[GR-19220] Implement new MatchData methods for 3.2 (#3393)

PullRequest: truffleruby/4124

CHANGELOG.md

@@ -31,6 +31,7 @@ Compatibility:
 * Implement the `Data` class from Ruby 3.2 (#3039, @moste00, @eregon).
 * Make `Coverage.start` and `Coverage.result` accept parameters (#3149, @mtortonesi, @andrykonchin).
 * Implement `rb_check_funcall()` (@eregon).
+* Implement `MatchData#{byteoffset,deconstruct,deconstruct_keys}` from Ruby 3.2 (#3039, @rwstauner).
 
 Performance:
 

spec/ruby/core/matchdata/begin_spec.rb

@@ -36,6 +36,18 @@
       match_data = /(.)(.)(\d+)(\d)/.match("THX1138.")
       match_data.begin(obj).should == 2
     end
+
+    it "raises IndexError if index is out of bounds" do
+      match_data = /(?<f>foo)(?<b>bar)/.match("foobar")
+
+      -> {
+        match_data.begin(-1)
+      }.should raise_error(IndexError, "index -1 out of matches")
+
+      -> {
+        match_data.begin(3)
+      }.should raise_error(IndexError, "index 3 out of matches")
+    end
   end
 
   context "when passed a String argument" do
@@ -68,6 +80,14 @@
       match_data = /(?<æ>.)(.)(?<b>\d+)(\d)/.match("THX1138.")
       match_data.begin("æ").should == 1
     end
+
+    it "raises IndexError if there is no group with the provided name" do
+      match_data = /(?<f>foo)(?<b>bar)/.match("foobar")
+
+      -> {
+        match_data.begin("y")
+      }.should raise_error(IndexError, "undefined group name reference: y")
+    end
   end
 
   context "when passed a Symbol argument" do
@@ -100,5 +120,13 @@
       match_data = /(?<æ>.)(.)(?<b>\d+)(\d)/.match("THX1138.")
       match_data.begin(:æ).should == 1
     end
+
+    it "raises IndexError if there is no group with the provided name" do
+      match_data = /(?<f>foo)(?<b>bar)/.match("foobar")
+
+      -> {
+        match_data.begin(:y)
+      }.should raise_error(IndexError, "undefined group name reference: y")
+    end
   end
 end

spec/ruby/core/matchdata/byteoffset_spec.rb

@@ -60,7 +60,7 @@ def obj.to_int; 2; end
       m.byteoffset(obj).should == [3, 6]
     end
 
-    it "raises IndexError if there is no group with provided name" do
+    it "raises IndexError if there is no group with the provided name" do
       m = /(?<f>foo)(?<b>bar)/.match("foobar")
 
       -> {
@@ -72,7 +72,7 @@ def obj.to_int; 2; end
       }.should raise_error(IndexError, "undefined group name reference: y")
     end
 
-    it "raises IndexError if index is out of matches" do
+    it "raises IndexError if index is out of bounds" do
       m = /(?<f>foo)(?<b>bar)/.match("foobar")
 
       -> {

src/main/java/org/truffleruby/core/regexp/MatchDataNodes.java

@@ -627,6 +627,13 @@ Object byteBegin(RubyMatchData matchData, int index,
             }
         }
 
+        @Specialization(guards = "!inBounds(matchData, index)")
+        Object byteBeginError(RubyMatchData matchData, int index) {
+            throw new RaiseException(
+                    getContext(),
+                    coreExceptions().indexError(StringUtils.format("index %d out of matches", index), this));
+        }
+
         protected boolean inBounds(RubyMatchData matchData, int index) {
             return index >= 0 && index < matchData.region.numRegs;
         }
@@ -649,6 +656,13 @@ Object byteEnd(RubyMatchData matchData, int index,
             }
         }
 
