+    assert sample_blob("public/javascripts/jquery-1.7.min.js").vendored?
+    assert sample_blob("public/javascripts/jquery-1.5.2.js").vendored?
+    assert sample_blob("public/javascripts/jquery-1.6.1.js").vendored?
+    assert sample_blob("public/javascripts/jquery-1.6.1.min.js").vendored?
+    assert sample_blob("public/javascripts/jquery-1.10.1.js").vendored?
+    assert sample_blob("public/javascripts/jquery-1.10.1.min.js").vendored?
+    assert !sample_blob("public/javascripts/jquery.github.menu.js").vendored?
+
+    # jQuery UI
+    assert sample_blob("themes/ui-lightness/jquery-ui.css").vendored?
+    assert sample_blob("themes/ui-lightness/jquery-ui-1.8.22.custom.css").vendored?
+    assert sample_blob("themes/ui-lightness/jquery.ui.accordion.css").vendored?
+    assert sample_blob("ui/i18n/jquery.ui.datepicker-ar.js").vendored?
+    assert sample_blob("ui/i18n/jquery-ui-i18n.js").vendored?
+    assert sample_blob("ui/jquery.effects.blind.js").vendored?
+    assert sample_blob("ui/jquery-ui-1.8.22.custom.js").vendored?
+    assert sample_blob("ui/jquery-ui-1.8.22.custom.min.js").vendored?
+    assert sample_blob("ui/jquery-ui-1.8.22.js").vendored?
+    assert sample_blob("ui/jquery-ui-1.8.js").vendored?
+    assert sample_blob("ui/jquery-ui.min.js").vendored?
+    assert sample_blob("ui/jquery.ui.accordion.js").vendored?
+    assert sample_blob("ui/minified/jquery.effects.blind.min.js").vendored?
+    assert sample_blob("ui/minified/jquery.ui.accordion.min.js").vendored?
+
+    # jQuery Gantt
+    assert sample_blob("web-app/jquery-gantt/js/jquery.fn.gantt.js").vendored?
+
+    # jQuery fancyBox
+    assert sample_blob("web-app/fancybox/jquery.fancybox.js").vendored?
+
+    # Fuel UX
+    assert sample_blob("web-app/fuelux/js/fuelux.js").vendored?
+
+    # jQuery File Upload
+    assert sample_blob("fileupload-9.0.0/jquery.fileupload-process.js").vendored?
+
+    # Slick
+    assert sample_blob("web-app/slickgrid/controls/slick.columnpicker.js").vendored?
+
+    # Leaflet plugins
+    assert sample_blob("leaflet-plugins/Leaflet.Coordinates-0.5.0.src.js").vendored?
+    assert sample_blob("leaflet-plugins/leaflet.draw-src.js").vendored?
+    assert sample_blob("leaflet-plugins/leaflet.spin.js").vendored?
+
+    # MooTools
+    assert sample_blob("public/javascripts/mootools-core-1.3.2-full-compat.js").vendored?
+    assert sample_blob("public/javascripts/mootools-core-1.3.2-full-compat-yc.js").vendored?
+
+    # Dojo
+    assert sample_blob("public/javascripts/dojo.js").vendored?
+
+    # MochiKit
+    assert sample_blob("public/javascripts/MochiKit.js").vendored?
+
+    # YUI
+    assert sample_blob("public/javascripts/yahoo-dom-event.js").vendored?
+    assert sample_blob("public/javascripts/yahoo-min.js").vendored?
+    assert sample_blob("public/javascripts/yuiloader-dom-event.js").vendored?
+
+    # WYS editors
+    assert sample_blob("public/javascripts/ckeditor.js").vendored?
+    assert sample_blob("public/javascripts/tiny_mce.js").vendored?
+    assert sample_blob("public/javascripts/tiny_mce_popup.js").vendored?
+    assert sample_blob("public/javascripts/tiny_mce_src.js").vendored?
+
+    # AngularJS
+    assert sample_blob("public/javascripts/angular.js").vendored?
+    assert sample_blob("public/javascripts/angular.min.js").vendored?
+
+    # D3.js
+    assert sample_blob("public/javascripts/d3.v3.js").vendored?
+    assert sample_blob("public/javascripts/d3.v3.min.js").vendored?
+
+    # Modernizr
+    assert sample_blob("public/javascripts/modernizr-2.7.1.js").vendored?
+    assert sample_blob("public/javascripts/modernizr.custom.01009.js").vendored?
+
+    # Fabric
+    assert sample_blob("fabfile.py").vendored?
+
+    # WAF
+    assert sample_blob("waf").vendored?
+
+    # Visual Studio IntelliSense
+    assert sample_blob("Scripts/jquery-1.7-vsdoc.js").vendored?
+
+    # Microsoft Ajax
+    assert sample_blob("Scripts/MicrosoftAjax.debug.js").vendored?
+    assert sample_blob("Scripts/MicrosoftAjax.js").vendored?
+    assert sample_blob("Scripts/MicrosoftMvcAjax.debug.js").vendored?
+    assert sample_blob("Scripts/MicrosoftMvcAjax.js").vendored?
+    assert sample_blob("Scripts/MicrosoftMvcValidation.debug.js").vendored?
+    assert sample_blob("Scripts/MicrosoftMvcValidation.js").vendored?
+
+    # jQuery validation plugin (MS bundles this with asp.net mvc)
+    assert sample_blob("Scripts/jquery.validate.js").vendored?
+    assert sample_blob("Scripts/jquery.validate.min.js").vendored?
+    assert sample_blob("Scripts/jquery.validate.unobtrusive.js").vendored?
+    assert sample_blob("Scripts/jquery.validate.unobtrusive.min.js").vendored?
+    assert sample_blob("Scripts/jquery.unobtrusive-ajax.js").vendored?
+    assert sample_blob("Scripts/jquery.unobtrusive-ajax.min.js").vendored?
+
+    # NuGet Packages
+    assert sample_blob("packages/Modernizr.2.0.6/Content/Scripts/modernizr-2.0.6-development-only.js").vendored?
+
+    # Font Awesome
+    assert sample_blob("some/asset/path/font-awesome.min.css").vendored?
+    assert sample_blob("some/asset/path/font-awesome.css").vendored?
+
+    # Normalize
+    assert sample_blob("some/asset/path/normalize.css").vendored?
+
+    # Carthage
+    assert sample_blob('Carthage/blah').vendored?
+
+    # Cocoapods
+    assert sample_blob('Pods/blah').vendored?
+
+    # Html5shiv
+    assert sample_blob("Scripts/html5shiv.js").vendored?
+    assert sample_blob("Scripts/html5shiv.min.js").vendored?
+
+    # Test fixtures
+    assert sample_blob("test/fixtures/random.rkt").vendored?
+    assert sample_blob("Test/fixtures/random.rkt").vendored?
+    assert sample_blob("tests/fixtures/random.rkt").vendored?
+
+    # Cordova/PhoneGap
+    assert sample_blob("cordova.js").vendored?
+    assert sample_blob("cordova.min.js").vendored?
+    assert sample_blob("cordova-2.1.0.js").vendored?
+    assert sample_blob("cordova-2.1.0.min.js").vendored?
+
+    # Foundation js
+    assert sample_blob("foundation.js").vendored?
+    assert sample_blob("foundation.min.js").vendored?
+    assert sample_blob("foundation.abide.js").vendored?
+
+    # Vagrant
+    assert sample_blob("Vagrantfile").vendored?
+
+    # Gradle
+    assert sample_blob("gradlew").vendored?
+    assert sample_blob("gradlew.bat").vendored?
+    assert sample_blob("gradle/wrapper/gradle-wrapper.properties").vendored?
+    assert sample_blob("subproject/gradlew").vendored?
+    assert sample_blob("subproject/gradlew.bat").vendored?
+    assert sample_blob("subproject/gradle/wrapper/gradle-wrapper.properties").vendored?
+
+    # Octicons
+    assert sample_blob("octicons.css").vendored?
+    assert sample_blob("public/octicons.min.css").vendored?
+    assert sample_blob("public/octicons/sprockets-octicons.scss").vendored?
+
+    # Typesafe Activator
+    assert sample_blob("activator").vendored?
+    assert sample_blob("activator.bat").vendored?
+    assert sample_blob("subproject/activator").vendored?
+    assert sample_blob("subproject/activator.bat").vendored?
+
+    assert_predicate fixture_blob(".google_apis/bar.jar"), :vendored?
+    assert_predicate fixture_blob("foo/.google_apis/bar.jar"), :vendored?
+
+    # Sphinx docs
+    assert sample_blob("docs/_build/asset.doc").vendored?
+    assert sample_blob("docs/theme/file.css").vendored?
+
+    # Vagrant
+    assert sample_blob("puphpet/file.pp").vendored?
+
+    # Fabric.io
+    assert sample_blob("Fabric.framework/Fabric.h").vendored?
+
+    # Crashlytics
+    assert sample_blob("Crashlytics.framework/Crashlytics.h").vendored?
+    assert sample_blob("myapp/My Template.xctemplate/___FILEBASENAME___.h").vendored?
+    assert sample_blob("myapp/My Images.xcassets/some/stuff.imageset/Contents.json").vendored?
+    assert !sample_blob("myapp/MyData.json").vendored?
+  end
+
+  def test_documentation
+    assert_predicate fixture_blob("doc/foo.html"), :documentation?
+    assert_predicate fixture_blob("docs/foo.html"), :documentation?
+    refute_predicate fixture_blob("project/doc/foo.html"), :documentation?
+    refute_predicate fixture_blob("project/docs/foo.html"), :documentation?
+
+    assert_predicate fixture_blob("Documentation/foo.md"), :documentation?
+    assert_predicate fixture_blob("documentation/foo.md"), :documentation?
+    assert_predicate fixture_blob("project/Documentation/foo.md"), :documentation?
+    assert_predicate fixture_blob("project/documentation/foo.md"), :documentation?
+
+    assert_predicate fixture_blob("javadoc/foo.html"), :documentation?
+    assert_predicate fixture_blob("project/javadoc/foo.html"), :documentation?
+
+    assert_predicate fixture_blob("man/foo.html"), :documentation?
+    refute_predicate fixture_blob("project/man/foo.html"), :documentation?
+
+    assert_predicate fixture_blob("README"), :documentation?
+    assert_predicate fixture_blob("README.md"), :documentation?
+    assert_predicate fixture_blob("README.txt"), :documentation?
+    assert_predicate fixture_blob("Readme"), :documentation?
+    assert_predicate fixture_blob("readme"), :documentation?
+    assert_predicate fixture_blob("foo/README"), :documentation?
+
+    assert_predicate fixture_blob("CHANGE"), :documentation?
+    assert_predicate fixture_blob("CHANGE.md"), :documentation?
+    assert_predicate fixture_blob("CHANGE.txt"), :documentation?
+    assert_predicate fixture_blob("foo/CHANGE"), :documentation?
+
+    assert_predicate fixture_blob("CHANGELOG"), :documentation?
+    assert_predicate fixture_blob("CHANGELOG.md"), :documentation?
+    assert_predicate fixture_blob("CHANGELOG.txt"), :documentation?
+    assert_predicate fixture_blob("foo/CHANGELOG"), :documentation?
+
+    assert_predicate fixture_blob("CHANGES"), :documentation?
+    assert_predicate fixture_blob("CHANGES.md"), :documentation?
+    assert_predicate fixture_blob("CHANGES.txt"), :documentation?
+    assert_predicate fixture_blob("foo/CHANGES"), :documentation?
+
+    assert_predicate fixture_blob("CONTRIBUTING"), :documentation?
+    assert_predicate fixture_blob("CONTRIBUTING.md"), :documentation?
+    assert_predicate fixture_blob("CONTRIBUTING.txt"), :documentation?
+    assert_predicate fixture_blob("foo/CONTRIBUTING"), :documentation?
