ec2_metadata_facts - Handle decompression when EC2 instance user-data is compressed

changelogs/fragments/20230526-ec2_mertadata_facts-handle_compressed_user_data.yml

@@ -0,0 +1,3 @@
+---
+bugfixes:
+  - ec2_metadata_facts - Handle decompression when EC2 instance user-data is gzip compressed. The fetch_url method from ansible.module_utils.urls does not decompress the user-data unless the header explicitly contains ``Content-Encoding: gzip`` (https://github.com/ansible-collections/amazon.aws/pull/1575).

plugins/modules/ec2_metadata_facts.py

@@ -438,6 +438,7 @@
 import re
 import socket
 import time
+import zlib
 
 from ansible.module_utils.basic import AnsibleModule
 from ansible.module_utils._text import to_text
@@ -484,11 +485,42 @@ def __init__(
         self._token = None
         self._prefix = "ansible_ec2_%s"
 
+    def _decode(self, data):
+        try:
+            return data.decode("utf-8")
+        except UnicodeDecodeError:
+            # Decoding as UTF-8 failed, return data without raising an error
+            self.module.warn("Decoding user-data as UTF-8 failed, return data as is ignoring any error")
+            return data.decode("utf-8", errors="ignore")
+
+    def decode_user_data(self, data):
+        is_compressed = False
+
+        # Check if data is compressed using zlib header
+        if data.startswith(b"\x78\x9c") or data.startswith(b"\x1f\x8b"):
+            is_compressed = True
+
+        if is_compressed:
+            # Data is compressed, attempt decompression and decode using UTF-8
+            try:
+                decompressed = zlib.decompress(data, zlib.MAX_WBITS | 32)
+                return self._decode(decompressed)
+            except zlib.error:
+                # Unable to decompress, return original data
+                self.module.warn(
+                    "Unable to decompress user-data using zlib, attempt to decode original user-data as UTF-8"
+                )
+                return self._decode(data)
+        else:
+            # Data is not compressed, decode using UTF-8
+            return self._decode(data)
+
     def _fetch(self, url):
         encoded_url = quote(url, safe="%/:=&?~#+!$,;'@()*[]")
         headers = {}
         if self._token:
             headers = {"X-aws-ec2-metadata-token": self._token}
+
         response, info = fetch_url(self.module, encoded_url, headers=headers, force=True)
 
         if info.get("status") in (401, 403):
@@ -505,6 +537,8 @@ def _fetch(self, url):
                 )
         if response and info["status"] < 400:
             data = response.read()
+            if "user-data" in encoded_url:
+                return to_text(self.decode_user_data(data))
         else:
             data = None
         return to_text(data)

tests/unit/plugins/modules/test_ec2_metadata_facts.py

@@ -1,6 +1,7 @@
 # This file is part of Ansible
 # GNU General Public License v3.0+ (see COPYING or https://www.gnu.org/licenses/gpl-3.0.txt)
 
+import gzip
 import io
 import pytest
 from unittest.mock import MagicMock
@@ -59,3 +60,41 @@ def test_fetch_recusive(m_fetch_url, ec2_instance):
     ]
     ec2_instance.fetch("http://169.254.169.254/latest/meta-data/")
     assert ec2_instance._data == {"http://169.254.169.254/latest/meta-data/whatever/my-key": "my-value"}
+
+
+@patch(module_name + ".fetch_url")
+def test__fetch_user_data_compressed(m_fetch_url, ec2_instance):
+    user_data = b"""Content-Type: multipart/mixed; boundary="MIMEBOUNDARY"
+MIME-Version: 1.0
+
+--MIMEBOUNDARY
+Content-Transfer-Encoding: 7bit
+Content-Type: text/cloud-config
+Mime-Version: 1.0
+
+packages: ['httpie']
+
+--MIMEBOUNDARY--
+"""
+
+    m_fetch_url.return_value = (io.BytesIO(gzip.compress(user_data)), {"status": 200})
+    assert ec2_instance._fetch("http://169.254.169.254/latest/user-data") == user_data.decode("utf-8")
+
+
+@patch(module_name + ".fetch_url")
+def test__fetch_user_data_plain(m_fetch_url, ec2_instance):
+    user_data = b"""Content-Type: multipart/mixed; boundary="MIMEBOUNDARY"
+MIME-Version: 1.0
+
+--MIMEBOUNDARY
+Content-Transfer-Encoding: 7bit
+Content-Type: text/cloud-config
+Mime-Version: 1.0
+
+packages: ['httpie']
+
+--MIMEBOUNDARY--
+"""
+
+    m_fetch_url.return_value = (io.BytesIO(user_data), {"status": 200})
+    assert ec2_instance._fetch("http://169.254.169.254/latest/user-data") == user_data.decode("utf-8")