from BeautifulSoup import BeautifulSoup
import re
import urllib2

HACKAGE_PKG_PREFIX = 'http://hackage.haskell.org/package/'
PKGDB_PREFIX= 'https://apps.fedoraproject.org/packages/'
ERROR_URL = PKGDB_PREFIX + 'error'

def find_deps(pkg, knowns=set(), founds=set()):
    print "DEBUG, checking", pkg
    print "DEBUG", knowns, founds
    if pkg in knowns:
        print "package is a known dependency"
        return (knowns, founds)
    elif pkg in founds:
        print "package already exists"
        return (knowns, founds)
    # else check the package
    res = urllib2.urlopen(PKGDB_PREFIX + 'ghc-' + pkg)
    if res.url != ERROR_URL:
        return (knowns, founds.union(set([pkg])))
    knowns.add(pkg)
    # package not found, check its deps
    page = urllib2.urlopen(HACKAGE_PKG_PREFIX + pkg)
    soup = BeautifulSoup(page)
    deps = [d.text
            for d in soup.find('th', text='Dependencies').next()
            if d.name == 'a']
    print "skipping known deps:", set(deps).intersection(knowns)
    print "... and found deps:", set(deps).intersection(founds)
    unknowns = set(deps).difference(knowns).difference(founds)
    print "unknown deps for", pkg, ":", unknowns

    # recursively check their dependencies
    for u in unknowns:
        if u in knowns or u in founds:
            # check again since after each serial recursion,
            # p might be added to a set already
            print "skipping", u
            continue
        (knowns1, founds1) = find_deps(u, knowns, founds)
        knowns = knowns.union(knowns1)
        founds = founds.union(founds1)
    return (knowns, founds)