remove email address detection in string literals (#876)

Signed-off-by: Max Fisher <[email protected]>

internal/staticanalysis/analyze_test.go

@@ -62,7 +62,6 @@ func makeDesiredResult(files ...testFile) *Result {
 				SuspiciousIdentifiers: []signals.SuspiciousIdentifier{},
 				EscapedStrings:        []signals.EscapedString{},
 				Base64Strings:         []string{},
-				EmailAddresses:        []string{},
 				HexStrings:            []string{},
 				IPAddresses:           []string{},
 				URLs:                  []string{},

internal/staticanalysis/signals/analyze.go

@@ -40,7 +40,6 @@ func AnalyzeSingle(parseData parsing.SingleResult) FileSignals {
 		SuspiciousIdentifiers: []SuspiciousIdentifier{},
 		URLs:                  []string{},
 		IPAddresses:           []string{},
-		EmailAddresses:        []string{},
 	}
 
 	for _, name := range identifierNames {
@@ -57,7 +56,6 @@ func AnalyzeSingle(parseData parsing.SingleResult) FileSignals {
 		signals.HexStrings = append(signals.HexStrings, detections.FindHexSubstrings(sl.Value)...)
 		signals.URLs = append(signals.URLs, detections.FindURLs(sl.Value)...)
 		signals.IPAddresses = append(signals.IPAddresses, detections.FindIPAddresses(sl.Value)...)
-		signals.EmailAddresses = append(signals.EmailAddresses, detections.FindEmailAddresses(sl.Value)...)
 		if detections.IsHighlyEscaped(sl, 8, 0.25) {
 			escapedString := EscapedString{
 				Value:           sl.Value,

internal/staticanalysis/signals/detections/addresses.go

@@ -61,9 +61,6 @@ var urlPathAndQuery = regexp.MustCompile(`(?:/(?:[^.]\S*)?)?(?:\?\S*)?`)
 // urlRegex is a fairly permissive url regex. Parts: scheme, subdomains, TLD, port, url query
 var urlRegexp = regexp.MustCompile(fmt.Sprintf(`%s:(?://%s)%s`, urlSchemes, urlAuthority, urlPathAndQuery))
 
-var emailUsername = regexp.MustCompile(`[^\s@]{1,64}`)
-var emailRegexp = regexp.MustCompile(fmt.Sprintf(`(?:mailto)?%s@[^\s@]{1,255}`, emailUsername))
-
 var ipv4URLRegexp = regexp.MustCompile(fmt.Sprintf(`%s://(?:%s)%s?%s`, urlSchemes, ipv4Regexp, port, urlPathAndQuery))
 var ipv6URLRegexp = regexp.MustCompile(fmt.Sprintf(`%s://\[(?:%s)]%s?%s`, urlSchemes, ipv6Regexp, port, urlPathAndQuery))
 
@@ -75,10 +72,6 @@ func FindURLs(s string) []string {
 	return urls
 }
 
-func FindEmailAddresses(s string) []string {
-	return emailRegexp.FindAllString(s, -1)
-}
-
 func findIPv4Addresses(s string) []string {
 	return ipv4Regexp.FindAllString(s, -1)
 }

internal/staticanalysis/signals/detections/addresses_test.go

@@ -285,21 +285,3 @@ func TestIPv6Regexp(t *testing.T) {
 		}
 	}
 }
-
-// TestEmailRegexp tests exact matching on single email addresses
-// The regexp is fairly lenient so some technically invalid email addresses
-// may get picked up as valid.
-func TestEmailRegexp(t *testing.T) {
-	for _, addr := range validEmailAddresses {
-		result := FindEmailAddresses(addr)
-		if !(len(result) == 1 && addr == result[0]) {
-			t.Errorf("expected to detect valid email address %s, got %v", addr, result)
-		}
-	}
-	for _, addr := range invalidEmailAddresses {
-		result := FindEmailAddresses(addr)
-		if len(result) == 1 && addr == result[0] {
-			t.Errorf("expected not to detect invalid email address %s, got %v", addr, result)
-		}
-	}
-}

internal/staticanalysis/signals/file_signals.go

@@ -30,9 +30,6 @@ type FileSignals struct {
 	// length in order to reduce the number of false positives.
 	Base64Strings []string `json:"base64_strings"`
 
-	// EmailAddresses contains any email addresses found in string literals
-	EmailAddresses []string `json:"email_addresses"`
-
 	// HexStrings holds a list of (substrings of) string literals found in the
 	// file that contain long (>8 digits) hexadecimal digit sequences.
 	HexStrings []string `json:"hex_strings"`
@@ -53,7 +50,6 @@ func (s FileSignals) String() string {
 		fmt.Sprintf("escaped strings: %v", s.EscapedStrings),
 		fmt.Sprintf("potential base64 strings: %v", s.Base64Strings),
 		fmt.Sprintf("hex strings: %v", s.HexStrings),
-		fmt.Sprintf("email addresses: %v", s.EmailAddresses),
 		fmt.Sprintf("hex strings: %v", s.HexStrings),
 		fmt.Sprintf("IP addresses: %v", s.IPAddresses),
 		fmt.Sprintf("URLs: %v", s.URLs),

internal/staticanalysis/signals/file_signals_test.go

@@ -25,7 +25,6 @@ var fileSignalsTestCases = []fileSignalsTestCase{
 			SuspiciousIdentifiers: []SuspiciousIdentifier{},
 			EscapedStrings:        []EscapedString{},
 			Base64Strings:         []string{},
-			EmailAddresses:        []string{},
 			HexStrings:            []string{},
 			IPAddresses:           []string{},
 			URLs:                  []string{},
@@ -49,7 +48,6 @@ var fileSignalsTestCases = []fileSignalsTestCase{
 			SuspiciousIdentifiers: []SuspiciousIdentifier{{Name: "a", Rule: "single"}},
 			EscapedStrings:        []EscapedString{},
 			Base64Strings:         []string{},
-			EmailAddresses:        []string{},
 			HexStrings:            []string{},
 			IPAddresses:           []string{},
 			URLs:                  []string{},
@@ -85,7 +83,6 @@ var fileSignalsTestCases = []fileSignalsTestCase{
 			},
 			EscapedStrings: []EscapedString{},
 			Base64Strings:  []string{},
-			EmailAddresses: []string{},
 			HexStrings:     []string{},
 			IPAddresses:    []string{},
 			URLs:           []string{},
@@ -120,7 +117,6 @@ var fileSignalsTestCases = []fileSignalsTestCase{
 			},
 			EscapedStrings: []EscapedString{},
 			Base64Strings:  []string{"aGVsbG8gd29ybGQK"},
-			EmailAddresses: []string{"[email protected]"},
 			HexStrings:     []string{"21323492394"},
 			IPAddresses:    []string{"8.8.8.8", "e3fc:234a:2341::abcd"},
 			URLs:           []string{"https://this.is.a.website.com"},
@@ -141,7 +137,6 @@ var fileSignalsTestCases = []fileSignalsTestCase{
 			IdentifierLengths:     valuecounts.New(),
 			SuspiciousIdentifiers: []SuspiciousIdentifier{},
 			Base64Strings:         []string{},
-			EmailAddresses:        []string{},
 			HexStrings:            []string{},
 			IPAddresses:           []string{},
 			URLs:                  []string{},