nemunaire/nemubot
1
0
Fork 0
Code Releases Activity

tools/feed: hardened parser

Browse Source
This commit is contained in:
nemunaire 2015-10-10 23:11:50 +02:00
parent 04dcf07fb2
commit 7102e08000
1 changed files with 30 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
nemubot/tools/feed.py
View File

@ -11,11 +11,16 @@ from xml.dom.minidom import getDOMImplementation
class AtomEntry: class AtomEntry:
def __init__(self, node): def __init__(self, node):
if len(node.getElementsByTagName("id")) > 0 and node.getElementsByTagName("id")[0].firstChild is not None:
self.id = node.getElementsByTagName("id")[0].firstChild.nodeValue self.id = node.getElementsByTagName("id")[0].firstChild.nodeValue
if node.getElementsByTagName("title")[0].firstChild is not None: else:
self.id = None
if len(node.getElementsByTagName("title")) > 0 and node.getElementsByTagName("title")[0].firstChild is not None:
self.title = node.getElementsByTagName("title")[0].firstChild.nodeValue self.title = node.getElementsByTagName("title")[0].firstChild.nodeValue
else: else:
self.title = "" self.title = ""
try: try:
self.updated = time.strptime(node.getElementsByTagName("updated")[0].firstChild.nodeValue[:19], "%Y-%m-%dT%H:%M:%S") self.updated = time.strptime(node.getElementsByTagName("updated")[0].firstChild.nodeValue[:19], "%Y-%m-%dT%H:%M:%S")
except: except:
@ -25,26 +30,32 @@ class AtomEntry:
print(node.getElementsByTagName("updated")[0].firstChild.nodeValue[:10]) print(node.getElementsByTagName("updated")[0].firstChild.nodeValue[:10])
self.updated = time.localtime() self.updated = time.localtime()
self.updated = datetime.datetime(*self.updated[:6]) self.updated = datetime.datetime(*self.updated[:6])
if len(node.getElementsByTagName("summary")) > 0 and node.getElementsByTagName("summary")[0].firstChild is not None: if len(node.getElementsByTagName("summary")) > 0 and node.getElementsByTagName("summary")[0].firstChild is not None:
self.summary = node.getElementsByTagName("summary")[0].firstChild.nodeValue self.summary = node.getElementsByTagName("summary")[0].firstChild.nodeValue
else: else:
self.summary = None self.summary = None
if len(node.getElementsByTagName("link")) > 0:
if len(node.getElementsByTagName("link")) > 0 and node.getElementsByTagName("link")[0].hasAttribute("href"):
self.link = node.getElementsByTagName("link")[0].getAttribute("href") self.link = node.getElementsByTagName("link")[0].getAttribute("href")
else: else:
self.link = None self.link = None
if len(node.getElementsByTagName("category")) >= 1:
if len(node.getElementsByTagName("category")) >= 1 and node.getElementsByTagName("category")[0].hasAttribute("term"):
self.category = node.getElementsByTagName("category")[0].getAttribute("term") self.category = node.getElementsByTagName("category")[0].getAttribute("term")
else: else:
self.category = None self.category = None
if len(node.getElementsByTagName("link")) > 1:
if len(node.getElementsByTagName("link")) > 1 and node.getElementsByTagName("link")[1].hasAttribute("href"):
self.link2 = node.getElementsByTagName("link")[1].getAttribute("href") self.link2 = node.getElementsByTagName("link")[1].getAttribute("href")
else: else:
self.link2 = None self.link2 = None
def __repr__(self): def __repr__(self):
return "<AtomEntry title='%s' updated='%s'>" % (self.title, self.updated) return "<AtomEntry title='%s' updated='%s'>" % (self.title, self.updated)
def __cmp__(self, other): def __cmp__(self, other):
return not (self.id == other.id) return not (self.id == other.id)
@ -52,26 +63,36 @@ class AtomEntry:
class RSSEntry: class RSSEntry:
def __init__(self, node): def __init__(self, node):
if len(node.getElementsByTagName("guid")) > 0 and node.getElementsByTagName("guid")[0].firstChild is not None:
self.id = node.getElementsByTagName("guid")[0].firstChild.nodeValue self.id = node.getElementsByTagName("guid")[0].firstChild.nodeValue
if node.getElementsByTagName("title")[0].firstChild is not None: else:
self.id = None
if len(node.getElementsByTagName("title")) > 0 and node.getElementsByTagName("title")[0].firstChild is not None:
self.title = node.getElementsByTagName("title")[0].firstChild.nodeValue self.title = node.getElementsByTagName("title")[0].firstChild.nodeValue
else: else:
self.title = "" self.title = ""
if len(node.getElementsByTagName("pubDate")) > 0 and node.getElementsByTagName("pubDate")[0].firstChild is not None:
self.pubDate = node.getElementsByTagName("pubDate")[0].firstChild.nodeValue self.pubDate = node.getElementsByTagName("pubDate")[0].firstChild.nodeValue
else:
self.pubDate = ""
if len(node.getElementsByTagName("description")) > 0 and node.getElementsByTagName("description")[0].firstChild is not None: if len(node.getElementsByTagName("description")) > 0 and node.getElementsByTagName("description")[0].firstChild is not None:
self.summary = node.getElementsByTagName("description")[0].firstChild.nodeValue self.summary = node.getElementsByTagName("description")[0].firstChild.nodeValue
else: else:
self.summary = None self.summary = None
if len(node.getElementsByTagName("link")) > 0:
if len(node.getElementsByTagName("link")) > 0 and node.getElementsByTagName("link")[0].hasAttribute("href"):
self.link = node.getElementsByTagName("link")[0].getAttribute("href") self.link = node.getElementsByTagName("link")[0].getAttribute("href")
else: else:
self.link = None self.link = None
def __repr__(self): def __repr__(self):
return "<RSSEntry title='%s' updated='%s'>" % (self.title, self.pubDate) return "<RSSEntry title='%s' updated='%s'>" % (self.title, self.pubDate)
def __cmp__(self, other): def __cmp__(self, other):
return not (self.id == other.id) return not (self.id == other.id)