Add log analysis script for homepage response time percentiles (#73)

Author: Kushagra1122

scripts/log.py

@@ -0,0 +1,208 @@
+import re
+import random
+import argparse
+from pathlib import Path
+from datetime import datetime
+from collections import defaultdict
+import numpy as np
+import matplotlib.pyplot as plt
+
+# keep max samples per day (adjust if needed)
+MAX_SAMPLES = 200000
+
+# Log format looks like:
+# [pid: ...] <ip> ... [Tue Feb 10 14:47:06 2026] GET /... => generated ... in 2740 msecs
+# Use a greedy prefix so we capture the *last* [...] date block before GET.
+pattern = re.compile(
+    r".*\[(?P<date>(?:Mon|Tue|Wed|Thu|Fri|Sat|Sun)\s+[A-Za-z]{3}\s+\d{1,2}\s+\d{2}:\d{2}:\d{2}\s+\d{4})\]\s+GET\s+(?P<path>/[^\s]*)\s+=> generated .* in (?P<time>\d+) msecs"
+)
+
+# reservoir per day
+samples = defaultdict(list)
+counts = defaultdict(int)
+
+def reservoir_add(day, value):
+    counts[day] += 1
+    c = counts[day]
+
+    if len(samples[day]) < MAX_SAMPLES:
+        samples[day].append(value)
+    else:
+        # replace elements with decreasing probability
+        j = random.randint(0, c - 1)
+        if j < MAX_SAMPLES:
+            samples[day][j] = value
+
+
+def _default_log_file() -> Path | None:
+    downloads = Path.home() / "Downloads"
+    if not downloads.exists():
+        return None
+    # Prefer the exact historical naming scheme, but fall back to any similar file.
+    matches = list(downloads.glob("orumsV3_spoken.log-*"))
+    if not matches:
+        matches = list(downloads.glob("*orumsV3_spoken*"))
+    matches = sorted(matches, key=lambda p: p.stat().st_mtime, reverse=True)
+    return matches[0] if matches else None
+
+
+def _parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description="Compute homepage response-time percentiles from a log file.")
+    parser.add_argument(
+        "--log-file",
+        type=str,
+        default=None,
+        help="Path to the log file. If omitted, uses the newest ~/Downloads/orumsV3_spoken.log-* file.",
+    )
+    parser.add_argument(
+        "--max-samples",
+        type=int,
+        default=MAX_SAMPLES,
+        help="Max reservoir samples per day (default: %(default)s).",
+    )
+    parser.add_argument(
+        "--days",
+        type=str,
+        default="10,11,12,13,14,15",
+        help="Comma-separated day-of-month numbers to include for Feb (default: %(default)s).",
+    )
+    parser.add_argument(
+        "--homepage-match",
+        choices=["strict", "allow-query"],
+        default="allow-query",
+        help="Homepage matching: strict='/' only; allow-query='/' and '/?…' (default: %(default)s).",
+    )
+    parser.add_argument(
+        "--plot",
+        choices=["compare", "hist", "both", "none"],
+        default="compare",
+        help="Plot type (default: %(default)s).",
+    )
+    return parser.parse_args()
+
+
+args = _parse_args()
+MAX_SAMPLES = args.max_samples
+days = {int(x.strip()) for x in args.days.split(",") if x.strip()}
+
+def _is_homepage(path: str) -> bool:
+    if args.homepage_match == "strict":
+        return path == "/"
+    # allow-query
+    return path == "/" or path.startswith("/?")
+
+log_file = Path(args.log_file).expanduser() if args.log_file else _default_log_file()
+if not log_file or not log_file.exists():
+    downloads = Path.home() / "Downloads"
+    default_hint = downloads / "orumsV3_spoken.log-*"
+    raise SystemExit(
+        "Log file not found.\n"
+        f"Tried auto-detecting the newest file matching `{default_hint}`.\n"
+        "Pass it explicitly, e.g.:\n"
+        "  python log.py --log-file ~/Downloads/orumsV3_spoken.log-20260215\n"
+    )
+
+
+with open(log_file, "r", errors="ignore") as f:
+    for line in f:
+        m = pattern.search(line)
+        if not m:
+            continue
+
+        path = m.group("path").strip()
+
+        # homepage only
+        if not _is_homepage(path):
+            continue
+
+        raw_date = m.group("date")
+        response_time = int(m.group("time"))
+
+        dt = datetime.strptime(raw_date, "%a %b %d %H:%M:%S %Y")
+
+        # only Feb 10–15
+        if dt.month == 2 and dt.day in days:
+            day_key = dt.strftime("%Y-%m-%d")
+            reservoir_add(day_key, response_time)
+
+
+# compute percentiles & print a comparison table
+print("\n📊 Homepage response-time percentiles (Feb comparison)\n")
+print(f"Log file: {log_file}")
+print(f"Homepage match: {args.homepage_match}")
+print("")
+
+ordered_days = [f"2026-02-{d:02d}" for d in sorted(days)]
+
+rows = []
+results = {}  # day -> np.ndarray (for optional plotting)
+
+for day in ordered_days:
+    arr_list = samples.get(day, [])
+    if not arr_list:
+        rows.append((day, 0, counts.get(day, 0), None, None, None))
+        continue
+
+    arr = np.array(arr_list)
+    results[day] = arr
+
+    p50 = float(np.percentile(arr, 50))
+    p80 = float(np.percentile(arr, 80))
+    p90 = float(np.percentile(arr, 90))
+    rows.append((day, len(arr), counts.get(day, len(arr)), p50, p80, p90))
+
+header = ("Date", "samples_used", "total_seen", "p50_ms", "p80_ms", "p90_ms")
+print(f"{header[0]:<12} {header[1]:>12} {header[2]:>10} {header[3]:>10} {header[4]:>10} {header[5]:>10}")
+print("-" * 70)
+for (day, n, total, p50, p80, p90) in rows:
+    if p50 is None:
+        print(f"{day:<12} {n:>12} {total:>10} {'-':>10} {'-':>10} {'-':>10}")
+    else:
+        print(f"{day:<12} {n:>12} {total:>10} {p50:>10.1f} {p80:>10.1f} {p90:>10.1f}")
+
+
+def _plot_histograms() -> None:
+    for day, arr in results.items():
+        plt.figure()
+        plt.hist(arr, bins=40)
+        plt.title(f"Homepage Response Time Distribution ({day})")
+        plt.xlabel("Response Time (ms)")
+        plt.ylabel("Requests")
+        plt.grid(True)
+
+
+def _plot_comparison() -> None:
+    xs = []
+    p50s = []
+    p80s = []
+    p90s = []
+    for (day, n, _total, p50, p80, p90) in rows:
+        if n == 0:
+            continue
+        xs.append(day)
+        p50s.append(p50)
+        p80s.append(p80)
+        p90s.append(p90)
+
+    if not xs:
+        return
+
+    plt.figure()
+    plt.plot(xs, p50s, marker="o", label="P50")
+    plt.plot(xs, p80s, marker="o", label="P80")
+    plt.plot(xs, p90s, marker="o", label="P90")
+    plt.title("Homepage Response Time Percentiles (Feb 10–15)")
+    plt.xlabel("Date")
+    plt.ylabel("Response time (ms)")
+    plt.grid(True)
+    plt.legend()
+    plt.xticks(rotation=45, ha="right")
+    plt.tight_layout()
+
+
+if args.plot in ("hist", "both"):
+    _plot_histograms()
+if args.plot in ("compare", "both"):
+    _plot_comparison()
+if args.plot != "none":
+    plt.show()