+import datetime
import os
import random
import re
import constants
import trigger
+
class chooser(object):
"""Base class of a thing that chooses pages"""
+
def get_page_list(self):
now = time.time()
valid_filename = re.compile("([^_]+)_(\d+)_([^\.]+)\.html")
filenames = []
- pages = [ f for f in os.listdir(constants.pages_dir)
- if os.path.isfile(os.path.join(constants.pages_dir, f))]
+ pages = [
+ f
+ for f in os.listdir(constants.pages_dir)
+ if os.path.isfile(os.path.join(constants.pages_dir, f))
+ ]
for page in pages:
result = re.match(valid_filename, page)
if result != None:
print(('chooser: candidate page: "%s"' % page))
- if (result.group(3) != "none"):
+ if result.group(3) != "none":
freshness_requirement = int(result.group(3))
- last_modified = int(os.path.getmtime(
- os.path.join(constants.pages_dir, page)))
- age = (now - last_modified)
- if (age > freshness_requirement):
+ last_modified = int(
+ os.path.getmtime(os.path.join(constants.pages_dir, page))
+ )
+ age = now - last_modified
+ if age > freshness_requirement:
print(('chooser: "%s" is too old.' % page))
continue
filenames.append(page)
def choose_next_page(self):
pass
+
class weighted_random_chooser(chooser):
- """Chooser that does it via weighted RNG"""
- def __init__(self):
+ """Chooser that does it via weighted RNG."""
+
+ def dont_choose_page_twice_in_a_row_filter(self, choice):
+ if choice == self.last_choice:
+ return False
+ self.last_choice = choice
+ return True
+
+ def __init__(self, filter_list):
self.last_choice = ""
self.valid_filename = re.compile("([^_]+)_(\d+)_([^\.]+)\.html")
self.pages = None
self.count = 0
+ self.filter_list = filter_list
+ if filter_list is None:
+ self.filter_list = []
+ self.filter_list.append(self.dont_choose_page_twice_in_a_row_filter)
def choose_next_page(self):
- if (self.pages == None or
- self.count % 100 == 0):
+ if self.pages == None or self.count % 100 == 0:
self.pages = self.get_page_list()
total_weight = 0
weight = int(result.group(2))
weights.append(weight)
total_weight += weight
-
- if (total_weight <= 0):
+ if total_weight <= 0:
raise error
while True:
- pick = random.randrange(0, total_weight - 1)
+ random_pick = random.randrange(0, total_weight - 1)
so_far = 0
for x in range(0, len(weights)):
so_far += weights[x]
- if (so_far > pick and
- self.pages[x] != self.last_choice):
- self.last_choice = self.pages[x]
- self.count += 1
- return self.pages[x]
+ if so_far > random_pick:
+ break
+ choice = self.pages[x]
+
+ # Allow filter list to suppress pages.
+ choice_is_filtered = False
+ for f in self.filter_list:
+ if not f(choice):
+ print("chooser: %s filtered by %s" % (choice, f.__name__))
+ choice_is_filtered = True
+ break
+ if choice_is_filtered:
+ continue
+
+ # We're good...
+ self.count += 1
+ return choice
+
class weighted_random_chooser_with_triggers(weighted_random_chooser):
"""Same as WRC but has trigger events"""
- def __init__(self, trigger_list):
- weighted_random_chooser.__init__(self)
+
+ def __init__(self, trigger_list, filter_list):
+ weighted_random_chooser.__init__(self, filter_list)
self.trigger_list = trigger_list
+ if trigger_list is None:
+ self.trigger_list = []
self.page_queue = set(())
def check_for_triggers(self):
return triggered
def choose_next_page(self):
- if (self.pages == None or
- self.count % 100 == 0):
+ if self.pages == None or self.count % 100 == 0:
self.pages = self.get_page_list()
triggered = self.check_for_triggers()
- # First try to satisfy from the page queue
- if (len(self.page_queue) > 0):
- print("Pulling page from queue")
+ # First try to satisfy from the page queue.
+ if len(self.page_queue) > 0:
+ print("chooser: Pulling page from queue...")
page = None
priority = None
for t in self.page_queue:
else:
return weighted_random_chooser.choose_next_page(self), False
+
class rotating_chooser(chooser):
"""Chooser that does it in a rotation"""
+
def __init__(self):
self.valid_filename = re.compile("([^_]+)_(\d+)_([^\.]+)\.html")
self.pages = None
self.count = 0
def choose_next_page(self):
- if (self.pages == None or
- self.count % 100 == 0):
+ if self.pages == None or self.count % 100 == 0:
self.pages = self.get_page_list()
if len(self.pages) == 0:
raise error
- if (self.current >= len(self.pages)):
+ if self.current >= len(self.pages):
self.current = 0
page = self.pages[self.current]
self.count += 1
return page
-#x = weighted_random_chooser_with_triggers(None)
+
+# Test
+def filter_news_during_dinnertime(page):
+ now = datetime.datetime.now()
+ is_dinnertime = now.hour >= 17 and now.hour <= 20
+ return not is_dinnertime or not (
+ "cnn" in page
+ or "news" in page
+ or "mynorthwest" in page
+ or "seattle" in page
+ or "stranger" in page
+ or "twitter" in page
+ or "wsj" in page
+ )
+
+
+# x = weighted_random_chooser_with_triggers([], [ filter_news_during_dinnertime ])
+# print(x.choose_next_page())