|
| 1 | +#!/usr/bin/env python3 |
| 2 | +"""Compare Strata example output catalogs against reference results. |
| 3 | +
|
| 4 | +Usage: |
| 5 | + compare_examples.py <strata_binary> <example_dir> [--rtol=2e-2] [--atol=1e-6] |
| 6 | + compare_examples.py <strata_binary> <example_dir> --update |
| 7 | +
|
| 8 | +Workflow for each example-NN.json with existing results: |
| 9 | + 1. Copy the file to a temporary directory |
| 10 | + 2. Run strata in batch mode on the copy |
| 11 | + 3. Compare the outputCatalog from the result against the original |
| 12 | +
|
| 13 | +Use --update to regenerate reference results in-place. |
| 14 | +""" |
| 15 | + |
| 16 | +import argparse |
| 17 | +import glob |
| 18 | +import json |
| 19 | +import os |
| 20 | +import shutil |
| 21 | +import subprocess |
| 22 | +import sys |
| 23 | +import tempfile |
| 24 | + |
| 25 | + |
| 26 | +def compare_values(ref, actual, path, rtol, atol, errors): |
| 27 | + """Recursively compare two JSON values, collecting mismatches.""" |
| 28 | + if isinstance(ref, dict) and isinstance(actual, dict): |
| 29 | + for key in ref: |
| 30 | + if key == "log": |
| 31 | + # Skip log text — messages vary between runs |
| 32 | + continue |
| 33 | + if key not in actual: |
| 34 | + errors.append(f"{path}.{key}: missing in actual output") |
| 35 | + continue |
| 36 | + compare_values(ref[key], actual[key], f"{path}.{key}", rtol, atol, errors) |
| 37 | + return |
| 38 | + |
| 39 | + if isinstance(ref, list) and isinstance(actual, list): |
| 40 | + if len(ref) != len(actual): |
| 41 | + errors.append( |
| 42 | + f"{path}: list length mismatch: expected {len(ref)}, got {len(actual)}" |
| 43 | + ) |
| 44 | + return |
| 45 | + for i, (r, a) in enumerate(zip(ref, actual)): |
| 46 | + compare_values(r, a, f"{path}[{i}]", rtol, atol, errors) |
| 47 | + return |
| 48 | + |
| 49 | + if isinstance(ref, (int, float)) and isinstance(actual, (int, float)): |
| 50 | + if ref == actual: |
| 51 | + return |
| 52 | + abs_err = abs(ref - actual) |
| 53 | + if abs_err <= atol: |
| 54 | + return |
| 55 | + denom = max(abs(ref), abs(actual), 1e-15) |
| 56 | + rel_err = abs_err / denom |
| 57 | + if rel_err > rtol: |
| 58 | + errors.append( |
| 59 | + f"{path}: value mismatch: expected {ref}, got {actual} " |
| 60 | + f"(rel_err={rel_err:.2e}, abs_err={abs_err:.2e})" |
| 61 | + ) |
| 62 | + return |
| 63 | + |
| 64 | + if isinstance(ref, str) and isinstance(actual, str): |
| 65 | + if ref != actual: |
| 66 | + errors.append(f"{path}: string mismatch: expected {ref!r}, got {actual!r}") |
| 67 | + return |
| 68 | + |
| 69 | + if isinstance(ref, bool) and isinstance(actual, bool): |
| 70 | + if ref != actual: |
| 71 | + errors.append(f"{path}: bool mismatch: expected {ref}, got {actual}") |
| 72 | + return |
| 73 | + |
| 74 | + if ref is None and actual is None: |
| 75 | + return |
| 76 | + |
| 77 | + if type(ref) != type(actual): |
| 78 | + errors.append( |
| 79 | + f"{path}: type mismatch: expected {type(ref).__name__}, " |
| 80 | + f"got {type(actual).__name__}" |
| 81 | + ) |
| 82 | + |
| 83 | + |
| 84 | +def extract_output_data(output_catalog): |
| 85 | + """Extract only the result data fields from the outputCatalog for comparison.""" |
