mirror of
https://github.com/moparisthebest/SickRage
synced 2024-11-11 03:45:01 -05:00
0d9fbc1ad7
This version of SickBeard uses both TVDB and TVRage to search and gather it's series data from allowing you to now have access to and download shows that you couldn't before because of being locked into only what TheTVDB had to offer. Also this edition is based off the code we used in our XEM editon so it does come with scene numbering support as well as all the other features our XEM edition has to offer. Please before using this with your existing database (sickbeard.db) please make a backup copy of it and delete any other database files such as cache.db and failed.db if present, we HIGHLY recommend starting out with no database files at all to make this a fresh start but the choice is at your own risk! Enjoy!
97 lines
4.4 KiB
Python
97 lines
4.4 KiB
Python
"""A collection of modules for building different kinds of tree from
|
|
HTML documents.
|
|
|
|
To create a treebuilder for a new type of tree, you need to do
|
|
implement several things:
|
|
|
|
1) A set of classes for various types of elements: Document, Doctype,
|
|
Comment, Element. These must implement the interface of
|
|
_base.treebuilders.Node (although comment nodes have a different
|
|
signature for their constructor, see treebuilders.simpletree.Comment)
|
|
Textual content may also be implemented as another node type, or not, as
|
|
your tree implementation requires.
|
|
|
|
2) A treebuilder object (called TreeBuilder by convention) that
|
|
inherits from treebuilders._base.TreeBuilder. This has 4 required attributes:
|
|
documentClass - the class to use for the bottommost node of a document
|
|
elementClass - the class to use for HTML Elements
|
|
commentClass - the class to use for comments
|
|
doctypeClass - the class to use for doctypes
|
|
It also has one required method:
|
|
getDocument - Returns the root node of the complete document tree
|
|
|
|
3) If you wish to run the unit tests, you must also create a
|
|
testSerializer method on your treebuilder which accepts a node and
|
|
returns a string containing Node and its children serialized according
|
|
to the format used in the unittests
|
|
|
|
The supplied simpletree module provides a python-only implementation
|
|
of a full treebuilder and is a useful reference for the semantics of
|
|
the various methods.
|
|
"""
|
|
|
|
treeBuilderCache = {}
|
|
|
|
import sys
|
|
|
|
def getTreeBuilder(treeType, implementation=None, **kwargs):
|
|
"""Get a TreeBuilder class for various types of tree with built-in support
|
|
|
|
treeType - the name of the tree type required (case-insensitive). Supported
|
|
values are "simpletree", "dom", "etree" and "beautifulsoup"
|
|
|
|
"simpletree" - a built-in DOM-ish tree type with support for some
|
|
more pythonic idioms.
|
|
"dom" - A generic builder for DOM implementations, defaulting to
|
|
a xml.dom.minidom based implementation for the sake of
|
|
backwards compatibility (as releases up until 0.10 had a
|
|
builder called "dom" that was a minidom implemenation).
|
|
"etree" - A generic builder for tree implementations exposing an
|
|
elementtree-like interface (known to work with
|
|
ElementTree, cElementTree and lxml.etree).
|
|
"beautifulsoup" - Beautiful soup (if installed)
|
|
|
|
implementation - (Currently applies to the "etree" and "dom" tree types). A
|
|
module implementing the tree type e.g.
|
|
xml.etree.ElementTree or lxml.etree."""
|
|
|
|
treeType = treeType.lower()
|
|
if treeType not in treeBuilderCache:
|
|
if treeType == "dom":
|
|
import dom
|
|
# XXX: Keep backwards compatibility by using minidom if no implementation is given
|
|
if implementation == None:
|
|
from xml.dom import minidom
|
|
implementation = minidom
|
|
# XXX: NEVER cache here, caching is done in the dom submodule
|
|
return dom.getDomModule(implementation, **kwargs).TreeBuilder
|
|
elif treeType == "simpletree":
|
|
import simpletree
|
|
treeBuilderCache[treeType] = simpletree.TreeBuilder
|
|
elif treeType == "beautifulsoup":
|
|
import soup
|
|
treeBuilderCache[treeType] = soup.TreeBuilder
|
|
elif treeType == "lxml":
|
|
import etree_lxml
|
|
treeBuilderCache[treeType] = etree_lxml.TreeBuilder
|
|
elif treeType == "etree":
|
|
# Come up with a sane default
|
|
if implementation == None:
|
|
try:
|
|
import xml.etree.cElementTree as ET
|
|
except ImportError:
|
|
try:
|
|
import xml.etree.ElementTree as ET
|
|
except ImportError:
|
|
try:
|
|
import cElementTree as ET
|
|
except ImportError:
|
|
import elementtree.ElementTree as ET
|
|
implementation = ET
|
|
import etree
|
|
# NEVER cache here, caching is done in the etree submodule
|
|
return etree.getETreeModule(implementation, **kwargs).TreeBuilder
|
|
else:
|
|
raise ValueError("""Unrecognised treebuilder "%s" """%treeType)
|
|
return treeBuilderCache.get(treeType)
|