Support pattern matching

ext/re2/re2.cc

@@ -687,6 +687,112 @@ static VALUE re2_matchdata_inspect(VALUE self) {
   return result;
 }
 
+/*
+ * Returns the array of submatches for pattern matching.
+ *
+ * @return [Array<String, nil>] the array of submatches
+ * @example
+ *   m = RE2::Regexp.new('(\d+)').match("bob 123")
+ *   m.deconstruct    #=> ["123"]
+ *
+ * @example pattern matching
+ *   case RE2::Regexp.new('(\d+) (\d+)').match("bob 123 456")
+ *   in x, y
+ *     puts "Matched #{x} #{y}"
+ *   else
+ *     puts "Unrecognised match"
+ *   end
+ */
+static VALUE re2_matchdata_deconstruct(VALUE self) {
+  int i;
+  re2_matchdata *m;
+  re2_pattern *p;
+  re2::StringPiece *match;
+  VALUE array;
+
+  Data_Get_Struct(self, re2_matchdata, m);
+  Data_Get_Struct(m->regexp, re2_pattern, p);
+
+  array = rb_ary_new2(m->number_of_matches - 1);
+  for (i = 1; i < m->number_of_matches; i++) {
+    match = &m->matches[i];
+
+    if (match->empty()) {
+      rb_ary_push(array, Qnil);
+    } else {
+      rb_ary_push(array, ENCODED_STR_NEW(match->data(), match->size(),
+            p->pattern->options().encoding() == RE2::Options::EncodingUTF8 ? "UTF-8" : "ISO-8859-1"));
+    }
+  }
+
+  return array;
+}
+
+/*
+ * Returns a hash of capturing group names to submatches for pattern matching.
+ *
+ * As this is used by Ruby's pattern matching, it will return an empty hash if given
+ * more keys than there are capturing groups. Given keys will populate the hash in
+ * order but an invalid name will cause the hash to be immediately returned.
+ *
+ * @return [Hash] a hash of capturing group names to submatches
+ * @param [Array<Symbol>, nil] keys an array of Symbol capturing group names or nil to return all names
+ * @example
+ *   m = RE2::Regexp.new('(?P<numbers>\d+) (?P<letters>[a-zA-Z]+)').match('123 abc')
+ *   m.deconstruct_keys(nil)                  #=> {:numbers => "123", :letters => "abc"}
+ *   m.deconstruct_keys([:numbers])           #=> {:numbers => "123"}
+ *   m.deconstruct_keys([:fruit])             #=> {}
+ *   m.deconstruct_keys([:letters, :fruit])   #=> {:letters => "abc"}
+ *
+ * @example pattern matching
+ *   case RE2::Regexp.new('(?P<numbers>\d+) (?P<letters>[a-zA-Z]+)').match('123 abc')
+ *   in numbers:, letters:
+ *     puts "Numbers: #{numbers}, letters: #{letters}"
+ *   else
+ *     puts "Unrecognised match"
+ *   end
+ */
+static VALUE re2_matchdata_deconstruct_keys(VALUE self, VALUE keys) {
+  int i;
+  VALUE capturing_groups, key;
+  re2_matchdata *m;
+  re2_pattern *p;
+  map<string, int> groups;
+  map<string, int>::iterator iterator;
+
+  Data_Get_Struct(self, re2_matchdata, m);
+  Data_Get_Struct(m->regexp, re2_pattern, p);
+
+  groups = p->pattern->NamedCapturingGroups();
+  capturing_groups = rb_hash_new();
+
+  if (NIL_P(keys)) {
+    for (iterator = groups.begin(); iterator != groups.end(); iterator++) {
+      rb_hash_aset(capturing_groups,
+          ID2SYM(rb_intern(iterator->first.data())),
+          re2_matchdata_nth_match(iterator->second, self));
+    }
+  } else {
+    Check_Type(keys, T_ARRAY);
+
+    if (p->pattern->NumberOfCapturingGroups() >= RARRAY_LEN(keys)) {
+      for (i = 0; i < RARRAY_LEN(keys); i++) {
+        key = rb_ary_entry(keys, i);
+        Check_Type(key, T_SYMBOL);
+        string name(rb_id2name(SYM2ID(key)));
+
+        if (groups.count(name) == 0) {
+          break;
+        }
+
+        rb_hash_aset(capturing_groups, key, re2_matchdata_nth_match(groups[name], self));
+      }
+    }
+  }
+
+  return capturing_groups;
+}
+
 /*
  * Returns a new RE2 object with a compiled version of
  * +pattern+ stored inside. Equivalent to +RE2.new+.
@@ -1666,6 +1772,10 @@ void Init_re2(void) {
         RUBY_METHOD_FUNC(re2_matchdata_to_s), 0);
   rb_define_method(re2_cMatchData, "inspect",
       RUBY_METHOD_FUNC(re2_matchdata_inspect), 0);
+  rb_define_method(re2_cMatchData, "deconstruct",
+      RUBY_METHOD_FUNC(re2_matchdata_deconstruct), 0);
+  rb_define_method(re2_cMatchData, "deconstruct_keys",
+      RUBY_METHOD_FUNC(re2_matchdata_deconstruct_keys), 1);
 
