json load pip results, then aggregate if that fails

Author: zsimjee

.github/workflows/install_from_hub.yml

@@ -0,0 +1,23 @@
+name: Notebook Execution and Error Check
+
+on:
+  push:
+    branches:
+      - main
+      - dev
+  workflow_dispatch:  # This enables manual triggering
+
+jobs:
+  install_from_hub:
+    runs-on: ubuntu-latest
+    steps:
+    - name: Set up Python
+      uses: actions/setup-python@v5
+      with: 
+        python-version: 3.11.x
+    - name: pip install from main
+      run: pip install git+https://github.com/guardrails-ai/guardrails.git@main
+    - name: Install PII validator
+      run: guardrails hub install hub://guardrails/detect_pii
+    - name: Verify PII validator is addressable
+      run: echo 'from guardrails.hub import DetectPII' | python 

guardrails/cli/hub/install.py

@@ -1,3 +1,4 @@
+import json
 import os
 import subprocess
 import sys
@@ -44,6 +45,13 @@ def pip_process(
         logger.debug(f"decoding output from pip {action} {package}")
         if format == json_format:
             parsed = BytesHeaderParser().parsebytes(output)
+            try:
+                return json.loads(str(parsed))
+            except Exception as e:
+                logger.debug(
+                    f"json parse exception in decoding output from pip {action} {package}. Falling back to accumulating the byte stream",  # noqa
+                    e,
+                )
             accumulator = {}
             for key, value in parsed.items():
                 accumulator[key] = value

tests/unit_tests/cli/hub/test_install.py

@@ -134,7 +134,7 @@ def parsebytes(self, *args):
 
         response = pip_process("show", "pip", format="json")
 
-        assert mock_logger_debug.call_count == 2
+        assert mock_logger_debug.call_count == 3
         debug_calls = [
             call("running pip show  pip"),
             call("decoding output from pip show pip"),