added admin functions (remove spam content)

use efficient sql SELECT for frontpage poll list
added link counter for spam detection
improve efficiency of post deletion by using sqlobject's "cascade" attribute

wortschlucker/src/wortschlucker.py

@@ -43,7 +43,6 @@ loader = genshi.template.TemplateLoader(os.path.join(BASE_DIR, 'templates'), aut
 
 BLOG_DIR = os.path.join(BASE_DIR, "blog")
 
-
 BASE_DICT = {
     "base_url": "/",    # the trailing slash is necessary
     "show_navbar": True,
@@ -51,6 +50,7 @@ BASE_DICT = {
     "authlevel": "public",	#authentiction level of template: one of "public" (e.g. frontpage), "poll_public" (e.g. poll_details),"poll_admin" (poll_admin_details","admin" (admin interface)
 }
 
+
 # used as the default setting for expose/close dates
 DEFAULT_DAYS_AHEAD = 7
 DATE_FORMAT = "%d.%m.%Y"
@@ -85,7 +85,7 @@ POLL_SETTING_TEMPLATES = {
 class ContentSubmission(sqlobject.SQLObject):
     submitter = sqlobject.UnicodeCol()
     content = sqlobject.UnicodeCol()
-    poll_id = sqlobject.ForeignKey("Poll")
+    poll_id = sqlobject.ForeignKey("Poll", cascade=True)
     timestamp_creation = sqlobject.DateTimeCol()
 
     def get_creation_time_string(self):
@@ -133,20 +133,20 @@ class Profile(sqlobject.SQLObject):
 
 
 class ProfilePolls(sqlobject.SQLObject):
-    user = sqlobject.ForeignKey("Profile")
+    user = sqlobject.ForeignKey("Profile", cascade=True)
-    poll = sqlobject.ForeignKey("Poll")
+    poll = sqlobject.ForeignKey("Poll", cascade=True)
     is_admin = sqlobject.BoolCol()
 
 
 class PollSetting(sqlobject.SQLObject):
-    poll_id = sqlobject.ForeignKey("Poll")
+    poll_id = sqlobject.ForeignKey("Poll", cascade=True)
     key = sqlobject.UnicodeCol()
     value = sqlobject.UnicodeCol()
 
 
 class PollRelation(sqlobject.SQLObject):
-    first = sqlobject.ForeignKey("Poll")
+    first = sqlobject.ForeignKey("Poll", cascade=True)
-    second = sqlobject.ForeignKey("Poll")
+    second = sqlobject.ForeignKey("Poll", cascade=True)
 
 
 class Poll(sqlobject.SQLObject):
@@ -252,15 +252,6 @@ class Poll(sqlobject.SQLObject):
     def get_submissions(self):
         return ContentSubmission.selectBy(poll_id=self.id).orderBy("timestamp_creation")
 
-    def delete_poll(self):
-        submissions = ContentSubmission.selectBy(poll_id=self.id)
-        settings = PollSetting.selectBy(poll_id=self.id)
-        for submission in submissions:
-            submission.destroySelf()
-        for setting in settings:
-            setting.destroySelf()
-        self.destroySelf()
-
     def get_url(self, absolute=False):
         return get_url_string("%s%s" % (BASE_DICT["base_url"], self.hash_key), absolute)
 
@@ -589,12 +580,15 @@ def is_spam_submitter_name(name, errors_dict):
     else:
         return False
 
-def is_spam_content(text, errors_dict):
+def count_urls(text, errors_dict):
+    hits = re.findall(r"(\swww\.|http://|https://)", text)
+    return len(hits)
+
+def check_spam_content(text):
     if re.search(r"(<a\s|\shref=|</a>)", text.lower()):
-        errors_dict["content"] = \
-                "Spam-Verdacht: Inhalt darf keine HTML-Tags enthalten"
         return True
-    else:
+    if count_urls(text) > 2:
+        return True
     return False
 
 @bobo.query('/profile/logout')
@@ -866,7 +860,11 @@ def submit_content(bobo_request, hash_key=None, submitter=None, content=None):
     value_dict["errors"] = {}
     value_dict["authlevel"] = "poll_public"
     data = {}
-    if content and (not is_spam_content(content, value_dict["errors"])):
+    if content and check_spam_content(content):
+        value_dict["errors"]["content"] = \
+                "Spam-Verdacht: Inhalt darf keine HTML-Tags und nicht " + \
+                "zuviele Links enthalten"
+    else:
         data["content"] = content
     if submitter and (not is_spam_submitter_name(submitter, value_dict["errors"])):
         data["submitter"] = submitter
@@ -899,7 +897,7 @@ def delete_poll(bobo_request, admin_hash_key=None):
     admin_poll_id = get_poll_admin_id(admin_hash_key)
     if not admin_poll_id is None:
         poll = Poll.get(admin_poll_id)
-        poll.delete_poll()
+        poll.destroySelf()
     return bobo.redirect(BASE_DICT["base_url"])
 