+
+    assert_predicate fixture_blob("examples/some-file.pl"), :documentation?
+    assert_predicate fixture_blob("Examples/some-example-file.rb"), :documentation?
+
+    assert_predicate fixture_blob("LICENSE"), :documentation?
+    assert_predicate fixture_blob("LICENCE.md"), :documentation?
+    assert_predicate fixture_blob("License.txt"), :documentation?
+    assert_predicate fixture_blob("LICENSE.txt"), :documentation?
+    assert_predicate fixture_blob("foo/LICENSE"), :documentation?
+
+    assert_predicate fixture_blob("COPYING"), :documentation?
+    assert_predicate fixture_blob("COPYING.md"), :documentation?
+    assert_predicate fixture_blob("COPYING.txt"), :documentation?
+    assert_predicate fixture_blob("foo/COPYING"), :documentation?
+
+    assert_predicate fixture_blob("INSTALL"), :documentation?
+    assert_predicate fixture_blob("INSTALL.md"), :documentation?
+    assert_predicate fixture_blob("INSTALL.txt"), :documentation?
+    assert_predicate fixture_blob("foo/INSTALL"), :documentation?
+
+    refute_predicate fixture_blob("foo.md"), :documentation?
+
+    # Samples
+    assert sample_blob("Samples/Ruby/foo.rb").documentation?
+
+    assert_predicate fixture_blob("INSTALL.txt"), :documentation?
+  end
+
+  def test_language
+    Samples.each do |sample|
+      blob = sample_blob(sample[:path])
+      assert blob.language, "No language for #{sample[:path]}"
+      assert_equal sample[:language], blob.language.name, blob.name
+    end
+
+    # Test language detection for files which shouldn't be used as samples
+    root = File.expand_path('../fixtures', __FILE__)
+    Dir.entries(root).each do |language|
+      next if language == '.' || language == '..' || language == 'Binary' ||
+              File.basename(language) == 'ace_modes.json'
+
+      # Each directory contains test files of a language
+      dirname = File.join(root, language)
+      Dir.entries(dirname).each do |filename|
+        # By default blob search the file in the samples;
+        # thus, we need to give it the absolute path
+        filepath = File.join(dirname, filename)
+        next unless File.file?(filepath)
+
+        blob = fixture_blob(filepath)
+        if language == 'Data'
+          assert blob.language.nil?, "A language was found for #{filepath}"
+        elsif language == 'Generated'
+          assert blob.generated?, "#{filepath} is not a generated file"
+        else
+          assert blob.language, "No language for #{filepath}"
+          assert_equal language, blob.language.name, blob.name
+        end
+      end
+    end
+  end
+
+  def test_minified_files_not_safe_to_highlight
+    assert !sample_blob("JavaScript/jquery-1.6.1.min.js").safe_to_colorize?
+  end
+
+  def test_empty
+    blob = Struct.new(:data) { include Linguist::BlobHelper }
+
+    assert blob.new("").empty?
+    assert blob.new(nil).empty?
+    refute blob.new(" ").empty?
+    refute blob.new("nope").empty?
+  end
+
+  def test_include_in_language_stats
+    vendored = sample_blob("bower_components/custom/custom.js")
+    assert_predicate vendored, :vendored?
+    refute_predicate vendored, :include_in_language_stats?
+
+    documentation = fixture_blob("README")
+    assert_predicate documentation, :documentation?
+    refute_predicate documentation, :include_in_language_stats?
+
+    generated = sample_blob("CSS/bootstrap.min.css")
+    assert_predicate generated, :generated?
+    refute_predicate generated, :include_in_language_stats?
+
+    data = sample_blob("Ant Build System/filenames/ant.xml")
+    assert_equal :data, data.language.type
+    refute_predicate data, :include_in_language_stats?
+
+    prose = sample_blob("Markdown/tender.md")
+    assert_equal :prose, prose.language.type
+    refute_predicate prose, :include_in_language_stats?
+
+    included = sample_blob("HTML/pages.html")
+    assert_predicate included, :include_in_language_stats?
+  end
 end