Add Regex::MatchData#captures, named_captures, to_a, to_h (#3783)

crystal-lang · Jun 12, 2017 · 61f21ae · 61f21ae
1 parent aaf000c
commit 61f21ae
Show file tree

Hide file tree

Showing 2 changed files with 153 additions and 0 deletions.
diff --git a/spec/std/match_data_spec.cr b/spec/std/match_data_spec.cr
@@ -107,6 +107,74 @@ describe "Regex::MatchData" do
     end
   end
 
+  describe "#captures" do
+    it "gets an array of unnamed captures" do
+      "Crystal".match(/(Cr)y/).not_nil!.captures.should eq(["Cr"])
+      "Crystal".match(/(Cr)(?<name1>y)(st)(?<name2>al)/).not_nil!.captures.should eq(["Cr", "st"])
+    end
+
+    it "gets an array of unnamed captures with optional" do
+      "Crystal".match(/(Cr)(s)?/).not_nil!.captures.should eq(["Cr", nil])
+      "Crystal".match(/(Cr)(?<name1>s)?(tal)?/).not_nil!.captures.should eq(["Cr", nil])
+    end
+  end
+
+  describe "#named_captures" do
+    it "gets a hash of named captures" do
+      "Crystal".match(/(?<name1>Cr)y/).not_nil!.named_captures.should eq({"name1" => "Cr"})
+      "Crystal".match(/(Cr)(?<name1>y)(st)(?<name2>al)/).not_nil!.named_captures.should eq({"name1" => "y", "name2" => "al"})
+    end
+
+    it "gets a hash of named captures with optional" do
+      "Crystal".match(/(?<name1>Cr)(?<name2>s)?/).not_nil!.named_captures.should eq({"name1" => "Cr", "name2" => nil})
+      "Crystal".match(/(Cr)(?<name1>s)?(t)?(?<name2>al)?/).not_nil!.named_captures.should eq({"name1" => nil, "name2" => nil})
+    end
+  end
+
+  describe "#to_a" do
+    it "converts into an array" do
+      "Crystal".match(/(?<name1>Cr)(y)/).not_nil!.to_a.should eq(["Cry", "Cr", "y"])
+      "Crystal".match(/(Cr)(?<name1>y)(st)(?<name2>al)/).not_nil!.to_a.should eq(["Crystal", "Cr", "y", "st", "al"])
+    end
+
+    it "converts into an array having nil" do
+      "Crystal".match(/(?<name1>Cr)(s)?/).not_nil!.to_a.should eq(["Cr", "Cr", nil])
+      "Crystal".match(/(Cr)(?<name1>s)?(yst)?(?<name2>al)?/).not_nil!.to_a.should eq(["Crystal", "Cr", nil, "yst", "al"])
+    end
+  end
+
+  describe "#to_h" do
+    it "converts into a hash" do
+      "Crystal".match(/(?<name1>Cr)(y)/).not_nil!.to_h.should eq({
+              0 => "Cry",
+        "name1" => "Cr",
+              2 => "y",
+      })
+      "Crystal".match(/(Cr)(?<name1>y)(st)(?<name2>al)/).not_nil!.to_h.should eq({
+              0 => "Crystal",
+              1 => "Cr",
+        "name1" => "y",
+              3 => "st",
+        "name2" => "al",
+      })
+    end
+
+    it "converts into a hash having nil" do
+      "Crystal".match(/(?<name1>Cr)(s)?/).not_nil!.to_h.should eq({
+              0 => "Cr",
+        "name1" => "Cr",
+              2 => nil,
+      })
+      "Crystal".match(/(Cr)(?<name1>s)?(yst)?(?<name2>al)?/).not_nil!.to_h.should eq({
+              0 => "Crystal",
+              1 => "Cr",
+        "name1" => nil,
+              3 => "yst",
+        "name2" => "al",
+      })
+    end
+  end
+
   it "can check equality" do
     re = /((?<hello>he)llo)/
     m1 = re.match("hello")

diff --git a/src/regex/match_data.cr b/src/regex/match_data.cr
@@ -182,6 +182,91 @@ class Regex
       @string.byte_slice(byte_end(0))
     end
 
+    # Returns an array of unnamed capture groups.
+    #
+    # It is a difference from `to_a` that the result array does not contain the match for the entire `Regex` (`self[0]`).
+    #
+    # ```
+    # match = "Crystal".match(/(Cr)(?<name1>y)(st)(?<name2>al)/).not_nil!
+    # match.captures # => ["Cr", "st"]
+    #
+    # # When this regex has an optional group, result array may contain
+    # # a `nil` if this group is not matched.
+    # match = "Crystal".match(/(Cr)(stal)?/).not_nil!
+    # match.captures # => ["Cr", nil]
+    # ```
+    def captures
+      name_table = @regex.name_table
+
+      caps = [] of String?
+      (1..size).each do |i|
+        caps << self[i]? unless name_table.has_key? i
+      end
+
+      caps
+    end
+
+    # Returns a hash of named capture groups.
+    #
+    # ```
+    # match = "Crystal".match(/(Cr)(?<name1>y)(st)(?<name2>al)/).not_nil!
+    # match.named_captures # => {"name1" => "y", "name2" => "al"}
+    #
+    # # When this regex has an optional group, result hash may contain
+    # # a `nil` if this group is not matched.
+    # match = "Crystal".match(/(?<name1>Cr)(?<name2>stal)?/).not_nil!
+    # match.named_captures # => {"name1" => "Cr", "name2" => nil}
+    # ```
+    def named_captures
+      name_table = @regex.name_table
+
+      caps = {} of String => String?
+      (1..size).each do |i|
+        if name = name_table[i]?
+          caps[name] = self[i]?
+        end
+      end
+
+      caps
+    end
+
+    # Convert this match data into an array.
+    #
+    # ```
+    # match = "Crystal".match(/(Cr)(?<name1>y)(st)(?<name2>al)/).not_nil!
+    # match.to_a # => ["Crystal", "Cr", "y", "st", "al"]
+    #
+    # # When this regex has an optional group, result array may contain
+    # # a `nil` if this group is not matched.
+    # match = "Crystal".match(/(Cr)(?<name1>stal)?/).not_nil!
+    # match.to_a # => ["Cr", "Cr", nil]
+    # ```
+    def to_a
+      (0..size).map { |i| self[i]? }
+    end
+
+    # Convert this match data into a hash.
+    #
+    # ```
+    # match = "Crystal".match(/(Cr)(?<name1>y)(st)(?<name2>al)/).not_nil!
+    # match.to_h # => {0 => "Crystal", 1 => "Cr", "name1" => "y", 3 => "st", "name2" => "al"}
+    #
+    # # When this regex has an optional group, result array may contain
+    # # a `nil` if this group is not matched.
+    # match = "Crystal".match(/(Cr)(?<name1>stal)?/).not_nil!
+    # match.to_h # => {0 => "Cr", 1 => "Cr", "name1" => nil}
+    # ```
+    def to_h
+      name_table = @regex.name_table
+
+      hash = {} of (String | Int32) => String?
+      (0..size).each do |i|
+        hash[name_table.fetch(i) { i }] = self[i]?
+      end
+
+      hash
+    end
+
     def inspect(io : IO)
       to_s(io)
     end