Version and test flow #35

springfall2008 · 2025-12-30T15:42:06Z

No description provided.

Copilot

Pull request overview

This PR introduces a comprehensive testing infrastructure and version compatibility improvements for PredAI. The changes enable automated testing with CI/CD integration while addressing PyTorch 2.6+ compatibility issues.

Key changes:

Added comprehensive unit tests for Prophet predictions with sine wave and incrementing sensor data
Implemented CI/CD workflow with GitHub Actions for automated testing
Updated PyTorch compatibility handling to support version 2.6+ with monkey-patched torch.load behavior

Reviewed changes

Copilot reviewed 6 out of 7 changed files in this pull request and generated 10 comments.

Show a summary per file

File	Description
test_predai.py	New comprehensive unit test suite with mocked HAInterface and synthetic data generation
setup.csh	Setup script for creating virtual environment (has shell syntax issues)
run_all	Bash script to execute unit tests (has logic issue with error handling)
predai/rootfs/predai.py	Added PyTorch 2.6+ compatibility fixes and proper main guard
predai/requirements.txt	Removed version pinning from neuralprophet dependency
.gitignore	Added test artifacts and PyTorch Lightning log directories
.github/workflows/test.yml	New CI workflow for automated testing on PR and push events

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2025-12-30T15:45:47Z

test_predai.py

+# Fix for PyTorch 2.6+ weights_only=True default - must be done BEFORE importing pred ai
+try:
+    import torch
+    _original_torch_load = torch.load
+    def _patched_torch_load(*args, **kwargs):
+        if 'weights_only' not in kwargs:
+            kwargs['weights_only'] = False
+        return _original_torch_load(*args, **kwargs)
+    torch.load = _patched_torch_load
+except (ImportError, AttributeError):
+    pass
+


The monkey-patching of torch.load is duplicated in both test_predai.py (lines 9-19) and predai.py (lines 46-53). This creates maintenance overhead and potential inconsistency. Consider consolidating this fix into a single location, preferably in predai.py since it's the main module, and have the test file simply import from there.

Suggested change

# Fix for PyTorch 2.6+ weights_only=True default - must be done BEFORE importing pred ai

try:

import torch

_original_torch_load = torch.load

def _patched_torch_load(*args, **kwargs):

if 'weights_only' not in kwargs:

kwargs['weights_only'] = False

return _original_torch_load(*args, **kwargs)

torch.load = _patched_torch_load

except (ImportError, AttributeError):

pass

Copilot · 2025-12-30T15:45:47Z

test_predai.py

+
+        # Generate one data point per hour
+        for i in range(hours):
+            # Create sine wave: one complete cycle per 24 hours
+            value = offset + amplitude * np.sin(2 * np.pi * i / 24)
+
+            timestamp = current_time.strftime("%Y-%m-%dT%H:%M:%S%z")
+            data.append({
+                "state": str(value),
+                "last_updated": timestamp
+            })
+            current_time += timedelta(hours=1)
+


The comment states "Generate one data point per hour" but this accurately describes the implementation, so this is correct. However, the function is called generate_sine_wave_data with a parameter hours that generates hourly data points, yet the test uses a period attribute set to 30 (minutes). This mismatch between the test data generation frequency (hourly) and the expected period (30 minutes) could lead to unexpected behavior in the Prophet model which expects data at 30-minute intervals.

Suggested change

# Generate one data point per hour

for i in range(hours):

# Create sine wave: one complete cycle per 24 hours

value = offset + amplitude * np.sin(2 * np.pi * i / 24)

timestamp = current_time.strftime("%Y-%m-%dT%H:%M:%S%z")

data.append({

"state": str(value),

"last_updated": timestamp

})

current_time += timedelta(hours=1)

# Determine sampling interval in minutes from the test's period (default 30 minutes)

interval_minutes = getattr(self, "period", 60)

if 60 % interval_minutes != 0:

# Fallback to hourly data if the period does not divide evenly into an hour

interval_minutes = 60

# Number of samples per hour and total samples for the requested duration

samples_per_hour = 60 // interval_minutes

total_samples = hours * samples_per_hour

# Generate one complete sine cycle per 24 hours, sampled every `interval_minutes`

samples_per_day = 24 * samples_per_hour

for i in range(total_samples):

# Create sine wave: one complete cycle per 24 hours

value = offset + amplitude * np.sin(2 * np.pi * i / samples_per_day)

timestamp = current_time.strftime("%Y-%m-%dT%H:%M:%S%z")

data.append({

"state": str(value),

"last_updated": timestamp

})

current_time += timedelta(minutes=interval_minutes)

Copilot · 2025-12-30T15:45:47Z

setup.csh

+python3 -m venv venv
+
+echo "Activating virtual environment..."
+source venv/bin/activate


The source command used here is a bash/sh command, but this is a csh script (indicated by the .csh extension). In csh/tcsh, you should use 'source venv/bin/activate.csh' instead of 'source venv/bin/activate'. The bash activation script won't work properly in a csh shell.

