re2 2.20.0-arm64-darwin → 2.22.0-arm64-darwin

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: ea87fb563c30fc2263189b72f3e50a73047c81d5dbda36806b1fb52e841e02d3
-  data.tar.gz: b1ff51e0db7548b4da42958aba50e398d7f8d3ae35c4df581920dbe93e90d18b
+  metadata.gz: a3c9a376891ca658c808ff093720cd2f78872dc03247a5fd65093982136a90a5
+  data.tar.gz: 5b0c7326f37ba5f02be27c0215b986fc007d832e1ad1d313a07f2c495d90e186
 SHA512:
-  metadata.gz: ae207fdcb0f368bfa2d290df3293fbb58dd296b6eb8d6f70ac19b679ed97b0e1261fcb3d87e4ecc3518f1d3edbe0d91cdbee8540b7180de82f499be3f1888ed6
-  data.tar.gz: 79c24308c834d407d45d3bd21964b4c3a745b6c34969cf650c7b84ff3ac6021435aed8f24ec1757cee299141b0bcf2b0f6ce3146b526f5978c2ac1fc70618bb6
+  metadata.gz: a406d71e277f509396eaec77798e0837b6f9af5d660c432d4ff367a5a42c2af96891feac35304573aaa94a195361f9fa9137621af79eaeb6e2e56beaf6c31512
+  data.tar.gz: 205f3f81aa5fad63990bc21a0d8e7216c837faf4668370652887ec48d22941cdca66698bf3ad8c21d389a665d57f1e2f1507a752216031df1d3efefa9f1f3585

data/README.md

@@ -6,8 +6,8 @@ Python".
 
 [![Build Status](https://github.com/mudge/re2/actions/workflows/tests.yml/badge.svg?branch=main)](https://github.com/mudge/re2/actions)
 
-**Current version:** 2.20.0  
-**Bundled RE2 version:** libre2.11 (2025-08-12)  
+**Current version:** 2.22.0  
+**Bundled RE2 version:** libre2.11 (2025-11-05)  
 
 ```ruby
 RE2('h.*o').full_match?("hello")    #=> true
@@ -264,7 +264,7 @@ This gem requires the following to run:
 
 It supports the following RE2 ABI versions:
 
-* libre2.0 (prior to release 2020-03-02) to libre2.11 (2023-07-01 to 2025-08-12)
+* libre2.0 (prior to release 2020-03-02) to libre2.11 (2023-07-01 to 2025-11-05)
 
 ### Native gems
 
@@ -378,6 +378,8 @@ Alternatively, you can set the `RE2_USE_SYSTEM_LIBRARIES` environment variable i
   improvements in 2.4.0.
 * Thanks to [Manuel Jacob](https://github.com/manueljacob) for reporting a bug
   when passing strings with null bytes.
+* Thanks to [Maciej Gajewski](https://github.com/konieczkow) for helping
+  confirm issues with GC compaction and mutable strings.
