|
| 1 | +import json |
| 2 | +import os |
| 3 | + |
| 4 | +import click |
| 5 | +import numpy as np |
| 6 | + |
| 7 | +BIG_WIDTH = 80 |
| 8 | +SMALL_WIDTH = 8 |
| 9 | + |
| 10 | + |
| 11 | +def find_json_files(base_dir): |
| 12 | + """Recursively find all JSON files in subdirectories.""" |
| 13 | + json_files = [] |
| 14 | + for root, _, files in os.walk(base_dir): |
| 15 | + for file in files: |
| 16 | + if file.endswith(".jsonl"): |
| 17 | + json_files.append(os.path.join(root, file)) |
| 18 | + return json_files |
| 19 | + |
| 20 | + |
| 21 | +def read_json_file(file_path): |
| 22 | + """Read a JSON file and return its content as a list of test configurations.""" |
| 23 | + with open(file_path, "r", encoding="utf-8") as f: |
| 24 | + try: |
| 25 | + data = [json.loads(line) for line in f] |
| 26 | + return data |
| 27 | + except json.JSONDecodeError as e: |
| 28 | + print(f"Error reading {file_path}: {e}") |
| 29 | + return [] |
| 30 | + |
| 31 | + |
| 32 | +def extract_tests_with_tags(json_files): |
| 33 | + """Extract test data and assign a tag based on the directory name.""" |
| 34 | + tests = [] |
| 35 | + |
| 36 | + for file_path in json_files: |
| 37 | + directory_name = os.path.basename(os.path.dirname(file_path)) |
| 38 | + test_data = read_json_file(file_path) |
| 39 | + |
| 40 | + for test in test_data: |
| 41 | + if test.get("outcome", "").lower() == "passed" and test.get("duration"): |
| 42 | + nodeid = test.get("nodeid") |
| 43 | + if nodeid.startswith("tests/"): |
| 44 | + nodeid = nodeid[6:] |
| 45 | + |
| 46 | + when = test.get("when") |
| 47 | + duration = test["duration"] |
| 48 | + tags = directory_name.split("-") |
| 49 | + tags.remove("logs") |
| 50 | + id_ = f"{nodeid}({when})" |
| 51 | + |
| 52 | + tests.append( |
| 53 | + { |
| 54 | + "tags": tags, |
| 55 | + "id": id_, |
| 56 | + "nodeid": nodeid, |
| 57 | + "duration": duration, |
| 58 | + "when": when, |
| 59 | + } |
| 60 | + ) |
| 61 | + return tests |
| 62 | + |
| 63 | + |
| 64 | +def compute_statistics(tests): |
| 65 | + """Compute average duration and standard deviation per test ID.""" |
| 66 | + test_stats = {} |
| 67 | + |
| 68 | + for test in tests: |
| 69 | + test_id = test["id"] |
| 70 | + if test_id not in test_stats: |
| 71 | + test_stats[test_id] = { |
| 72 | + "durations": [], |
| 73 | + "n_tests": 0, |
| 74 | + } |
| 75 | + |
| 76 | + test_stats[test_id]["durations"].append(test["duration"]) |
| 77 | + test_stats[test_id]["n_tests"] += 1 |
| 78 | + |
| 79 | + summary = [] |
| 80 | + |
| 81 | + for test_id, data in test_stats.items(): |
| 82 | + durations = np.array(data["durations"]) |
| 83 | + |
| 84 | + if durations.size == 0: |
| 85 | + continue |
| 86 | + |
| 87 | + avg_duration = np.mean(durations) |
| 88 | + std_dev = np.std(durations) |
| 89 | + |
| 90 | + mask_99 = durations < np.percentile(durations, 99) |
| 91 | + if mask_99.sum() == 0: |
| 92 | + avg_duration_minus_one = np.nan |
| 93 | + std_dev_minus_one = np.nan |
| 94 | + else: |
| 95 | + avg_duration_minus_one = np.mean(durations[mask_99]) |
| 96 | + std_dev_minus_one = np.std(durations[mask_99]) |
| 97 | + |
| 98 | + mask_75 = durations < np.percentile(durations, 75) |
| 99 | + if mask_75.sum() == 0: |
| 100 | + avg_duration_minus_34 = np.nan |
| 101 | + std_dev_minus_34 = np.nan |
| 102 | + else: |
| 103 | + avg_duration_minus_34 = np.mean(durations[mask_75]) |
| 104 | + std_dev_minus_34 = np.std(durations[mask_75]) |
| 105 | + |
| 106 | + summary.append( |
| 107 | + { |
| 108 | + "id": test_id, |
| 109 | + "n_tests": data["n_tests"], |
| 110 | + "average_duration": avg_duration, |
| 111 | + "std_dev": std_dev, |
| 112 | + "avg_duration_minus_one": avg_duration_minus_one, |
| 113 | + "std_dev_minus_one": std_dev_minus_one, |
| 114 | + "avg_duration_minus_34": avg_duration_minus_34, |
| 115 | + "std_dev_minus_34": std_dev_minus_34, |
| 116 | + } |
| 117 | + ) |
| 118 | + |
| 119 | + return summary |
| 120 | + |
| 121 | + |
| 122 | +def print_table(data, keys, headers, title=""): |
| 123 | + JUNCTION = "|" |
| 124 | + |
| 125 | + def make_bold(s): |
| 126 | + return click.style(s, bold=True) |
| 127 | + |
| 128 | + h = [headers[0].ljust(BIG_WIDTH)] |
