metadata
dataset_info:
features:
- name: level
dtype: int64
- name: level_id
dtype: string
- name: category
dtype: string
- name: words
sequence: string
splits:
- name: train
num_bytes: 55055
num_examples: 711
download_size: 32798
dataset_size: 55055
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
license: mit
language:
- en
pretty_name: NYT Connections Answers
size_categories:
- n<1K
Made with the following script using a local copy of this website:
import re
from bs4 import BeautifulSoup
from datasets import Dataset
with open("Today’s NYT Connections Answers Jan 5, #574 - Daily Updates & Hints - Word Tips.htm", encoding="utf-8") as f:
html = f.read()
soup = BeautifulSoup(html, "html.parser")
texts = re.findall(r'"([^"]*)"', "".join(soup.find_all("script")[9]))
texts = [" ".join(text.split()).replace(" ,", ", ") for text in texts if ":" in text and (text.startswith("🟡") or text.startswith("🟢") or text.startswith("🔵") or text.startswith("🟣"))]
levels = {
"🟡": 1,
"🟢": 2,
"🔵": 3,
"🟣": 4
}
def gen():
for group in texts:
level_id = group[:1]
group = group[2:]
category, group = group.split(":")
entry = {
"level": levels[level_id],
"level_id": level_id,
"category": category,
"words": [word.strip() for word in group.split(",")]
}
#pprint(entry)
yield entry
dataset = Dataset.from_generator(gen)
dataset.push_to_hub("T145/connections")