[python] Add hashes option, refactor Python vs. RedHatPython

Adds a new `hashes` option to the `python` plugin. When specified, this
option will generate a hash for every python file discovered in the
`/usr/lib*` and `/usr/local/lib*` canonical paths. This information is
then written to the archive as `digests.json`

Additionally, refactors the differences between the base `Python` and
`RedHatPython` classes when dealing with the system-default python
installation.

Resolves: #1801

Signed-off-by: Bissenbay Dauletbayev <bissenbay@gmail.com>
Signed-off-by: Jake Hunsaker <jhunsake@redhat.com>

1 files changed, 55 insertions, 8 deletions

diff --git a/sos/plugins/python.py b/sos/plugins/python.py
index e355f518..f2d16c1e 100644
--- a/sos/plugins/python.py
+++ b/sos/plugins/python.py
@@ -10,6 +10,9 @@
 
 from sos.plugins import Plugin, RedHatPlugin, DebianPlugin, UbuntuPlugin
 import sos.policies
+import os
+import json
+import hashlib
 
 
 class Python(Plugin, DebianPlugin, UbuntuPlugin):
@@ -21,25 +24,69 @@ class Python(Plugin, DebianPlugin, UbuntuPlugin):
 
     packages = ('python',)
 
+    python_version = "python -V"
+
     def setup(self):
-        self.add_cmd_output("python -V", suggest_filename="python-version")
+        self.add_cmd_output(
+            self.python_version, suggest_filename="python-version"
+        )
         self.add_cmd_output("pip list")
 
 
 class RedHatPython(Python, RedHatPlugin):
 
     packages = ('python', 'python36', 'python2', 'platform-python')
+    option_list = [
+        ('hashes', "gather hashes for all python files", 'slow',
+         False)]
 
     def setup(self):
         self.add_cmd_output(['python2 -V', 'python3 -V'])
-        self.add_cmd_output("pip list")
         if isinstance(self.policy, sos.policies.redhat.RHELPolicy) and \
                 self.policy.dist_version() > 7:
-            self.add_cmd_output(
-                '/usr/libexec/platform-python -V',
-                suggest_filename='python-version'
-            )
-        else:
-            self.add_cmd_output('python -V', suggest_filename='python-version')
+            self.python_version = "/usr/libexec/platform-python -V"
+        super(RedHatPython, self).setup()
+
+        if self.get_option('hashes'):
+            digests = {
+                'digests': []
+            }
+
+            py_paths = [
+                '/usr/lib',
+                '/usr/lib64',
+                '/usr/local/lib',
+                '/usr/local/lib64'
+            ]
+
+            for py_path in py_paths:
+                for root, _, files in os.walk(py_path):
+                    for file_ in files:
+                        filepath = os.path.join(root, file_)
+                        if filepath.endswith('.py'):
+                            try:
+                                with open(filepath, 'rb') as f:
+                                    digest = hashlib.sha256()
+                                    chunk = 1024
+                                    while True:
+                                        data = f.read(chunk)
+                                        if data:
+                                            digest.update(data)
+                                        else:
+                                            break
+
+                                    digest = digest.hexdigest()
+
+                                    digests['digests'].append({
+                                        'filepath': filepath,
+                                        'sha256': digest
+                                    })
+                            except IOError:
+                                self._log_error(
+                                    "Unable to read python file at %s" %
+                                    filepath
+                                )
+
+            self.add_string_as_file(json.dumps(digests), 'digests.json')
 
 # vim: set et ts=4 sw=4 :