+        @Specialization(guards = "!inBounds(matchData, index)")
+        Object byteEndError(RubyMatchData matchData, int index) {
+            throw new RaiseException(
+                    getContext(),
+                    coreExceptions().indexError(StringUtils.format("index %d out of matches", index), this));
+        }
+
         protected boolean inBounds(RubyMatchData matchData, int index) {
             return index >= 0 && index < matchData.region.numRegs;
         }

src/main/ruby/truffleruby/core/match_data.rb

@@ -40,6 +40,11 @@ class << self
     undef_method :allocate
   end
 
+  def byteoffset(idx)
+    backref = backref_from_arg(idx)
+    [Primitive.match_data_byte_begin(self, backref), Primitive.match_data_byte_end(self, backref)]
+  end
+
   def offset(idx)
     [self.begin(idx), self.end(idx)]
   end
@@ -61,6 +66,26 @@ def string
   def captures
     to_a[1..-1]
   end
+  alias_method :deconstruct, :captures
+
+  def deconstruct_keys(array_of_names)
+    Truffle::Type.rb_check_type(array_of_names, Array) unless Primitive.nil?(array_of_names)
+
+    hash = named_captures.transform_keys(&:to_sym)
+    return hash if Primitive.nil?(array_of_names)
+
+    ret = {}
+    return ret if array_of_names.size > hash.size
+
+    array_of_names.each do |key|
+      Truffle::Type.rb_check_type(key, Symbol)
+      value = Primitive.hash_get_or_undefined(hash, key)
+      break if Primitive.undefined?(value)
+      ret[key] = value
+    end
+
+    ret
+  end
 
   def names
     regexp.names
@@ -71,26 +96,12 @@ def named_captures
   end
 
   def begin(index)
-    backref = if Primitive.is_a?(index, String) || Primitive.is_a?(index, Symbol)
-                names_to_backref = Hash[Primitive.regexp_names(self.regexp)]
-                names_to_backref[index.to_sym].last
-              else
-                Truffle::Type.coerce_to(index, Integer, :to_int)
-              end
-
-
+    backref = backref_from_arg(index)
     Primitive.match_data_begin(self, backref)
   end
 
   def end(index)
-    backref = if Primitive.is_a?(index, String) || Primitive.is_a?(index, Symbol)
-                names_to_backref = Hash[Primitive.regexp_names(self.regexp)]
-                names_to_backref[index.to_sym].last
-              else
-                Truffle::Type.coerce_to(index, Integer, :to_int)
-              end
-
-
+    backref = backref_from_arg(index)
     Primitive.match_data_end(self, backref)
   end
 
@@ -153,6 +164,21 @@ def match_length(n)
   def to_s
     self[0]
   end
+
+  private
+
+  def backref_from_arg(index)
+    if Primitive.is_a?(index, String) || Primitive.is_a?(index, Symbol)
+      names_to_backref = Hash[Primitive.regexp_names(self.regexp)]
+      array = names_to_backref[index.to_sym]
+
+      raise IndexError, "undefined group name reference: #{index}" unless array
+
+      return array.last
+    end
+
+    Primitive.rb_to_int(index)
+  end
 end
 
 Truffle::KernelOperations.define_hooked_variable(

test/mri/excludes/TestRegexp.rb

@@ -39,10 +39,7 @@
 exclude :test_match_control_meta_escape, "<0> expected but was"
 exclude :test_initialize_option, "<//m> expected but was"
 exclude :test_initialize_bool_warning, "expected: /expected true or false as ignorecase/"
-exclude :test_match_byteoffset_begin_end, "NoMethodError: undefined method `byteoffset' for #<MatchData \"bar\" x:\"bar\">"
-exclude :test_match_data_deconstruct, "NoMethodError: undefined method `deconstruct' for #<MatchData \"foobarbaz\">"
 exclude :test_linear_time_p, "NoMethodError: undefined method `linear_time?' for Regexp:Class"
-exclude :test_match_data_deconstruct_keys, "NoMethodError: undefined method `deconstruct_keys' for #<MatchData \"foobarbaz\">"
 exclude :test_extended_comment_invalid_escape_bug_18294, "assert_separately failed with error message"
 exclude :test_timeout_nil, "NoMethodError: undefined method `timeout=' for Regexp:Class"
 exclude :test_timeout_shorter_than_global, "NoMethodError: undefined method `timeout=' for Regexp:Class"