From 7c39c5478a3934ae862aa5d86718d7414ecb3a3c Mon Sep 17 00:00:00 2001
From: Udo Klimaschewski <udo.klimaschewski@gmx.net>
Date: Thu, 8 Jun 2023 10:22:58 +0200
Subject: [PATCH] fixes OutOfMemoryError/memory leak on e.<number> input #374
 (#385)

* fixes OutOfMemoryError/memory leak on e.<number> input #374
---
 .../com/ezylang/evalex/parser/Tokenizer.java  | 65 ++++++++++++-------
 .../parser/ShuntingYardExceptionsTest.java    |  4 +-
 .../parser/TokenizerNumberLiteralTest.java    |  2 +-
 3 files changed, 47 insertions(+), 24 deletions(-)

diff --git a/src/main/java/com/ezylang/evalex/parser/Tokenizer.java b/src/main/java/com/ezylang/evalex/parser/Tokenizer.java
index abb20856..08d2979d 100644
--- a/src/main/java/com/ezylang/evalex/parser/Tokenizer.java
+++ b/src/main/java/com/ezylang/evalex/parser/Tokenizer.java
@@ -351,33 +351,54 @@ private boolean infixOperatorAllowed() {
   }
 
   private Token parseNumberLiteral() throws ParseException {
-    int tokenStartIndex = currentColumnIndex;
-    StringBuilder tokenValue = new StringBuilder();
     int nextChar = peekNextChar();
     if (currentChar == '0' && (nextChar == 'x' || nextChar == 'X')) {
-      // hexadecimal number, consume "0x"
+      return parseHexNumberLiteral();
+    } else {
+      return parseDecimalNumberLiteral();
+    }
+  }
+
+  private Token parseDecimalNumberLiteral() throws ParseException {
+    int tokenStartIndex = currentColumnIndex;
+    StringBuilder tokenValue = new StringBuilder();
+
+    int lastChar = -1;
+    boolean scientificNotation = false;
+    while (currentChar != -1 && isAtNumberChar()) {
+      if (currentChar == 'e' || currentChar == 'E') {
+        scientificNotation = true;
+      }
       tokenValue.append((char) currentChar);
+      lastChar = currentChar;
       consumeChar();
+    }
+    // illegal scientific format literal
+    if (scientificNotation
+        && (lastChar == 'e'
+            || lastChar == 'E'
+            || lastChar == '+'
+            || lastChar == '-'
+            || lastChar == '.')) {
+      throw new ParseException(
+          new Token(tokenStartIndex, tokenValue.toString(), TokenType.NUMBER_LITERAL),
+          "Illegal scientific format");
+    }
+    return new Token(tokenStartIndex, tokenValue.toString(), TokenType.NUMBER_LITERAL);
+  }
+
+  private Token parseHexNumberLiteral() {
+    int tokenStartIndex = currentColumnIndex;
+    StringBuilder tokenValue = new StringBuilder();
+
+    // hexadecimal number, consume "0x"
+    tokenValue.append((char) currentChar);
+    consumeChar();
+    tokenValue.append((char) currentChar);
+    consumeChar();
+    while (currentChar != -1 && isAtHexChar()) {
       tokenValue.append((char) currentChar);
       consumeChar();
-      while (currentChar != -1 && isAtHexChar()) {
-        tokenValue.append((char) currentChar);
-        consumeChar();
-      }
-    } else {
-      // decimal number
-      int lastChar = -1;
-      while (currentChar != -1 && isAtNumberChar()) {
-        tokenValue.append((char) currentChar);
-        lastChar = currentChar;
-        consumeChar();
-      }
-      // illegal scientific format literal
-      if (lastChar == 'e' || lastChar == 'E' || lastChar == '+' || lastChar == '-') {
-        throw new ParseException(
-            new Token(tokenStartIndex, tokenValue.toString(), TokenType.NUMBER_LITERAL),
-            "Illegal scientific format");
-      }
     }
     return new Token(tokenStartIndex, tokenValue.toString(), TokenType.NUMBER_LITERAL);
   }
@@ -485,7 +506,7 @@ private boolean isAtNumberStart() {
   private boolean isAtNumberChar() {
     int previousChar = peekPreviousChar();
 
-    if (previousChar == 'e' || previousChar == 'E') {
+    if ((previousChar == 'e' || previousChar == 'E') && currentChar != '.') {
       return Character.isDigit(currentChar) || currentChar == '+' || currentChar == '-';
     }
 
diff --git a/src/test/java/com/ezylang/evalex/parser/ShuntingYardExceptionsTest.java b/src/test/java/com/ezylang/evalex/parser/ShuntingYardExceptionsTest.java
index 8f237231..9a261b12 100644
--- a/src/test/java/com/ezylang/evalex/parser/ShuntingYardExceptionsTest.java
+++ b/src/test/java/com/ezylang/evalex/parser/ShuntingYardExceptionsTest.java
@@ -151,7 +151,9 @@ void testFunctionTooManyParameters() {
         "Hello 1 + 1",
         "Hello World",
         "Hello 1",
-        "1 2"
+        "1 2",
+        "e.1",
+        "E.1"
       })
   void testTooManyOperands(String expressionString) {
     Expression expression = new Expression(expressionString);
diff --git a/src/test/java/com/ezylang/evalex/parser/TokenizerNumberLiteralTest.java b/src/test/java/com/ezylang/evalex/parser/TokenizerNumberLiteralTest.java
index 2c41fcbf..a27c2296 100644
--- a/src/test/java/com/ezylang/evalex/parser/TokenizerNumberLiteralTest.java
+++ b/src/test/java/com/ezylang/evalex/parser/TokenizerNumberLiteralTest.java
@@ -83,7 +83,7 @@ void testSciOK() throws ParseException {
   }
 
   @ParameterizedTest
-  @ValueSource(strings = {"2e", "2E", "2e+", "2E+", "2e-", "2E-", "2e."})
+  @ValueSource(strings = {"2e", "2E", "2e+", "2E+", "2e-", "2E-", "2e.", "2E.", "2ex", "2Ex"})
   void testScientificLiteralsParseException(String expression) {
     assertThatThrownBy(() -> new Tokenizer(expression, configuration).parse())
         .isInstanceOf(ParseException.class)