go-inbucket/smtpd/utils.go

package smtpd

import (
	"bytes"
	"container/list"
	"crypto/sha1"
	"fmt"
	"io"
	"strings"
)

// ParseMailboxName takes a localPart string (ex: "user+ext" without "@domain")
// and returns just the mailbox name (ex: "user").  Returns an error if
// localPart contains invalid characters; it won't accept any that must be
// quoted according to RFC3696.
func ParseMailboxName(localPart string) (result string, err error) {
	if localPart == "" {
		return "", fmt.Errorf("Mailbox name cannot be empty")
	}
	result = strings.ToLower(localPart)

	invalid := make([]byte, 0, 10)

	for i := 0; i < len(result); i++ {
		c := result[i]
		switch {
		case 'a' <= c && c <= 'z':
		case '0' <= c && c <= '9':
		case bytes.IndexByte([]byte("!#$%&'*+-=/?^_`.{|}~"), c) >= 0:
		default:
			invalid = append(invalid, c)
		}
	}

	if len(invalid) > 0 {
		return "", fmt.Errorf("Mailbox name contained invalid character(s): %q", invalid)
	}

	if idx := strings.Index(result, "+"); idx > -1 {
		result = result[0:idx]
	}
	return result, nil
}

// HashMailboxName accepts a mailbox name and hashes it.  Inbucket uses this as
// the directory to house the mailbox
func HashMailboxName(mailbox string) string {
	h := sha1.New()
	if _, err := io.WriteString(h, mailbox); err != nil {
		// This shouldn't ever happen
		return ""
	}
	return fmt.Sprintf("%x", h.Sum(nil))
}

// JoinStringList joins a List containing strings by commas
func JoinStringList(listOfStrings *list.List) string {
	if listOfStrings.Len() == 0 {
		return ""
	}
	s := make([]string, 0, listOfStrings.Len())
	for e := listOfStrings.Front(); e != nil; e = e.Next() {
		s = append(s, e.Value.(string))
	}
	return strings.Join(s, ",")
}

// ValidateDomainPart returns true if the domain part complies to RFC3696, RFC1035
func ValidateDomainPart(domain string) bool {
	if len(domain) == 0 {
		return false
	}
	if len(domain) > 255 {
		return false
	}
	if domain[len(domain)-1] != '.' {
		domain += "."
	}
	prev := '.'
	labelLen := 0
	hasAlphaNum := false

	for _, c := range domain {
		switch {
		case ('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z') ||
			('0' <= c && c <= '9') || c == '_':
			// Must contain some of these to be a valid label
			hasAlphaNum = true
			labelLen++
		case c == '-':
			if prev == '.' {
				// Cannot lead with hyphen
				return false
			}
		case c == '.':
			if prev == '.' || prev == '-' {
				// Cannot end with hyphen or double-dot
				return false
			}
			if labelLen > 63 {
				return false
			}
			if !hasAlphaNum {
				return false
			}
			labelLen = 0
			hasAlphaNum = false
		default:
			// Unknown character
			return false
		}
		prev = c
	}

	return true
}

// ParseEmailAddress unescapes an email address, and splits the local part from the domain part.
// An error is returned if the local or domain parts fail validation following the guidelines
// in RFC3696.
func ParseEmailAddress(address string) (local string, domain string, err error) {
	if address == "" {
		return "", "", fmt.Errorf("Empty address")
	}
	if len(address) > 320 {
		return "", "", fmt.Errorf("Address exceeds 320 characters")
	}
	if address[0] == '@' {
		return "", "", fmt.Errorf("Address cannot start with @ symbol")
	}
	if address[0] == '.' {
		return "", "", fmt.Errorf("Address cannot start with a period")
	}

	// Loop over address parsing out local part
	buf := new(bytes.Buffer)
	prev := byte('.')
	inCharQuote := false
	inStringQuote := false
LOOP:
	for i := 0; i < len(address); i++ {
		c := address[i]
		switch {
		case ('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z'):
			// Letters are OK
			_ = buf.WriteByte(c)
			inCharQuote = false
		case '0' <= c && c <= '9':
			// Numbers are OK
			_ = buf.WriteByte(c)
			inCharQuote = false
		case bytes.IndexByte([]byte("!#$%&'*+-/=?^_`{|}~"), c) >= 0:
			// These specials can be used unquoted
			_ = buf.WriteByte(c)
			inCharQuote = false
		case c == '.':
			// A single period is OK
			if prev == '.' {
				// Sequence of periods is not permitted
				return "", "", fmt.Errorf("Sequence of periods is not permitted")
			}
			_ = buf.WriteByte(c)
			inCharQuote = false
		case c == '\\':
			inCharQuote = true
		case c == '"':
			if inCharQuote {
				_ = buf.WriteByte(c)
				inCharQuote = false
			} else if inStringQuote {
				inStringQuote = false
			} else {
				if i == 0 {
					inStringQuote = true
				} else {
					return "", "", fmt.Errorf("Quoted string can only begin at start of address")
				}
			}
		case c == '@':
			if inCharQuote || inStringQuote {
				_ = buf.WriteByte(c)
				inCharQuote = false
			} else {
				// End of local-part
				if i > 63 {
					return "", "", fmt.Errorf("Local part must not exceed 64 characters")
				}
				if prev == '.' {
					return "", "", fmt.Errorf("Local part cannot end with a period")
				}
				domain = address[i+1:]
				break LOOP
			}
		case c > 127:
			return "", "", fmt.Errorf("Characters outside of US-ASCII range not permitted")
		default:
			if inCharQuote || inStringQuote {
				_ = buf.WriteByte(c)
				inCharQuote = false
			} else {
				return "", "", fmt.Errorf("Character %q must be quoted", c)
			}
		}
		prev = c
	}
	if inCharQuote {
		return "", "", fmt.Errorf("Cannot end address with unterminated quoted-pair")
	}
	if inStringQuote {
		return "", "", fmt.Errorf("Cannot end address with unterminated string quote")
	}

	if !ValidateDomainPart(domain) {
		return "", "", fmt.Errorf("Domain part validation failed")
	}

	return buf.String(), domain, nil
}