Traces the bloodlines of academic royalty by sifting through Wikipedia.

royalty.py

#!/usr/bin/python
# Traces the bloodlines of academic royalty by sifting through Wikipedia.

import re
import urllib2
from BeautifulSoup import BeautifulSoup

opener = urllib2.build_opener()
opener.addheaders = [('User-agent', 'Mozilla/5.0')]
advisor = re.compile("advisor", re.IGNORECASE)
prefix = "http://en.wikipedia.org"

def procLink(link, level=0):
	try:
		page = opener.open(link).read()
		soup = BeautifulSoup(page)
	except:
		return []
	name = soup.find(id="firstHeading").string
	vcard = soup.find("table", "infobox vcard")
	if not vcard:
		return []
	row = vcard.find("tr", text=advisor)
	if not row:
		return name
	mentors = row.findNext("td").findChildren("a")
	refs = [prefix + elt['href'] for elt in mentors]	
	print level * ">> ", name
	children = [procLink(ref, level + 1) for ref in refs]
	return [name] + [children]

if __name__ == '__main__':
	print procLink("http://en.wikipedia.org/wiki/Feynman")
	print procLink("http://en.wikipedia.org/wiki/John_McCarthy_(computer_scientist)")

sample = """
 Richard Feynman
>>  John Archibald Wheeler
>> >>  Karl Herzfeld
>> >> >>  Friedrich Hasenöhrl
>> >> >> >>  Franz S. Exner
>> >> >> >> >>  August Kundt
>> >> >> >> >> >>  Heinrich Gustav Magnus
>> >> >> >> >> >> >>  Eilhard Mitscherlich
>> >> >> >> >> >> >> >>  Friedrich Stromeyer
>> >> >> >> >> >> >> >> >>  Johann Friedrich Gmelin
>> >> >> >> >> >> >> >> >> >>  Philipp Friedrich Gmelin
>> >> >> >> >> >> >> >> >> >> >>  Burchard Mauchart
>> >> >> >> >> >> >> >> >> >> >> >>  Elias Rudolph Camerarius Jr.
>> >> >> >> >> >> >> >> >> >> >> >> >>  Elias Rudolph Camerarius, Sr.
>> >> >> >> >> >> >> >> >> >> >> >> >> >>  Georg Balthasar Metzger
>> >> >> >> >> >> >> >> >> >> >> >> >> >> >>  Johann Georg Macasius
>> >> >> >> >> >> >> >> >> >> >> >> >> >> >> >>  Johannes Musaeus
>> >> >> >> >> >> >> >> >> >> >> >> >> >> >>  Emmanuel Stupanus
>> >> >> >> >> >> >> >> >>  Louis Nicolas Vauquelin
>> >> >> >> >> >> >> >> >> >>  Antoine François, comte de Fourcroy
[u'Richard Feynman', [[u'John Archibald Wheeler', [[u'Karl Herzfeld', [[u'Friedrich Hasen\xf6hrl', [[u'Franz S. Exner', [[u'August Kundt', [[u'Heinrich Gustav Magnus', [[u'Eilhard Mitscherlich', [[u'Friedrich Stromeyer', [[u'Johann Friedrich Gmelin', [[u'Philipp Friedrich Gmelin', [[u'Burchard Mauchart', [[u'Elias Rudolph Camerarius Jr.', [[u'Elias Rudolph Camerarius, Sr.', [[u'Georg Balthasar Metzger', [[u'Johann Georg Macasius', [[u'Johannes Musaeus', [[]]], []]], [u'Emmanuel Stupanus', [[]]]]]]]]]]]]], []]], [u'Louis Nicolas Vauquelin', [[u'Antoine Fran\xe7ois, comte de Fourcroy', [[]]]]]]]]]]]]]]]]]]]]]]]

 John McCarthy (computer scientist)
>>  Solomon Lefschetz
>> >>  William Edward Story
>> >> >>  Felix Klein
>> >> >> >>  Julius Plücker
>> >> >> >> >>  Christian Ludwig Gerling
>> >> >> >> >> >>  Carl Friedrich Gauss
>> >> >> >> >> >> >>  Johann Friedrich Pfaff
>> >> >> >> >> >> >> >>  Abraham Gotthelf Kästner
>> >> >> >> >> >> >> >> >>  Christian August Hausen
>> >> >> >> >> >> >> >> >> >>  Johann Christoph Wichmannshausen
>> >> >> >> >> >> >> >> >> >> >>  Otto Mencke
>> >> >> >> >> >> >> >> >> >> >> >>  Jakob Thomasius
>> >> >> >>  Rudolf Lipschitz
>> >> >> >> >>  Johann Peter Gustav Lejeune Dirichlet
>> >> >> >> >> >>  Siméon Denis Poisson
>> >> >> >> >> >> >>  Joseph Louis Lagrange
>> >> >> >> >> >> >> >>  Leonhard Euler
>> >> >> >> >> >> >> >> >>  Johann Bernoulli
>> >> >> >> >> >> >>  Pierre-Simon Laplace
>> >> >> >> >> >>  Joseph Fourier
>> >> >> >> >> >> >>  Joseph Louis Lagrange
>> >> >> >> >> >> >> >>  Leonhard Euler
>> >> >> >> >> >> >> >> >>  Johann Bernoulli
>> >> >>  Carl Neumann
[u'John McCarthy (computer scientist)', [[u'Solomon Lefschetz', [[u'William Edward Story', [[u'Felix Klein', [[u'Julius Pl\xfccker', [[u'Christian Ludwig Gerling', [[u'Carl Friedrich Gauss', [[u'Johann Friedrich Pfaff', [[u'Abraham Gotthelf K\xe4stner', [[u'Christian August Hausen', [[u'Johann Christoph Wichmannshausen', [[u'Otto Mencke', [[u'Jakob Thomasius', [[]]]]]]]]]]]]]]]]]]], [u'Rudolf Lipschitz', [[u'Johann Peter Gustav Lejeune Dirichlet', [[u'Sim\xe9on Denis Poisson', [[u'Joseph Louis Lagrange', [[u'Leonhard Euler', [[u'Johann Bernoulli', [u'Jacob Bernoulli']]]]]], [u'Pierre-Simon Laplace', [u"Jean le Rond d'Alembert", []]]]], [u'Joseph Fourier', [[u'Joseph Louis Lagrange', [[u'Leonhard Euler', [[u'Johann Bernoulli', [u'Jacob Bernoulli']]]]]]]]]], []]]]], [u'Carl Neumann', [[], []]]]]]]]]
"""