#!/usr/bin/env python

"""
This module parses Slashdot comment headers for analysis. 
"""

# # content = '''   Re:one could argue (Score:1)
# #   by [108]Dwonis (52652) [109]Alter Relationship
# #   <[110]dlitzNO@SPAMdlitz.net> on 10:44 PM January 14th, 2003
# # 
# #     Re:one could argue (Score:2)
# #   by [108]Dwonis (52652) [109]Alter Relationship
# #   <[110]dlitzNO@SPAMdlitz.net> on 10:44 PM January 15th, 2003  
# #     
# #     Re:one could argue (Score:3)
# #   by [108]Dwonis (52652) [109]Alter Relationship
# #   <[110]dlitzNO@SPAMdlitz.net> on 10:44 PM January 15th, 2003'''

import os, re, string
from time import *

results = {}

# http://developers.slashdot.org/article.pl?sid=03/01/14/1514205
content = open("slash-comments.txt", "rb").read()
story_posted = "01:43 PM January  14th, 2003"
n_story_posted = strptime(story_posted, "%H:%M %p %b %dth, %Y")
n_story_posted = mktime(n_story_posted)

prog = re.compile('\(Score:(.*?)[\,,\)].*? on (.*?2003)',re.DOTALL)
headers = prog.findall(content)

for (score, date) in headers:
	score = int(score)
	comment = string.join(string.split(date))
	n_comment = strptime(comment, "%H:%M %p %b %dth, %Y")
	n_comment = mktime(n_comment)
	difference = (n_comment - n_story_posted) / 3660
# 	print "score: %s, elapsed time: %s" % (score, difference)
	results.setdefault(score,[]).append(difference)

import stats


print "Score with Mean, Minimum, and Maximum Elapsed Time in Hours"
print "score	 mean			minimum	maximum"
for key in [5,4,3,2,1,0,-1]:
	print "%+2d	 %2.2f			%2.2f	%2.2f" \
		% (key, stats.mean(results[key]), min(results[key]), max(results[key]))
	#print stats.lhistogram(results[key])