Merge branch 'prevent-blob-loads'

This commit is contained in:
Aman Gupta
2011-11-23 17:26:26 -08:00
19 changed files with 83 additions and 34 deletions

View File

@@ -90,6 +90,15 @@ module Linguist
@detect_encoding ||= CharlockHolmes::EncodingDetector.new.detect(data) if data
end
# Public: Is the blob binary according to its mime type
#
# Return true or false
def binary_mime_type?
if mime_type = Mime.lookup_mime_type_for(pathname.extname)
mime_type.binary?
end
end
# Public: Is the blob binary?
#
# Return true or false
@@ -126,6 +135,22 @@ module Linguist
['.png', '.jpg', '.jpeg', '.gif'].include?(extname)
end
# Public: Is the blob a possible drupal php file?
#
# Return true or false
def drupal_extname?
['.module', '.install', '.test', '.inc'].include?(extname)
end
# Public: Is the blob likely to have a shebang?
#
# Return true or false
def shebang_extname?
extname.empty? &&
mode &&
(mode.to_i(8) & 05) == 05
end
MEGABYTE = 1024 * 1024
# Public: Is the blob too big to load?
@@ -141,7 +166,7 @@ module Linguist
#
# Return true or false
def viewable?
text? && !large?
!large? && text?
end
vendored_paths = YAML.load_file(File.expand_path("../vendor.yml", __FILE__))
@@ -359,7 +384,7 @@ module Linguist
#
# Returns a Language or nil
def guess_language
return if binary?
return if binary_mime_type?
# Disambiguate between multiple language extensions
disambiguate_extension_language ||
@@ -503,10 +528,13 @@ module Linguist
# Internal: Guess language from the first line.
#
# Look for leading "<?php"
# Look for leading "<?php" in Drupal files
#
# Returns a Language.
def first_line_language
# Only check files with drupal php extensions
return unless drupal_extname?
# Fail fast if blob isn't viewable?
return unless viewable?
@@ -573,6 +601,9 @@ module Linguist
#
# Returns the Language or nil
def shebang_language
# Skip file extensions unlikely to have shebangs
return unless shebang_extname?
if script = shebang_script
Language[script]
end

View File

@@ -32,6 +32,13 @@ module Linguist
# Returns a String
attr_reader :name
# Public: Read file permissions
#
# Returns a String like '100644'
def mode
File.stat(@path).mode.to_s(8)
end
# Public: Read file contents.
#
# Returns a String.

View File

@@ -67,6 +67,9 @@ module Linguist
return if @computed_stats
@enum.each do |blob|
# Skip binary file extensions
next if blob.binary_mime_type?
# Skip vendored or generated blobs
next if blob.vendored? || blob.generated? || blob.language.nil?

View File

@@ -57,6 +57,8 @@
- (^|/)ckeditor\.js$
- (^|/)tiny_mce([^.]*)\.js$
# MathJax
- (^|/)MathJax/
## Python ##