| 86 | + data = {} |
| 87 | + for catalog_key in [ |
| 88 | + "profilesOutputCatalog", |
| 89 | + "ratiosOutputCatalog", |
| 90 | + "soilTypesOutputCatalog", |
| 91 | + "spectraOutputCatalog", |
| 92 | + "timeSeriesOutputCatalog", |
| 93 | + ]: |
| 94 | + entries = output_catalog.get(catalog_key, []) |
| 95 | + catalog_data = [] |
| 96 | + for entry in entries: |
| 97 | + if entry.get("enabled") and entry.get("data"): |
| 98 | + catalog_data.append( |
| 99 | + { |
| 100 | + "className": entry.get("className"), |
| 101 | + "data": entry["data"], |
| 102 | + } |
| 103 | + ) |
| 104 | + if catalog_data: |
| 105 | + data[catalog_key] = catalog_data |
| 106 | + return data |
| 107 | + |
| 108 | + |
| 109 | +def run_example(strata_bin, example_path, rtol, atol, update=False): |
| 110 | + """Run strata on an example file and compare results. Returns (name, errors, skipped).""" |
| 111 | + name = os.path.basename(example_path) |
| 112 | + |
| 113 | + # Load reference |
| 114 | + with open(example_path) as f: |
| 115 | + ref_doc = json.load(f) |
| 116 | + |
| 117 | + if not update and not ref_doc.get("hasResults", False): |
| 118 | + return name, [], True # skip |
| 119 | + |
| 120 | + ref_output = extract_output_data(ref_doc.get("outputCatalog", {})) |
| 121 | + if not update and not ref_output: |
| 122 | + return name, ["No enabled output data in reference file"], False |
| 123 | + |
| 124 | + # Copy to temp directory and run |
| 125 | + with tempfile.TemporaryDirectory() as tmpdir: |
| 126 | + tmp_file = os.path.join(tmpdir, name) |
| 127 | + shutil.copy2(example_path, tmp_file) |
| 128 | + |
| 129 | + # Also copy any motion files referenced (same directory) |
| 130 | + example_dir = os.path.dirname(example_path) |
| 131 | + motions_dir = os.path.join(example_dir, "motions") |
| 132 | + if os.path.isdir(motions_dir): |
| 133 | + shutil.copytree(motions_dir, os.path.join(tmpdir, "motions")) |
| 134 | + |
| 135 | + # Run strata in batch mode |
| 136 | + try: |
| 137 | + result = subprocess.run( |
| 138 | + [strata_bin, "-b", tmp_file], |
| 139 | + capture_output=True, |
| 140 | + text=True, |
| 141 | + timeout=300, |
| 142 | + ) |
| 143 | + except subprocess.TimeoutExpired: |
| 144 | + return name, ["Strata timed out after 300 seconds"], False |
| 145 | + |
| 146 | + if result.returncode != 0: |
| 147 | + return ( |
| 148 | + name, |
| 149 | + [f"Strata exited with code {result.returncode}: {result.stderr}"], |
| 150 | + False, |
| 151 | + ) |
| 152 | + |
| 153 | + # Load results |
| 154 | + with open(tmp_file) as f: |
| 155 | + actual_doc = json.load(f) |
| 156 | + |
| 157 | + if not actual_doc.get("hasResults", False): |
| 158 | + return name, ["Strata did not produce results"], False |
| 159 | + |
| 160 | + if update: |
| 161 | + # Overwrite the original file with new results |
| 162 | + shutil.copy2(tmp_file, example_path) |
| 163 | + return name, [], False |
| 164 | + |
| 165 | + actual_output = extract_output_data(actual_doc.get("outputCatalog", {})) |
| 166 | + |
| 167 | + # Compare |
| 168 | + errors = [] |
| 169 | + compare_values(ref_output, actual_output, "outputCatalog", rtol, atol, errors) |
| 170 | + return name, errors, False |
| 171 | + |
| 172 | + |
