Wikipedia page lengths

lengths.py

import requests, json, re

def main():
	states = [l.split(",") for l in open("states.txt", "r").read().split("\n")]
	state_codes = {l.split(",")[0].lower(): l.split(",")[1].lower() for l in open("state_codes.csv", "r").read().split("\n")}
	styling = ""
	for state in states:
		length = page_length(state[0], True)
		print(f"{state[1].title()}: {length}")
		styling += f".{state_codes[state[1]]} {{fill: {get_color(length)};}}\n"
	print(styling)

def page_length(title, recursive=False):
	req = requests.get(f"https://en.wikipedia.org/w/api.php?action=parse&page={title}&format=json")
	content = json.loads(req.content)["parse"]["text"]["*"]
	sum_len = len(content)

	if not recursive:
		return sum_len

	links = get_links(content)
	for link in links:
		sum_len += page_length(link)
	
	return sum_len

def get_links(content):
	regex = r"Main articles?: (?:<a href=\"\/wiki\/(.+?)\".+?>.+?</a> ?a?n?d? ?)?(?:<a href=\"\/wiki\/(.+?)\".+?>.+?</a> ?a?n?d? ?){1,}"
	matches = re.findall(regex, content)
	new_matches = []
	for match in matches:
		for rm in match:
			if len(rm) > 0:
				new_matches.append(rm.split("#")[0])
	return new_matches

def get_color(length):
	ratio = max(min(length / 5800883, 1), 0)
	r = round(255 - ratio * 128)
	gb = round(255 - ratio * 255)
	return f"rgba({r}, {gb}, {gb}, 1)"

if __name__ == "__main__":
	main()

state_codes.csv

states.txt

California,california
Texas,texas
Florida,florida
New_York_(state),new york
Pennsylvania,pennsylvania
Illinois,illinois
Ohio,ohio
Georgia_(U.S._state),georgia
North_Carolina,north carolina
Michigan,michigan
New_Jersey,new jersey
Virginia,virginia
Washington_(state),washington
Arizona,arizona
Massachusetts,massachusetts
Tennessee,tennessee
Indiana,indiana
Missouri,missouri
Maryland,maryland
Wisconsin,wisconsin
Colorado,colorado
Minnesota,minnesota
South_Carolina,south carolina
Alabama,alabama
Louisiana,louisiana
Kentucky,kentucky
Oregon,oregon
Oklahoma,oklahoma
Connecticut,connecticut
Utah,utah
Iowa,iowa
Nevada,nevada
Arkansas,arkansas
Mississippi,mississippi
Kansas,kansas
New_Mexico,new mexico
Nebraska,nebraska
Idaho,idaho
West_Virginia,west virginia
Hawaii,hawaii
New_Hampshire,new hampshire
Maine,maine
Montana,montana
Rhode_Island,rhode island
Delaware,delaware
South_Dakota,south dakota
North_Dakota,north dakota
Alaska,alaska
Vermont,vermont
Wyoming,wyoming