Merge pull request #12890 from danielnachun/add_binary_grep

Add binary grep method to keg_relocate

Library/Homebrew/extend/os/mac/keg_relocate.rb

@@ -183,4 +183,10 @@ class Keg
     # it's wrong. -O is a BSD-grep-only option.
     "-lrO"
   end
+
+  def egrep_args
+    grep_bin = "egrep"
+    grep_args = recursive_fgrep_args
+    [grep_bin, grep_args]
+  end
 end

Library/Homebrew/keg_relocate.rb

@@ -8,6 +8,8 @@ class Keg
   LIBRARY_PLACEHOLDER = "@@HOMEBREW_LIBRARY@@"
   PERL_PLACEHOLDER = "@@HOMEBREW_PERL@@"
   JAVA_PLACEHOLDER = "@@HOMEBREW_JAVA@@"
+  NULL_BYTE = "\x00"
+  NULL_BYTE_STRING = "\\x00"
 
   class Relocation
     extend T::Sig
@@ -173,16 +175,42 @@ class Keg
   end
   alias generic_recursive_fgrep_args recursive_fgrep_args
 
-  def each_unique_file_matching(string)
+  def egrep_args
+    grep_bin = "grep"
+    grep_args = recursive_fgrep_args
+    grep_args += "Pa"
+    [grep_bin, grep_args]
+  end
+  alias generic_egrep_args egrep_args
+
+  def each_unique_file(io, block)
+    hardlinks = Set.new
+
+    until io.eof?
+      file = Pathname.new(io.readline.chomp)
+      # Don't return symbolic links.
+      next if file.symlink?
+
+      # To avoid returning hardlinks, only return files with unique inodes.
+      # Hardlinks will have the same inode as the file they point to.
+      block.call file if hardlinks.add? file.stat.ino
+    end
+  end
+
+  def each_unique_file_matching(string, &block)
     Utils.popen_read("fgrep", recursive_fgrep_args, string, to_s) do |io|
-      hardlinks = Set.new
+      each_unique_file(io, block)
+    end
+  end
 
-      until io.eof?
+  def each_unique_binary_file(&block)
-        file = Pathname.new(io.readline.chomp)
+    grep_bin, grep_args = egrep_args
-        next if file.symlink?
 
-        yield file if hardlinks.add? file.stat.ino
+    # We need to pass NULL_BYTE_STRING, the literal string "\x00", to grep
-      end
+    # rather than NULL_BYTE, a literal null byte, because grep will internally
+    # convert the literal string "\x00" to a null byte.
+    Utils.popen_read(grep_bin, grep_args, NULL_BYTE_STRING, to_s) do |io|
+      each_unique_file(io, block)
     end
   end
 

Library/Homebrew/test/keg_relocate/grep_spec.rb

@@ -0,0 +1,58 @@
+# typed: false
+# frozen_string_literal: true
+
+require "keg_relocate"
+
+describe Keg do
+  subject(:keg) { described_class.new(HOMEBREW_CELLAR/"foo/1.0.0") }
+
+  let(:dir) { HOMEBREW_CELLAR/"foo/1.0.0" }
+  let(:text_file) { dir/"file.txt" }
+  let(:binary_file) { dir/"file.bin" }
+
+  before do
+    dir.mkpath
+  end
+
+  def setup_text_file
+    text_file.atomic_write <<~EOS
+      #{dir}/file.txt
+      /foo#{dir}/file.txt
+      foo/bar:#{dir}/file.txt
+      foo/bar:/foo#{dir}/file.txt
+      #{dir}/bar.txt:#{dir}/baz.txt
+    EOS
+  end
+
+  def setup_binary_file
+    binary_file.atomic_write <<~EOS
+      \x00
+    EOS
+  end
+
+  describe "#each_unique_file_matching" do
+    specify "find string matches to path" do
+      setup_text_file
+
+      string_matches = Set.new
+      keg.each_unique_file_matching(dir) do |file|
+        string_matches << file
+      end
+
+      expect(string_matches.size).to eq 1
+    end
+  end
+
+  describe "#each_unique_binary_file" do
+    specify "find null bytes in binaries" do
+      setup_binary_file
+
+      binary_matches = Set.new
+      keg.each_unique_binary_file do |file|
+        binary_matches << file
+      end
+
+      expect(binary_matches.size).to eq 1
+    end
+  end
+end