Question

Following an advice in the answer: subclassing beautifulsoup html parser, getting type error, I’m trying to use class composition instead of subclassing BeautifulSoup.

The basic Scraper class works fine on it’s own (at least to my limited testing).

The Scraper class:

from BeautifulSoup import BeautifulSoup
import urllib2

class Scrape():
    """base class to be subclassed
    basically a  wrapper that providers basic url fetching with urllib2
    and the basic html parsing with beautifulsoupץ
    some useful methods are provided with class composition with BeautifulSoup.
    for direct access to the soup class you can use the _soup property."""

    def __init__(self,file):
        self._file = file
        #very basic input validation
        #import re

        #import urllib2
        #from BeautifulSoup import BeautifulSoup
        try:
            self._page = urllib2.urlopen(self._file) #fetching the page
        except (urllib2.URLError):
            print ('please enter a valid url starting with http/https/ftp/file')

        self._soup = BeautifulSoup(self._page) #calling the html parser

        #BeautifulSoup.__init__(self,self._page)
        # the next part is the class compostion part - we transform attribute and method calls to the BeautifulSoup class
        #search functions:
        self.find = self._soup.find
        self.findAll = self._soup.findAll

        self.__iter__ = self._soup.__iter__ #enables iterating,looping in the object

        self.__len__ = self._soup.__len__
        self.__contains__ = self._soup.__contains__
        #attribute fetching and setting - __getattr__ implented by the scraper class
        self.__setattr__ = self._soup.__setattr__
        self.__getattribute__ = self._soup.__getattribute__

        #Called to implement evaluation of self[key]
        self.__getitem__ = self._soup.__getitem__
        self.__setitem__ = self._soup.__setitem__
        self.__delitem__ = self._soup.__delitem__

        self.__call__ = self._soup.__call__#Called when the instance is “called” as a function

        self._getAttrMap = self._soup._getAttrMap
        self.has_key = self._soup.has_key

        #walking the html document methods
        self.contents = self._soup.contents
        self.text = self._soup.text
        self.extract = self._soup.extract
        self.next = self._soup.next
        self.parent = self._soup.parent
        self.fetch = self._soup.fetch
        self.fetchText = self._soup.fetchText
        self.findAllNext = self._soup.findAllNext
        self.findChild = self._soup.findChild
        self.findChildren = self._soup.findChildren
        self.findNext = self._soup.findNext
        self.findNextSibling = self._soup.findNextSibling
        self.first = self._soup.first
        self.name = self._soup.name
        self.get = self._soup.get
        self.getString = self._soup.getString


        # comparison operators or similiar boolean checks
        self.__eq__ = self._soup.__eq__
        self.__ne__ = self._soup.__ne__
        self.__hash__ = self._soup.__hash__
        self.__nonezero__ = self._soup.__nonzero__ #not sure



        # the class represntation magic methods:
        self.__str__ = self._soup.__str__
        self.__repr__ =self._soup.__repr__
        #self.__dict__ = self._soup.__dict__


    def __getattr__(self,method):
        """basically this 'magic' method transforms calls for unknown attributes to
        and enables to traverse the html document with the .notation.
        for example - using instancename.div will return the first div.
        explantion: python calls __getattr__ if It didn't find any method or attribute correspanding to the call.
        I'm not sure this is a good or the right use for the method """

        return self._soup.find(method)

    def clean(self,work=False,element=False):
        """clean method that provides:basic cleaning of head,scripts etc
        input 'work' soup object to clean from unneccesary parts:scripts,head,style
        has optional variable:'element' that can get a tuple of element
        that enables to override what element to clean"""
        self._work = work or self._soup
        self._cleanelements=element or ("head","style","script")

        #for elem in self._work.findAll(self._cleanelements):
        for elem in self.findAll(self._cleanelements):
            elem.extract()

But when I subclass it I get some sort of recursion loop, I just can figure.

Here is the subclass (the relevant parts):

class MainTraffic(Scrape):
    """class traffic - subclasses the Scrape class
    inputs a page url and a category"""

    def __init__(self, file, cat, caller = False):
        if not caller:
            self._file = file
            #import urllib2
            #self._request = urllib2.Request(self._file)# request to post the show all questions
            Scrape.__init__(self,self._file)
            self.pagecat = cat
            self.clean(self)
            self.cleansoup = self.cleantotable(self)
            self.fetchlinks(self.cleansoup)
            #self.populatequestiondic()
            #del (self.cleansoup)

    def cleantotable(self):
        pass

    def fetchlinks(self,fetch):
        pass

    def length(self):
        from sqlalchemy import func
        self.len = session.query(func.count(Question.id)).scalar()
        return int(self.len)

    def __len__(self):
        return self.length()

    def __repr__(self):
        self.repr = "traffic theory question, current number of questions:{0}".format(self.length())
        return self.repr

    def  __getitem__(self,key):
        try:
            self._item = session.query(Question).filter_by(question_num=key).first()
            return self._item
        except (IndexError, KeyError):
            print "no such key:{0}".format(key)

and here is the error message:

File "C:\Python27\learn\traffic.py", line 117, in __init__
    Scrape.__init__(self,self._file)
  File "C:\Python27\learn\traffic.py", line 26, in __init__
    self._soup = BeautifulSoup(self._page) #calling the html parser
  File "C:\Python27\learn\traffic.py", line 92, in __getattr__
    return self._soup.find(method)
  File "C:\Python27\learn\traffic.py", line 92, in __getattr__
    return self._soup.find(method)
  File "C:\Python27\learn\traffic.py", line 92, in __getattr__
    return self._soup.find(method)
RuntimeError: maximum recursion depth exceeded

I suspect the problem is with me misusing the __getattr__, but I couldn’t figure out what should I change.

Answer 1

Part 1

Your code doesn’t work because __getattr__() accesses self._soup before it has been initialized. This happens due to four innocuous-looking lines:

try:
  self._page = urllib2.urlopen(self._file)
except (urllib2.URLError):
  print ('please enter a valid url starting with http/https/ftp/file') 

Why do you catch the exception and not actually handle it?

The next line accesses self._page, which has not been set yet if urlopen() threw an exception:

self._soup = BeautifulSoup(self._page)

Since it hasn’t been set, accessing it calls __getattr__(), which accesses self._soup, which has not been set yet so it accesses __getattr__.

The easiest “fix” is to special-case _soup to prevent infinite recursion. Additionally, it seems to make more sense for __getattr__ to simply do normal attribute lookup on soup:

def __getattr__(self,attr):
  if attr == "_soup":
    raise AttributeError()
  return getattr(self._soup,attr)

Part 2

Copying all the methods over is unlikely to work very well, and seems to miss the point of class composition entirely.