 
 ## Contact
 

data/Rakefile

@@ -76,7 +76,7 @@ namespace :gem do
         sudo apt-get install -y cmake=3.22.2-0kitware1ubuntu20.04.1 cmake-data=3.22.2-0kitware1ubuntu20.04.1 &&
         rbenv shell 3.1.6 &&
         gem install bundler --no-document &&
-        bundle &&
+        bundle install &&
         bundle exec rake native:#{platform} pkg/#{re2_gemspec.full_name}-#{Gem::Platform.new(platform)}.gem PATH="/usr/local/bin:$PATH"
       SCRIPT
     end

data/dependencies.yml

@@ -1,7 +1,7 @@
 ---
 libre2:
-  version: '2025-08-12'
-  sha256: 2f3bec634c3e51ea1faf0d441e0a8718b73ef758d7020175ed7e352df3f6ae12
+  version: '2025-11-05'
+  sha256: 87f6029d2f6de8aa023654240a03ada90e876ce9a4676e258dd01ea4c26ffd67
 abseil:
   version: '20250814.1'
   sha256: 1692f77d1739bacf3f94337188b78583cf09bab7e420d2dc6c5605a4f86785a1

data/ext/re2/re2.cc

@@ -128,13 +128,12 @@ static void parse_re2_options(RE2::Options* re2_options, const VALUE options) {
 static void re2_matchdata_mark(void *ptr) {
   re2_matchdata *m = reinterpret_cast<re2_matchdata *>(ptr);
   rb_gc_mark_movable(m->regexp);
-  rb_gc_mark_movable(m->text);
+  rb_gc_mark(m->text);
 }
 
 static void re2_matchdata_compact(void *ptr) {
   re2_matchdata *m = reinterpret_cast<re2_matchdata *>(ptr);
   m->regexp = rb_gc_location(m->regexp);
-  m->text = rb_gc_location(m->text);
 }
 
 static void re2_matchdata_free(void *ptr) {
@@ -173,13 +172,12 @@ static const rb_data_type_t re2_matchdata_data_type = {
 static void re2_scanner_mark(void *ptr) {
   re2_scanner *s = reinterpret_cast<re2_scanner *>(ptr);
   rb_gc_mark_movable(s->regexp);
-  rb_gc_mark_movable(s->text);
+  rb_gc_mark(s->text);
 }
 
 static void re2_scanner_compact(void *ptr) {
   re2_scanner *s = reinterpret_cast<re2_scanner *>(ptr);
   s->regexp = rb_gc_location(s->regexp);
-  s->text = rb_gc_location(s->text);
 }
 
 static void re2_scanner_free(void *ptr) {
@@ -278,10 +276,12 @@ static VALUE re2_matchdata_string(const VALUE self) {
 }
 
 /*
- * Returns the text supplied when incrementally matching with
+ * Returns a frozen copy of the text supplied when incrementally matching with
  * {RE2::Regexp#scan}.
  *
- * @return [String] the original string passed to {RE2::Regexp#scan}
+ * If the text was already a frozen string, returns the original.
+ *
+ * @return [String] a frozen string with the text passed to {RE2::Regexp#scan}
  * @example
  *   c = RE2::Regexp.new('(\d+)').scan("foo")
  *   c.string #=> "foo"
@@ -326,6 +326,11 @@ static VALUE re2_scanner_rewind(VALUE self) {
   delete c->input;
   c->input = new(std::nothrow) re2::StringPiece(
       RSTRING_PTR(c->text), RSTRING_LEN(c->text));
+  if (c->input == 0) {
+    rb_raise(rb_eNoMemError,
+             "not enough memory to allocate StringPiece for input");
+  }
+
   c->eof = false;
 
   return self;
@@ -412,8 +417,8 @@ static re2::StringPiece *re2_matchdata_find_match(VALUE idx, const VALUE self) {
 
   int id;
 
-  if (FIXNUM_P(idx)) {
-    id = FIX2INT(idx);
+  if (RB_INTEGER_TYPE_P(idx)) {
+    id = NUM2INT(idx);
   } else if (SYMBOL_P(idx)) {
     const std::map<std::string, int>& groups = p->pattern->NamedCapturingGroups();
     std::map<std::string, int>::const_iterator search = groups.find(rb_id2name(SYM2ID(idx)));
@@ -681,10 +686,10 @@ static VALUE re2_matchdata_aref(int argc, VALUE *argv, const VALUE self) {
         std::string(RSTRING_PTR(idx), RSTRING_LEN(idx)), self);
   } else if (SYMBOL_P(idx)) {
     return re2_matchdata_named_match(rb_id2name(SYM2ID(idx)), self);
-  } else if (!NIL_P(rest) || !FIXNUM_P(idx) || FIX2INT(idx) < 0) {