| 173 | +def main(): |
| 174 | + parser = argparse.ArgumentParser( |
| 175 | + description="Run Strata examples and compare output catalogs" |
| 176 | + ) |
| 177 | + parser.add_argument("strata_binary", help="Path to the strata executable") |
| 178 | + parser.add_argument("example_dir", help="Path to the example directory") |
| 179 | + parser.add_argument( |
| 180 | + "--rtol", |
| 181 | + type=float, |
| 182 | + default=2e-2, |
| 183 | + help="Relative tolerance for numerical comparison (default: 2e-2)", |
| 184 | + ) |
| 185 | + parser.add_argument( |
| 186 | + "--atol", |
| 187 | + type=float, |
| 188 | + default=1e-6, |
| 189 | + help="Absolute tolerance for numerical comparison (default: 1e-6)", |
| 190 | + ) |
| 191 | + parser.add_argument( |
| 192 | + "--update", |
| 193 | + action="store_true", |
| 194 | + help="Regenerate reference results by running strata and saving output", |
| 195 | + ) |
| 196 | + args = parser.parse_args() |
| 197 | + |
| 198 | + strata_bin = os.path.abspath(args.strata_binary) |
| 199 | + example_dir = os.path.abspath(args.example_dir) |
| 200 | + |
| 201 | + if not os.path.isfile(strata_bin): |
| 202 | + print(f"ERROR: Strata binary not found: {strata_bin}", file=sys.stderr) |
| 203 | + sys.exit(1) |
| 204 | + |
| 205 | + # Find example files (example-NN.json pattern, excluding -test variants) |
| 206 | + pattern = os.path.join(example_dir, "example-[0-9][0-9].json") |
| 207 | + example_files = sorted(glob.glob(pattern)) |
| 208 | + |
| 209 | + if not example_files: |
| 210 | + print(f"ERROR: No example files found matching {pattern}", file=sys.stderr) |
| 211 | + sys.exit(1) |
| 212 | + |
| 213 | + print(f"Found {len(example_files)} example file(s)") |
| 214 | + if args.update: |
| 215 | + print("Mode: updating reference results") |
| 216 | + else: |
| 217 | + print(f"Tolerances: rtol={args.rtol}, atol={args.atol}") |
| 218 | + print() |
| 219 | + |
| 220 | + failed = [] |
| 221 | + passed = [] |
| 222 | + skipped = [] |
| 223 | + |
| 224 | + for example_path in example_files: |
| 225 | + name, errors, was_skipped = run_example( |
| 226 | + strata_bin, example_path, args.rtol, args.atol, update=args.update |
| 227 | + ) |
| 228 | + if was_skipped: |
| 229 | + skipped.append(name) |
| 230 | + print(f" SKIP: {name} (no reference results)") |
| 231 | + elif args.update: |
| 232 | + if errors: |
| 233 | + failed.append((name, errors)) |
| 234 | + print(f" FAIL: {name}") |
| 235 | + for err in errors: |
| 236 | + print(f" {err}") |
| 237 | + else: |
| 238 | + passed.append(name) |
| 239 | + print(f" UPDATED: {name}") |
| 240 | + elif errors: |
| 241 | + failed.append((name, errors)) |
| 242 | + print(f" FAIL: {name}") |
| 243 | + for err in errors[:10]: |
| 244 | + print(f" {err}") |
| 245 | + if len(errors) > 10: |
| 246 | + print(f" ... and {len(errors) - 10} more errors") |
| 247 | + else: |
| 248 | + passed.append(name) |
| 249 | + print(f" PASS: {name}") |
| 250 | + |
| 251 | + print() |
| 252 | + print(f"Results: {len(passed)} passed, {len(failed)} failed, {len(skipped)} skipped") |
| 253 | + |
| 254 | + if failed: |
| 255 | + sys.exit(1) |
| 256 | + |
| 257 | + |
| 258 | +if __name__ == "__main__": |
| 259 | + main() |
0 commit comments