More effective use of incremental MD5 to eliminate loading entire tarball into memory

Signed-off-by: Adam Vandenberg <flangy@gmail.com>

3 files changed, 16 insertions, 3 deletions

diff --git a/Library/Homebrew/brew.h.rb b/Library/Homebrew/brew.h.rb
index ff091fea3..78f84119f 100644
--- a/Library/Homebrew/brew.h.rb
+++ b/Library/Homebrew/brew.h.rb
@@ -52,7 +52,7 @@ def __make url, name
     if strategy == CurlDownloadStrategy
       d = strategy.new url, name, version, nil
       the_tarball = d.fetch
-      md5 = Digest::MD5.hexdigest(the_tarball.read)
+      md5 = the_tarball.md5
       puts "MD5 is #{md5}"
     else
       puts "--cache requested, but we can only cache formulas that use Curl."
diff --git a/Library/Homebrew/extend/pathname.rb b/Library/Homebrew/extend/pathname.rb
index 4416e0efe..e2489a51b 100644
--- a/Library/Homebrew/extend/pathname.rb
+++ b/Library/Homebrew/extend/pathname.rb
@@ -157,7 +157,13 @@ class Pathname
 
   def md5
     require 'digest'
-    Digest::MD5.hexdigest(File.read(self))
+    incr_md5 = Digest::MD5.new
+    self.open('r') do |f|
+      f.each_line do |line|
+        incr_md5 << line
+      end
+    end
+    incr_md5.hexdigest
   end
 
   if '1.9' <= RUBY_VERSION
diff --git a/Library/Homebrew/formula.rb b/Library/Homebrew/formula.rb
index 9f7137507..2605cd7a8 100644
--- a/Library/Homebrew/formula.rb
+++ b/Library/Homebrew/formula.rb
@@ -325,7 +325,14 @@ private
     type ||= :md5
     supplied=instance_variable_get("@#{type}")
     type=type.to_s.upcase
-    hash=Digest.const_get(type).hexdigest(fn.read)
+    hasher = Digest.const_get(type)
+
+    # Most are MD5 and it should be efficient.
+    if hasher == Digest::MD5
+      hash = fn.md5
+    else
+      hash = hasher.hexdigest(fn.read)
+    end
 
     if supplied and not supplied.empty?
       raise "#{type} mismatch\nExpected: #{hash}\nArchive: #{fn}" unless supplied.upcase == hash.upcase