0.1.2 - Merge pull request #8 from ronaldstoner/0.1.2

Code cleanup, comments, better error handling

Author: Ron Stoner

selfhash/__init__.py

@@ -1,49 +1,80 @@
 #!/usr/bin/python
-# Hash: d758637770374a601330b7e5dc60215505a338702c6188b8563034c3e487ab2b
+# Hash: 7133c3ec57f1dbdd2b35a409abebd34ea0736fde377056706b32bf955ae7d313
+# Author: Ron Stoner
+# Github: ronaldstoner
+# Website: stoner.com
 
-# No password is set for this hash as it is used to verify the selfhash module code itself and can be checked against the github repo
+# No password is set for this hash as it is used to verify the selfhash module code itself
+# This can be checked against the GitHub repository
 
-"""SelfHash - Self hashing and verification python script"""
+"""SelfHash - Self-hashing and verification Python script"""
 
 import hashlib
 import getpass
 import sys
 
+
 class SelfHash:
-    """Class for SelfHash"""
+
+    """Class to handle the self-hashing and verification of Python source code."""
     def __init__(self, bypass_salt=False):
-        """Init function"""
-        self.file_data_hash = None
-        self.source_code_hash = None
-        self.known_hash = None
-        self.bypass_salt = bypass_salt
+        """
+        Initializes the SelfHash object.
+        
+        :param bypass_salt: Flag to bypass the salt input (default is False).
+        """
+        self.file_data_hash = None      # Holds the hash of the file data without the hash line
+        self.source_code_hash = None    # Holds the calculated hash of the source code
+        self.known_hash = None          # Holds the known hash from the source code
+        self.bypass_salt = bypass_salt  # Flag to bypass the salt input
 
     def hash(self, file):
-        """Function that hashes the source code"""
+        """
+        Reads the file, calculates its hash, and verifies the integrity of the code by comparing
+        it with the known hash embedded in the source.
+        
+        :param file: The Python file to be hashed and verified.
+        """
         fail = False
 
+        # Open the file and read its contents into a list of lines
         with open(file, 'r', encoding="utf-8") as source_file:
             file_data = source_file.readlines()
 
+        # Ensure the file has at least 2 lines (the second line will contain the hash)
+        if len(file_data) < 2:
+            print("Error: The file is too short. It should have at least 2 lines.")
+            fail = True
+            sys.exit(1)  # Exit immediately if the file is too short
+
+        # Extract the hash from the line and fail if not found
         try:
             hash_line_index = [i for i, line in enumerate(file_data) if line.strip().startswith("# Hash:")][0]
         except IndexError:
             print("The '# Hash:' line was not found in the file.")
             print("Please add '# Hash: INSERT_HASH_HERE' at \nthe top of your python file and try again.")
             fail = True
+            sys.exit(1) # Exit immediately if the '# Hash: ' is not found
 
-        if fail:
-            sys.exit(1)
-
+        # Remove the hash line from the source file data to compute the rest of the code's hash
         self.file_data_hash = ''.join([line for i, line in enumerate(file_data) if i != hash_line_index])
 
+        # Prompt for a salt if not bypassing it
         if not self.bypass_salt:
             salt = getpass.getpass(prompt='This python script is protected by SelfHash.\nPlease provide a salt for the hash calculation.\nIf you do not want to provide one, just press Enter: ')
-            self.file_data_hash += salt
+            if salt:  # If a salt is provided, add it to the file data hash
+                self.file_data_hash += salt
 
+        # Calculate the SHA-256 hash of the source code (with salt if provided)
         self.source_code_hash = hashlib.sha256(self.file_data_hash.encode()).hexdigest()
 
+        # Compare the known hash to the calculated hash
         self.known_hash = file_data[hash_line_index].strip().split(' ')[-1]
+        self.known_hash = self.known_hash.strip()  # Clean up extra spaces
+
+        if len(self.known_hash) != 64:  # Ensure it is a valid SHA256 hash
+            print("Invalid hash format found in the file.")
+            fail = True
 
         if self.known_hash in ("Hash:", "INSERT_HASH_HERE"):
             print("The hash of the source code is not set yet.\nPlease run the script once and then replace INSERT_HASH_HERE with the hash.")
@@ -55,5 +86,6 @@ def hash(self, file):
             print("\033[91mFAIL\033[0m: The source code may have been tampered with or the salt/passphrase is incorrect.")
             fail = True
 
+        # Exit with error if there was any failure
         if fail:
             sys.exit(1)

selfhash/selfhash.py

@@ -1,22 +1,31 @@
 #!/usr/bin/python
-# Hash: 193dfdda9e08fb513338a76bd17dacf8eb80e6a0fb8cef90bb838466e073f78f
+# Hash: 8b74bf84000993da1f0fbb10e97197c7443b978b43edd1fa5f3957e729e49718
+# Author: Ron Stoner
+# Github: ronaldstoner
+# Website: stoner.com
 
 """SelfHash setup.py file package metadata"""
 
 from setuptools import setup, find_packages
 
 setup(
     name='selfhash',
-    version='0.1.1',
+    version='0.1.2',
     packages=find_packages(),
     author='Ron Stoner',
     author_email='ron@stoner.com',
     description='A package to self hash and verify a python script',
-    long_description_content_type='text/markdown',
+    long_description_content_type='text/markdown; charset=UTF-8',
     long_description=open('README.md').read(),
     url='https://github.com/ronaldstoner/selfhash-python',
     classifiers=[
+        'Development Status :: 4 - Beta',
+        'Intended Audience :: Developers',
+        'Topic :: Software Development :: Build Tools',
         'Programming Language :: Python :: 3.6',
+        'Programming Language :: Python :: 3.7',
+        'Programming Language :: Python :: 3.8',
         'License :: OSI Approved :: MIT License',
     ],
+    keywords='self hash verify python script integrity checksum ',
 )

setup.py

@@ -1,28 +1,50 @@
 #!/usr/bin/python
-# Hash: ff09080b23d744ebcc446a3fdce8dc0da5ae1b127517a4d29c3be46219a012c4
+# Hash: 4d8077bf0fa508b5211a94521fdf436dc1316d3da6ea8d2570470b25ed2330db
+# Author: Ron Stoner
+# Github: ronaldstoner
+# Website: stoner.com
 
