Le site arthur.bebou.netlib.re - retour accueil
git clone git://bebou.netlib.re/arthur.bebou
Log | Files | Refs |
sub.py (2988B)
1 #!/usr/bin/env python3
2 import re
3 import sys
4 import time
5
6 # Constants
7 _NB_RUN = 10
8 _PATTERN = re.compile(r"https://sws.geonames.org/(\d+)/about.rdf")
9 # _PATH_TXT = "./data/input_big.xml"
10 _PATH_TXT = "./A"
11 _PATH_MAP = "corres"
12 _PATH_RES = "output.xml"
13
14
15 def build_map(data):
16 """
17 Build a mapping dictionary from a list of data.
18
19 Args:
20 data (list): A list of strings where each even-indexed element is a key
21 and each odd-indexed element is a value.
22
23 Returns:
24 dict: A dictionary mapping keys to values.
25 """
26 return {data[k][:-1]: data[k - 1][:-1] for k in range(1, len(data), 2)}
27
28
29 def replace_match(match, mapping):
30 """
31 Replace a matched string using a mapping dictionary.
32
33 Args:
34 match (re.Match): A match object containing the matched string.
35 mapping (dict): A dictionary mapping keys to replacement values.
36
37 Returns:
38 str: The replacement string.
39 """
40 key = match.group(1)
41 return mapping.get(key, match.group(0))
42
43
44 def read_file(file_path):
45 """
46 Read the content of a file.
47
48 Args:
49 file_path (str): The path to the file to be read.
50
51 Returns:
52 str: The content of the file as a string.
53 """
54 try:
55 with open(file_path, "rb") as file:
56 return file.read().decode("utf-8")
57 except FileNotFoundError:
58 sys.exit(f"Error: File '{file_path}' not found.")
59 except Exception as e:
60 sys.exit(f"Error reading file '{file_path}': {str(e)}")
61
62
63 def write_file(file_path, content):
64 """
65 Write content to a file.
66
67 Args:
68 file_path (str): The path to the file to be written.
69 content (str): The content to write to the file.
70 """
71 try:
72 with open(file_path, "wb") as file:
73 file.write(content.encode("utf-8"))
74 except Exception as e:
75 sys.exit(f"Error writing to file '{file_path}': {str(e)}")
76
77
78 def process(input_file, map_file, pattern):
79 """
80 Process input data using a pattern and a mapping dictionary and write the
81 result to an output file.
82
83 Args:
84 input_file (str): The path to the input file.
85 map_file (str): The path to the map file.
86 pattern (re.Pattern): A regular expression pattern to match.
87
88 """
89 txt = read_file(input_file)
90 mapping = build_map(read_file(map_file))
91 output = pattern.sub(lambda match: replace_match(match, mapping), txt)
92 write_file(_PATH_RES, output)
93
94
95 if __name__ == "__main__":
96 times = []
97 print(f"Starting batch of {_NB_RUN} runs...")
98 for run, _ in enumerate(range(_NB_RUN)):
99 start_time = time.time()
100 process(_PATH_TXT, _PATH_MAP, _PATTERN)
101 sys.stdout.write(f"\rRun {run + 1}/{_NB_RUN} completed")
102 sys.stdout.flush()
103 times.append(time.time() - start_time)
104
105 print("\nExecution times:")
106 print(f" - min: {min(times):.5f} sec")
107 print(f" - avg: {sum(times) / _NB_RUN:.5f} sec")
108 print(f" - max: {max(times):.5f} sec")