 @bobo.query('/:admin_hash_key/vote_enable')
@@ -1111,15 +1109,21 @@ def serve_blog(bobo_request, blog_id=None):
     value_dict["blog_list"] = blog_list
     return render("blog_list.html", **value_dict)
 
-def show_poll_list(bobo_request, render_file, page_size, page=None, filter_private=True):
+def show_poll_list(bobo_request, render_file, page_size, page=None,
-    value_dict = get_default_values(bobo_request)
+        filter_private=True, value_dict=None):
-    polls = Poll.select().orderBy("-timestamp_creation")
+    if value_dict is None:
-    # TODO: speed the filtering up by using SQL statements (see sqlobject "filter")
+        value_dict = {}
+    value_dict.update(get_default_values(bobo_request))
+    true_string = get_poll_setting_string("public", True)
     if filter_private:
-        polls = [poll for poll in polls if poll.get_settings()["public"]]
+        polls = Poll.select(sqlobject.sqlbuilder.AND(
+                PollSetting.q.key == "public",
+                PollSetting.q.value == true_string,
+                PollSetting.q.poll_id == Poll.q.id))
     else:
-        # convert the sql query into a list (probably an expensive operation)
+        polls = Poll.select()
-        polls = [poll for poll in polls]
+    polls = polls.orderBy("-timestamp_creation")
+    poll_count = polls.count()
     if page is None:
         page = 1
     else:
@@ -1130,15 +1134,16 @@ def show_poll_list(bobo_request, render_file, page_size, page=None, filter_priva
     # "page" should at least be 1 - zero shows an empty list
     page = max(1, page)
     start = (page - 1) * page_size
-    if start >= len(polls):
+    if start >= poll_count:
-        start = len(polls) - page_size
+        start = poll_count - page_size
-        page = (len(polls) + page_size - 1) / page_size
+        page = (poll_count + page_size - 1) / page_size
     end = start + page_size - 1
     value_dict["polls"] = polls[start : end + 1]
     # show a link for the next page, if more polls are available
-    value_dict["show_next_link"] = (end + 1 < len(polls))
+    value_dict["show_next_link"] = (end + 1 < poll_count)
     value_dict["show_previous_link"] = (start > 0)
     value_dict["page"] = page
+    value_dict["page_size"] = page_size
     return render(render_file, **value_dict)
 
 def render_poll_admin(bobo_request, poll, add_related, del_related, count, page, show_delete):
@@ -1176,22 +1181,47 @@ def render_poll_admin(bobo_request, poll, add_related, del_related, count, page,
     value_dict["authlevel"] = "poll_admin"
     return render("poll_admin_details.html", **value_dict)
 
+@bobo.query('/admin/maintenance')
+def admin_maintenance(age_days=60, keyword="viagra", keyword_submission="viagra", method=None):
+    if method == "Remove old":
+        age_days = int(age_days)
+        now = datetime.datetime.now()
+        oldest = now - datetime.timedelta(days=age_days)
+        for poll in Poll.select(Poll.q.timestamp_creation < oldest):
+            if poll.get_num_of_submissions() == 0:
+                poll.destroySelf()
+    elif method == "Remove poll by keyword" and keyword:
+        keyword = keyword.lower()
+        for poll in Poll.select():
+            if (keyword in poll.title.lower()) or (keyword in poll.description.lower()):
+                poll.destroySelf()
+    elif method == "Remove submission by keyword" and keyword:
+        keyword = keyword_submission.lower()
+        for submission in ContentSubmission.select():
+            if keyword in submission.content.lower():
+                submission.destroySelf()
+    return bobo.redirect("../admin")
+
 @bobo.query('/admin')
 @bobo.query('/admin/')
 @bobo.query('/admin/page/:page')
-def show_admin_page(bobo_request, page=None, page_size=20):
+def show_admin_page(bobo_request, page=None, page_size=50):
     try:
         page_size = int(page_size)
     except ValueError:
-        page_size = 30
+        page_size = 50
-    return show_poll_list(bobo_request, "admin.html", page_size, page, filter_private=False)
+    value_dict = {}
+    value_dict["poll_table"] = Poll
+    value_dict["submission_table"] = ContentSubmission
+    return show_poll_list(bobo_request, "admin.html", page_size, page=page,
+            filter_private=False, value_dict=value_dict)
 
 @bobo.query('/')
 @bobo.query('/public')
 @bobo.query('/public/')
 @bobo.query('/public/page/:page')
 def show_frontpage(bobo_request, page=None):
-    return show_poll_list(bobo_request, "frontpage.html", 20, page)
+    return show_poll_list(bobo_request, "frontpage.html", 20, page=page)
 
 @bobo.query('')
 def base():