Skip to content
120 changes: 109 additions & 11 deletions docs/Secure-Coding-Guide-for-Python/CWE-697/CWE-595/README.md
Original file line number Diff line number Diff line change
@@ -1,14 +1,71 @@
# CWE-595: Comparison of Object References Instead of Object Contents

In Python, the `==` operator is implemented by the `__eq__` method on an object [[python.org data model 2023](https://docs.python.org/3/reference/datamodel.html?highlight=__eq__#object.__eq__)]. For built-in types like `int` and `str`, the comparison is implemented in the interpreter. The main issue comes when implementing custom classes, where the default implementation compares object references using the `is` operator. The `is` operator compares the identities of the objects, equivalent to `id(obj1) == id(obj2)`. The `id` function is built into Python, and in the CPython interpreter, the standard implementation, it returns the object's memory address [[de Langen 2023](https://realpython.com/python-is-identity-vs-equality/)].
Prevent unexpected results by knowing the differences between comparison operators such as `==` and `is`.

In Python, the `==` operator is implemented by the `__eq__` method on an object [[python.org data model 2023](https://docs.python.org/3/reference/datamodel.html?highlight=__eq__#object.__eq__)]. For built-in types like `int` and `str`, the comparison is implemented in the interpreter. The main issue comes when implementing custom classes, where the default implementation compares object references using the `is` operator. The `is` operator compares the identities of the objects, equivalent to `id(obj1) == id(obj2)`.
In CPython, this is their memory address. Everything in Python is an object, and each object is stored at a specific memory location [[de Langen 2023](https://realpython.com/python-is-identity-vs-equality/)].

You want to implement the `__eq__` method on a class if you believe you ever want to compare it to another object or find it in a list of objects. Actually, it is so common that the `dataclasses.dataclass` decorator by default implements it for you [[dataclasses — Data Classes — Python 3.11.4 documentation](https://docs.python.org/3/library/dataclasses.html#dataclasses.dataclass)].

Be aware of Python's memory optimization for strings and numbers as demonstrated in `example01.py` code.
Python tries to avoid allocating more memory for the same string. The process of reusing already existing strings is a Python optimization technique known as **String interning** [[sys — System-specific parameters and functions — Python 3.11.4 documentation](https://docs.python.org/3/library/sys.html#sys.intern)] According to the documentation, "CPython keeps an array of integer objects for all integers between `-5` and `256`. When you create an `int` in that range you actually just get back a reference to the existing object." [[Integer objects — Python 3.11.4 documentation](https://docs.python.org/3/c-api/long.html#c.PyLong_FromLong)]

_[example01.py:](example01.py)_

```py
""" Code Example """

print("-" * 10 + "Memory optimization with strings" + 10 * "-")
a = "foobar"
b = "foobar"
c = ''.join(["foo", "bar"])
print(f"a is b: {a} is {b}?", a is b)
print(f"a is c: {a} is {c}?", a is c)
print(f"a == c: {a} == {c}?", a == c)
print(f"size? len(a)={len(a)} len(b)={len(b)} len(c)={len(c)}")

print("-" * 10 + "Memory optimization with numbers" + 10 * "-")
a = b = 256
print (f"{a} is {b}?", a is b)
a = b = 257
print (f"{a} is {b}?", a is b)

print("-" * 10 + "Memory optimization with numbers in a loop" + 10 * "-")
a = b = 255
while(a is b):
a += 1
b += 1
print (f"{a} is {b}?", a is b)
```

**Output of example01.py:**

```bash
----------Memory optimization with strings----------
a is b: foobar is foobar? True
a is c: foobar is foobar? False
a == c: foobar == foobar? True
size? len(a)=6 len(b)=6 len(c)=6
----------Memory optimization with numbers----------
256 is 256? True
257 is 257? True
----------Memory optimization with numbers in a loop----------
256 is 256? True
257 is 257? False
```

The first set of print statements illustrates string interning. While `a` and `b` reuse the same object, `c` is created by joining two new strings, which results in an object with a different `id()`. The variables in the middle example both point to the same number object, which is why comparing them after `a = b = 257` still returns `True` even though `257` falls outside of the cached range. However, when assigning values in a loop, Python needs to allocate new objects for numbers greater than `256` and thus will create two separate objects as soon as it hits `257`. The way caching and interning works may differ between running a Python script from a file and using REPL, which may produce different results when running `example01.py` in Python's interactive mode.

## Non-Compliant Code Example

The non-compliant code shows how the default comparison operator compares object references rather than the object values. Furthermore, it displays how this causes issues when comparing lists of objects, although it applies to other types of collections as well. Finally, it shows how the `in` operator also depends on the behavior of the `__eq__` method and, therefore, also returns a non-desirable result.
The `noncompliant01.py` code demonstrates potentially unexpected outcomes when using different comparisons.

[*noncompliant01.py:*](noncompliant01.py)
* The `==` operator using `__eq__`, checks value equality for most build-in types, checks for reference equality if the `__eq__` is missing in a custom class. So `12 == 12` is `True` and `Integer(12) == Integer(12)` is `False`.
* The `==` comparing lists of objects, that also applies to other types of collections.
* The `in` operator also depends on the behavior of the `__eq__` method
* The `is` operator that checks the references point to the same object regardless of the stored value.