0
test/fixtures/script.bash vendored Normal file → Executable file
View File

0
test/fixtures/script.foo vendored Normal file → Executable file
View File

0
test/fixtures/script.groovy vendored Normal file → Executable file
View File

0
test/fixtures/script.js vendored Normal file → Executable file
View File

0
test/fixtures/script.mrb vendored Normal file → Executable file
View File

0
test/fixtures/script.nu vendored Normal file → Executable file
View File

0
test/fixtures/script.pl vendored Normal file → Executable file
View File

0
test/fixtures/script.py vendored Normal file → Executable file
View File

0
test/fixtures/script.rake vendored Normal file → Executable file
View File

0
test/fixtures/script.rb vendored Normal file → Executable file
View File

0
test/fixtures/script.rkt vendored Normal file → Executable file
View File

0
test/fixtures/script.scala vendored Normal file → Executable file
View File

0
test/fixtures/script.sh vendored Normal file → Executable file
View File

0
test/fixtures/script.zsh vendored Normal file → Executable file
View File

View File

@@ -17,6 +17,12 @@ class TestBlob < Test::Unit::TestCase
FileBlob.new(File.join(fixtures_path, name), fixtures_path)
end
def script_blob(name)
blob = blob(name)
blob.instance_variable_set(:@name, 'script')
blob
end
def test_name
assert_equal "foo.rb", blob("foo.rb").name
end
@@ -385,40 +391,40 @@ class TestBlob < Test::Unit::TestCase
end
def test_shebang_script
assert_equal 'sh', blob("script.sh").shebang_script
assert_equal 'bash', blob("script.bash").shebang_script
assert_equal 'zsh', blob("script.zsh").shebang_script
assert_equal 'perl', blob("script.pl").shebang_script
assert_equal 'ruby', blob("script.rb").shebang_script
assert_equal 'ruby', blob("script2.rb").shebang_script
assert_equal 'python', blob("script.py").shebang_script
assert_equal 'node', blob("script.js").shebang_script
assert_equal 'groovy', blob("script.groovy").shebang_script
assert_equal 'macruby', blob("script.mrb").shebang_script
assert_equal 'rake', blob("script.rake").shebang_script
assert_equal 'foo', blob("script.foo").shebang_script
assert_equal 'nush', blob("script.nu").shebang_script
assert_equal 'scala', blob("script.scala").shebang_script
assert_equal 'racket', blob("script.rkt").shebang_script
assert_equal nil, blob("foo.rb").shebang_script
assert_equal 'sh', script_blob("script.sh").shebang_script
assert_equal 'bash', script_blob("script.bash").shebang_script
assert_equal 'zsh', script_blob("script.zsh").shebang_script
assert_equal 'perl', script_blob("script.pl").shebang_script
assert_equal 'ruby', script_blob("script.rb").shebang_script
assert_equal 'ruby', script_blob("script2.rb").shebang_script
assert_equal 'python', script_blob("script.py").shebang_script
assert_equal 'node', script_blob("script.js").shebang_script
assert_equal 'groovy', script_blob("script.groovy").shebang_script
assert_equal 'macruby', script_blob("script.mrb").shebang_script
assert_equal 'rake', script_blob("script.rake").shebang_script
assert_equal 'foo', script_blob("script.foo").shebang_script
assert_equal 'nush', script_blob("script.nu").shebang_script
assert_equal 'scala', script_blob("script.scala").shebang_script
assert_equal 'racket', script_blob("script.rkt").shebang_script
assert_equal nil, script_blob("foo.rb").shebang_script
end
def test_shebang_language
assert_equal Language['Shell'], blob("script.sh").shebang_language
assert_equal Language['Shell'], blob("script.bash").shebang_language
assert_equal Language['Shell'], blob("script.zsh").shebang_language
assert_equal Language['Perl'], blob("script.pl").shebang_language
assert_equal Language['Ruby'], blob("script.rb").shebang_language
assert_equal Language['Python'], blob("script.py").shebang_language
assert_equal Language['JavaScript'], blob("script.js").shebang_language
assert_equal Language['Groovy'], blob("script.groovy").shebang_language
assert_equal Language['Ruby'], blob("script.mrb").shebang_language
assert_equal Language['Ruby'], blob("script.rake").shebang_language
assert_equal Language['Nu'], blob("script.nu").shebang_language
assert_equal Language['Scala'], blob("script.scala").shebang_language
assert_equal Language['Racket'], blob("script.rkt").shebang_language
assert_equal nil, blob("script.foo").shebang_language
assert_equal nil, blob("foo.rb").shebang_language
assert_equal Language['Shell'], script_blob("script.sh").shebang_language
assert_equal Language['Shell'], script_blob("script.bash").shebang_language
assert_equal Language['Shell'], script_blob("script.zsh").shebang_language
assert_equal Language['Perl'], script_blob("script.pl").shebang_language
assert_equal Language['Ruby'], script_blob("script.rb").shebang_language
assert_equal Language['Python'], script_blob("script.py").shebang_language
assert_equal Language['JavaScript'], script_blob("script.js").shebang_language
assert_equal Language['Groovy'], script_blob("script.groovy").shebang_language
assert_equal Language['Ruby'], script_blob("script.mrb").shebang_language
assert_equal Language['Ruby'], script_blob("script.rake").shebang_language
assert_equal Language['Nu'], script_blob("script.nu").shebang_language
assert_equal Language['Scala'], script_blob("script.scala").shebang_language
assert_equal Language['Racket'], script_blob("script.rkt").shebang_language
assert_equal nil, script_blob("script.foo").shebang_language
assert_equal nil, script_blob("foo.rb").shebang_language
end
def test_colorize