Harden potentially insecure deserialization operations
defectdojo:python/avoid-insecure-deserialization​
| Importance | Review Guidance | Requires Scanning Tool |
|---|---|---|
| High | Merge After Cursory Review | Yes (DefectDojo) |
Use of insecure deserialization can potentially lead to arbitrary code execution. This codemod addresses this issue with two different serialization providers:
Each is described in more detail below.
PyYAML​
The default loaders in PyYAML are not safe to use with untrusted data. They potentially make your application vulnerable to arbitrary code execution attacks. If you open a YAML file from an untrusted source, and the file is loaded with the default loader, an attacker could execute arbitrary code on your machine.
Calling yaml.load() without an explicit loader argument is equivalent to calling it with Loader=yaml.Loader, which is unsafe. This usage [has been deprecated](https://github.com/yaml/pyyaml/wiki/PyYAML-yaml.load(input\)-Deprecation) since PyYAML 5.1. This codemod will add an explicit SafeLoader argument to all yaml.load() calls that don't use an explicit loader.
The changes from this codemod look like the following:
import yaml
data = b'!!python/object/apply:subprocess.Popen \\n- ls'
- deserialized_data = yaml.load(data, yaml.Loader)
+ deserialized_data = yaml.load(data, Loader=yaml.SafeLoader)
Pickle​
Python's pickle module is notoriouly insecure. While it is very useful for serializing and deserializing Python objects, it is not safe to use pickle to load data from untrusted sources. This is because pickle can execute arbitrary code when loading data. This can be exploited by an attacker to execute arbitrary code on your system. Unlike yaml there is no concept of a "safe" loader in pickle. Therefore, it is recommended to avoid pickle and to use a different serialization format such as json or yaml when working with untrusted data.
However, if you must use pickle to load data from an untrusted source, we recommend using the open-source fickling library. fickling is a drop-in replacement for pickle that validates the data before loading it and checks for the possibility of code execution. This makes it much safer (although still not entirely safe) to use pickle to load data from untrusted sources.
This codemod replaces calls to pickle.load with fickling.load in Python code. It also adds an import statement for fickling if it is not already present.
The changes look like the following:
- import pickle
+ import fickling
- data = pickle.load(file)
+ data = fickling.load(file)
If you have feedback on this codemod, please let us know!
F.A.Q.​
Why is this codemod marked as Merge After Cursory Review?​
This change is generally safe and will prevent deserialization vulnerabilities.
Codemod Settings​
N/A
References​
N/A