Pathname: add Mach-O module

The MachO module contains methods for learning about Mach-O binaries,
and can be used where one might normally shell out to file(1).

Signed-off-by: Jack Nagel <jacknagel@gmail.com>

7 files changed, 240 insertions, 0 deletions

diff --git a/Library/Homebrew/extend/pathname.rb b/Library/Homebrew/extend/pathname.rb
index 98618be41..bd0e6154e 100644
--- a/Library/Homebrew/extend/pathname.rb
+++ b/Library/Homebrew/extend/pathname.rb
@@ -1,8 +1,11 @@
 require 'pathname'
 require 'bottles'
+require 'mach'
 
 # we enhance pathname to make our code more readable
 class Pathname
+  include MachO
+
   def install *sources
     results = []
     sources.each do |src|
@@ -268,6 +271,10 @@ class Pathname
     end
   end
 
+  def text_executable?
+    %r[#!\s*(/.+)+] === open('r') { |f| f.readline }
+  end
+
   def incremental_hash(hasher)
     incr_hash = hasher.new
     self.open('r') do |f|
diff --git a/Library/Homebrew/mach.rb b/Library/Homebrew/mach.rb
new file mode 100644
index 000000000..e24cc1bc5
--- /dev/null
+++ b/Library/Homebrew/mach.rb
@@ -0,0 +1,93 @@
+module MachO
+  # Mach-O binary methods, see:
+  # /usr/include/mach-o/loader.h
+  # /usr/include/mach-o/fat.h
+
+  def mach_data
+    @mach_data ||= begin
+      offsets = []
+      mach_data = []
+
+      header = read(8).unpack("N2")
+      case header[0]
+      when 0xcafebabe # universal
+        header[1].times do |i|
+          # header[1] is the number of struct fat_arch in the file.
+          # Each struct fat_arch is 20 bytes, and the 'offset' member
+          # begins 8 bytes into the struct, with an additional 8 byte
+          # offset due to the struct fat_header at the beginning of
+          # the file.
+          offsets << read(4, 20*i + 16).unpack("N")[0]
+        end
+      when 0xcefaedfe, 0xcffaedfe, 0xfeedface, 0xfeedfacf # Single arch
+        offsets << 0
+      else
+        raise "Not a Mach-O binary."
+      end
+
+      offsets.each do |offset|
+        arch = case read(8, offset).unpack("N2")
+          when [0xcefaedfe, 0x07000000] then :i386
+          when [0xcffaedfe, 0x07000001] then :x86_64
+          when [0xfeedface, 0x00000012] then :ppc7400
+          when [0xfeedfacf, 0x01000012] then :ppc64
+          else :dunno
+          end
+
+        type = case read(4, offset + 12).unpack("N")[0]
+         when 0x00000002, 0x02000000 then :executable
+         when 0x00000006, 0x06000000 then :dylib
+         else :dunno
+         end
+
+        mach_data << { :arch => arch, :type => type }
+      end
+      mach_data
+    rescue
+      # read() will raise if it sees EOF, which should only happen if the
+      # file is < 8 bytes. Otherwise, we raise if the file is not a Mach-O 
+      # binary. In both cases, we want to return an empty array.
+      []
+    end
+  end
+
+  def archs
+    mach_data.map{ |m| m.fetch :arch }
+  end
+
+  def arch
+    case archs.length
+    when 0 then :dunno
+    when 1 then archs.first
+    else :universal
+    end
+  end
+
+  def universal?
+    arch == :universal
+  end
+
+  def i386?
+    arch == :i386
+  end
+
+  def x86_64?
+    arch == :x86_64
+  end
+
+  def ppc7400?
+    arch == :ppc7400
+  end
+
+  def ppc64?
+    arch == :ppc64
+  end
+
+  def dylib?
+    mach_data.map{ |m| m.fetch :type }.include? :dylib
+  end
+
+  def mach_o_executable?
+    mach_data.map{ |m| m.fetch :type }.include? :executable
+  end
+end
diff --git a/Library/Homebrew/test/mach/a.out b/Library/Homebrew/test/mach/a.out
new file mode 100755
index 000000000..18e0e982f
--- /dev/null
+++ b/Library/Homebrew/test/mach/a.out
diff --git a/Library/Homebrew/test/mach/fat.dylib b/Library/Homebrew/test/mach/fat.dylib
new file mode 100644
index 000000000..6886a6a8b
--- /dev/null
+++ b/Library/Homebrew/test/mach/fat.dylib
diff --git a/Library/Homebrew/test/mach/i386.dylib b/Library/Homebrew/test/mach/i386.dylib
new file mode 100644
index 000000000..0304dca01
--- /dev/null
+++ b/Library/Homebrew/test/mach/i386.dylib
diff --git a/Library/Homebrew/test/mach/x86_64.dylib b/Library/Homebrew/test/mach/x86_64.dylib
new file mode 100644
index 000000000..781b41cb4
--- /dev/null
