#! /usr/bin/env python

#
# Baidu Blog and Forum Timed Corpus sample script (word temporal statistics)
#
# Copyright (c) 2010 Baidu Japan, Inc.
# URL: http://www.baidu.jp/corpus/
#
# Usage: python bbftf_sample.py <corpus dir> <word1> <word2> ...
# 

import sys, codecs

sys.stdout = codecs.getwriter('utf_8')(sys.stdout)

corpusdir = sys.argv[1] if len(sys.argv)>1 else "./"
query = ' '.join(sys.argv[2:]).decode('utf-8') if len(sys.argv)>2 else u"\u85C1"
gram =  len(query.split(' '))

# maximum scales -> 0.001, 0.002, 0.005, 0.01, 0.02, 0.05, ...
scales = [r*f for r in [10**x for x in xrange(-3, 2)] for f in [1,2,5]]

# screen width = 80
WIDTH = 80
GRAPHWIDTH = WIDTH-8-2-8 # WIDTH-len("YYYY-MM ")-2*len("|")-len("00.000 %")

print 'Word temporal statistics of: "%s"' % query

res = []
print "Reading %dgm ..." % gram,
for year in xrange(2000, 2011):
    print "%d ..." % year,
    sys.stdout.flush()
    
    for month in xrange(1, 13):

        if year == 2010 and month == 8: break
        ymstr = "%d-%02d" % (year, month)
        
        # read 1gm file
        query_freq = 0
        total_freq = 0
        for line in open("%s/%s.%dgm" % (corpusdir, ymstr, gram)):
            w, c_str = line.decode('utf-8').split('\t')
            c = int(c_str[:-1])
            total_freq += c
            
            if query == w: query_freq = c

        percent = query_freq * 100.0 / total_freq
        res.append( [ymstr, percent] )

# find the maximum fitting scale
max_percent = max(r[1] for r in res)
scale = min(s for s in scales if s > max_percent)

# draw the axis
print ""
print "        | 0 %% %s % 7.3f %% |" % (" " * (GRAPHWIDTH - len(" 0 %   00.000 % ")), scale)

# draw the graph
for r in res:
    width = int(GRAPHWIDTH * r[1] / scale)
    print "%s |%s%s| % 7.3f %%" % (r[0], "*"*width, " "*(GRAPHWIDTH-width), r[1] )