-"""Script to verify all python files in a directory recursively"""
+"""Script to verify all Python files in a directory recursively"""
 
 import glob
 import selfhash
 
 def hash_and_verify_files(directory):
-    """Hash and verify all found .py files"""
-    # Use glob to find all .py files in the directory and its subdirectories
-    for filename in glob.iglob(directory + '**/*.py', recursive=True):
-        print(f"Processing {filename}...")
+    """Hash and verify all .py files in the given directory recursively."""
 
-        # Instantiate a new selfhash class for each file
+    # Find all Python files in the directory and its subdirectories
+    py_files = glob.iglob(directory + '**/*.py', recursive=True)
+    total_files = sum(1 for _ in py_files)  # Count the total number of .py files
+    print(f"\nTotal Python files found: {total_files}")
+
+    # Re-run glob to get the file list again after counting
+    py_files = glob.iglob(directory + '**/*.py', recursive=True)
+
+    for filename in py_files:
+        print(f"\nProcessing {filename}...")
+
+        # Instantiate the SelfHash class for each file
         hasher = selfhash.SelfHash(bypass_salt=True)
+
+        # Perform the hash verification for each file
         hasher.hash(filename)
 
+        # Check the hash against the known value and display appropriate messages
         if hasher.known_hash == "INSERT_HASH_HERE":
-            print(f"Generated Hash for {filename}: {hasher.source_code_hash}")
-            print("Please replace INSERT_HASH_HERE with this hash and run the script again.")
+            print(f"\033[93mWARNING\033[0m: The hash is not set for {filename}.")
+            print(f"Generated Hash: {hasher.source_code_hash}")
+            print("Please replace 'INSERT_HASH_HERE' with the generated hash and run the script again.")
         elif hasher.known_hash == hasher.source_code_hash:
-            print(f"\033[92mPASS\033[0m: The program {filename} is verified and true.")
+            print(f"Expected Hash: {hasher.known_hash}")
+            print(f"Actual Hash: {hasher.source_code_hash}")
         else:
-            print(f"\033[91mFAIL\033[0m: The source code of {filename} may have been tampered with.")
+            print(f"\033[91mFAIL\033[0m: The source code of {filename} may have been tampered with or the hash does not match.")
+            print(f"Expected Hash: {hasher.known_hash}")
+            print(f"Actual Hash: {hasher.source_code_hash}")
+            print("Please investigate this file.\n")
+
+    print("\nVerification process complete.")
 
-# Run the function starting from the current directory
-hash_and_verify_files("./")
+# Run the verification function starting from the current directory
+if __name__ == "__main__":
+    hash_and_verify_files("./")

verify_all.py

@@ -1,14 +1,38 @@
-#!/usr/bin/python
-# Hash: 48717260d6a313c4cc1b3d361a852a34bd24e8ba2786c2a08ecfdaf7f1e556c2
+#!/usr/bin/python3
+# Hash: 7171f49aa3395c9177ee86c621c5635ad5f62cc52a8564e5adb6b7f2e50b786b
+# Author: Ron Stoner
+# Github: ronaldstoner
+# Website: stoner.com
 
 """Script to verify the selfhash/selfhash.py module hash"""
 
+import sys
 import selfhash
 
-# Load selfhash into a hasher and perform the verification check
-hasher = selfhash.SelfHash(bypass_salt=True)
-hasher.hash("selfhash/selfhash.py")  # replace with the actual path to the selfhash.py file
+def verify_selfhash():
+    """Verifies the hash of the selfhash.py module."""
 
-print(hasher.hash)
-# This should only run if the hash matches and the program is 'verified'
-print("The hash inside of selfhash/selfhash.py matches and looks correct.")
+    # Instantiate the SelfHash class with bypass_salt=True to skip salt prompt
+    hasher = selfhash.SelfHash(bypass_salt=True)
+
+    # Perform the hash verification check on the selfhash.py module
+    print("\nVerifying the hash of the 'selfhash/selfhash.py' module...")
+    hasher.hash("selfhash/selfhash.py")  # Replace with the actual path to selfhash.py if needed
+
+    # Output the calculated hash and comparison results
+    print("Expected Hash: ", hasher.known_hash)
+    print("Actual Hash: ", hasher.source_code_hash)
+    if hasher.known_hash != hasher.source_code_hash:
+        print("\033[91mFAIL\033[0m: The selfhash.py module has been tampered with or the hash does not match.")
+        print("Expected Hash: ", hasher.known_hash)
+        print("Actual Hash: ", hasher.source_code_hash)
+        sys.exit(1)
+    elif hasher.known_hash in ("INSERT_HASH_HERE", "Hash:"):
+        print("\033[93mWARNING\033[0m: The hash is not set yet. Please replace 'INSERT_HASH_HERE' with the calculated hash.")
+        print("Generated Hash: ", hasher.source_code_hash)
+        sys.exit(1)
+    print("\nVerification complete.")
+
+# Run the verification function
+if __name__ == "__main__":
+    verify_selfhash()