from os.path import isfile, isdir
from shutil import rmtree
-from os import mkdir, remove
+from os import mkdir, remove, utime
import pickle
import md5
import feedparser
import urllib2
from BeautifulSoup import BeautifulSoup
from urlparse import urljoin
+import logging
+logger = logging.getLogger(__name__)
#CONFIGDIR="/home/user/.feedingit/"
class ImageHandler:
def __init__(self, configdir):
- self.configdir = configdir
- self.images = {}
-
- def addImage(self, key, baseurl, url):
- filename = self.configdir+key+".d/"+getId(url)
+ pass
+
+class Feed:
+ def __init__(self, uniqueId, name, url):
+ self.titles = []
+ self.entries = {}
+ self.ids = []
+ self.readItems = {}
+ self.name = name
+ self.url = url
+ self.countUnread = 0
+ self.updateTime = "Never"
+ self.updateStamp = 0
+ self.uniqueId = uniqueId
+ self.etag = None
+ self.modified = None
+
+ def addImage(self, configdir, key, baseurl, url):
+ filename = configdir+key+".d/"+getId(url)
if not isfile(filename):
try:
#if url.startswith("http"):
f.close()
outf.close()
except:
- print "Could not download " + url
+ logger.error("Could not download " + url)
else:
- open(filename,"a").close() # "Touch" the file
- if filename in self.images:
- self.images[filename] += 1
- else:
- self.images[filename] = 1
+ #open(filename,"a").close() # "Touch" the file
+ file = open(filename,"a")
+ utime(filename, None)
+ file.close()
return filename
-
- def removeImage(self, key, filename):
- #filename = self.configdir+key+".d/"+getId(url)
- try:
- self.images[filename] -= 1
- except:
- self.images[filename] = 0 #Delete image
- try:
- if self.images[filename] == 0:
- remove(filename) #os.remove
- del self.images[filename]
- except:
- print "Could not remove image %s" % filename
-
-class Feed:
- def __init__(self, uniqueId, name, url, imageHandler):
- self.titles = []
- self.entries = {}
- self.ids = []
- self.readItems = {}
- self.name = name
- self.url = url
- self.countUnread = 0
- self.updateTime = "Never"
- self.uniqueId = uniqueId
- self.imageHandler = imageHandler
def editFeed(self, url):
self.url = url
def updateFeed(self, configdir, expiryTime=24, proxy=None, imageCache=False):
# Expiry time is in hours
if proxy == None:
- tmp=feedparser.parse(self.url)
+ tmp=feedparser.parse(self.url, etag = self.etag, modified = self.modified)
else:
- tmp=feedparser.parse(self.url, handlers = [proxy])
+ tmp=feedparser.parse(self.url, etag = self.etag, modified = self.modified, handlers = [proxy])
expiry = float(expiryTime) * 3600.
+
# Check if the parse was succesful (number of entries > 0, else do nothing)
if len(tmp["entries"])>0:
- #reversedEntries = self.getEntries()
- #reversedEntries.reverse()
+ # The etag and modified value should only be updated if the content was not null
+ try:
+ self.etag = tmp["etag"]
+ except KeyError:
+ self.etag = None
+ try:
+ self.modified = tmp["modified"]
+ except KeyError:
+ self.modified = None
+ #if len(tmp["entries"])>0:
if not isdir(configdir+self.uniqueId+".d"):
mkdir(configdir+self.uniqueId+".d")
+ try:
+ f = urllib2.urlopen(urljoin(tmp["feed"]["link"],"/favicon.ico"))
+ data = f.read()
+ f.close()
+ outf = open(configdir+self.uniqueId+".d/favicon.ico", "w")
+ outf.write(data)
+ outf.close()
+ del data
+ except:
+ #import traceback
+ #traceback.print_exc()
+ pass
+
+
+ #reversedEntries = self.getEntries()
+ #reversedEntries.reverse()
+
currentTime = time.time()
tmpEntries = {}
tmpIds = []
if imageCache:
for img in images:
try:
- filename = self.imageHandler.addImage(self.uniqueId, baseurl, img['src'])
+ filename = self.addImage(configdir, self.uniqueId, baseurl, img['src'])
img['src']=filename
