complete string implementation

html/html.go

@@ -31,7 +31,7 @@ var (
 		return Node{Result: ret}
 	})
 
-	attr  = And(identifier, "=", Any(String('"'), String('\'')))
+	attr  = And(identifier, "=", String(`"'`))
 	attrs = Map(Kleene(attr), func(node Node) Node {
 		attr := map[string]string{}
 

json/json.go

@@ -16,7 +16,7 @@ var (
 		}
 		return Node{Result: ret}
 	})
-	properties = Kleene(And(String('"'), ":", &value), ",")
+	properties = Kleene(And(String(`"`), ":", &value), ",")
 	_object    = Map(And("{", properties, "}"), func(n Node) Node {
 		ret := map[string]interface{}{}
 
@@ -31,7 +31,7 @@ var (
 	_true  = Bind("true", true)
 	_false = Bind("false", false)
 
-	_string = Map(String('"'), func(n Node) Node {
+	_string = Map(String(`"`), func(n Node) Node {
 		return Node{Result: n.Token}
 	})
 )

parser.go

@@ -201,39 +201,92 @@ func charsImpl(matcher string, stopOn bool, repetition ...int) Parser {
 	}
 }
 
-func String(quote rune) Parser {
+func String(allowedQuotes string) Parser {
 	return NewParser("string", func(ps *State) Node {
 		ps.AutoWS()
-		var r rune
+
-		var w int
+		for i := 0; i < len(allowedQuotes); i++ {
-		var matched int
+			if ps.Input[ps.Pos] == allowedQuotes[i] {
-		r, matched = utf8.DecodeRuneInString(ps.Input[ps.Pos:])
+
-		if r != quote {
+			}
-			ps.ErrorHere("\"")
+		}
+		if !stringContainsByte(allowedQuotes, ps.Input[ps.Pos]) {
+			ps.ErrorHere(allowedQuotes)
+			return Node{}
+		}
+		quote := ps.Input[ps.Pos]
+
+		var end int = ps.Pos + 1
+
+		inputLen := len(ps.Input)
+		result := &bytes.Buffer{}
+
+		for end < inputLen {
+			switch ps.Input[end] {
+			case '\\':
+				if end+1 >= inputLen {
+					ps.ErrorHere(string(quote))
 					return Node{}
 				}
 
-		result := &bytes.Buffer{}
+				c := ps.Input[end+1]
-
+				if c == 'u' {
-		for ps.Pos+matched < len(ps.Input) {
+					if end+6 >= inputLen {
-			r, w = utf8.DecodeRuneInString(ps.Input[ps.Pos+matched:])
+						ps.Error.Expected = "[a-f0-9]{4}"
-			matched += w
+						ps.Error.pos = end + 2
-
+						return Node{}
-			if r == '\\' {
-				r, w = utf8.DecodeRuneInString(ps.Input[ps.Pos+matched:])
-				result.WriteRune(r)
-				matched += w
-				continue
 					}
 
-			if r == quote {
+					r, ok := unhex(ps.Input[end+2 : end+6])
-				ps.Advance(matched)
+					if !ok {
+						ps.Error.Expected = "[a-f0-9]"
+						ps.Error.pos = end + 2
+						return Node{}
+					}
+					result.WriteRune(r)
+					end += 6
+				} else {
+					result.WriteByte(c)
+					end += 2
+				}
+			case quote:
+				ps.Pos = end + 1
 				return Node{Token: result.String()}
-			}
+			default:
+				r, w := utf8.DecodeRuneInString(ps.Input[end:])
 				result.WriteRune(r)
+				end += w
+			}
 		}
 
-		ps.ErrorHere("\"")
+		ps.ErrorHere(string(quote))
 		return Node{}
 	})
 }
+
+func stringContainsByte(s string, b byte) bool {
+	for i := 0; i < len(s); i++ {
+		if b == s[i] {
+			return true
+		}
+	}
+	return false
+}
+
+func unhex(b string) (v rune, ok bool) {
+	for _, c := range b {
+		v <<= 4
+		switch {
+		case '0' <= c && c <= '9':
+			v |= c - '0'
+		case 'a' <= c && c <= 'f':
+			v |= c - 'a' + 10
+		case 'A' <= c && c <= 'F':
+			v |= c - 'A' + 10
+		default:
+			return 0, false
+		}
+	}
+
+	return v, true
+}

parser_test.go

@@ -152,29 +152,97 @@ func TestParseString(t *testing.T) {
 }
 
 func TestString(t *testing.T) {
-	t.Run("test basic match", func(t *testing.T) {
+	parser := String(`"'`)
-		result, p := runParser(`"hello"`, String('"'))
+	t.Run("test double match", func(t *testing.T) {
+		result, p := runParser(`"hello"`, parser)
 		require.Equal(t, `hello`, result.Token)
 		require.Equal(t, "", p.Get())
 	})
 
