Merge pull request #5 from breml/quote-string

Add Quote for Logstash strings

ast/astutil/quote.go

@@ -0,0 +1,91 @@
+package astutil
+
+import (
+	"bytes"
+	"errors"
+	"regexp"
+
+	"github.com/breml/logstash-config/ast"
+)
+
+var barewordRe = regexp.MustCompile("(?s:^[A-Za-z_][A-Za-z0-9_]+$)")
+
+// Quote returns a a string with quotes for Logstash. Supported quote types
+// are ast.DoubleQuoted, ast.SingleQuoted and ast.Bareword.
+// If escape is false and the result is not a valid quoted value, an error
+// is returned. If escape is true, the value will be escaped such, that the
+// returned value is a valid quoted Logstash string.
+// For ast.DoubleQuoted, all double quotes (`"`) are escaped to `\"`.
+// For ast.SingleQuoted, all single quotes (`'`) are escaped to `\'`.
+// For ast.Bareword, all characters not matching "[A-Za-z_][A-Za-z0-9_]+" are
+// replaced with `_`.
+func Quote(value string, quoteType ast.StringAttributeType, escape bool) (string, error) {
+	var hasDoubleQuote bool
+	var hasSingleQuote bool
+
+	for i, chr := range value {
+		if chr == '"' && i > 1 && value[i-1] != '\\' {
+			hasDoubleQuote = true
+		}
+		if chr == '\'' && i > 1 && value[i-1] != '\\' {
+			hasSingleQuote = true
+		}
+	}
+
+	switch quoteType {
+	case ast.DoubleQuoted:
+		if hasDoubleQuote && !escape {
+			return "", errors.New("value %q contains unescaped double quotes and can not be quoted with double quotes without escaping")
+		}
+		return `"` + escapeQuotes(value, '"') + `"`, nil
+	case ast.SingleQuoted:
+		if hasSingleQuote && !escape {
+			return "", errors.New("value %q contains unescaped single quotes and can not be quoted with double quotes without escaping")
+		}
+		return `'` + escapeQuotes(value, '\'') + `'`, nil
+	case ast.Bareword:
+		if !barewordRe.MatchString(value) && !escape {
+			return "", errors.New("value %q contains non bareword characters and can not be quoted as bareword without escaping")
+		}
+		return escapeBareword(value), nil
+	default:
+		panic("quote type not supported")
+	}
+}
+
+func escapeQuotes(value string, quote byte) string {
+	b := []byte(value)
+
+	for i := 0; i < len(b); i++ {
+		if b[i] == quote && (i == 0 || i > 1 && b[i-1] != '\\') {
+			b = append(b[:i], append([]byte{'\\'}, b[i:]...)...)
+		}
+	}
+
+	return string(b)
+}
+
+func escapeBareword(value string) string {
+	if len(value) == 0 {
+		return ""
+	}
+	b := []byte(value)
+	if b[0] >= '0' && b[0] <= '9' {
+		b[0] = '_'
+	}
+	barewordMap := func(r rune) rune {
+		switch {
+		case r >= '0' && r <= '9':
+			return r
+		case r >= 'A' && r <= 'Z':
+			return r
+		case r >= 'a' && r <= 'z':
+			return r
+		default:
+			return '_'
+		}
+	}
+	b = bytes.Map(barewordMap, b)
+
+	return string(b)
+}

ast/astutil/quote_internal_test.go

@@ -0,0 +1,96 @@
+package astutil
+
+import "testing"
+
+func TestEscapeQuotes(t *testing.T) {
+	tt := []struct {
+		name       string
+		wantDouble string
+		wantSingle string
+	}{
+		{
+			name:       ``,
+			wantDouble: ``,
+			wantSingle: ``,
+		},
+		{
+			name:       `"`,
+			wantDouble: `\"`,
+			wantSingle: `"`,
+		},
+		{
+			name:       `"foo"bar"`,
+			wantDouble: `\"foo\"bar\"`,
+			wantSingle: `"foo"bar"`,
+		},
+		{
+			name:       `'`,
+			wantDouble: `'`,
+			wantSingle: `\'`,
+		},
+		{
+			name:       `'foo'bar'`,
+			wantDouble: `'foo'bar'`,
+			wantSingle: `\'foo\'bar\'`,
+		},
+	}
+
+	for _, tc := range tt {
+		t.Run(tc.name, func(t *testing.T) {
+			got := escapeQuotes(tc.name, '"')
+			if tc.wantDouble != got {
+				t.Errorf("want: %q, got: %q", tc.wantDouble, got)
+			}
+
+			got = escapeQuotes(tc.name, '\'')
+			if tc.wantSingle != got {
+				t.Errorf("want: %q, got: %q", tc.wantSingle, got)
+			}
+		})
+	}
+}
+
+func TestEscapeBareword(t *testing.T) {
+	tt := []struct {
+		name string
+		want string
+	}{
+		{
+			name: "",
+			want: "",
+		},
+		{
+			name: "0",
+			want: "_",
+		},
+		{
+			name: "bareword",
+			want: "bareword",
+		},
+		{
+			name: "BAREWORD",
+			want: "BAREWORD",
+		},
+		{
+			name: "_bare_word_",
+			want: "_bare_word_",
+		},
+		{
+			name: "0bare1word9",
+			want: "_bare1word9",
+		},
+		{
+			name: "-() ",
+			want: "____",
+		},
+	}
+
+	for _, tc := range tt {
+		t.Run(tc.name, func(t *testing.T) {
+			got := escapeBareword(tc.name)
+			if tc.want != got {
+				t.Errorf("want: %q, got: %q", tc.want, got)
+			}
+		})
+	}
+}