+  } else if (!NIL_P(rest) || !RB_INTEGER_TYPE_P(idx) || NUM2INT(idx) < 0) {
     return rb_ary_aref(argc, argv, re2_matchdata_to_a(self));
   } else {
-    return re2_matchdata_nth_match(FIX2INT(idx), self);
+    return re2_matchdata_nth_match(NUM2INT(idx), self);
   }
 }
 
@@ -1421,7 +1426,7 @@ static VALUE re2_regexp_match(int argc, VALUE *argv, const VALUE self) {
   RE2::Anchor anchor = RE2::UNANCHORED;
 
   if (RTEST(options)) {
-    if (FIXNUM_P(options)) {
+    if (RB_INTEGER_TYPE_P(options)) {
       n = NUM2INT(options);
 
       if (n < 0) {
@@ -1435,8 +1440,6 @@ static VALUE re2_regexp_match(int argc, VALUE *argv, const VALUE self) {
       VALUE endpos_option = rb_hash_aref(options, ID2SYM(id_endpos));
       if (!NIL_P(endpos_option)) {
 #ifdef HAVE_ENDPOS_ARGUMENT
-        Check_Type(endpos_option, T_FIXNUM);
-
         endpos = NUM2INT(endpos_option);
 
         if (endpos < 0) {
@@ -1465,8 +1468,6 @@ static VALUE re2_regexp_match(int argc, VALUE *argv, const VALUE self) {
 
       VALUE submatches_option = rb_hash_aref(options, ID2SYM(id_submatches));
       if (!NIL_P(submatches_option)) {
-        Check_Type(submatches_option, T_FIXNUM);
-
         n = NUM2INT(submatches_option);
 
         if (n < 0) {
@@ -1482,8 +1483,6 @@ static VALUE re2_regexp_match(int argc, VALUE *argv, const VALUE self) {
 
       VALUE startpos_option = rb_hash_aref(options, ID2SYM(id_startpos));
       if (!NIL_P(startpos_option)) {
-        Check_Type(startpos_option, T_FIXNUM);
-
         startpos = NUM2INT(startpos_option);
 
         if (startpos < 0) {
@@ -1515,37 +1514,43 @@ static VALUE re2_regexp_match(int argc, VALUE *argv, const VALUE self) {
 #endif
     return BOOL2RUBY(matched);
   } else {
+    if (n == INT_MAX) {
+      rb_raise(rb_eRangeError, "number of matches should be < %d", INT_MAX);
+    }
+
     /* Because match returns the whole match as well. */
     n += 1;
 
-    VALUE matchdata = rb_class_new_instance(0, 0, re2_cMatchData);
-    TypedData_Get_Struct(matchdata, re2_matchdata, &re2_matchdata_data_type, m);
-    m->matches = new(std::nothrow) re2::StringPiece[n];
-    RB_OBJ_WRITE(matchdata, &m->regexp, self);
-    if (!RTEST(rb_obj_frozen_p(text))) {
-      text = rb_str_freeze(rb_str_dup(text));
-    }
-    RB_OBJ_WRITE(matchdata, &m->text, text);
-
-    if (m->matches == 0) {
+    re2::StringPiece *matches = new(std::nothrow) re2::StringPiece[n];
+    if (matches == 0) {
       rb_raise(rb_eNoMemError,
                "not enough memory to allocate StringPieces for matches");
     }
 
-    m->number_of_matches = n;
+    text = rb_str_new_frozen(text);
 
 #ifdef HAVE_ENDPOS_ARGUMENT
     bool matched = p->pattern->Match(
-        re2::StringPiece(RSTRING_PTR(m->text), RSTRING_LEN(m->text)),
-        startpos, endpos, anchor, m->matches, n);
+        re2::StringPiece(RSTRING_PTR(text), RSTRING_LEN(text)),
+        startpos, endpos, anchor, matches, n);
 #else
     bool matched = p->pattern->Match(
-        re2::StringPiece(RSTRING_PTR(m->text), RSTRING_LEN(m->text)),
-        startpos, anchor, m->matches, n);
+        re2::StringPiece(RSTRING_PTR(text), RSTRING_LEN(text)),
+        startpos, anchor, matches, n);
 #endif
     if (matched) {
+      VALUE matchdata = rb_class_new_instance(0, 0, re2_cMatchData);
+      TypedData_Get_Struct(matchdata, re2_matchdata, &re2_matchdata_data_type, m);
+
+      RB_OBJ_WRITE(matchdata, &m->regexp, self);
+      RB_OBJ_WRITE(matchdata, &m->text, text);
+      m->matches = matches;
+      m->number_of_matches = n;
+
       return matchdata;
     } else {
+      delete[] matches;
+
       return Qnil;
     }
   }