+++ b/Library/Homebrew/test/mach/x86_64.dylib
diff --git a/Library/Homebrew/test/test_mach.rb b/Library/Homebrew/test/test_mach.rb
new file mode 100644
index 000000000..2da4e429a
--- /dev/null
+++ b/Library/Homebrew/test/test_mach.rb
@@ -0,0 +1,140 @@
+require 'testing_env'
+
+require 'extend/ARGV' # needs to be after test/unit to avoid conflict with OptionsParser
+ARGV.extend(HomebrewArgvExtension)
+
+class MachOPathnameTests < Test::Unit::TestCase
+  def test_fat_dylib
+    pn = Pathname.new("#{TEST_FOLDER}/mach/fat.dylib")
+    assert pn.universal?
+    assert !pn.i386?
+    assert !pn.x86_64?
+    assert !pn.ppc7400?
+    assert !pn.ppc64?
+    assert pn.dylib?
+    assert !pn.mach_o_executable?
+    assert !pn.text_executable?
+    assert pn.arch == :universal
+    assert_match /Mach-O (64-bit )?dynamically linked shared library/,
+      `/usr/bin/file -h '#{pn}'`.chomp
+  end
+
+  def test_i386_dylib
+    pn = Pathname.new("#{TEST_FOLDER}/mach/i386.dylib")
+    assert !pn.universal?
+    assert pn.i386?
+    assert !pn.x86_64?
+    assert !pn.ppc7400?
+    assert !pn.ppc64?
+    assert pn.dylib?
+    assert !pn.mach_o_executable?
+    assert !pn.text_executable?
+    assert_match /Mach-O (64-bit )?dynamically linked shared library/,
+      `/usr/bin/file -h '#{pn}'`.chomp
+  end
+
+  def test_x86_64_dylib
+    pn = Pathname.new("#{TEST_FOLDER}/mach/x86_64.dylib")
+    assert !pn.universal?
+    assert !pn.i386?
+    assert pn.x86_64?
+    assert !pn.ppc7400?
+    assert !pn.ppc64?
+    assert pn.dylib?
+    assert !pn.mach_o_executable?
+    assert !pn.text_executable?
+    assert_match /Mach-O (64-bit )?dynamically linked shared library/,
+      `/usr/bin/file -h '#{pn}'`.chomp
+  end
+
+  def test_mach_o_executable
+    pn = Pathname.new("#{TEST_FOLDER}/mach/a.out")
+    assert pn.universal?
+    assert !pn.i386?
+    assert !pn.x86_64?
+    assert !pn.ppc7400?
+    assert !pn.ppc64?
+    assert !pn.dylib?
+    assert pn.mach_o_executable?
+    assert !pn.text_executable?
+    assert_match /Mach-O (64-bit )?executable/,
+      `/usr/bin/file -h '#{pn}'`.chomp
+  end
+
+  def test_non_mach_o
+    pn = Pathname.new("#{TEST_FOLDER}/tarballs/testball-0.1.tbz")
+    assert !pn.universal?
+    assert !pn.i386?
+    assert !pn.x86_64?
+    assert !pn.ppc7400?
+    assert !pn.ppc64?
+    assert !pn.dylib?
+    assert !pn.mach_o_executable?
+    assert !pn.text_executable?
+    assert pn.arch == :dunno
+    assert_no_match /Mach-O (64-bit )?dynamically linked shared library/,
+      `/usr/bin/file -h '#{pn}'`.chomp
+    assert_no_match /Mach-O [^ ]* ?executable/,
+      `/usr/bin/file -h '#{pn}'`.chomp
+  end
+end
+
+class TextExecutableTests < Test::Unit::TestCase
+  TMPDIR = HOMEBREW_PREFIX/'tmp'
+
+  def setup
+    FileUtils.mkdir_p TMPDIR
+  end
+
+  def test_simple_shebang
+    pn = Pathname.new('foo')
+    pn.write '#!/bin/sh'
+    assert !pn.universal?
+    assert !pn.i386?
+    assert !pn.x86_64?
+    assert !pn.ppc7400?
+    assert !pn.ppc64?
+    assert !pn.dylib?
+    assert !pn.mach_o_executable?
+    assert pn.text_executable?
+    assert_equal [], pn.archs
+    assert pn.arch == :dunno
+    assert_match /text executable/, `/usr/bin/file -h '#{pn}'`.chomp
+  end
+
+  def test_shebang_with_options
+    pn = Pathname.new('bar')
+    pn.write '#! /usr/bin/perl -w'
+    assert !pn.universal?
+    assert !pn.i386?
+    assert !pn.x86_64?
+    assert !pn.ppc7400?
+    assert !pn.ppc64?
+    assert !pn.dylib?
+    assert !pn.mach_o_executable?
+    assert pn.text_executable?
+    assert_equal [], pn.archs
+    assert pn.arch == :dunno
+    assert_match /text executable/, `/usr/bin/file -h '#{pn}'`.chomp
+  end
+
+  def test_malformed_shebang
+    pn = Pathname.new('baz')
+    pn.write '#! '
+    assert !pn.universal?
+    assert !pn.i386?
+    assert !pn.x86_64?
+    assert !pn.ppc7400?
+    assert !pn.ppc64?
+    assert !pn.dylib?
+    assert !pn.mach_o_executable?
+    assert !pn.text_executable?
+    assert_equal [], pn.archs
+    assert pn.arch == :dunno
+    assert_match /text executable/, `/usr/bin/file -h '#{pn}'`.chomp
+  end
+
+  def teardown
+    TMPDIR.rmtree
+  end
+end