diff --git a/histories/models.py b/histories/models.py index 5411fa7..2f33e2e 100644 --- a/histories/models.py +++ b/histories/models.py @@ -10,6 +10,7 @@ import profiles.models as profiles from crawl import crawl from pinocchio.settings import HISTORY_MIN + class HistoryEntry(models.Model): """ A history entry, aka a url, and a timestamp. """ @@ -62,12 +63,16 @@ class History(models.Model): self.save() -def generate_partial_history(user, t_start, url, history): +def generate_partial_history(user, t_start, history): """ Generate the part of the history resulting from the crawl starting at the given url. """ + result = [] + basis = generate_first_url(user) + result.append((basis, t_start)) + t_start += 5* random.weibullvariate(1, 1.5) #crawler = crawl.CrawlingThread() - return [] + return result def generate_first_url(user): """ Generate the first url of a partial history, based on the user @@ -77,7 +82,7 @@ def generate_first_url(user): user.interests.websites.all(), user.interests.events.all() ] ) - search_term = random.choice(interset) + search_term = random.choice(interest) url = search_term.generate_url(user) return url @@ -98,12 +103,13 @@ def generate_history(user, ts_start): history_line = 0 while history_line < length: - ts_start += random.uniform(1, 10) - history = generate_partial_history(user, ts_start, url) - ts_start = history[-1].timestamp + 5 * weilbullvariate(1, 5) - for (url, ts) in history: + ts_start += 5 * random.weibullvariate(1, 2.8) + history_list = generate_partial_history(user, ts_start, history) + ts_start = history_list[-1].timestamp + 5 * random.weibullvariate(1, 5) + for (url, timestamp) in history_list: new_line = HistoryEntry( search=url, - timestamp=ts, + timestamp=timestamp, history=history ) + new_line.save()