+	t.Run("test single match", func(t *testing.T) {
+		result, p := runParser(`"hello"`, parser)
+		require.Equal(t, `hello`, result.Token)
+		require.Equal(t, "", p.Get())
+	})
+
+	t.Run("test nested quotes", func(t *testing.T) {
+		result, p := runParser(`"hello 'world'"`, parser)
+		require.Equal(t, `hello 'world'`, result.Token)
+		require.Equal(t, "", p.Get())
+	})
+
 	t.Run("test non match", func(t *testing.T) {
-		_, p := runParser(`1`, String('"'))
+		_, p := runParser(`1`, parser)
-		require.Equal(t, `"`, p.Error.Expected)
+		require.Equal(t, `"'`, p.Error.Expected)
 		require.Equal(t, `1`, p.Get())
 	})
 
 	t.Run("test unterminated string", func(t *testing.T) {
-		_, p := runParser(`"hello `, String('"'))
+		_, p := runParser(`"hello `, parser)
 		require.Equal(t, `"`, p.Error.Expected)
 		require.Equal(t, `"hello `, p.Get())
 	})
 
+	t.Run("test unmatched quotes", func(t *testing.T) {
+		_, p := runParser(`"hello '`, parser)
+		require.Equal(t, `"`, p.Error.Expected)
+		require.Equal(t, 0, p.Pos)
+	})
+
+	t.Run("test unterminated escape", func(t *testing.T) {
+		_, p := runParser(`"hello \`, parser)
+		require.Equal(t, `"`, p.Error.Expected)
+		require.Equal(t, 0, p.Pos)
+	})
+
 	t.Run("test escaping", func(t *testing.T) {
-		result, p := runParser(`"hello \"world\""`, String('"'))
+		result, p := runParser(`"hello \"world\""`, parser)
 		require.Equal(t, `hello "world"`, result.Token)
 		require.Equal(t, ``, p.Get())
 	})
+
+	t.Run("test escaped unicode", func(t *testing.T) {
+		result, p := runParser(`"hello \ubeef cake"`, parser)
+		require.Equal(t, "", p.Error.Expected)
+		require.Equal(t, "hello \uBEEF cake", result.Token)
+		require.Equal(t, ``, p.Get())
+	})
+
+	t.Run("test invalid escaped unicode", func(t *testing.T) {
+		_, p := runParser(`"hello \ucake"`, parser)
+		require.Equal(t, "offset 9: Expected [a-f0-9]", p.Error.Error())
+		require.Equal(t, 0, p.Pos)
+	})
+
+	t.Run("test incomplete escaped unicode", func(t *testing.T) {
+		_, p := runParser(`"hello \uca"`, parser)
+		require.Equal(t, "offset 9: Expected [a-f0-9]{4}", p.Error.Error())
+		require.Equal(t, 0, p.Pos)
+	})
+}
+
+func TestUnhex(t *testing.T) {
+	tests := map[int64]string{
+		0xF:        "F",
+		0x5:        "5",
+		0xFF:       "FF",
+		0xFFF:      "FFF",
+		0xA4B:      "a4b",
+		0xFFFF:     "FFFF",
+		0xBEEFCAFE: "beeFCAfe",
+	}
+	for expected, input := range tests {
+		t.Run(input, func(t *testing.T) {
+			r, ok := unhex(input)
+			require.True(t, ok)
+			require.EqualValues(t, expected, r)
+		})
+	}
+
+	t.Run("Fails on non hex chars", func(t *testing.T) {
+		_, ok := unhex("hello")
+		require.False(t, ok)
+	})
 }
 
 func runParser(input string, parser Parser) (Node, *State) {