correction

0x03-log_parsing/0-generator.py

@@ -1,23 +1,16 @@
 #!/usr/bin/python3
-'''A script that generates random HTTP request logs.
-'''
+"""Log Parsing"""
 import random
 import sys
-import datetime
 from time import sleep
-
+import datetime
 
 for i in range(10000):
     sleep(random.random())
-    sys.stdout.write("{:d}.{:d}.{:d}.{:d} - [{}] \"GET {} {}\" {} {}\n".format(
-        random.randint(1, 255),
-        random.randint(1, 255),
-        random.randint(1, 255),
-        random.randint(1, 255),
+    sys.stdout.write("{:d}.{:d}.{:d}.{:d} - [{}] \"GET /projects/260 HTTP/1.1\" {} {}\n".format(
+        random.randint(1, 255), random.randint(1, 255), random.randint(1, 255), random.randint(1, 255),
         datetime.datetime.now(),
-        '/projects/1216',
-        'HTTP/1.1',
         random.choice([200, 301, 400, 401, 403, 404, 405, 500]),
         random.randint(1, 1024)
     ))
-    sys.stdout.flush()
+    sys.stdout.flush()

0x03-log_parsing/0-stats.py

@@ -1,53 +1,87 @@
-#!/usr/bin/env python3
-"""Log Parsing"""
+#!/usr/bin/python3
+"""A script for parsing HTTP request logs."""
+import re
 
-import sys
 
-def print_stats(total_file_size, status_codes):
-    """Prints the stats"""
-    print("File size:", total_file_size)
-    for code in sorted(status_codes.keys()):
-        print(f"{code}: {status_codes[code]}")
-
-def main():
-    """Main function"""
-    total_file_size = 0
-    status_codes = {
-        200: 0,
-        301: 0,
-        400: 0,
-        401: 0,
-        403: 0,
-        404: 0,
-        405: 0,
-        500: 0
+def extract_input(input_line):
+    '''Extracts sections of a line of an HTTP request log.
+    '''
+    fp = (
+        r'\s*(?P<ip>\S+)\s*',
+        r'\s*\[(?P<date>\d+\-\d+\-\d+ \d+:\d+:\d+\.\d+)\]',
+        r'\s*"(?P<request>[^"]*)"\s*',
+        r'\s*(?P<status_code>\S+)',
+        r'\s*(?P<file_size>\d+)'
+    )
+    info = {
+        'status_code': 0,
+        'file_size': 0,
     }
+    log_fmt = '{}\\-{}{}{}{}\\s*'.format(fp[0], fp[1], fp[2], fp[3], fp[4])
+    resp_match = re.fullmatch(log_fmt, input_line)
+    if resp_match is not None:
+        status_code = resp_match.group('status_code')
+        file_size = int(resp_match.group('file_size'))
+        info['status_code'] = status_code
+        info['file_size'] = file_size
+    return info
 
-    try:
-        for idx, line in enumerate(sys.stdin, start=1):
-            line = line.strip()
-            parts = line.split()
 
-            if len(parts) != 7:
-                continue
+def print_statistics(total_file_size, status_codes_stats):
+    '''Prints the accumulated statistics of the HTTP request log.
+    '''
+    print('File size: {:d}'.format(total_file_size), flush=True)
+    for status_code in sorted(status_codes_stats.keys()):
+        num = status_codes_stats.get(status_code, 0)
+        if num > 0:
+            print('{:s}: {:d}'.format(status_code, num), flush=True)
+
 
-            _, _, _, status_code_str, file_size_str = parts
+def update_metrics(line, total_file_size, status_codes_stats):
+    '''Updates the metrics from a given HTTP request log.
 
-            try:
-                status_code = int(status_code_str)
-                file_size = int(file_size_str)
-            except ValueError:
-                continue
+    Args:
+        line (str): The line of input from which to retrieve the metrics.
 
-            total_file_size += file_size
-            if status_code in status_codes:
-                status_codes[status_code] += 1
+    Returns:
+        int: The new total file size.
+    '''
+    line_info = extract_input(line)
+    status_code = line_info.get('status_code', '0')
+    if status_code in status_codes_stats.keys():
+        status_codes_stats[status_code] += 1
+    return total_file_size + line_info['file_size']
 
-            if idx % 10 == 0:
-                print_stats(total_file_size, status_codes)
 
-    except KeyboardInterrupt:
-        print_stats(total_file_size, status_codes)
+def run():
+    '''Starts the log parser.
+    '''
+    line_num = 0
+    total_file_size = 0
+    status_codes_stats = {
+        '200': 0,
+        '301': 0,
+        '400': 0,
+        '401': 0,
+        '403': 0,
+        '404': 0,
+        '405': 0,
+        '500': 0,
+    }
+    try:
+        while True:
+            line = input()
+            total_file_size = update_metrics(
+                line,
+                total_file_size,
+                status_codes_stats,
+            )
+            line_num += 1
+            if line_num % 10 == 0:
+                print_statistics(total_file_size, status_codes_stats)
+    except (KeyboardInterrupt, EOFError):
+        print_statistics(total_file_size, status_codes_stats)
+
 
-if __name__ == "__main__":
-    main()
+if __name__ == '__main__':
+    run()