tmpEntry["images"].append(filename)
except:
- print "Error downloading image %s" % img
+ logger.error("Error downloading image %s" % img)
tmpEntry["contentLink"] = configdir+self.uniqueId+".d/"+id+".html"
file = open(tmpEntry["contentLink"], "w")
file.write(soup.prettify())
if id not in self.readItems:
self.readItems[id] = False
else:
- tmpEntries[id] = self.entries[id]
- tmpIds.append(id)
+ try:
+ filename = configdir+self.uniqueId+".d/"+id+".html"
+ file = open(filename,"a")
+ utime(filename, None)
+ file.close()
+ for image in self.entries[id]["images"]:
+ file = open(image,"a")
+ utime(image, None)
+ file.close()
+ except:
+ pass
+ tmpEntries[id] = self.entries[id]
+ tmpIds.append(id)
oldIds = self.ids[:]
for entryId in oldIds:
tmpEntries[entryId] = self.entries[entryId]
tmpIds.append(entryId)
except:
- print "Error purging old articles %s" % entryId
+ logger.error("Error purging old articles %s" % entryId)
self.removeEntry(entryId)
self.entries = tmpEntries
del tmp
self.countUnread = tmpUnread
self.updateTime = time.asctime()
+ self.updateStamp = currentTime
self.saveFeed(configdir)
+ from glob import glob
+ from os import stat
+ for file in glob(configdir+self.uniqueId+".d/*"):
+ #
+ stats = stat(file)
+ #
+ # put the two dates into matching format
+ #
+ lastmodDate = stats[8]
+ #
+ expDate = time.time()-expiry*3
+ # check if image-last-modified-date is outdated
+ #
+ if expDate > lastmodDate:
+ #
+ try:
+ #
+ #print 'Removing', file
+ #
+ remove(file) # commented out for testing
+ #
+ except OSError:
+ #
+ logger.error('Could not remove %s' % file)
+
def extractContent(self, entry):
content = ""
self.readItems[id] = False
def isEntryRead(self, id):
- return self.readItems[id]
+ # Check if an entry is read; return False if the read
+ # status of an entry is unknown (id not in readItems)
+ return self.readItems.get(id, False)
def getTitle(self, id):
return self.entries[id]["title"]
def getUpdateTime(self):
return self.updateTime
+ def getUpdateStamp(self):
+ try:
+ return self.updateStamp
+ except:
+ self.updateStamp = 0
+ return self.updateStamp
+
def getEntries(self):
return self.entries
return self.entries[id]
except:
return []
+
+ def getImages(self, id):
+ return self.entries[id]["images"]
def getContent(self, id):
if self.entries[id].has_key("contentLink"):
#try:
if self.entries.has_key(id):
entry = self.entries[id]
- if entry.has_key("images"):
- for img in entry["images"]:
- self.imageHandler.removeImage(self.uniqueId, img)
if entry.has_key("contentLink"):
try:
remove(entry["contentLink"]) #os.remove
except:
- print "File not found for deletion: %s" % entry["contentLink"]
+ logger.error("File not found for deletion: %s"
+ % entry["contentLink"])
del self.entries[id]
else:
- print "Entries has no %s key" % id
+ logger.error("Entries has no %s key" % id)
if id in self.ids:
self.ids.remove(id)
else:
- print "Ids has no %s key" % id
+ logger.error("Ids has no %s key" % id)
if self.readItems.has_key(id):
if self.readItems[id]==False:
self.countUnread = self.countUnread - 1
del self.readItems[id]
else:
- print "ReadItems has no %s key" % id
+ logger.error("ReadItems has no %s key" % id)
#except:
# print "Error removing entry %s" %id
images = soup('img')
baseurl = entry["link"]
for img in images:
- filename = self.imageHandler.addImage(self.uniqueId, baseurl, img['src'])
+ filename = self.addImage(configdir, self.uniqueId, baseurl, img['src'])
img['src']=filename
- entry["images"].append(filename)
entry["contentLink"] = configdir+self.uniqueId+".d/"+id+".html"
file = open(entry["contentLink"], "w")
file.write(soup.prettify())
