Performance fix for StringScanner#peek and #peek_behind (#16593)

Co-authored-by: jneen <jneen@jneen.net>

Author: jneen

spec/std/string_scanner_spec.cr

@@ -417,6 +417,45 @@ describe StringScanner do
       s.peek(7).should eq("this is")
       s.offset.should eq(0)
     end
+
+    it "shows the next len characters for multi-byte strings" do
+      s = StringScanner.new("これは文字列である")
+      s.offset.should eq(0)
+      s.peek(3).should eq("これは")
+      s.offset.should eq(0)
+      s.peek(6).should eq("これは文字列")
+      s.offset.should eq(0)
+    end
+
+    it "errors on negative input" do
+      s = StringScanner.new("abcde")
+      s.scan(2)
+      expect_raises(ArgumentError, "Negative lookahead count: -1") { s.peek(-1) }
+    end
+  end
+
+  describe "#peek_behind" do
+    it "shows characters behind the scan head" do
+      s = StringScanner.new("abcdefg")
+      s.peek_behind(10).should eq("")
+      s.scan(3)
+      s.peek_behind(10).should eq("abc")
+      s.peek_behind(2).should eq("bc")
+    end
+
+    it "shows characters behind the scan head for multi-byte strings" do
+      s = StringScanner.new("あいうえお")
+      s.peek_behind(10).should eq("")
+      s.scan(3)
+      s.peek_behind(10).should eq("あいう")
+      s.peek_behind(2).should eq("いう")
+    end
+
+    it "errors on negative input" do
+      s = StringScanner.new("abcde")
+      s.scan(3)
+      expect_raises(ArgumentError, "Negative lookbehind count") { s.peek_behind(-1) }
+    end
   end
 
   describe "#reset" do

src/string_scanner.cr

@@ -41,8 +41,9 @@
 # * `#skip`
 # * `#skip_until`
 #
-# Methods that look ahead:
+# Methods that look ahead or behind:
 # * `#peek`
+# * `#peek_behind`
 # * `#check`
 # * `#check_until`
 #
@@ -418,10 +419,20 @@ class StringScanner
     @str
   end
 
-  # Extracts a string corresponding to string[offset,*len*], without advancing
-  # the scan offset.
+  # Extracts a string by looking ahead *len* characters, without advancing the
+  # scan offset. The return value has at most *len* characters, but may have fewer
+  # if the scan head is close to the end of the string.
   def peek(len) : String
-    @str[offset, len]
+    byte_len = lookahead_byte_length(len) || @str.bytesize
+    @str.byte_slice(@byte_offset, byte_len)
+  end
+
+  # Extracts a string by looking behind *len* characters, without moving the
+  # scan offset. The return value has at most *len* characters, but may have fewer
+  # if the scan head is close to the beginning of the string.
+  def peek_behind(len) : String
+    byte_len = lookbehind_byte_length(len) || @byte_offset
+    @str.byte_slice(@byte_offset - byte_len, byte_len)
   end
 
   # Returns the remainder of the string after the scan offset.