Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix counting of escape sequences when splitting TXT strings #1540

Merged
merged 1 commit into from
Apr 17, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
39 changes: 31 additions & 8 deletions parse_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -1098,18 +1098,41 @@ func TestTXT(t *testing.T) {
}
}

// Test TXT record with chunk larger than 255 bytes, they should be split up, by the parser
s := ""
for i := 0; i < 255; i++ {
s += "a"
// Test TXT record with string larger than 255 bytes that should be split
// up by the parser. Add some escape sequences too to ensure their length
// is counted correctly.
s := `"\;\\\120` + strings.Repeat("a", 255) + `b"`
rr, err = NewRR(`test.local. 60 IN TXT ` + s)
if err != nil {
t.Error("failed to parse empty-string TXT record", err)
}
if rr.(*TXT).Txt[1] != "aaab" {
t.Errorf("Txt should have two strings, last one must be 'aaab', but is %s", rr.(*TXT).Txt[1])
}
s += "b"
rr, err = NewRR(`test.local. 60 IN TXT "` + s + `"`)
rrContent := strings.Replace(rr.String(), rr.Header().String(), "", 1)
expectedRRContent := `";\\x` + strings.Repeat("a", 252) + `" "aaab"`
if expectedRRContent != rrContent {
t.Errorf("Expected TXT RR content to be %#q but got %#q", expectedRRContent, rrContent)
}

// Test TXT record that is already split up into strings of len <= 255.
s = fmt.Sprintf(
"%q %q %q %q %q %q",
strings.Repeat(`a`, 255),
strings.Repeat("b", 255),
strings.Repeat("c", 255),
strings.Repeat("d", 0),
strings.Repeat("e", 1),
strings.Repeat("f", 123),
)
rr, err = NewRR(`test.local. 60 IN TXT ` + s)
if err != nil {
t.Error("failed to parse empty-string TXT record", err)
}
if rr.(*TXT).Txt[1] != "b" {
t.Errorf("Txt should have two chunk, last one my be 'b', but is %s", rr.(*TXT).Txt[1])
rrContent = strings.Replace(rr.String(), rr.Header().String(), "", 1)
expectedRRContent = s // same as input
if expectedRRContent != rrContent {
t.Errorf("Expected TXT RR content to be %#q but got %#q", expectedRRContent, rrContent)
}
}

Expand Down
59 changes: 42 additions & 17 deletions scan_rr.go
Original file line number Diff line number Diff line change
Expand Up @@ -51,25 +51,21 @@ func endingToTxtSlice(c *zlexer, errstr string) ([]string, *ParseError) {
switch l.value {
case zString:
empty = false
if len(l.token) > 255 {
// split up tokens that are larger than 255 into 255-chunks
sx := []string{}
p, i := 0, 255
for {
if i <= len(l.token) {
sx = append(sx, l.token[p:i])
} else {
sx = append(sx, l.token[p:])
break

}
p, i = p+255, i+255
// split up tokens that are larger than 255 into 255-chunks
sx := []string{}
p := 0
for {
i := escapedStringOffset(l.token[p:], 255)
if i != -1 && p+i != len(l.token) {
sx = append(sx, l.token[p:p+i])
} else {
sx = append(sx, l.token[p:])
break

}
s = append(s, sx...)
break
p += i
}

s = append(s, l.token)
s = append(s, sx...)
case zBlank:
if quote {
// zBlank can only be seen in between txt parts.
Expand Down Expand Up @@ -1920,3 +1916,32 @@ func (rr *APL) parse(c *zlexer, o string) *ParseError {
rr.Prefixes = prefixes
return nil
}

// escapedStringOffset finds the offset within a string (which may contain escape
// sequences) that corresponds to a certain byte offset. If the input offset is
// out of bounds, -1 is returned.
func escapedStringOffset(s string, byteOffset int) int {
if byteOffset == 0 {
return 0
}

offset := 0
for i := 0; i < len(s); i++ {
offset += 1

// Skip escape sequences
if s[i] != '\\' {
// Not an escape sequence; nothing to do.
} else if isDDD(s[i+1:]) {
i += 3
} else {
i++
}

if offset >= byteOffset {
return i + 1
}
}

return -1
}
28 changes: 28 additions & 0 deletions scan_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -427,3 +427,31 @@ func BenchmarkZoneParser(b *testing.B) {
}
}
}

func TestEscapedStringOffset(t *testing.T) {
var cases = []struct {
input string
inputOffset int
expectedOffset int
}{
{"simple string with no escape sequences", 20, 20},
{"simple string with no escape sequences", 500, -1},
{`\;\088\\\;\120\\`, 0, 0},
{`\;\088\\\;\120\\`, 1, 2},
{`\;\088\\\;\120\\`, 2, 6},
{`\;\088\\\;\120\\`, 3, 8},
{`\;\088\\\;\120\\`, 4, 10},
{`\;\088\\\;\120\\`, 5, 14},
{`\;\088\\\;\120\\`, 6, 16},
{`\;\088\\\;\120\\`, 7, -1},
}
for i, test := range cases {
outputOffset := escapedStringOffset(test.input, test.inputOffset)
if outputOffset != test.expectedOffset {
t.Errorf(
"Test %d (input %#q offset %d) returned offset %d but expected %d",
i, test.input, test.inputOffset, outputOffset, test.expectedOffset,
)
}
}
}
Loading