Suggested change

source venv/bin/activate

source venv/bin/activate.csh

Copilot · 2025-12-30T15:45:48Z

setup.csh

+# Setup script for PredAI test environment
+# Creates a virtual environment and installs dependencies
+
+echo "Creating Python virtual environment..."
+python3 -m venv venv
+
+echo "Activating virtual environment..."
+source venv/bin/activate
+
+echo "Installing dependencies..."
+python -m pip install --upgrade pip
+python -m pip install -r predai/requirements.txt
+
+echo ""
+echo "Setup complete!"
+echo "To activate the environment manually, run:"
+echo "  source venv/bin/activate.csh"
+echo ""
+echo "To run tests, execute:"
+echo "  ./run_all"
+


This file uses 'echo' commands which are bash/sh syntax, but the file has a .csh extension indicating it should be a C shell script. C shell uses different syntax for commands and doesn't support echo in the same way. Either rename this file to setup.sh and add a proper shebang (#!/bin/bash), or rewrite it using proper csh syntax.

Copilot · 2025-12-30T15:45:48Z

run_all

+python -m unittest test_predai.py -v
+
+# Check test result
+if [ $? -eq 0 ]; then
+    echo ""
+    echo "================================"
+    echo "✅ All tests passed successfully!"
+else
+    echo ""
+    echo "================================"
+    echo "❌ Some tests failed"
+    exit 1
+fi


The set -e flag on line 4 causes the script to exit immediately if any command fails. This means if the unittest command on line 21 fails, the script will exit before reaching this conditional block. The test result check (lines 24-33) will never execute in case of test failure. Either remove set -e or capture the exit status before the script exits (e.g., using || true after the unittest command and then checking $?).

Copilot · 2025-12-30T15:45:48Z

.github/workflows/test.yml

+        python -m unittest test_predai.py -v
+        
+    - name: Test summary
+      if: always()
+      run: |
+        if [ $? -eq 0 ]; then


Similar to the issue in run_all script, this conditional references $? after other commands have executed (the echo statements on lines 43-45). The $? will reflect the exit status of the echo command, not the unittest command on line 37. The exit status should be captured immediately after the test command completes.

Suggested change

python -m unittest test_predai.py -v

- name: Test summary

if: always()

run: |

if [ $? -eq 0 ]; then

if python -m unittest test_predai.py -v; then

Copilot · 2025-12-30T15:45:48Z

test_predai.py

@@ -0,0 +1,258 @@
+import unittest
+from unittest.mock import AsyncMock, MagicMock, patch


Import of 'MagicMock' is not used.
Import of 'patch' is not used.

Suggested change

from unittest.mock import AsyncMock, MagicMock, patch

from unittest.mock import AsyncMock

Copilot · 2025-12-30T15:45:49Z

test_predai.py

+# Add the rootfs directory to the path
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), 'predai', 'rootfs'))
+
+from predai import HAInterface, Prophet, timestr_to_datetime


Import of 'timestr_to_datetime' is not used.

Suggested change

from predai import HAInterface, Prophet, timestr_to_datetime

from predai import HAInterface, Prophet

Copilot · 2025-12-30T15:45:49Z

test_predai.py

+            kwargs['weights_only'] = False
+        return _original_torch_load(*args, **kwargs)
+    torch.load = _patched_torch_load
+except (ImportError, AttributeError):


'except' clause does nothing but pass and there is no explanatory comment.

Suggested change

except (ImportError, AttributeError):

except (ImportError, AttributeError):

# Torch is not installed or does not expose torch.load as expected; continue without patching.

Copilot · 2025-12-30T15:45:49Z

predai/requirements.txt

@@ -1,4 +1,4 @@
-neuralprophet==0.9.0
+neuralprophet


The change from a pinned neuralprophet==0.9.0 to an unpinned neuralprophet introduces a supply-chain risk: future installs will always pull the latest version from PyPI, which could be compromised or contain breaking changes and will run with the same privileges as your test/build environment. An attacker who gains control over the neuralprophet package (or its latest release) could execute arbitrary code during pip install -r predai/requirements.txt in CI or developer environments. To reduce this risk, pin neuralprophet to a specific version (or a carefully managed range) and update it intentionally after review.

Version and test flow

e5a6041

Copilot AI review requested due to automatic review settings December 30, 2025 15:42

Copilot started reviewing on behalf of springfall2008 December 30, 2025 15:42 View session

Copilot AI reviewed Dec 30, 2025

View reviewed changes

springfall2008 merged commit b8d3a49 into main Dec 30, 2025
7 checks passed

springfall2008 deleted the update branch December 30, 2025 15:46

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Version and test flow #35

Version and test flow #35

Uh oh!

springfall2008 commented Dec 30, 2025

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Dec 30, 2025

Uh oh!

Copilot AI Dec 30, 2025

Uh oh!

Copilot AI Dec 30, 2025

Uh oh!

Copilot AI Dec 30, 2025

Uh oh!

Copilot AI Dec 30, 2025

Uh oh!

Copilot AI Dec 30, 2025

Uh oh!

Copilot AI Dec 30, 2025

Uh oh!

Copilot AI Dec 30, 2025

Uh oh!

Copilot AI Dec 30, 2025

Uh oh!

Copilot AI Dec 30, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		@@ -0,0 +1,258 @@
		import unittest
		from unittest.mock import AsyncMock, MagicMock, patch

	from unittest.mock import AsyncMock, MagicMock, patch
	from unittest.mock import AsyncMock

	from predai import HAInterface, Prophet, timestr_to_datetime
	from predai import HAInterface, Prophet

	except (ImportError, AttributeError):
	except (ImportError, AttributeError):
	# Torch is not installed or does not expose torch.load as expected; continue without patching.

Version and test flow #35

Version and test flow #35

Uh oh!

Conversation

springfall2008 commented Dec 30, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants