Merge pull request #24321 from Homebrew/strings-c-locale

keg_relocate: strip high bytes before matched paths

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

@@ -10,6 +10,15 @@ module OS
 
       requires_ancestor { ::Keg }
 
+      module ClassMethods
+        # GNU `strings` splits runs at non-ASCII bytes, such as those in a
+        # prefix like `/Users/José`, unless asked to keep 8-bit bytes.
+        sig { returns(T::Array[String]) }
+        def strings_encoding_args
+          ["-e", "S"]
+        end
+      end
+
       sig {
         params(relocation: ::Keg::Relocation, with_placeholders: T::Boolean,
                files: T.nilable(T::Array[::Pathname])).returns(T::Array[::Pathname])
@@ -126,4 +135,5 @@ module OS
   end
 end
 
+Keg.singleton_class.prepend(OS::Linux::Keg::ClassMethods)
 Keg.prepend(OS::Linux::Keg)

Library/Homebrew/keg_relocate.rb

@@ -614,7 +614,8 @@ class Keg
       # Some binaries contain strings with lists of files
       # e.g. `/usr/local/lib/foo:/usr/local/share/foo:/usr/lib/foo`
       # Each item in the list should be checked separately
-      match.split(":").each do |sub_match|
+      # A high byte just before a path can be reported as part of the same run.
+      match.split(":").map { it.b.sub(%r{\A[\x80-\xFF]+(?=/)}n, "") }.each do |sub_match|
         # Not all items in the list may be matches
         next unless sub_match.match? path_regex
         next if linked_libraries.include? sub_match # Don't bother reporting a string if it was found by otool
@@ -648,7 +649,7 @@ class Keg
       return
     end
 
-    Utils.popen_read("strings", "-t", "x", "-", file.to_s) do |io|
+    Utils.popen_read("strings", *strings_encoding_args, "-t", "x", "-", file.to_s) do |io|
       until io.eof?
         line = io.readline.chomp
         offset, match = line.split(" ", 2)
@@ -823,6 +824,11 @@ class Keg
     []
   end
 
+  sig { returns(T::Array[String]) }
+  def self.strings_encoding_args
+    []
+  end
+
   private
 
   sig { returns(T::Boolean) }

Library/Homebrew/test/keg_relocate/text_spec.rb

@@ -48,6 +48,22 @@ RSpec.describe Keg do
     expect(result.count).to eq 2
   end
 
+  specify "::text_matches_in_file finds a path preceded by a non-ASCII byte" do
+    ENV["LC_ALL"] = "en_US.UTF-8"
+    file.binwrite("\0\xD6#{dir}/file.txt\0".b)
+
+    expect(described_class.text_matches_in_file(file, dir.to_s, [], [], nil).count).to eq 1
+  end
+
+  specify "::text_matches_in_file finds a path containing non-ASCII characters",
+          unless: ENV.fetch("HOMEBREW_TEST_GENERIC_OS", nil) do
+    ENV["LC_ALL"] = "en_US.UTF-8"
+    (dir/"José").mkpath
+    file.binwrite("\0#{dir}/José\0".b)
+
+    expect(described_class.text_matches_in_file(file, dir.to_s, [], [], nil).count).to eq 1
+  end
+
   specify "::each_candidate_string still yields runs that are not valid UTF-8" do
     setup_file