解析ftrace-block

例子：

             fio-30749 [036] 5651360.257707: block_rq_issue: 8,0 WS 4096 () 1367650688 + 8 [fio]
             <idle>-0 [036] 5651360.257768: block_rq_complete: 8,0 WS () 1367650688 + 8 [0]

import os
import sys
import re
from openpyxl import Workbook

def parse_ftrace_file(file_path):
    #issue_pattern = re.compile(r'(\w+-\d+) \[(\d+)\] (\d+\.\d+): block_rq_issue: (\d+),(\d+) (\w+) (\d+) \(\) (\d+) \+ (\d+)')
    issue_pattern = re.compile(r'(\d+)\.(\d+): block_rq_issue: (\d+),(\d+) (\w+) (\d+) \(\) (\d+) \+ (\d+)')

    complete_pattern = re.compile(r'(\d+)\.(\d+): block_rq_complete: (\d+),(\d+) (\w+) \(\) (\d+) \+ (\d+)')

    with open(file_path, 'r') as f:
        lines = f.readlines()

    request_details = []
    request_counts = {}
    for line in lines:
        print(line)
        issue_match = issue_pattern.search(line)
        if issue_match:
            issue_time = float(issue_match.group(1)) + float(issue_match.group(2)) / 1e6
            operation = issue_match.group(5)
            lba = int(issue_match.group(7))
            size = int(issue_match.group(8))  # 获取请求的大小
            request_details.append((operation, lba, size, issue_time, None, None))
            request_counts[operation] = request_counts.get(operation, 0) + 1
            continue

        complete_match = complete_pattern.search(line)
        if complete_match:
            complete_time = float(complete_match.group(1)) + float(complete_match.group(2)) / 1e6
            complete_request_id = (complete_match.group(5), int(complete_match.group(6)), int(complete_match.group(7)))
            print("complete_request_id:",complete_request_id)

            for i, details in enumerate(request_details):
                if details[0:3] == complete_request_id:
                    latency = complete_time - details[3]
                    request_details[i] = (details[0], details[1], details[2], details[3], complete_time, latency)
                    break

    return request_details, request_counts

def calculate_percentages(request_counts):
    total_count = sum(request_counts.values())
    percentages = {operation: count / total_count * 100 for operation, count in request_counts.items()}
    return percentages

def save_to_excel(request_details, request_counts, output_file):
    wb = Workbook()
    ws = wb.active

    ws['A1'] = 'Operation'
    ws['B1'] = 'Count'
    ws['C1'] = 'Percentage'

    percentages = calculate_percentages(request_counts)
    for i, (operation, count) in enumerate(request_counts.items(), start=2):
        ws[f'A{i}'] = operation
        ws[f'B{i}'] = count

    for i, (operation, percentage) in enumerate(percentages.items(), start=2):
        ws[f'C{i}'] = percentage

    ws['A{}'.format(len(request_counts) + 2)] = 'Operation'
    ws['B{}'.format(len(request_counts) + 2)] = 'LBA'
    ws['C{}'.format(len(request_counts) + 2)] = 'Size'
    ws['D{}'.format(len(request_counts) + 2)] = 'Issue Time'
    ws['E{}'.format(len(request_counts) + 2)] = 'Complete Time'
    ws['F{}'.format(len(request_counts) + 2)] = 'Latency'

    for i, details in enumerate(request_details, start=len(request_counts) + 3):
        ws[f'A{i}'] = details[0]
        ws[f'B{i}'] = details[1]
        ws[f'C{i}'] = details[2]
        ws[f'D{i}'] = details[3]
        ws[f'E{i}'] = details[4]
        ws[f'F{i}'] = details[5]

    wb.save(output_file)

if __name__ == "__main__":
    if len(sys.argv) != 3:
        print("Usage: python script.py <path_to_ftrace_file<output_excel_file>")
        sys.exit(1)

    ftrace_file_path = sys.argv[1]
    output_excel_file = sys.argv[2]

    if not os.path.exists(ftrace_file_path):
        print(f"Error: File '{ftrace_file_path}' does not exist.")
        sys.exit(1)

    request_details, request_counts = parse_ftrace_file(ftrace_file_path)
    print(f"Request details: {request_details}")
    print(f"Request counts: {request_counts}")

    save_to_excel(request_details, request_counts, output_excel_file)
    print(f"Results saved to '{output_excel_file}'")

import re
import pandas as pd

读取ftrace文件内容
with open("path/to/your/ftrace_file.txt", "r") as f:
ftrace_output = f.read()

定义一个正则表达式模式，用于匹配block_rq_issue和block_rq_complete事件
pattern = r"(\d+.\d+):\s+(block_rq_issue|block_rq_complete):\s+(\d+),\s*(\d+)\s+([A-Z]+)\s+(\d+)\s+()\s+(\d+)\s++\s+(\d+)\s+[([a-zA-Z0-9_-]+)]"

使用正则表达式匹配ftrace输出
matches = re.finditer(pattern, ftrace_output)

初始化变量
request_data = {}
data = []

遍历匹配结果并提取信息
for match in matches:
timestamp = match.group(1)
event_type = match.group(2)
dev_major = match.group(3)
dev_minor = match.group(4)
rwbs = match.group(5)
sector = match.group(6)
nr_sector = match.group(7)
comm = match.group(8)

if event_type == "block_rq_issue":
request_id = (dev_major, dev_minor, sector, nr_sector, comm)
request_data[request_id] = {
"Timestamp Issue": timestamp,
"Dev Major": dev_major,
"Dev Minor": dev_minor,
"RWBS": rwbs,
"Sector": sector,
"Nr Sector": nr_sector,
"Comm": comm
}
elif event_type == "block_rq_complete":
request_id = (dev_major, dev_minor, sector, nr_sector, comm)
if request_id in request_data:
request_datarequest_id = timestamp
data.append(request_data[request_id])
del request_data[request_id]

将字典列表转换为DataFrame
df = pd.DataFrame(data)

计算延迟并将其添加到DataFrame
df["Latency"] = df.apply(lambda row: float(row["Timestamp Complete"]) - float(row["Timestamp Issue"]), axis=1)

将DataFrame写入Excel文件
df.to_excel("output.xlsx", index=False)