fix: Support string escape sequences in matching definitions

pact-foundation · Jun 21, 2023 · a2f7fbf · a2f7fbf
1 parent 3c78dc5
commit a2f7fbf
Show file tree

Hide file tree

Showing 2 changed files with 133 additions and 6 deletions.
diff --git a/...rc/main/kotlin/au/com/dius/pact/core/model/matchingrules/expressions/MatcherDefinition.kt b/...rc/main/kotlin/au/com/dius/pact/core/model/matchingrules/expressions/MatcherDefinition.kt
@@ -467,22 +467,98 @@ class MatcherDefinitionParser(private val lexer: MatcherDefinitionLexer) {
   fun string(): Result<String?, String> {
     lexer.skipWhitespace()
     return if (lexer.matchChar('\'')) {
-      var ch = lexer.peekNextChar()
+      var ch = lexer.nextChar()
+      var ch2 = lexer.peekNextChar()
       var stringResult = ""
-      while (ch != '\'' && ch != null) {
+      while (ch != null && ((ch == '\\' && ch2 == '\'') || (ch != '\''))) {
         stringResult += ch
-        lexer.advance()
-        ch = lexer.peekNextChar()
+        if (ch == '\\' && ch2 == '\'') {
+          stringResult += ch2
+          lexer.advance()
+        }
+        ch = lexer.nextChar()
+        ch2 = lexer.peekNextChar()
       }
 
       if (ch == '\'') {
-        lexer.advance()
-        Result.Ok(stringResult)
+        processRawString(stringResult)
       } else {
         Result.Err("Unterminated string found at index ${lexer.index}")
       }
     } else {
       Result.Err("Was expecting a string at index ${lexer.index}")
     }
   }
+
+  fun processRawString(rawString: String): Result<String, String> {
+    val buffer = StringBuilder(rawString.length)
+    val chars = rawString.chars().iterator()
+    while (chars.hasNext()) {
+      val ch = chars.nextInt().toChar()
+      if (ch == '\\') {
+        if (chars.hasNext()) {
+          when (val ch2 = chars.nextInt().toChar()) {
+            '\\' -> buffer.append(ch)
+            'b' -> buffer.append('\u0008')
+            'f' -> buffer.append('\u000C')
+            'n' -> buffer.append('\n')
+            'r' -> buffer.append('\r')
+            't' -> buffer.append('\t')
+            'u' -> {
+              if (!chars.hasNext()) {
+                return Result.Err("Invalid unicode escape found at index ${lexer.index}")
+              }
+              val code1 = chars.nextInt().toChar()
+              val b = StringBuilder(4)
+              if (code1 == '{') {
+                var c: Char? = null
+                while (chars.hasNext()) {
+                  c = chars.nextInt().toChar()
+                  if (c == '}') {
+                    break
+                  }
+                  b.append(c)
+                }
+                if (c != '}') {
+                  return Result.Err("Invalid unicode escape found at index ${lexer.index}")
+                }
+              } else {
+                b.append(code1)
+                if (!chars.hasNext()) {
+                  return Result.Err("Invalid unicode escape found at index ${lexer.index}")
+                }
+                val code2 = chars.nextInt().toChar()
+                b.append(code2)
+                if (!chars.hasNext()) {
+                  return Result.Err("Invalid unicode escape found at index ${lexer.index}")
+                }
+                val code3 = chars.nextInt().toChar()
+                b.append(code3)
+                if (!chars.hasNext()) {
+                  return Result.Err("Invalid unicode escape found at index ${lexer.index}")
+                }
+                val code4 = chars.nextInt().toChar()
+                b.append(code4)
+              }
+              val code = try {
+                b.toString().toInt(16)
+              } catch (e: NumberFormatException) {
+                return Result.Err("Invalid unicode escape found at index ${lexer.index}")
+              }
+              buffer.append(Character.toString(code))
+            }
+            else -> {
+              buffer.append(ch)
+              buffer.append(ch2)
+            }
+          }
+        } else {
+          buffer.append(ch)
+        }
+      } else {
+        buffer.append(ch)
+      }
+    }
+    return Result.Ok(buffer.toString())
+  }
 }
diff --git a/...au/com/dius/pact/core/model/matchingrules/expressions/MatchingDefinitionParserSpec.groovy b/...au/com/dius/pact/core/model/matchingrules/expressions/MatchingDefinitionParserSpec.groovy
@@ -165,4 +165,55 @@ class MatchingDefinitionParserSpec extends Specification {
     "notEmpty( 'true' )" | 'true' | ValueType.String
     'notEmpty(true)'     | 'true' | ValueType.Boolean
   }
+
+  def 'parsing string values'() {
+    expect:
+    new MatcherDefinitionParser(new MatcherDefinitionLexer(expression)).string().value == result
+
+    where:
+
+    expression                                      | result
+    "''"                                            | ''
+    "'Example value'"                               | 'Example value'
+    "'yyyy-MM-dd HH:mm:ssZZZZZ'"                    | 'yyyy-MM-dd HH:mm:ssZZZZZ'
+    "'2020-05-21 16:44:32+10:00'"                   | '2020-05-21 16:44:32+10:00'
+    "'\\w{3}\\d+'"                                  | "\\w{3}\\d+"
+    "'<?xml?><test/>'"                              | '<?xml?><test/>'
+    "'\\\$(\\.\\w+)+'"                              | "\\\$(\\.\\w+)+"
+    "'we don\\'t currently support parallelograms'" | "we don\\'t currently support parallelograms"
+    "'\\b backspace'"                               | "\b backspace"
+    "'\\f formfeed'"                                | "\f formfeed"
+    "'\\n linefeed'"                                | "\n linefeed"
+    "'\\r carriage return'"                         | "\r carriage return"
+    "'\\t tab'"                                     | "\t tab"
+    "'\\u0109 unicode hex code'"                    | "\u0109 unicode hex code"
+    "'\\u{1DF0B} unicode hex code'"                 | "${Character.toString(0x1DF0B)} unicode hex code"
+    "'\\u{1D400} unicode hex code'"                 | "𝐀 unicode hex code"
+  }
+
+  def 'process raw string'() {
+    expect:
+    new MatcherDefinitionParser(new MatcherDefinitionLexer("")).processRawString(expression).value == result
+
+    where:
+
+    expression                  | result
+    ''                          | ""
+    'Example value'             | 'Example value'
+    'not escaped \\$(\\.\\w+)+' | 'not escaped \\$(\\.\\w+)+'
+    'escaped \\\\'              | 'escaped \\'
+    'slash at end \\'           | 'slash at end \\'
+  }
+
+  def "process raw string error test"() {
+    given:
+    def parser = new MatcherDefinitionParser(new MatcherDefinitionLexer("'invalid escape \\u in string'"))
+
+    expect:
+    parser.processRawString("'invalid escape \\u in string'").errorValue() == "Invalid unicode escape found at index 0"
+    parser.processRawString('\\u0') instanceof Result.Err
+    parser.processRawString('\\u00') instanceof Result.Err
+    parser.processRawString('\\u000') instanceof Result.Err
+    parser.processRawString('\\u{000') instanceof Result.Err
+  }
 }