ast/astutil/quote_test.go

@@ -0,0 +1,169 @@
+package astutil_test
+
+import (
+	"testing"
+
+	"github.com/breml/logstash-config/ast"
+	"github.com/breml/logstash-config/ast/astutil"
+)
+
+func TestQuote(t *testing.T) {
+	tt := []struct {
+		name string
+		in   string
+
+		want        []string
+		wantErr     []bool
+		wantEscaped []string
+	}{
+		{
+			name: "bareword",
+			in:   `bareword`,
+
+			want: []string{
+				ast.DoubleQuoted: `"bareword"`,
+				ast.SingleQuoted: `'bareword'`,
+				ast.Bareword:     `bareword`,
+			},
+			wantErr: []bool{
+				ast.DoubleQuoted: false,
+				ast.SingleQuoted: false,
+				ast.Bareword:     false,
+			},
+			wantEscaped: []string{
+				ast.DoubleQuoted: `"bareword"`,
+				ast.SingleQuoted: `'bareword'`,
+				ast.Bareword:     `bareword`,
+			},
+		},
+		{
+			name: "multiple words",
+			in:   `multiple words`,
+
+			want: []string{
+				ast.DoubleQuoted: `"multiple words"`,
+				ast.SingleQuoted: `'multiple words'`,
+				ast.Bareword:     ``,
+			},
+			wantErr: []bool{
+				ast.DoubleQuoted: false,
+				ast.SingleQuoted: false,
+				ast.Bareword:     true,
+			},
+			wantEscaped: []string{
+				ast.DoubleQuoted: `"multiple words"`,
+				ast.SingleQuoted: `'multiple words'`,
+				ast.Bareword:     `multiple_words`,
+			},
+		},
+		{
+			name: "double quote",
+			in:   `value with " (double quote)`,
+
+			want: []string{
+				ast.DoubleQuoted: ``,
+				ast.SingleQuoted: `'value with " (double quote)'`,
+				ast.Bareword:     ``,
+			},
+			wantErr: []bool{
+				ast.DoubleQuoted: true,
+				ast.SingleQuoted: false,
+				ast.Bareword:     true,
+			},
+			wantEscaped: []string{
+				ast.DoubleQuoted: `"value with \" (double quote)"`,
+				ast.SingleQuoted: `'value with " (double quote)'`,
+				ast.Bareword:     `value_with____double_quote_`,
+			},
+		},
+		{
+			name: "escaped double quote",
+			in:   `value with \" (escaped double quote)`,
+
+			want: []string{
+				ast.DoubleQuoted: `"value with \" (escaped double quote)"`,
+				ast.SingleQuoted: `'value with \" (escaped double quote)'`,
+				ast.Bareword:     ``,
+			},
+			wantErr: []bool{
+				ast.DoubleQuoted: false,
+				ast.SingleQuoted: false,
+				ast.Bareword:     true,
+			},
+			wantEscaped: []string{
+				ast.DoubleQuoted: `"value with \" (escaped double quote)"`,
+				ast.SingleQuoted: `'value with \" (escaped double quote)'`,
+				ast.Bareword:     `value_with_____escaped_double_quote_`,
+			},
+		},
+		{
+			name: "single quote",
+			in:   `value with ' (single quote)`,
+
+			want: []string{
+				ast.DoubleQuoted: `"value with ' (single quote)"`,
+				ast.SingleQuoted: ``,
+				ast.Bareword:     ``,
+			},
+			wantErr: []bool{
+				ast.DoubleQuoted: false,
+				ast.SingleQuoted: true,
+				ast.Bareword:     true,
+			},
+			wantEscaped: []string{
+				ast.DoubleQuoted: `"value with ' (single quote)"`,
+				ast.SingleQuoted: `'value with \' (single quote)'`,
+				ast.Bareword:     `value_with____single_quote_`,
+			},
+		},
+		{
+			name: "escaped single quote",
+			in:   `value with \' (escaped single quote)`,
+
+			want: []string{
+				ast.DoubleQuoted: `"value with \' (escaped single quote)"`,
+				ast.SingleQuoted: `'value with \' (escaped single quote)'`,
+				ast.Bareword:     ``,
+			},
+			wantErr: []bool{
+				ast.DoubleQuoted: false,
+				ast.SingleQuoted: false,
+				ast.Bareword:     true,
+			},
+			wantEscaped: []string{
+				ast.DoubleQuoted: `"value with \' (escaped single quote)"`,
+				ast.SingleQuoted: `'value with \' (escaped single quote)'`,
+				ast.Bareword:     `value_with_____escaped_single_quote_`,
+			},
+		},
+	}
+
+	quoteTypes := map[string]ast.StringAttributeType{
+		"double quote": ast.DoubleQuoted,
+		"single quote": ast.SingleQuoted,
+		"bareword":     ast.Bareword,
+	}
+
+	for _, tc := range tt {
+		t.Run(tc.name, func(t *testing.T) {
+			if len(tc.want) != 4 && len(tc.wantErr) != 4 {
+			}
+			for name, quoteType := range quoteTypes {
+				t.Run(name, func(t *testing.T) {
+					got, err := astutil.Quote(tc.in, quoteType, false)
+					if tc.wantErr[quoteType] != (err != nil) {
+						t.Errorf("wantErr %t, err: %v", tc.wantErr[quoteType], err)
+					}
+					if tc.want[quoteType] != got {
+						t.Errorf("want: %q, got: %q", tc.want[quoteType], got)
+					}
+
+					gotEscaped, _ := astutil.Quote(tc.in, quoteType, true)
+					if tc.wantEscaped[quoteType] != gotEscaped {
+						t.Errorf("want: %q, got: %q", tc.wantEscaped[quoteType], gotEscaped)
+					}
+				})
+			}
+		})
+	}
+}