@@ -1614,10 +1619,14 @@ static VALUE re2_regexp_scan(const VALUE self, VALUE text) {
   VALUE scanner = rb_class_new_instance(0, 0, re2_cScanner);
   TypedData_Get_Struct(scanner, re2_scanner, &re2_scanner_data_type, c);
 
-  c->input = new(std::nothrow) re2::StringPiece(
-      RSTRING_PTR(text), RSTRING_LEN(text));
   RB_OBJ_WRITE(scanner, &c->regexp, self);
-  RB_OBJ_WRITE(scanner, &c->text, text);
+  RB_OBJ_WRITE(scanner, &c->text, rb_str_new_frozen(text));
+  c->input = new(std::nothrow) re2::StringPiece(
+      RSTRING_PTR(c->text), RSTRING_LEN(c->text));
+  if (c->input == 0) {
+    rb_raise(rb_eNoMemError,
+             "not enough memory to allocate StringPiece for input");
+  }
 
   if (p->pattern->ok()) {
     c->number_of_capturing_groups = p->pattern->NumberOfCapturingGroups();
@@ -1913,21 +1922,19 @@ static VALUE re2_set_add(VALUE self, VALUE pattern) {
   re2_set *s;
   TypedData_Get_Struct(self, re2_set, &re2_set_data_type, s);
 
-  /* To prevent the memory of the err string leaking when we call rb_raise,
-   * take a copy of it and let it go out of scope.
-   */
-  char msg[100];
   int index;
+  VALUE msg;
 
   {
     std::string err;
     index = s->set->Add(
         re2::StringPiece(RSTRING_PTR(pattern), RSTRING_LEN(pattern)), &err);
-    strlcpy(msg, err.c_str(), sizeof(msg));
+    msg = rb_str_new(err.data(), err.size());
   }
 
   if (index < 0) {
-    rb_raise(rb_eArgError, "str rejected by RE2::Set->Add(): %s", msg);
+    rb_raise(rb_eArgError,
+             "str rejected by RE2::Set->Add(): %s", RSTRING_PTR(msg));
   }
 
   return INT2FIX(index);

data/lib/re2/version.rb

@@ -10,5 +10,5 @@
 
 
 module RE2
-  VERSION = "2.20.0"
+  VERSION = "2.22.0"
 end

data/spec/re2/match_data_spec.rb

@@ -133,6 +133,13 @@ RSpec.describe RE2::MatchData do
       expect(md["name"].encoding.name).to eq("ISO-8859-1")
       expect(md[:name].encoding.name).to eq("ISO-8859-1")
     end
+
+    it "supports GC compaction" do
+      md = RE2::Regexp.new('(wo{2})').match('woohoo' * 5)
+      GC.compact
+
+      expect(md[1]).to eq("woo")
+    end
   end
 
   describe "#string" do
@@ -287,6 +294,13 @@ RSpec.describe RE2::MatchData do
 
       expect { md.begin(nil) }.to raise_error(TypeError)
     end
+
+    it "supports GC compaction" do
+      md = RE2::Regexp.new('(wo{2})').match('woohoo' * 5)
+      GC.compact
+
+      expect(md.string[md.begin(0)..-1]).to eq('woohoo' * 5)
+    end
   end
 
   describe "#end" do
@@ -349,6 +363,13 @@ RSpec.describe RE2::MatchData do
 
       expect { md.end(nil) }.to raise_error(TypeError)
     end
+
+    it "supports GC compaction" do
+      md = RE2::Regexp.new('(wo{2})').match('woohoo' * 5)
+      GC.compact
+
+      expect(md.string[0...md.end(0)]).to eq('woo')
+    end
   end
 
   describe "#deconstruct" do

data/spec/re2/regexp_spec.rb

@@ -1,6 +1,10 @@
 # frozen_string_literal: true
 
+require "rbconfig/sizeof"
+
 RSpec.describe RE2::Regexp do
+  INT_MAX = 2**(RbConfig::SIZEOF.fetch("int") * 8 - 1) - 1
+
   describe "#initialize" do
     it "returns an instance given only a pattern" do
       re = RE2::Regexp.new('woo')
@@ -566,6 +570,12 @@ RSpec.describe RE2::Regexp do
       expect { re.match("one two three", submatches: :invalid) }.to raise_error(TypeError)
     end
 
+    it "raises an exception when given too large a number of submatches" do
+      re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
+
+      expect { re.match("one two three", submatches: INT_MAX) }.to raise_error(RangeError, "number of matches should be < #{INT_MAX}")
+    end
+
     it "defaults to extracting all submatches when given nil", :aggregate_failures do
