Dictiofuzzy - recursive words meaning stats

README.md

sources

dictionary_compact.json: https://github.com/matthewreagan/WebstersEnglishDictionary/blob/master/dictionary_compact.json

installation

$ pip install rich

main.py

import re
import json
import random

from rich.console import Console
from rich.markup import escape


with open('dictionary_compact.json', 'r') as json_file:
    dictionary = json.load(json_file)


WORD_RE = re.compile(r"[a-z']+", re.UNICODE | re.IGNORECASE)
known_words: set[str] = set()
unknown_words: set[str] = set()


console = Console()


def debug(*args, **kwargs):
    # console.print(*args, **kwargs)
    pass


def analyze(sentence: str):
    # iterative version

    words = WORD_RE.findall(sentence.lower())
    i = 0

    with console.status("Analyzing...") as status:
        while words:
            i += 1
            if i % 100 == 0:
                status.update(f"[bold blue]Total: {i}[reset] [green]|[reset] [bold]Analyzing...[reset] {len(words)} words left + {len(unknown_words)} unknown words")

            word = words.pop().lower()
            if word in known_words:
                continue
            elif word in unknown_words:
                continue
            if word not in dictionary:
                debug(f"Storing '{word}' in known_words, total: {len(known_words)}")
                unknown_words.add(word)
                # raise ValueError(f"Word '{word}' not found in dictionary")
                continue

            known_words.add(word)
            debug("\n[bold red]" + word)
            debug(escape(dictionary[word]))
            for word in reversed(WORD_RE.findall(dictionary[word].lower())):
                if word not in known_words:
                    words.append(word)


sentence = input("Enter a sentence: ")
analyze(sentence)

known_words = sorted(known_words)
unknown_words = sorted(unknown_words)

console.print(f":sparkles: To totally understand [bold blue]'{sentence}'[reset], you need to know: [green bold]{len(known_words)}[reset] words, plus [red bold]{len(unknown_words)}[reset] unknown words.")
console.print(f":bulb: Random known words: {random.sample(known_words, k=5)}", highlight=True)
console.print(f":fire: Random unknown words: {random.sample(unknown_words, k=5)}", highlight=True)