Replace lxml with Python's built in html.parser

[orenyomtov]

After my server has been running for some time, fbchat starts throwing the following exception:

Failed loading session
Traceback (most recent call last):
  File "/usr/local/lib/python2.7/dist-packages/fbchat/client.py", line 366, in setSession
    self._postLogin()
  File "/usr/local/lib/python2.7/dist-packages/fbchat/client.py", line 212, in _postLogin
    soup = bs(r.text, "lxml")
  File "/usr/lib/python2.7/dist-packages/bs4/__init__.py", line 215, in __init__
    self._feed()
  File "/usr/lib/python2.7/dist-packages/bs4/__init__.py", line 239, in _feed
    self.builder.feed(self.markup)
  File "/usr/lib/python2.7/dist-packages/bs4/builder/_lxml.py", line 240, in feed
    self.parser.feed(markup)
  File "src/lxml/parser.pxi", line 1194, in lxml.etree._FeedParser.feed (src/lxml/lxml.etree.c:119773)
  File "src/lxml/parser.pxi", line 1237, in lxml.etree._FeedParser.feed (src/lxml/lxml.etree.c:118685)
  File "src/lxml/parser.pxi", line 827, in lxml.etree._BaseParser._getPushParserContext (src/lxml/lxml.etree.c:113960)
  File "src/lxml/parser.pxi", line 844, in lxml.etree._BaseParser._createContext (src/lxml/lxml.etree.c:114186)
  File "src/lxml/parsertarget.pxi", line 110, in lxml.etree._TargetParserContext._setTarget (src/lxml/
  File "src/lxml/parsertarget.pxi", line 41, in lxml.etree._PythonSaxParserTarget.__cinit__ (src/lxml/
  File "/usr/lib/python2.7/inspect.py", line 814, in getargspec
    func = func.im_func
RuntimeError: restricted attribute

Restarting the server fixes the problem temporarily, but it returns after a while.

Researching the issue brought up this problem being reported several times and considered "The unpredictable BS behavior again."

The only remedy being offered is replacing lxml.

This pull request replaces lxml with Python's built in html.parser.

[madsmtm]

The main reason for using lxml is the speed benefit, but I went ahead and ran a few tests against data from the three Facebook urls we're be parsing with BeautifulSoup:

from requests import get
from timeit import timeit
from bs4 import BeautifulSoup as bs
n = 2000

d = get("https://m.facebook.com/").text
timeit('bs(d, "html.parser")', number=n, globals=globals())/n
# -> 0.0039508302819886015
timeit('bs(d, "lxml")', number=n, globals=globals())/n
# -> 0.0026858313914999597

d = get("https://www.facebook.com/").text
timeit('bs(d, "html.parser")', number=n, globals=globals())/n
# -> 0.037392912552997586
timeit('bs(d, "lxml")', number=n, globals=globals())/n
# -> 0.03604348213601043

d = get("https://m.facebook.com/login.php?login_attempt=1").text
timeit('bs(d, "html.parser")', number=n, globals=globals())/n
# -> 0.0038544781890086596
timeit('bs(d, "lxml")', number=n, globals=globals())/n
# -> 0.002652162682003109

The speed benefit is mostly negligible (compared to IO), and non-existent in the most time consuming case, so I'd argue this was a classic example of premature optimisation.
I've also seen a lot of people having trouble installing lxml in the past, see #101, #119, #113, #230 and #272.
I'll test it with TravisCI when I get it up and running again, and I'll make sure to include it in the next release. Thanks ;)