       re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
       md = re.match("one two three", submatches: nil)
@@ -584,6 +594,13 @@ RSpec.describe RE2::Regexp do
       expect(md[3]).to be_nil
     end
 
+    it "raises an exception if given too large a number of submatches instead of options" do
+      re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
+      md = re.match("one two three", 2)
+
+      expect { re.match("one two three", INT_MAX) }.to raise_error(RangeError, "number of matches should be < #{INT_MAX}")
+    end
+
     it "raises an exception when given invalid options" do
       re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
 

data/spec/re2/scanner_spec.rb

@@ -11,7 +11,39 @@ RSpec.describe RE2::Scanner do
   end
 
   describe "#string" do
-    it "returns the original text for the scanner" do
+    it "returns the text for the scanner" do
+      re = RE2::Regexp.new('(\w+)')
+      text = "It is a truth"
+      scanner = re.scan(text)
+
+      expect(scanner.string).to eq("It is a truth")
+    end
+
+    it "returns a frozen string" do
+      re = RE2::Regexp.new('(\w+)')
+      text = "It is a truth"
+      scanner = re.scan(text)
+
+      expect(scanner.string).to be_frozen
+    end
+
+    it "freezes unfrozen strings" do
+      re = RE2::Regexp.new('(\w+)')
+      text = +"It is a truth"
+      scanner = re.scan(text)
+
+      expect(scanner.string).to be_frozen
+    end
+
+    it "copies unfrozen strings" do
+      re = RE2::Regexp.new('(\w+)')
+      text = +"It is a truth"
+      scanner = re.scan(text)
+
+      expect(scanner.string).to_not equal(text)
+    end
+
+    it "does not copy the string if it was already frozen" do
       re = RE2::Regexp.new('(\w+)')
       text = "It is a truth"
       scanner = re.scan(text)
@@ -162,6 +194,23 @@ RSpec.describe RE2::Scanner do
       expect(scanner.scan).to eq(["world"])
       expect(scanner.scan).to be_nil
     end
+
+    it "supports GC compaction" do
+      r = RE2::Regexp.new('(\w+)')
+      scanner = r.scan("Hello world" * 2)
+      GC.compact
+
+      expect(scanner.scan).to eq(["Hello"])
+    end
+
+    it "works even if the original input is mutated" do
+      r = RE2::Regexp.new('(\w+)')
+      text = +"It is a truth universally acknowledged"
+      scanner = r.scan(text)
+      text.upcase!
+
+      expect(scanner.scan).to eq(["It"])
+    end
   end
 
   it "is enumerable" do

data/spec/re2/set_spec.rb

@@ -66,10 +66,10 @@ RSpec.describe RE2::Set do
       expect { set.add("???") }.to raise_error(ArgumentError, /str rejected by RE2::Set->Add\(\)/)
     end
 
-    it "truncates error messages to 100 characters" do
+    it "includes the full error message" do
       set = RE2::Set.new(:unanchored, log_errors: false)
 
-      expect { set.add("(?P<#{'o' * 200}") }.to raise_error(ArgumentError, "str rejected by RE2::Set->Add(): invalid named capture group: (?P<oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo")
+      expect { set.add("(?P<#{'o' * 200}") }.to raise_error(ArgumentError, "str rejected by RE2::Set->Add(): invalid named capture group: (?P<oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo")
     end
 
     it "raises an error if called after #compile" do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: re2
 version: !ruby/object:Gem::Version
-  version: 2.20.0
+  version: 2.22.0
 platform: arm64-darwin
 authors:
 - Paul Mucur
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2025-09-26 00:00:00.000000000 Z
+date: 2025-11-30 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake-compiler