   rb_define_method(re2_cScanner, "string",
       RUBY_METHOD_FUNC(re2_scanner_string), 0);

spec/re2/match_data_spec.rb

@@ -241,4 +241,62 @@
       expect(md.end(:foo)).to be_nil
     end
   end
+
+  describe "#deconstruct" do
+    it "returns all capturing groups" do
+      md = RE2::Regexp.new('w(o)(o)').match('woo')
+
+      expect(md.deconstruct).to eq(['o', 'o'])
+    end
+
+    it "includes optional capturing groups as nil" do
+      md = RE2::Regexp.new('w(.)(.)(.)?').match('woo')
+
+      expect(md.deconstruct).to eq(['o', 'o', nil])
+    end
+  end
+
+  describe "#deconstruct_keys" do
+    it "returns all named captures if given nil" do
+      md = RE2::Regexp.new('(?P<numbers>\d+) (?P<letters>[a-zA-Z]+)').match('123 abc')
+
+      expect(md.deconstruct_keys(nil)).to eq(:numbers => '123', :letters => 'abc')
+    end
+
+    it "returns only named captures if given names" do
+      md = RE2::Regexp.new('(?P<numbers>\d+) (?P<letters>[a-zA-Z]+)').match('123 abc')
+
+      expect(md.deconstruct_keys([:numbers])).to eq(:numbers => '123')
+    end
+
+    it "returns named captures up until an invalid name is given" do
+      md = RE2::Regexp.new('(?P<numbers>\d+) (?P<letters>[a-zA-Z]+)').match('123 abc')
+
+      expect(md.deconstruct_keys([:numbers, :punctuation])).to eq(:numbers => '123')
+    end
+
+    it "returns an empty hash if given more capture names than exist" do
+      md = RE2::Regexp.new('(?P<numbers>\d+) (?P<letters>[a-zA-Z]+)').match('123 abc')
+
+      expect(md.deconstruct_keys([:numbers, :letters, :punctuation])).to eq({})
+    end
+
+    it "returns an empty hash if there are no named capturing groups" do
+      md = RE2::Regexp.new('(\d+) ([a-zA-Z]+)').match('123 abc')
+
+      expect(md.deconstruct_keys(nil)).to eq({})
+    end
+
+    it "raises an error if given a non-array of keys" do
+      md = RE2::Regexp.new('(?P<numbers>\d+) (?P<letters>[a-zA-Z]+)').match('123 abc')
+
+      expect { md.deconstruct_keys(0) }.to raise_error(TypeError)
+    end
+
+    it "raises an error if given keys as non-symbols" do
+      md = RE2::Regexp.new('(?P<numbers>\d+) (?P<letters>[a-zA-Z]+)').match('123 abc')
+
+      expect { md.deconstruct_keys([0]) }.to raise_error(TypeError)
+    end
+  end
 end