Merge pull request #24349 from Homebrew/elf-utf8-relocation-paths

Preserve UTF-8 paths in ELF loader strings

Library/Homebrew/keg_relocate.rb

@@ -743,9 +743,10 @@ class Keg
       end
 
       strings.filter_map do |offset, match|
-        next unless match.ascii_only?
+        match.force_encoding(Encoding::UTF_8)
+        next unless match.valid_encoding?
 
-        [offset.to_s(16), match.force_encoding(Encoding::UTF_8)]
+        [offset.to_s(16), match]
       end
     ensure
       stream.close

Library/Homebrew/test/dev-cmd/bottle_spec.rb

@@ -111,6 +111,14 @@ RSpec.describe Homebrew::DevCmd::Bottle do
     (formula.libexec/"raw-prefix").binwrite(
       "\0#{Array.new(Homebrew::DevCmd::Bottle::MAXIMUM_STRING_MATCHES + 1, formula.libexec.to_s).join("\0")}\0",
     )
+    require "patchelf"
+    library = formula.libexec/"José/libhello.so.0"
+    library.dirname.mkpath
+    FileUtils.cp TEST_FIXTURE_DIR/"elf/libhello.so.0", library
+    FileUtils.cp TEST_FIXTURE_DIR/"elf/hello", formula.libexec/"elf-prefix"
+    patcher = PatchELF::Patcher.new((formula.libexec/"elf-prefix").to_s, on_error: :silent)
+    patcher.add_needed(library.to_s.b)
+    patcher.save
     stderr = if !Sandbox.available?
       "Warning: Sandbox unavailable: processing downloaded files without sandboxing!\n"
     elsif Sandbox.nested_sandbox?
@@ -140,7 +148,7 @@ RSpec.describe Homebrew::DevCmd::Bottle do
       expect(tag.fetch("tab")).to include(
         "changed_files"           => be_an(Array),
         "linkage_files"           => be_an(Array),
-        "binary_relocation_files" => include("libexec/raw-prefix"),
+        "binary_relocation_files" => include("libexec/raw-prefix", "libexec/elf-prefix"),
         "built_prefix"            => HOMEBREW_PREFIX.to_s,
       )
       binary_relocation_diagnostics = tag.fetch("binary_relocation_diagnostics")

Library/Homebrew/test/keg_relocate/elf_checker_spec.rb

@@ -34,6 +34,24 @@ RSpec.describe Keg do
       expect(described_class.text_matches_in_file(file, dir.to_s, [], [], nil).size).to eq 1
     end
 
+    it "keeps referenced paths containing non-ASCII characters" do
+      rpath = dir/"José"
+      rpath.mkpath
+      patch_rpath rpath.to_s
+
+      expect(described_class.text_matches_in_file(file, dir.to_s, [], [], nil).map(&:first)).to eq [rpath.to_s]
+    end
+
+    it "drops referenced paths containing invalid UTF-8" do
+      rpath = "#{dir}/invalid-\xFF".b
+      patch_rpath rpath
+
+      candidates = []
+      described_class.each_candidate_string(file, dir.to_s) { |candidate| candidates << candidate.last.b }
+
+      expect(candidates).not_to include(rpath)
+    end
+
     it "drops dead loader strings nothing references any more" do
       # Shrinking the RPATH leaves an in-place `X` run inside `.dynstr`;
       # writing a prefix string over it recreates the dead bytes a moved or
@@ -96,6 +114,15 @@ RSpec.describe Keg do
       expect(described_class.text_matches_in_file(file, dir.to_s, [], [], nil).size).to eq 1
     end
 
+    it "keeps an interpreter containing non-ASCII characters" do
+      interpreter = dir/"ld-José.so"
+      FileUtils.touch interpreter
+      patch_rpath "#{Keg::PREFIX_PLACEHOLDER}/lib"
+      patch_interpreter interpreter.to_s.b
+
+      expect(described_class.text_matches_in_file(file, dir.to_s, [], [], nil).map(&:first)).to eq [interpreter.to_s]
+    end
+
     it "drops old interpreter bytes the loader no longer uses" do
       patch_rpath "#{Keg::PREFIX_PLACEHOLDER}/lib"
       # Growing the interpreter moves it; shrinking it back leaves an