Skip to content
TheChatArchiveThe Chat Archive
Mistralopenhermes-2.5-mistral-7bHugging Face Arena Human Preference 55K

How can I create a regex that will match the title of the movie, its rating (from 0.00…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with stripedhyena-nous-7b’s answer.
  1. User

    How can I create a regex that will match the title of the movie, its rating (from 0.00 to 5.00), and its year of release?

    I have the following: (?'title'.)\n(?'rating'\d.\d\d)\n(?'ratings'.)

    But there is a problem. My data has a flaw. Here's a sample of it:

    Annie Hall 3.94 6,776 54 20 April 1977

    Touch of Evil 3.97 3,282 48 30 March 1958

    Taiga 4.06 141 21 February 1992

    Trainspotting 3.93 10,950 68 23 February 1996

    Movies like 'Taiga' only have four lines rather than five. This makes extracting the year more annoying. Any help?

  2. Mistral Assistant

    You can modify your regex to handle movies with four lines instead of five. Here's an updated regex that should work:

    import re
    
    data = """Annie Hall
    3.94
    6,776
    54
    20 April 1977
    ----
    Touch of Evil
    3.97
    3,282
    48
    30 March 1958
    ----
    Taiga
    4.06
    141
    21 February 1992
    ----
    Trainspotting
    3.93
    10,950
    68
    23 February 1996
    ----"""
    
    pattern = r"(?'title'.*)\n(?'rating'\d+\.\d+)\n(?'audience'\d+)\n(?'date'.*)"
    
    for line in data.splitlines():
        match = re.search(pattern, line)
        if match:
            title = match.group("title")
            rating = float(match.group("rating"))
            audience = int(match.group("audience"))
            year = int(match.group("date").split()[5])
            print(f"Title: {title}, Rating: {rating}, Year: {year}")
    

    This will print the title, rating, and year for each movie in the data, even if there are only four lines per movie. The main change is the addition of \d+ to match the audience count, and the modification of the date extraction to only grab the year.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.