decklin/rawdog-contrib

view feedtime @ 13:900be58d356d

find changesets by author, revision, files, or words in the commit message
Big refactoring. Gather the deltas in one place (no dict, slow), and use a
smarter median after. Scale with sqrt again... in the absence of better info
everything will settle on DEF. Average in the existing interval value to avoid
crazy big adjustments, too.
author decklin
date Fri Jun 29 20:16:01 2007 +0000 (2 years ago)
parents 8f4b9824a87d
children
line source
1 #!/usr/bin/python
2
3 import sys
4 import math
5 import time
6 import pickle
7
8 from rawdoglib import rawdog
9
10 MIN = 60 * 30
11 DEF = 60 * 144
12
13 def scale(t):
14 return math.sqrt(DEF) * math.sqrt(t)
15
16 def ok_median(l):
17 mean = float(sum(l)) / len(l)
18 dsq = sum([(x - mean) ** 2 for x in l])
19 stdev = math.sqrt(dsq / len(l))
20
21 l = [x for x in l if abs(x - mean) <= 3*stdev]
22 l.sort()
23
24 if len(l) % 2:
25 return l[len(l)//2]
26 else:
27 return (l[len(l)//2-1] + l[len(l)//2]) / 2
28
29 def delta_dates(state, url):
30 arts = [a for a in state.articles.values() if a.feed == url]
31 dates = list(set([a.date or a.added for a in arts]))
32
33 if len(dates) == 1:
34 dates.append(time.time())
35 else:
36 dates.sort()
37
38 return [b-a for a, b in zip(dates, dates[1:]) if b-a > MIN/2]
39
40 if __name__ == '__main__':
41 state = pickle.load(file('state'))
42
43 for line in file('config'):
44 tokens = line.split()
45
46 if tokens and tokens[0] == 'feed' and 't=fix' not in tokens[3:]:
47 intervals = [rawdog.parse_time(tokens[1])]
48 intervals += delta_dates(state, tokens[2])
49 t = scale(ok_median(intervals))
50 print 'feed %dm %s' % (t/60, ' '.join(tokens[2:]))
51 else:
52 print line,
53
54 def unparse_time(t):
55 units = ((604800, 'w'), (86400, 'd'), (3600, 'h'), (60, 'm'), (1, 's'))
56 s = []
57
58 for u in units:
59 p, t = divmod(t, u[0])
60 if p or s:
61 s.append('%d%s' % (p, u[1]))
62 if t == 0:
63 break
64
65 return ''.join(s)