Merge pull request #364 from Mats-SX/no-antlr-in-tck

Support escaping in string literals
opencypher · Apr 11, 2019 · d38d6ff · d38d6ff
2 parents 0f580ef + 8f23266
commit d38d6ff
Show file tree

Hide file tree

Showing 2 changed files with 37 additions and 3 deletions.
diff --git a/tools/tck-api/src/main/scala/org/opencypher/tools/tck/values/CypherValueParser.scala b/tools/tck-api/src/main/scala/org/opencypher/tools/tck/values/CypherValueParser.scala
@@ -30,7 +30,7 @@ import fastparse.Parsed.{Failure, Success}
 import fastparse._
 import org.opencypher.tools.tck.values.Connection.{backward, forward}
 
-case class CypherValueParseException(msg: String) extends Exception(msg)
+case class CypherValueParseException(msg: String, expected: String) extends Exception(msg)
 
 class CypherValueParser(val orderedLists: Boolean) {
   def parse(s: String): CypherValue = {
@@ -50,7 +50,7 @@ class CypherValueParser(val orderedLists: Boolean) {
               |$locationPointer
               |
               |${extra.trace().msg}""".stripMargin
-        throw CypherValueParseException(msg)
+        throw CypherValueParseException(msg, expected)
     }
   }
 
@@ -102,7 +102,10 @@ class CypherValueParser(val orderedLists: Boolean) {
     }
 
   private def string[_: P]: P[CypherString] =
-    P("'" ~ CharsWhile(_ != ''', 0).!.map(CypherString) ~ "'")
+    P("'" ~/ (stringChunk | backslash | escape).rep.!.map { s =>
+      val escaped = s.replaceAllLiterally("\\'", "'").replaceAllLiterally("\\\\", "\\")
+      CypherString(escaped)
+    } ~ "'")
 
   private def float[_: P]: P[CypherFloat] =
     P("-".? ~ floatRepr).!.map { s =>
@@ -136,6 +139,25 @@ class CypherValueParser(val orderedLists: Boolean) {
 
   private def symbolicName[_: P]: P[String] = CharsWhileIn("a-zA-Z0-9$_").!
 
+  /**
+    * A 'simple' string chunk; without apostrophes or backslash (escape sequence)
+    */
+  private def stringChunk[_: P]: P[Unit] = {
+    CharsWhile(c => c != ''' && c != '\\')
+  }
+  /**
+    * We escape apostrophes inside strings using a backslash
+    */
+  private def escape[_: P]: P[Unit] = {
+    P("\\") ~/ P("'")
+  }
+  /**
+    * Since backslash is used
+    */
+  private def backslash[_: P]: P[Unit] = {
+    P("\\\\")
+  }
+
   private def digits[_: P]: P[Unit] = CharsWhileIn("0-9")
   private def floatRepr[_: P]: P[Unit] =
     (digits ~~ "." ~~ digits ~~ exponent.?) |

diff --git a/tools/tck-api/src/test/scala/org/opencypher/tools/tck/CypherValueParserTest.scala b/tools/tck-api/src/test/scala/org/opencypher/tools/tck/CypherValueParserTest.scala
@@ -64,6 +64,18 @@ class CypherValueParserTest extends FunSuite with Matchers {
     CypherValue("null") should equal(CypherNull)
   }
 
+  test("string escaping") {
+    CypherValue("'The Devil\\'s Advocate'") should equal(CypherString("The Devil's Advocate"))
+    CypherValue("'\\\\'") should equal(CypherString("\\"))
+    CypherValue("'\\''") should equal(CypherString("'"))
+    CypherValue("'\\'\\''") should equal(CypherString("''"))
+  }
+
+  test("incorrect escaping") {
+    (the[CypherValueParseException] thrownBy CypherValue("'\\'")).expected should equal("\"'\"")
+    (the[CypherValueParseException] thrownBy CypherValue("'''")).expected should equal("end-of-input")
+  }
+
   test("floats in exponent form") {
     CypherValue(".4e10") should equal(CypherFloat(.4e10))
     CypherValue(".4e-10") should equal(CypherFloat(.4e-10))