_[noncompliant01.py:](noncompliant01.py)_

```py
""" Non-compliant Code Example """
Expand All @@ -27,42 +84,81 @@ print(Integer(12) == Integer(12))
print([Integer(12)] == [Integer(12)])
# And this is equally this will always be False as well
print(Integer(12) in [Integer(10), Integer(12)])
# The 'is' will return True only if both references point to the same object
a = Integer(12)
b = a
# Here, a and b point to the same Integer, so 'is' returns True
print(a is b)

b = Integer(12)
# Even though b still points to an Integer of the same value, it is a new object, so 'is' returns False
print(a is b)

```

**Output of noncompliant01.py:**

```bash
False
False
False
True
False
```

## Compliant Solution

In this compliant solution the `__eq__` method is implemented and all the comparisons now correctly compares the object values, rather than the object reference.
In this compliant solution, the `__eq__` method is implemented and the comparisons that not use `is` now correctly compare the object values, rather than the object reference. The `is` operator does not call `__eq__`, hence the last print will still display `False`.

[*compliant01.py:*](compliant01.py)
_[compliant01.py:](compliant01.py)_

```py
""" Compliant Code Example """


class Integer:
def __init__(self, value):
self.value = value

def __eq__(self, other):
if isinstance(other, type(self)):
return self.value == other.value
if isinstance(other, int):
return self.value == other
return False


#####################
# exploiting above code example
#####################
# All these scenarios will now show True
print(Integer(12) == Integer(12))
print([Integer(12)] == [Integer(12)])
print(Integer(12) in [Integer(10), Integer(12)])

# By adding the handling for int we also support
print(Integer(12) == 12)
# The 'is' will return True only if both references point to the same object
a = Integer(12)
b = a
# Here, a and b point to the same Integer, so 'is' returns True
print(a is b)

b = Integer(12)
# Since the 'is' operator does not call __eq__, print below will still return False
print(a is b)

```

**Output of compliant01.py:**

```bash
True
True
True
True
True
False
```

## Automated Detection
Expand All @@ -86,3 +182,5 @@ print(Integer(12) == 12)
|[[python.org data model 2023](https://docs.python.org/3/reference/datamodel.html?highlight=__eq__#object.__eq__)]|[3. Data model — Python 3.11.3 documentation](https://docs.python.org/3/reference/datamodel.html?highlight=__eq__#object.__eq__)|
|[[de Langen 2023](https://realpython.com/python-is-identity-vs-equality/)]|[Python '!=' Is Not 'is not': Comparing Objects in Python – Real Python](https://realpython.com/python-is-identity-vs-equality/)|
|[[dataclasses — Data Classes — Python 3.11.4 documentation](https://docs.python.org/3/library/dataclasses.html#dataclasses.dataclass)]|[9. Classes — Python 3.11.3 documentation](https://docs.python.org/3/tutorial/classes.html)|
|[[sys — System-specific parameters and functions — Python 3.11.4 documentation](https://docs.python.org/3/library/sys.html#sys.intern)]|[sys — System-specific parameters and functions — Python 3.11.3 documentation](https://docs.python.org/3/library/sys.html#sys.intern)|
|[[Integer objects — Python 3.11.4 documentation](https://docs.python.org/3/c-api/long.html#c.PyLong_FromLong)]|[Integer objects — Python 3.11.4 documentation](https://docs.python.org/3/c-api/long.html#c.PyLong_FromLong)|
21 changes: 15 additions & 6 deletions docs/Secure-Coding-Guide-for-Python/CWE-697/CWE-595/compliant01.py
Original file line number Diff line number Diff line change
@@ -1,27 +1,36 @@
# SPDX-FileCopyrightText: OpenSSF project contributors
# SPDX-License-Identifier: MIT
""" Compliant Code Example """


class Integer:
def __init__(self, value):
self.value = value

def __eq__(self, other):
if isinstance(other, type(self)):
return self.value == other.value
if isinstance(other, int):
return self.value == other
return False


#####################
# exploiting above code example
#####################
# All these scenarios will now show True
print(Integer(12) == Integer(12))
print([Integer(12)] == [Integer(12)])
print(Integer(12) in [Integer(10), Integer(12)])

# By adding the handling for int we also support
print(Integer(12) == 12)
# The 'is' will return True only if both references point to the same object
a = Integer(12)
b = a
# Here, a and b point to the same Integer, so 'is' returns True
print(a is b)

b = Integer(12)
# Since the 'is' operator does not call __eq__, print below will still return False
print(a is b)
25 changes: 25 additions & 0 deletions docs/Secure-Coding-Guide-for-Python/CWE-697/CWE-595/example01.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,25 @@
# SPDX-FileCopyrightText: OpenSSF project contributors
# SPDX-License-Identifier: MIT
""" Code Example """

print("-" * 10 + "Memory optimization with strings" + 10 * "-")
a = "foobar"
b = "foobar"
c = ''.join(["foo", "bar"])
print(f"a is b: {a} is {b}?", a is b)
print(f"a is c: {a} is {c}?", a is c)
print(f"a == c: {a} == {c}?", a == c)
print(f"size? len(a)={len(a)} len(b)={len(b)} len(c)={len(c)}")

print("-" * 10 + "Memory optimization with numbers" + 10 * "-")
a = b = 256
print (f"{a} is {b}?", a is b)
a = b = 257
print (f"{a} is {b}?", a is b)

print("-" * 10 + "Memory optimization with numbers in a loop" + 10 * "-")
a = b = 255
while(a is b):
a += 1
b += 1
print (f"{a} is {b}?", a is b)
Original file line number Diff line number Diff line change
Expand Up @@ -16,3 +16,12 @@ def __init__(self, value):
print([Integer(12)] == [Integer(12)])
# And this is equally this will always be False as well
print(Integer(12) in [Integer(10), Integer(12)])
# The 'is' will return True only if both references point to the same object
a = Integer(12)
b = a
# Here, a and b point to the same Integer, so 'is' returns True
print(a is b)

b = Integer(12)
# Even though b still points to an Integer of the same value, it is a new object, so 'is' returns False
print(a is b)