uat_extractor.py

uat_extractor.py

import csv
import random
import re

"""
Helps us find the number in between two brackets ie (33)
"""
def find_between(text, start, end):
    return re.findall(re.escape(start) + "(.*)" + re.escape(end), text)[0].strip()

"""
Reads the input file then performs the following
1. read one line at a time
2. convert lines that have UVM_INFO, UVM_WARNING, UVM_ERROR into one line
3. extract the fields "message type, full path to file, line number, time, hierarchical location, message" from the rows
4. create a csv with fields "message type, full path to file, line number, time, hierarchical location, message"
https://www.w3schools.com/python/python_file_open.asp
"""
def read_input_file():
    k = open("input1.txt", "r")
    rows = []

    # read commands in full
    for y in k:
        if "UVM_INFO" in y or "UVM_WARNING" in y or "UVM_ERROR" in y:
            rows.append(y)
        elif rows:
            rows[-1] = rows[-1] + " " + y

    # extract the data from the commands using the structure
    # <message type> <full path to file>(<line number>) @ <time><timeunit>: <hierarchical location> <message>
    extracted_data = []
    for row in rows:
        try:
            first_part = row.split('@')[0].strip()
            last_part = row.split('@')[-1].strip().replace('\n', '')
            message_type = first_part.split(' ', 1)[0]
            path = first_part.split(' ', 1)[-1]
            line_number = find_between(first_part, '(', ')')
            time = last_part.split('ns:')[0]
            hl_message = last_part.split('ns:')[-1].strip()
            hl = hl_message.split(' ', 1)[0]
            message = hl_message.split(' ', 1)[-1]
            print(message_type, path, line_number, time, hl, message)
            extracted_data.append([message_type, path, line_number, time, hl, message])
        except Exception as e:
            print(e)
    return extracted_data


# https://www.pythontutorial.net/python-basics/python-write-csv-file/
def write_csv(data):
    header = ['message type', 'full path to file', 'line number', 'time', 'hierarchical location', 'message']
    with open(str(random.randint(100000, 999999)) + 'uart_monitor.csv', 'w', encoding='UTF8') as f:
        writer = csv.writer(f)
        # write the header
        writer.writerow(header)
        for row in data:
            # write the data
            writer.writerow(row)

def start():
    data = read_input_file()
    write_csv(data)

if __name__ == '__main__':
    start()