wiki-mad-libs/batch/process_wiki_article.py

120 lines
3.6 KiB
Python

import hashlib
import json
import random
import re
import requests
import time
from nltk import pos_tag, sent_tokenize, word_tokenize
from nltk.corpus import stopwords
from wikipediaapi import Wikipedia, WikipediaPage
# Info about the default pos_tag tags
# https://www.ling.upenn.edu/courses/Fall_2003/ling001/penn_treebank_pos.html
adlib_tags = {
"JJ": "Adjective",
"JJR": "Adjective ending in 'er'",
"JJS": "Adjective ending in 'est'",
"NN": "Noun",
"NNS": "Plural Noun",
"NNP": "Proper Noun",
"NNPS": "Plural Proper Noun",
"RB": "Adverb",
"RBR": "Adverb ending in 'er'",
"RBS": "Adverb ending in 'est'",
"VB": "Verb",
"VBD": "Past Tense Verb",
"VBG": "Verb ending in 'ing'",
"VBN": "Past Tense Verb",
"VBP": "Present Tense Verb",
"VBZ": "Present Tense Verb ending in 's'",
}
stop_words = set(stopwords.words("english"))
months = {
"january",
"february",
"march",
"april",
"may",
"june",
"july",
"august",
"september",
"october",
"november",
"december",
}
stop_words.update(months)
# More stop words: becomes, become, became, well
def get_random_wikipedia_title() -> str:
random_result = json.loads(requests.get('https://en.wikipedia.org/api/rest_v1/page/random/title').text)
return random_result['items'][0]['title']
def get_random_wikipedia_page(wikipedia: Wikipedia, min_length: int = None) -> WikipediaPage:
page = None
while(page is None):
page = wikipedia.page(get_random_wikipedia_title())
if(min_length and len(page.summary) < min_length):
print(f"{page.displaytitle} is too short. Retrying...")
page = None
time.sleep(3)
return page
def sentence_tokenize_and_tag(text: str):
text = re.sub(r"(\w+)\.([A-Z])", r"\1. \2", text) # Try to break up sentences mashed together by stripping strings
sentences = sent_tokenize(text)
tagged_sentences = []
for sentence in sentences:
tagged_sentences.append(pos_tag(word_tokenize(sentence)))
return tagged_sentences
def adlibify(wiki_page, min_adlib_rest):
lowered_title = wiki_page.title.lower()
i = adlib_word_counter = 0
output_tokens = []
for sentence in sentence_tokenize_and_tag(wiki_page.summary):
for token, tag in sentence:
if tag in ["NNP", "NNPS"] and i > 0 and output_tokens[-1]["tag"] in ["NNP", "NNPS"]:
output_tokens[-1]["token"] += f" {token}"
else:
output_token = {"id": i, "token": token, "tag": tag}
adlib_tag = adlib_tags.get(tag)
if adlib_tag is not None:
if random.randint(0, adlib_word_counter) > min_adlib_rest \
and token.lower() not in stop_words \
and token.lower() not in lowered_title:
output_token["adlib_tag"] = adlib_tag
adlib_word_counter = 0
else:
adlib_word_counter += 1
output_tokens.append(output_token)
i += 1
url = wiki_page.canonicalurl
url_hash = hashlib.sha1(url.encode('utf-8')).hexdigest()[:8]
article = {'title': wiki_page.displaytitle, 'url': url, 'hash': url_hash}
article['tokens'] = output_tokens
return article
def main():
wikipedia = Wikipedia('en')
wiki_page = get_random_wikipedia_page(wikipedia, 500)
print(wiki_page.title)
article = adlibify(wiki_page, 4)
with open("article.js", "w") as json_file:
json_file.write(f"article = {json.dumps(article)}")
if __name__ == '__main__':
main()