8285081: Improve XPath operators count accuracy

Backport-of: 8e0783917975075aae5d586f0076d5093afb0b62

Author: GoeLin

src/java.xml/share/classes/com/sun/java_cup/internal/runtime/lr_parser.java

@@ -137,7 +137,7 @@
  * @see     com.sun.java_cup.internal.runtime.virtual_parse_stack
  * @author  Frank Flannery
  *
- * @LastModified: Jan 2022
+ * @LastModified: June 2022
  */
 
 public abstract class lr_parser {
@@ -149,6 +149,7 @@ public abstract class lr_parser {
     private int grpCount = 0;
     private int opCount = 0;
     private int totalOpCount = 0;
+    private int lastSym;
 
   /*-----------------------------------------------------------*/
   /*--- Constructor(s) ----------------------------------------*/
@@ -377,13 +378,17 @@ public Symbol scan() throws Exception {
           opCount++; // function
           isLiteral = false;
       } else if (contains(sym.OPERATORS, s.sym)) {
-          opCount++;
+          // axis nodetest is counted as one step, so not counted if last=DCOLON
+          if (lastSym != sym.DCOLON) {
+              opCount++;
+          }
           isLiteral = false;
       }
 
       if (s.sym == sym.Literal || s.sym == sym.QNAME) {
           isLiteral = true;
       }
+      lastSym = s.sym;
 
     return s;
   }
@@ -588,6 +593,7 @@ public Symbol parse() throws java.lang.Exception
       isLiteral = false;
       grpCount = 0;
       opCount = 0;
+      lastSym = -1;
 
       /* get the first token */
       cur_token = scan();

src/java.xml/share/classes/com/sun/org/apache/xalan/internal/xsltc/compiler/sym.java

@@ -25,13 +25,11 @@
 
 package com.sun.org.apache.xalan.internal.xsltc.compiler;
 
-import java.util.Arrays;
-
 /**
  * CUP generated class containing symbol constants.
  * This class was generated by CUP v0.10j on Fri Feb 27 13:01:50 PST 2004.
  *
- * @LastModified: Jan 2022
+ * @LastModified: June 2022
  */
 public class sym {
   /* terminals */
@@ -92,9 +90,11 @@ public class sym {
   /*
     AXES: count once at DCOLON,
           these axes names are therefore not counted:
-            NAMESPACE, FOLLOWINGSIBLING, CHILD, DESCENDANTORSELF, DESCENDANT
-          , PRECEDINGSIBLING, SELF, ANCESTORORSELF, PRECEDING, ANCESTOROR, PARENT, FOLLOWING, ATTRIBUTE
+          NAMESPACE, FOLLOWINGSIBLING, CHILD, DESCENDANTORSELF, DESCENDANT,
+          PRECEDINGSIBLING, SELF, ANCESTORORSELF, PRECEDING, ANCESTOROR, PARENT,
+          FOLLOWING, ATTRIBUTE
   */
-  public static final int[] OPERATORS = {GE, SLASH, ATSIGN, LPAREN, DCOLON,
-      MINUS, STAR, LT, OR, DIV, PLUS, LE, VBAR, MOD, EQ, LBRACK, DOLLAR, NE, GT};
+  public static final int[] OPERATORS = {GT, GE, EQ, NE, LT, LE, SLASH, DSLASH,
+      DOT, DDOT, ATSIGN, DCOLON, PLUS, MINUS, STAR, DIV, MOD, AND, OR, LPAREN,
+      LBRACK, VBAR, DOLLAR, NODE, TEXT, PI, PIPARAM};
 }

src/java.xml/share/classes/com/sun/org/apache/xpath/internal/compiler/Lexer.java

@@ -32,7 +32,7 @@
  * This class is in charge of lexical processing of the XPath
  * expression into tokens.
  *
- * @LastModified: Apr 2022
+ * @LastModified: June 2022
  */
 class Lexer
 {
@@ -155,6 +155,7 @@ void tokenize(String pat, List<String> targetStrings)
     boolean isStartOfPat = true;
     boolean isAttrName = false;
     boolean isNum = false;
+    boolean isAxis = false;
 
     // Nesting of '[' so we can know if the given element should be
     // counted inside the m_patternMap.
@@ -254,8 +255,7 @@ void tokenize(String pat, List<String> targetStrings)
             // check operator symbol
             String s = pat.substring(startSubstring, i);
             if (Token.contains(s)) {
-                m_opCount++;
-                isLiteral = false;
+                incrementCount();
             }
             addToTokenQueue(s);
           }
@@ -339,23 +339,45 @@ else if (Token.STAR == c)
         {
           nesting--;
         }
-        else if ((Token.LPAREN == c) || (Token.LBRACK == c))
+        else if (Token.LBRACK == c)
         {
           nesting++;
-          if (!isLiteral && (Token.LPAREN == c)) {
-            m_grpCount++;
-            m_opCount++;
-            isLiteral = false;
+          incrementCount();
+          isAxis = false;
+        }
+        else if ((Token.LPAREN == c))
+        {
+          nesting++;
+          if (isLiteral) {
+              if (!isAxis) {
+                  incrementCount();
+              }
+          } else {
+              m_grpCount++;
+              incrementCount();
           }
+          isAxis = false;
         }
 
-        if ((Token.GT == c || Token.LT == c || Token.EQ == c) && Token.EQ != peekNext(pat, i)) {
-            m_opCount++;
-            isLiteral = false;
+        if ((Token.GT == c || Token.LT == c || Token.EQ == c || Token.EM == c)) {
+            if (Token.EQ != peekNext(pat, i)) {
+                incrementCount();
+            }
         }
-        else if ((Token.LPAREN != c) && (Token.RPAREN != c) && (Token.RBRACK != c)) {
-            m_opCount++;
-            isLiteral = false;
+        else if (Token.SLASH == c) {
+            isAxis = false;
+            if (Token.SLASH != peekNext(pat, i)) {
+                incrementCount();
+            }
+        }
+        // '(' and '[' already counted above; ':' is examined in case below
+        // ',' is part of a function
+        else if ((Token.LPAREN != c) && (Token.LBRACK != c) && (Token.RPAREN != c)
+                && (Token.RBRACK != c) && (Token.COLON != c) && (Token.COMMA != c)) {
+            if (Token.STAR != c || !isAxis) {
+                incrementCount();
+            }
+            isAxis = false;
         }
 
         addToTokenQueue(pat.substring(i, i + 1));
@@ -376,6 +398,7 @@ else if ((Token.LPAREN != c) && (Token.RPAREN != c) && (Token.RBRACK != c)) {
             startSubstring = -1;
             posOfNSSep = -1;
             m_opCount++;
+            isAxis = true;
             addToTokenQueue(pat.substring(i - 1, i + 1));
 
             break;
@@ -389,6 +412,9 @@ else if ((Token.LPAREN != c) && (Token.RPAREN != c) && (Token.RBRACK != c)) {
       // fall through on purpose
       default :
         isLiteral = true;
+        if (!isNum && Token.DOT == c && Token.DOT != peekNext(pat, i)) {
+            incrementCount();
+        }
         if (-1 == startSubstring)
         {
           startSubstring = i;
@@ -443,6 +469,11 @@ else if (null != targetStrings)
     m_processor.m_queueMark = 0;
   }
 
+  private void incrementCount() {
+      m_opCount++;
+      isLiteral = false;
+  }
+
   /**
    * Peeks at the next character without advancing the index.
    * @param s the input string