Merge #155131

155131: asim: add asimdiff r=wenyihu6 a=wenyihu6

This commit introduces a adds a python script that pipes git diffs to compare
certain metrics sum diff. Generated by claude.

Sample usage:
```
git diff | python3 pkg/kv/kvserver/asim/tests/cmd/asimdiff.py thrash_pct
git diff head~1 head | python3 pkg/kv/kvserver/asim/tests/cmd/asimdiff.py last  --metric mean
```

Sample output:
```
[skewed_cpu_skewed_write.txt] before = 249%,  after = 249%,  diff = +0%
  cpu_util#5: thrash_pct: [s1=40%, s2=76%, s3=49%, s4=30%, s5=33%, s6=21%]  (sum=249%)
...
total diff = -5374%  (total before = 498950%, total after = 493576%)
percent change = -1.1%
avg diff = -22.7%,  max diff = +7824%,  min diff = -5509%
changes: 136 regressions, 99 improvements, 2 unchanged
```

Epic: CRDB-55052
Release note: none

Co-authored-by: wenyihu6 <wenyi@cockroachlabs.com>

Author: craig[bot]

pkg/kv/kvserver/asim/tests/cmd/asimdiff.py

@@ -0,0 +1,163 @@
+# Copyright 2025 The Cockroach Authors.
+#
+# Use of this software is governed by the CockroachDB Software License
+# included in the /LICENSE file.
+
+#!/usr/bin/env python3
+import re
+import sys
+import argparse
+
+def extract_value(line, metric):
+    """Extract the value for a specific metric (sum, mean, or stddev) from a line"""
+    if metric == 'sum':
+        pattern = r'sum=(\d+)%?'
+    elif metric == 'mean':
+        pattern = r'mean=([\d.]+)%?'
+    elif metric == 'stddev':
+        pattern = r'stddev=([\d.]+)%?'
+    else:
+        return None
+
+    match = re.search(pattern, line)
+    if match:
+        value = match.group(1)
+        # Return as float for mean/stddev, int for sum
+        return float(value) if metric in ['mean', 'stddev'] else int(value)
+    return None
+
+def process_diff(diff_content, keyword, metric):
+    """Process git diff and extract changes for lines containing keyword"""
+    lines = diff_content.split('\n')
+    results = []
+    total_before = 0
+    total_after = 0
+    current_file = None
+
+    for line in lines:
+        # Track current file being processed
+        if line.startswith('diff --git') or line.startswith('+++'):
+            if line.startswith('+++'):
+                # Extract filename from +++ b/path/to/file
+                match = re.search(r'\+\+\+ b/(.+)', line)
+                if match:
+                    current_file = match.group(1)
+
+        if keyword not in line:
+            continue
+
+        if line.startswith('-') and not line.startswith('---'):
+            # Removed line (before)
+            value = extract_value(line, metric)
+            if value is not None:
+                total_before += value
+                results.append({
+                    'type': 'before',
+                    'line': line[1:].strip(),
+                    'value': value,
+                    'file': current_file
+                })
+        elif line.startswith('+') and not line.startswith('+++'):
+            # Added line (after)
+            value = extract_value(line, metric)
+            if value is not None:
+                total_after += value
+                results.append({
+                    'type': 'after',
+                    'line': line[1:].strip(),
+                    'value': value,
+                    'file': current_file
+                })
+
+    return results, total_before, total_after
+
+def print_results(results, total_before, total_after, metric):
+    """Print formatted results"""
+    # Group before/after pairs
+    before_lines = [r for r in results if r['type'] == 'before']
+    after_lines = [r for r in results if r['type'] == 'after']
+
+    # Collect individual diffs for statistics and sorting
+    entries = []
+
+    # Pair up changes and calculate diffs
+    max_len = max(len(before_lines), len(after_lines))
+    for i in range(max_len):
+        before_val = before_lines[i]['value'] if i < len(before_lines) else 0
+        after_val = after_lines[i]['value'] if i < len(after_lines) else 0
+        diff = after_val - before_val
+
+        full_path = after_lines[i]['file'] if i < len(after_lines) else (before_lines[i]['file'] if i < len(before_lines) else 'unknown')
+        filename = full_path.split('/')[-1] if full_path else 'unknown'
+        line_content = after_lines[i]['line'] if i < len(after_lines) else ''
+
+        entries.append({
+            'filename': filename,
+            'before': before_val,
+            'after': after_val,
+            'diff': diff,
+            'line': line_content
+        })
+
+    # Sort by absolute value of diff (largest to smallest)
+    entries.sort(key=lambda x: abs(x['diff']), reverse=True)
+
+    # Determine format string based on metric type
+    if metric == 'sum':
+        fmt = lambda x: f"{x:.0f}"
+    else:  # mean or stddev
+        fmt = lambda x: f"{x:.2f}"
+
+    # Print sorted entries
+    diffs = []
+    for entry in entries:
+        diffs.append(entry['diff'])
+        pct_diff = (entry['diff'] / entry['before'] * 100) if entry['before'] != 0 else 0
+        print(f"[{entry['filename']}] before = {fmt(entry['before'])},  after = {fmt(entry['after'])},  diff = {entry['diff']:+.2f} ({pct_diff:+.1f}%)")
+        print(f"  {entry['line']}")
+
+    # Calculate statistics
+    total_diff = total_after - total_before
+    avg_diff = sum(diffs) / len(diffs) if diffs else 0
+    max_diff = max(diffs) if diffs else 0
+    min_diff = min(diffs) if diffs else 0
+
+    # Calculate percent change
+    pct_change = (total_diff / total_before * 100) if total_before > 0 else 0
+
+    # Count regressions and improvements
+    regressions = sum(1 for d in diffs if d > 0)
+    improvements = sum(1 for d in diffs if d < 0)
+    unchanged = sum(1 for d in diffs if d == 0)
+
+    print(f"total diff = {total_diff:+.2f}  (total before = {fmt(total_before)}, total after = {fmt(total_after)})")
+    print(f"percent change = {pct_change:+.1f}%")
+    print(f"avg diff = {avg_diff:+.2f},  max diff = {max_diff:+.2f},  min diff = {min_diff:+.2f}")
+    print(f"changes: {regressions} regressions, {improvements} improvements, {unchanged} unchanged")
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description='Analyze git diff for lines containing a keyword with metric values',
+        epilog='Example: git diff | python %(prog)s thrash_pct --metric sum'
+    )
+    parser.add_argument('keyword', help='Keyword to search for in diff lines (e.g., thrash_pct)')
+    parser.add_argument('--metric', '-m', choices=['sum', 'mean', 'stddev'], default='sum',
+                        help='Metric to extract and compare (default: sum)')
+    parser.add_argument('--file', '-f', help='Diff file to read (if not provided, reads from stdin)')
+
+    args = parser.parse_args()
+
+    # Read from stdin or file
+    if args.file:
+        with open(args.file, 'r') as f:
+            diff_content = f.read()
+    else:
+        diff_content = sys.stdin.read()
+
+    results, total_before, total_after = process_diff(diff_content, args.keyword, args.metric)
+
+    if not results:
+        print(f"No lines containing '{args.keyword}' with {args.metric} values found in the diff!")
+        sys.exit(1)
+
+    print_results(results, total_before, total_after, args.metric)