Add a python script to plot the results

[Emmaaycoberry]

Add a python script that can be used as follow:
python shine/plot_shine_results.py --output my_output/

Features:
✅ Reads observation.npz and posterior.nc from the specified directory
✅ Saves all plots to the same directory
✅ Generates visualizations:

1. Observation visual (image, PSF, noise map)
2. Posterior distributions
3. Corner plot with 68% and 95% confidence intervals
4. Shear analysis (if g1/g2 present)
5. Summary statistics table (printed, not saved)
6. Trace plots (if MCMC chains present)
7. Correlation matrix

✅ Clean terminal output with progress indicators
✅ Proper error handling if files are missing
✅ Auto-installs the corner package if needed

[claude]

Code Review

I've reviewed this PR and found 3 issues that need to be addressed:

1. Missing Type Hints (CLAUDE.md violation)

Location: All functions in shine/plot_shine_results.py

The functions lack type hints, which violates CLAUDE.md Section 4.1:

• Type hints: Full PEP 484 compliance required

For example, setup_plot_style() at line 28 should be:

def setup_plot_style() -> None:
    """Set up matplotlib style for all plots."""
    plt.style.use('seaborn-v0_8-darkgrid')
    plt.rcParams['figure.figsize'] = (12, 8)

This applies to all 10 functions in the file.

2. Docstrings Not in Google-Style Format (CLAUDE.md violation)

Location: All functions in shine/plot_shine_results.py

The docstrings don't follow Google-style format as required by CLAUDE.md Section 4.1:

• Docstrings: Google-style

Current format (example from load_data at lines 36-43):

def load_data(output_dir):
    """Load observation and posterior data from the output directory.
    
    Args:
        output_dir: Path to directory containing observation.npz and posterior.nc
        
    Returns:
        tuple: (obs_data, idata, posterior)
    """

Should be:

def load_data(output_dir: Path) -> tuple[dict, az.InferenceData, xr.Dataset]:
    """Load observation and posterior data from the output directory.
    
    Args:
        output_dir (Path): Path to directory containing observation.npz and posterior.nc
        
    Returns:
        tuple: A tuple containing:
            - obs_data (dict): Loaded observation data from observation.npz
            - idata (az.InferenceData): ArviZ InferenceData object
            - posterior (xr.Dataset): Posterior dataset
            
    Raises:
        FileNotFoundError: If observation.npz or posterior.nc not found
    """

Google-style requires type annotations in the Args section and more detailed Returns sections.

3. Security Issue - Automatic Package Installation

Location: Lines 202-205 and 266-269 in shine/plot_shine_results.py

The code automatically installs the corner package without user consent:

SHINE/shine/plot_shine_results.py

Lines 199 to 207 in a2ce2d4

	try:
	import corner
	except ImportError:
	print("Installing corner package...")
	import subprocess
	subprocess.check_call([sys.executable, "-m", "pip", "install", "corner"])
	import corner
	# Prepare data: stack all parameters as columns

This is problematic because:

• It modifies the user's Python environment as a side effect of running a visualization script
• Users don't get a chance to confirm or opt out
• It could execute arbitrary code from PyPI in compromised environments
• It violates the principle of least surprise

Recommended fix: Either:

1. Declare corner as an optional dependency in pyproject.toml under [project.optional-dependencies] (e.g., plotting = ["corner>=2.0.0"])
2. Fail gracefully with a clear error message: raise ImportError("corner package is required. Install it with: pip install corner")

This same pattern appears twice in the file (in both plot_corner and plot_shear_analysis functions).

---

Summary: Please address these 3 issues before merging. The first two are code standards compliance issues from CLAUDE.md, and the third is a security/best-practices concern.

[EiffL]

@claude can you address the issues found during the review?

[claude]

Claude finished @EiffL's task in 2m 16s —— View job

---

Code Review Issues Fixed ✅

All 3 issues identified in the code review have been addressed and pushed to the dev-plot-results branch.

Changes Made

1. Added Type Hints (PEP 484 Compliance)

• Added type hints to all 10 functions in shine/plot_shine_results.py
• Imported required types: Any, Dict, List, Tuple from typing
• All function signatures now include parameter types and return types
• Example: def load_data(output_dir: Path) -> Tuple[Dict[str, Any], az.InferenceData, xr.Dataset]:

2. Converted Docstrings to Google-Style Format

• All docstrings now follow Google-style with proper formatting
• Added detailed Args sections with type annotations in parentheses
• Added comprehensive Returns sections with detailed descriptions
• Added Raises sections where applicable (e.g., FileNotFoundError, ImportError)
• Improved clarity and completeness of documentation

3. Removed Automatic Package Installation (Security Issue)

• Replaced automatic pip install corner with proper error handling
• Both plot_corner() and plot_shear_analysis() now raise ImportError with clear installation instructions
• Error message: "The 'corner' package is required for corner plots. Install it with: pip install corner"
• Users maintain control over their Python environment

Commit Details

Commit: dbdeaff
Branch: dev-plot-results

The changes ensure full compliance with CLAUDE.md requirements and follow security best practices.

---

[EiffL]

@claude corner is a great package, very useful, can you add it as a dependency of the package? we will certainly use it more that once

[claude]

Claude finished @EiffL's task in 58s —— View job

---

Adding corner as a Package Dependency ✅

Successfully added corner as a project dependency.

Tasks

• Read current pyproject.toml structure
• Add corner to dependencies
• Commit and push changes
• Update comment with results

Changes Made

Added corner>=2.0.0 to the main dependencies in pyproject.toml:34:

dependencies = [
    "jax>=0.4.0",
    "numpyro>=0.13.0",
    "galsim>=2.5.0",
    "jax-galsim>=2026.2.0",
    "pydantic>=2.0.0",
    "pyyaml>=6.0",
    "arviz>=0.17.0",
    "matplotlib>=3.7.0",
    "corner>=2.0.0",  # Added for corner plots and visualization
]

The corner package will now be automatically installed with SHINE, making it available for the plotting script and any future visualization needs.

Commit Details

• Commit: f3bb9df
• Branch: dev-plot-results

--- | Branch: dev-plot-results

[EiffL]

@claude hum, this is a plotting script, it shouldn't be in the package itself. Can you add a scripts folder where such utility scripts can be stored?

[claude]

Claude finished @EiffL's task in 2m 43s —— View job

---

Moving Plotting Script to Scripts Folder ✅

Successfully moved shine/plot_shine_results.py to a dedicated scripts/ folder for utility scripts.

Tasks

• Create scripts/ directory
• Move shine/plot_shine_results.py to scripts/plot_shine_results.py
• Commit and push changes
• Update comment with results

Changes Made

The plotting script has been relocated from the package to a utility scripts directory:

Old location: shine/plot_shine_results.py
New location: scripts/plot_shine_results.py

Git has preserved the file's history by detecting this as a rename operation (98% similarity).

Usage

The script can now be run using:

python scripts/plot_shine_results.py --output my_output/

Commit Details

• Commit: e909dd7
• Branch: dev-plot-results

---

| Branch: dev-plot-results

[EiffL]

lgtm