Apostrophe handling in CpsPathParser

Apostrophe is not currently handled correctly, and having apostrophe in
the xpath will lead to various errors.
For example, normalizing this xpath works:
  /path[@name="I'm quoted"] -> /path[@name='I\'m quoted']
However the resulting xpath will throw a PathParsingException if parsed!
(Thus path normalization is not idempotent.)

- Use '' for escaping apostrophe in single quoted leaf value,
  to comply with XPath standard (and use "" for escaping in ").
- Use Liquibase to make existing data comply with new rules.
- Leaf values in data leaves are now unescaped, e.g. "I'm quoted"
- Quoting is now consistent for leaf/text/contains conditions.

Issue-ID: CPS-1769
Signed-off-by: danielhanrahan <daniel.hanrahan@est.tech>
Change-Id: Iafc287f738254d7f99706c6bc548091c0ecd5aa0

3 files changed, 37 insertions, 22 deletions

diff --git a/cps-path-parser/src/main/antlr4/org/onap/cps/cpspath/parser/antlr4/CpsPath.g4 b/cps-path-parser/src/main/antlr4/org/onap/cps/cpspath/parser/antlr4/CpsPath.g4
index c88a82265..3aef120fe 100644
--- a/cps-path-parser/src/main/antlr4/org/onap/cps/cpspath/parser/antlr4/CpsPath.g4
+++ b/cps-path-parser/src/main/antlr4/org/onap/cps/cpspath/parser/antlr4/CpsPath.g4
@@ -1,6 +1,6 @@
 /*
  *  ============LICENSE_START=======================================================
- *  Copyright (C) 2021-2022 Nordix Foundation
+ *  Copyright (C) 2021-2023 Nordix Foundation
  *  Modifications Copyright (C) 2023 TechMahindra Ltd
  *  ================================================================================
  *  Licensed under the Apache License, Version 2.0 (the "License");
@@ -19,6 +19,12 @@
  *  ============LICENSE_END=========================================================
  */
 
+/*
+ * Parser Rules
+ * Some of the parser rules below are inspired by
+ * https://github.com/antlr/grammars-v4/blob/master/xpath/xpath31/XPath31Parser.g4
+ */
+
 grammar CpsPath ;
 
 cpsPath : ( prefix | descendant | incorrectPrefix ) multipleLeafConditions? textFunctionCondition? containsFunctionCondition? ancestorAxis? invalidPostFix?;
@@ -60,7 +66,7 @@ invalidPostFix : (AT | CB | COLONCOLON | comparativeOperators ).+ ;
 /*
  * Lexer Rules
  * Most of the lexer rules below are inspired by
- * https://raw.githubusercontent.com/antlr/grammars-v4/master/xpath/xpath31/XPath31.g4
+ * https://github.com/antlr/grammars-v4/blob/master/xpath/xpath31/XPath31Lexer.g4
  */
 
 AT : '@' ;
@@ -89,9 +95,9 @@ IntegerLiteral : FragDigits ;
 // Add below type definitions for leafvalue comparision in https://jira.onap.org/browse/CPS-440
 DecimalLiteral : ('.' FragDigits) | (FragDigits '.' [0-9]*) ;
 DoubleLiteral : (('.' FragDigits) | (FragDigits ('.' [0-9]*)?)) [eE] [+-]? FragDigits ;
-StringLiteral : ('"' (FragEscapeQuot | ~[^"])*? '"') | ('\'' (FragEscapeApos | ~['])*? '\'') ;
+StringLiteral : '"' (~["] | FragEscapeQuot)* '"' | '\'' (~['] | FragEscapeApos)* '\'' ;
 fragment FragEscapeQuot : '""' ;
-fragment FragEscapeApos : '\'' ;
+fragment FragEscapeApos : '\'\'';
 fragment FragDigits : [0-9]+ ;
 
 QName  : FragQName ;
diff --git a/cps-path-parser/src/main/java/org/onap/cps/cpspath/parser/CpsPathBuilder.java b/cps-path-parser/src/main/java/org/onap/cps/cpspath/parser/CpsPathBuilder.java
index 99135962f..de261e64b 100644
--- a/cps-path-parser/src/main/java/org/onap/cps/cpspath/parser/CpsPathBuilder.java
+++ b/cps-path-parser/src/main/java/org/onap/cps/cpspath/parser/CpsPathBuilder.java
@@ -79,18 +79,13 @@ public class CpsPathBuilder extends CpsPathBaseListener {
 
     @Override
     public void exitLeafCondition(final LeafConditionContext ctx) {
-        Object comparisonValue;
+        final Object comparisonValue;
         if (ctx.IntegerLiteral() != null) {
             comparisonValue = Integer.valueOf(ctx.IntegerLiteral().getText());
         } else if (ctx.StringLiteral() != null) {
-            final boolean wasWrappedInDoubleQuote = ctx.StringLiteral().getText().startsWith("\"");
-            comparisonValue = stripFirstAndLastCharacter(ctx.StringLiteral().getText());
-            if (wasWrappedInDoubleQuote) {
-                comparisonValue = String.valueOf(comparisonValue).replace("'", "\\'");
-            }
+            comparisonValue = unwrapQuotedString(ctx.StringLiteral().getText());
         } else {
-            throw new PathParsingException(
-                    "Unsupported comparison value encountered in expression" + ctx.getText());
+            throw new PathParsingException("Unsupported comparison value encountered in expression" + ctx.getText());
         }
         leafContext(ctx.leafName(), comparisonValue);
     }
@@ -140,13 +135,13 @@ public class CpsPathBuilder extends CpsPathBaseListener {
     @Override
     public void exitTextFunctionCondition(final TextFunctionConditionContext ctx) {
         cpsPathQuery.setTextFunctionConditionLeafName(ctx.leafName().getText());
-        cpsPathQuery.setTextFunctionConditionValue(stripFirstAndLastCharacter(ctx.StringLiteral().getText()));
+        cpsPathQuery.setTextFunctionConditionValue(unwrapQuotedString(ctx.StringLiteral().getText()));
     }
 
     @Override
     public void exitContainsFunctionCondition(final CpsPathParser.ContainsFunctionConditionContext ctx) {
         cpsPathQuery.setContainsFunctionConditionLeafName(ctx.leafName().getText());
-        cpsPathQuery.setContainsFunctionConditionValue(stripFirstAndLastCharacter(ctx.StringLiteral().getText()));
+        cpsPathQuery.setContainsFunctionConditionValue(unwrapQuotedString(ctx.StringLiteral().getText()));
     }
 
     @Override
@@ -173,10 +168,6 @@ public class CpsPathBuilder extends CpsPathBaseListener {
         return cpsPathQuery;
     }
 
-    private static String stripFirstAndLastCharacter(final String wrappedString) {
-        return wrappedString.substring(1, wrappedString.length() - 1);
-    }
-
     @Override
     public void exitContainerName(final CpsPathParser.ContainerNameContext ctx) {
         final String containerName = ctx.getText();
@@ -207,11 +198,25 @@ public class CpsPathBuilder extends CpsPathBaseListener {
                                     .append(name)
                                     .append(getLastElement(comparativeOperators))
                                     .append("'")
-                                    .append(value)
+                                    .append(value.toString().replace("'", "''"))
                                     .append("'");
     }
 
-    private String getLastElement(final List<String> listOfStrings) {
+    private static String getLastElement(final List<String> listOfStrings) {
         return listOfStrings.get(listOfStrings.size() - 1);
     }
+
+    private static String unwrapQuotedString(final String wrappedString) {
+        final boolean wasWrappedInSingleQuote = wrappedString.startsWith("'");
+        final String value = stripFirstAndLastCharacter(wrappedString);
+        if (wasWrappedInSingleQuote) {
+            return value.replace("''", "'");
+        } else {
+            return value.replace("\"\"", "\"");
+        }
+    }
+
+    private static String stripFirstAndLastCharacter(final String wrappedString) {
+        return wrappedString.substring(1, wrappedString.length() - 1);
+    }
 }
diff --git a/cps-path-parser/src/test/groovy/org/onap/cps/cpspath/parser/CpsPathQuerySpec.groovy b/cps-path-parser/src/test/groovy/org/onap/cps/cpspath/parser/CpsPathQuerySpec.groovy
index 78963033d..0017242ab 100644
--- a/cps-path-parser/src/test/groovy/org/onap/cps/cpspath/parser/CpsPathQuerySpec.groovy
+++ b/cps-path-parser/src/test/groovy/org/onap/cps/cpspath/parser/CpsPathQuerySpec.groovy
@@ -46,6 +46,10 @@ class CpsPathQuerySpec extends Specification {
             'spaces around ='      | '/parent/child[@common-leaf-name-int = 5]'                 || '/parent/child'                                 | 'common-leaf-name-int' | 5
             'key in top container' | '/parent[@common-leaf-name-int=5]'                         || '/parent'                                       | 'common-leaf-name-int' | 5
             'parent list'          | '/shops/shop[@id=1]/categories[@id=1]/book[@title="Dune"]' || "/shops/shop[@id='1']/categories[@id='1']/book" | 'title'                | 'Dune'
+            "' in double quote"    | '/parent[@common-leaf-name="leaf\'value"]'                 || '/parent'                                       | 'common-leaf-name'     | "leaf'value"
+            "' in single quote"    | "/parent[@common-leaf-name='leaf''value']"                 || '/parent'                                       | 'common-leaf-name'     | "leaf'value"
+            '" in double quote'    | '/parent[@common-leaf-name="leaf""value"]'                 || '/parent'                                       | 'common-leaf-name'     | 'leaf"value'
+            '" in single quote'    | '/parent[@common-leaf-name=\'leaf"value\']'                || '/parent'                                       | 'common-leaf-name'     | 'leaf"value'
     }
 
     def 'Parse cps path of type ends with a #scenario.'() {
@@ -80,8 +84,8 @@ class CpsPathQuerySpec extends Specification {
             'parent leaf of type Integer & child'                         | '/parent/child[@code=1]/child2'                                || "/parent/child[@code='1']/child2"
             'parent leaf with double quotes'                              | '/parent/child[@code="1"]/child2'                              || "/parent/child[@code='1']/child2"
             'parent leaf with double quotes inside single quotes'         | '/parent/child[@code=\'"1"\']/child2'                          || "/parent/child[@code='\"1\"']/child2"
-            'parent leaf with single quotes inside double quotes'         | '/parent/child[@code="\'1\'"]/child2'                          || "/parent/child[@code='\\\'1\\\'']/child2"
-            'leaf with single quotes inside double quotes'                | '/parent/child[@code="\'1\'"]'                                 || "/parent/child[@code='\\\'1\\\'']"
+            'parent leaf with single quotes inside double quotes'         | '/parent/child[@code="\'1\'"]/child2'                          || "/parent/child[@code='''1''']/child2"
+            'leaf with single quotes inside double quotes'                | '/parent/child[@code="\'1\'"]'                                 || "/parent/child[@code='''1''']"
             'leaf with more than one attribute'                           | '/parent/child[@key1=1 and @key2="abc"]'                       || "/parent/child[@key1='1' and @key2='abc']"
             'parent & child with more than one attribute'                 | '/parent/child[@key1=1 and @key2="abc"]/child2'                || "/parent/child[@key1='1' and @key2='abc']/child2"
             'leaf with more than one attribute has OR operator'           | '/parent/child[@key1=1 or @key2="abc"]'                        || "/parent/child[@key1='1' or @key2='abc']"