Implement MatchData#byteoffset for 3.2 compatibility

Author: rwstauner

spec/tags/core/matchdata/byteoffset_tags.txt

@@ -1,11 +1 @@
-fails:MatchData#byteoffset returns beginning and ending byte-based offset of whole matched substring for 0 element
-fails:MatchData#byteoffset returns beginning and ending byte-based offset of n-th match, all the subsequent elements are capturing groups
-fails:MatchData#byteoffset accepts String as a reference to a named capture
-fails:MatchData#byteoffset accepts Symbol as a reference to a named capture
-fails:MatchData#byteoffset returns [nil, nil] if a capturing group is optional and doesn't match
-fails:MatchData#byteoffset returns correct beginning and ending byte-based offset for multi-byte strings
-fails:MatchData#byteoffset returns [nil, nil] if a capturing group is optional and doesn't match for multi-byte string
-fails:MatchData#byteoffset converts argument into integer if is not String nor Symbol
-fails:MatchData#byteoffset raises IndexError if there is no group with provided name
-fails:MatchData#byteoffset raises IndexError if index is out of matches
 fails:MatchData#byteoffset raises TypeError if can't convert argument into Integer

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

@@ -627,6 +627,14 @@ Object byteBegin(RubyMatchData matchData, int index,
             }
         }
 
+        @TruffleBoundary
+        @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 +657,14 @@ Object byteEnd(RubyMatchData matchData, int index,
             }
         }
 
+        @TruffleBoundary
+        @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
@@ -71,26 +76,12 @@ 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 +144,20 @@ 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)]
+      unless names_to_backref.key?(index.to_sym)
+        raise IndexError, "undefined group name reference: #{index}"
+      end
+      names_to_backref[index.to_sym].last
+    else
+      Truffle::Type.coerce_to(index, Integer, :to_int)
+    end
+  end
 end
 
 Truffle::KernelOperations.define_hooked_variable(

test/mri/excludes/TestRegexp.rb

@@ -39,7 +39,6 @@
 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\">"