File size: 6,213 Bytes
23a7a20
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
#!/usr/bin/env python3
"""Compute crawler workload statistics for evaluation."""

import os
import csv
import numpy as np
from transformers import AutoTokenizer
import argparse
from pathlib import Path
from tqdm import tqdm
from functools import partial
import multiprocessing as mp


def tokenize_text(text: str, tokenizer_model: str) -> int:
    """Tokenize text and return token count."""
    try:
        tokenizer = AutoTokenizer.from_pretrained(tokenizer_model,
                                                   local_files_only=True)
    except:
        tokenizer = AutoTokenizer.from_pretrained(tokenizer_model)

    tokens = tokenizer.encode(text, truncation=False, add_special_tokens=True)
    return len(tokens)


def process_query_trace(csv_file: str, tokenizer_model: str):
    """Process a single query trace file and return statistics."""
    try:
        tokenizer = AutoTokenizer.from_pretrained(tokenizer_model,
                                                   local_files_only=True)
    except:
        tokenizer = AutoTokenizer.from_pretrained(tokenizer_model)

    total_tokens = 0
    start_time = None
    end_time = None
    page_count = 0

    try:
        with open(csv_file, 'r') as f:
            reader = csv.DictReader(f)
            for row in reader:
                if not row:
                    continue

                # Count tokens in page content
                if 'content' in row and row['content']:
                    tokens = tokenizer.encode(row['content'],
                                              truncation=False,
                                              add_special_tokens=True)
                    total_tokens += len(tokens)

                # Track start and end times
                if 'startTime' in row and row['startTime']:
                    try:
                        if start_time is None:
                            start_time = float(row['startTime'])
                    except:
                        pass

                if 'endTime' in row and row['endTime']:
                    try:
                        end_time = float(row['endTime'])
                    except:
                        pass

                page_count += 1

        if page_count > 0:
            total_time = 0.0
            if start_time is not None and end_time is not None:
                total_time = end_time - start_time
            return {'total_tokens': total_tokens, 'total_time': total_time}
    except Exception as e:
        pass

    return None


def main():
    parser = argparse.ArgumentParser(
        description="Compute crawler workload statistics")
    parser.add_argument("--input-dir",
                        "-i",
                        default="traces/simpleQA_ALL",
                        help="Directory containing crawler trace CSV files")
    parser.add_argument("--tokenizer-model",
                        "-t",
                        default="meta-llama/Llama-3.1-8B-Instruct",
                        help="HuggingFace tokenizer model")
    parser.add_argument("--cores",
                        type=int,
                        default=100,
                        help="Number of CPU cores to use")
    parser.add_argument("--max-queries",
                        type=int,
                        default=None,
                        help="Maximum number of queries to process")
    parser.add_argument("--output-dir",
                        default="tables",
                        help="Output directory for statistics file")

    args = parser.parse_args()

    # Find all CSV files
    input_dir = Path(args.input_dir)
    csv_files = list(input_dir.glob("*.csv"))
    if not csv_files:
        print(f"No CSV files found in {args.input_dir}")
        return

    if args.max_queries:
        csv_files = csv_files[:args.max_queries]

    print(f"Found {len(csv_files)} query files")
    print(f"Processing with {args.cores} cores...")

    # Process files
    worker_func = partial(process_query_trace,
                          tokenizer_model=args.tokenizer_model)

    total_tokens_list = []
    total_time_list = []

    if args.cores == 1:
        for csv_file in tqdm(csv_files, desc="Processing"):
            result = worker_func(str(csv_file))
            if result:
                total_tokens_list.append(result['total_tokens'])
                total_time_list.append(result['total_time'])
    else:
        with mp.Pool(args.cores) as pool:
            results = list(
                tqdm(pool.imap_unordered(worker_func,
                                         [str(f) for f in csv_files]),
                     total=len(csv_files),
                     desc="Processing"))
            for result in results:
                if result:
                    total_tokens_list.append(result['total_tokens'])
                    total_time_list.append(result['total_time'])

    # Compute statistics and save to file
    os.makedirs(args.output_dir, exist_ok=True)
    output_file = os.path.join(args.output_dir, "workload_stats_crawler.txt")

    with open(output_file, 'w') as f:
        f.write("\n" + "=" * 70 + "\n")
        f.write("CRAWLER WORKLOAD STATISTICS\n")
        f.write("=" * 70 + "\n")

        if total_tokens_list:
            total_tokens = np.array(total_tokens_list)
            f.write(f"\nQuery Total Tokens (n={len(total_tokens)})\n")
            f.write(f"  Mean: {total_tokens.mean():.0f} tokens\n")
            f.write(f"  P50: {np.percentile(total_tokens, 50):.0f} tokens\n")
            f.write(f"  P75: {np.percentile(total_tokens, 75):.0f} tokens\n")
            f.write(f"  P95: {np.percentile(total_tokens, 95):.0f} tokens\n")

        if total_time_list:
            total_time = np.array(total_time_list)
            f.write(f"\nTotal Collection Time (n={len(total_time)})\n")
            f.write(f"  Mean: {total_time.mean():.3f} seconds\n")
            f.write(f"  P50: {np.percentile(total_time, 50):.3f} seconds\n")
            f.write(f"  P75: {np.percentile(total_time, 75):.3f} seconds\n")
            f.write(f"  P95: {np.percentile(total_time, 95):.3f} seconds\n")

        f.write("=" * 70 + "\n")


if __name__ == "__main__":
    main()