obfuscate: add exception when parsing empty-string identifiers (#514)

obfuscate/sql_test.go

@@ -86,8 +86,6 @@ func TestSQLResourceWithError(t *testing.T) {
 }
 
 func TestSQLQuantizer(t *testing.T) {
-	assert := assert.New(t)
-
 	cases := []sqlTestCase{
 		{
 			"select * from users where id = 42",
@@ -340,6 +338,26 @@ FROM [Blogs] AS [b]
 ORDER BY [b].[Name]`,
 			`SELECT [ b ] . [ BlogId ], [ b ] . [ Name ] FROM [ Blogs ] ORDER BY [ b ] . [ Name ]`,
 		},
+		{
+			`SELECT * FROM users WHERE firstname=''`,
+			`SELECT * FROM users WHERE firstname = ?`,
+		},
+		{
+			`SELECT * FROM users WHERE firstname=' '`,
+			`SELECT * FROM users WHERE firstname = ?`,
+		},
+		{
+			`SELECT * FROM users WHERE firstname=""`,
+			`SELECT * FROM users WHERE firstname = ""`,
+		},
+		{
+			`SELECT * FROM users WHERE lastname=" "`,
+			`SELECT * FROM users WHERE lastname = ""`,
+		},
+		{
+			`SELECT * FROM users WHERE lastname="	 "`,
+			`SELECT * FROM users WHERE lastname = ""`,
+		},
 		{
 			`SELECT [b].[BlogId], [b].[Name]
 FROM [Blogs] AS [b
@@ -348,10 +366,12 @@ ORDER BY [b].[Name]`,
 		},
 	}
 
-	for _, c := range cases {
-		s := SQLSpan(c.query)
-		NewObfuscator(nil).Obfuscate(s)
-		assert.Equal(c.expected, s.Resource)
+	for i, c := range cases {
+		t.Run(strconv.Itoa(i), func(t *testing.T) {
+			s := SQLSpan(c.query)
+			NewObfuscator(nil).Obfuscate(s)
+			assert.Equal(t, c.expected, s.Resource)
+		})
 	}
 }
 

obfuscate/sql_tokenizer.go

@@ -3,6 +3,7 @@ package obfuscate
 import (
 	"bytes"
 	"strings"
+	"unicode"
 )
 
 // tokenizer.go implemenents a lexer-like iterator that tokenizes SQL and CQL
@@ -408,7 +409,15 @@ func (tkn *Tokenizer) scanString(delim uint16, typ int) (int, []byte) {
 		}
 		buffer.WriteByte(byte(ch))
 	}
-	return typ, buffer.Bytes()
+	buf := buffer.Bytes()
+	if typ == ID && len(buf) == 0 || bytes.IndexFunc(buf, func(r rune) bool { return !unicode.IsSpace(r) }) == -1 {
+		// This string is an empty or white-space only identifier.
+		// We should keep the start and end delimiters in order to
+		// avoid creating invalid queries.
+		// See: https://github.com/DataDog/datadog-trace-agent/issues/316
+		return typ, []byte{byte(delim), byte(delim)}
+	}
+	return typ, buf
 }
 
 func (tkn *Tokenizer) scanCommentType1(prefix string) (int, []byte) {