# if currentTime - entry["time"] > 2*expiry:
# self.removeEntry(id)
self.updateTime = time.asctime()
+ self.updateStamp = time.time()
self.saveFeed(configdir)
def purgeReadArticles(self):
self.listOfFeeds = pickle.load(file)
file.close()
else:
- self.listOfFeeds = {getId("Slashdot"):{"title":"Slashdot", "url":"http://rss.slashdot.org/Slashdot/slashdot", "unread":0, "updateTime":"Never"}, }
- try:
- file = open(self.configdir+"images.pickle")
- self.imageHandler = pickle.load(file)
- file.close()
- except:
- self.imageHandler = ImageHandler(self.configdir)
+ self.listOfFeeds = {getId("Maemo News"):{"title":"Maemo News", "url":"http://maemo.org/news/items.xml", "unread":0, "updateTime":"Never"}, }
if self.listOfFeeds.has_key("font"):
del self.listOfFeeds["font"]
if self.listOfFeeds.has_key("feedingit-order"):
file.close()
try:
feed.uniqueId
- feed.imageHandler
except AttributeError:
feed.uniqueId = getId(feed.name)
- feed.imageHandler = self.imageHandler
+ try:
+ del feed.imageHandler
+ except:
+ pass
+ try:
+ feed.etag
+ except AttributeError:
+ feed.etag = None
+ try:
+ feed.modified
+ except AttributeError:
+ feed.modified = None
#feed.reloadUnread(self.configdir)
else:
#print key
title = self.listOfFeeds[key]["title"]
url = self.listOfFeeds[key]["url"]
if key == "ArchivedArticles":
- feed = ArchivedArticles("ArchivedArticles", title, url, self.imageHandler)
+ feed = ArchivedArticles("ArchivedArticles", title, url)
else:
- feed = Feed(getId(title), title, url, self.imageHandler)
+ feed = Feed(getId(title), title, url)
return feed
def updateFeeds(self, expiryTime=24, proxy=None, imageCache=False):
feed.updateFeed(self.configdir, expiryTime, proxy, imageCache)
self.listOfFeeds[key]["unread"] = feed.getNumberOfUnreadItems()
self.listOfFeeds[key]["updateTime"] = feed.getUpdateTime()
+ self.listOfFeeds[key]["updateStamp"] = feed.getUpdateStamp()
def updateFeed(self, key, expiryTime=24, proxy=None, imageCache=False):
feed = self.getFeed(key)
feed.updateFeed(self.configdir, expiryTime, proxy, imageCache)
self.listOfFeeds[key]["unread"] = feed.getNumberOfUnreadItems()
self.listOfFeeds[key]["updateTime"] = feed.getUpdateTime()
+ self.listOfFeeds[key]["updateStamp"] = feed.getUpdateStamp()
def editFeed(self, key, title, url):
self.listOfFeeds[key]["title"] = title
feed.reloadUnread(self.configdir)
except:
# If the feed file gets corrupted, we need to reset the feed.
+ import traceback
+ logger.error("getFeed: %s" % traceback.format_exc())
import dbus
bus = dbus.SessionBus()
remote_object = bus.get_object("org.freedesktop.Notifications", # Connection name
self.listOfFeeds[key]["updateTime"] = "Never"
return self.listOfFeeds[key]["updateTime"]
+ def getFeedUpdateStamp(self, key):
+ #print self.listOfFeeds.has_key(key)
+ if not self.listOfFeeds[key].has_key("updateStamp"):
+ self.listOfFeeds[key]["updateStamp"] = 0
+ return self.listOfFeeds[key]["updateStamp"]
+
def getFeedNumberOfUnreadItems(self, key):
if not self.listOfFeeds[key].has_key("unread"):
self.listOfFeeds[key]["unread"] = 0
def getListOfFeeds(self):
return self.sortedKeys
+ def getFavicon(self, key):
+ filename = self.configdir+key+".d/favicon.ico"
+ if isfile(filename):
+ return filename
+ else:
+ return False
+
def addFeed(self, title, url):
if not self.listOfFeeds.has_key(getId(title)):
self.listOfFeeds[getId(title)] = {"title":title, "url":url, "unread":0, "updateTime":"Never"}
file = open(self.configdir+"feeds.pickle", "w")
pickle.dump(self.listOfFeeds, file)
file.close()
- file = open(self.configdir+"images.pickle", "w")
- pickle.dump(self.imageHandler, file)
- file.close()
def moveUp(self, key):
index = self.sortedKeys.index(key)