| 129 | + h.extend([each.center(SMALL_WIDTH)[:SMALL_WIDTH] for each in headers[1:]]) |
| 130 | + |
| 131 | + len_h = len("| " + " | ".join(h) + " |") |
| 132 | + |
| 133 | + sep = ( |
| 134 | + f"{JUNCTION}-" |
| 135 | + + f"-{JUNCTION}-".join(["-" * len(each) for each in h]) |
| 136 | + + f"-{JUNCTION}" |
| 137 | + ) |
| 138 | + top_sep = f"{JUNCTION}" + "-" * (len_h - 2) + f"{JUNCTION}" |
| 139 | + |
| 140 | + if title: |
| 141 | + # click.echo(top_sep) |
| 142 | + click.echo( |
| 143 | + "| " + make_bold(f"Top {len(data)} {title}".center(len_h - 4)) + " |" |
| 144 | + ) |
| 145 | + click.echo(sep) |
| 146 | + |
| 147 | + click.echo("| " + " | ".join([make_bold(each) for each in h]) + " |") |
| 148 | + click.echo(sep) |
| 149 | + |
| 150 | + for test in data: |
| 151 | + s = [] |
| 152 | + for i, each_key in enumerate(keys): |
| 153 | + |
| 154 | + if i == 0: |
| 155 | + id_ = test[each_key] |
| 156 | + |
| 157 | + id_ = ( |
| 158 | + id_.replace("(", "\(") |
| 159 | + .replace(")", "\)") |
| 160 | + .replace("[", "\[") |
| 161 | + .replace("]", "\]") |
| 162 | + ) |
| 163 | + if len(id_) >= BIG_WIDTH: |
| 164 | + id_ = id_[: BIG_WIDTH - 15] + "..." + id_[-12:] |
| 165 | + |
| 166 | + s.append(f"{id_}".ljust(BIG_WIDTH)[0:BIG_WIDTH]) |
| 167 | + |
| 168 | + elif each_key == "n_tests": |
| 169 | + s.append(f"{int(test[each_key])}".center(SMALL_WIDTH)) |
| 170 | + else: |
| 171 | + if np.isnan(test[each_key]): |
| 172 | + s.append("N/A".center(SMALL_WIDTH)) |
| 173 | + else: |
| 174 | + s.append(f"{test[each_key]:.4f}".center(SMALL_WIDTH)) |
| 175 | + |
| 176 | + click.echo("| " + " | ".join(s) + " |") |
| 177 | + # click.echo(sep) |
| 178 | + |
| 179 | + |
| 180 | +def print_summary(summary, num=10): |
| 181 | + """Print the top N longest tests and the top N most variable tests.""" |
| 182 | + longest_tests = sorted(summary, key=lambda x: -x["average_duration"])[:num] |
| 183 | + most_variable_tests = sorted(summary, key=lambda x: -x["std_dev"])[:num] |
| 184 | + |
| 185 | + print(f"\n## Top {num} Longest Running Tests\n") |
| 186 | + print_table( |
| 187 | + longest_tests, |
| 188 | + ["id", "n_tests", "average_duration", "std_dev"], |
| 189 | + ["Test ID", "N. tests", "Avg", "STD"], |
| 190 | + # "Longest Running Tests", |
| 191 | + ) |
| 192 | + |
| 193 | + print("") |
| 194 | + print(f"\n## Top {num} Most Variable Running Tests\n") |
| 195 | + print_table( |
| 196 | + most_variable_tests, |
| 197 | + [ |
| 198 | + "id", |
| 199 | + "n_tests", |
| 200 | + "std_dev", |
| 201 | + "average_duration", |
| 202 | + "std_dev_minus_one", |
| 203 | + "avg_duration_minus_one", |
| 204 | + "std_dev_minus_34", |
| 205 | + "avg_duration_minus_34", |
| 206 | + ], |
| 207 | + [ |
| 208 | + "Test ID", |
| 209 | + "N. tests", |
| 210 | + "Std", |
| 211 | + "Avg", |
| 212 | + "Std-99%", |
| 213 | + "Avg-99%", |
| 214 | + "Std-75%", |
| 215 | + "Avg-75%", |
| 216 | + ], |
| 217 | + # "Most Variable Running Tests", |
| 218 | + ) |
| 219 | + |
| 220 | + |
| 221 | +@click.command() |
| 222 | +@click.option( |
| 223 | + "--directory", |
| 224 | + type=click.Path(exists=True, file_okay=False, dir_okay=True), |
| 225 | + default=None, |
| 226 | +) |
| 227 | +@click.option( |
| 228 | + "--num", default=10, help="Number of top tests to display.", show_default=True |
| 229 | +) |
| 230 | +@click.option( |
| 231 | + "--save-file", |
| 232 | + default=None, |
| 233 | + help="File to save the test durations. Default 'tests_durations.json'.", |
| 234 | + show_default=True, |
| 235 | +) |
| 236 | +def analyze_tests(directory, num, save_file): |
| 237 | + directory = directory or os.getcwd() # Change this to your base directory |
| 238 | + json_files = find_json_files(directory) |
| 239 | + tests = extract_tests_with_tags(json_files) |
| 240 | + |
| 241 | + if save_file: |
| 242 | + with open(save_file, "a+") as f: |
| 243 | + for each_line in tests: |
| 244 | + json.dump(each_line, f, indent=2) |
| 245 | + |
| 246 | + summary = compute_statistics(tests) |
| 247 | + print_summary(summary, num=num) |
| 248 | + |
| 249 | + |
| 250 | +if __name__ == "__main__": |
| 251 | + analyze_tests() |
0 commit comments