#! /usr/bin/env python
# -*- coding: utf-8 -*-
#Format the pickles files dictOcc, dictOcc.old, listOcc and listOcc.old to make a table with corresponding colors for entries
import sys, re, getopt
import pickle
from string import strip
def compute_occ_subdict(extendedl,subdictTemplate={}):
"""Compute occurence of each item with a dictionary in a dictionary
Return a dict of dicts"""
occ = {}
for e in extendedl:
if occ.get(e[1],0) == 0: #Subdict not yet created
occ[e[1]] = dict(subdictTemplate) #Do not copy the reference but create a new one from template
occ[e[1]][e[0][0]] = occ.get(e[1],0).get(e[0][0],0) + 1
return occ
def file_to_list(nameFile):
f = open(nameFile,'r')
l =[]
for line in f:
l.append(line.strip('\n'))
f.close()
return l
def list_to_file(l,nameFile):
"""Print a list to a file.
For list not used in this module, the result can be unexpected"""
if type(l) != list:
print "Warning: The argument is not a list (list_to_file())"
print "Unexpected behavior can occur"
f = open(nameFile,'w')
for s in l:
if type(s) == tuple:
f.write(str(s[0])+"\t"+str(s[1])+'\n')
elif type(s) == str:
f.write(s+'\n')
else:
print "This format is not supported"
f.close()
def dict_to_file(d,nameFile):
"""Print a dict to a file.
For dict not used in this module, the result can be unexpected"""
if type(d) != dict:
print "Warning: The argument is not a dict (dict_to_file())"
print "Unexpected behavior can occur"
f = open(nameFile,'w')
for s in d:
f.write(s+' '+d[s]+'\n')
f.close()
def check_mapping(p):
mapping = [(50000, "bgcolor='#000000' style='color:white;' |"), (20000, "bgcolor='#20FF20'|"), (5000, "bgcolor='#6EFF6E'|"), (1000, "bgcolor='#F5FFF5'|"), (0,"") ] # Add all your values and returns here
for check, value in mapping:
if p >= check:
return value
def check_mapping_diff(p):
if p < 0:
return "bgcolor='red'|"
elif p==0:
return "bgcolor='grey'|"
mapping = [(5000, "bgcolor='#000000' style='color:white;' |+"), (2000, "bgcolor='#20FF20'|+"), (500, "bgcolor='#6EFF6E'|+"), (100, "bgcolor='#F5FFF5'|+"), (0,"+") ] # Add all your values and returns here
for check, value in mapping:
if p >= check:
return value
def write_item(it,f):
f.write(check_mapping(it))
f.write(" "+str(it))
def write_item_diff(it,f):
f.write(check_mapping_diff(it))
f.write(" "+str(it))
def write_line(dict,f,lineNumber):
f.write('|'+str(lineNumber)+'||bgcolor="#EEEEEE" |[[:Catégorie:{{nom langue|'+dict["lang"]+'}}|{{nom langue|'+dict["lang"]+'}}]]||')
write_item(dict["total"],f)
f.write(' ||')
write_item(dict["oldTotal"],f)
f.write(' ||')
write_item_diff(dict["total"]-dict["oldTotal"],f)
f.write(' ||')
write_item(dict["nom"],f)
f.write(' ||')
write_item(dict["nom-pr"],f)
f.write(' ||')
write_item(dict["adj"],f)
f.write(' ||')
write_item(dict["verb"],f)
f.write(' ||')
write_item(dict["adv"],f)
f.write(' ||')
write_item(dict["total"]-dict["flex"],f)
f.write(' ||')
write_item(dict["flex"],f)
f.write(' ||')
write_item(dict["loc"],f)
f.write(' ||')
write_item(dict["stub"],f)
f.write('\n')
#f.write('| %(lang)s || %(total)d || %(nom)d || %(nom-pr)d || %(adj)d || %(verb)d || %(adv)d\n' % dict)
def format_and_write(l,outf):
"""Format the list of tuple and write it to a file formated for a wikitable"""
fFinal = open(outf,'w')
for line in l:
fFinal.write('|-\n')
fFinal.write('| %(lang)s || %(number)d\n' % {'lang': line[1],'number': line[0]})
fFinal.close()
def format_and_write_more(d,l,outf):
"""Format the list of tuple and write it to a file formated for a wikitable"""
fFinal = open(outf,'w')
for line in l:
dictFull = {'lang' : line[1], 'total' : line[0]}
if d.has_key(line[1]):
dictFull.update(d[line[1]])
fFinal.write('|-\n')
write_line(dictFull,fFinal)
else:
print 'no dictionary for ',line[1]
fFinal.close()
def format_and_write_both(d,l,oldd,oldl,outf):
"""Format the list of tuple and write it to a file formated for a wikitable"""
lineNumber = 0
fFinal = open(outf,'w')
for line in l:
lineNumber = lineNumber + 1
dictFull = {'lang' : line[1], 'total' : line[0]}
dictFull.update({'oldTotal' : oldl.get(line[1],0)})
if d.has_key(line[1]):
dictFull.update(d[line[1]]) #Add the dictionary of type in a global dictionary
# fFinal.write('|-\n')
# write_line(dictFull,fFinal,lineNumber)
else:
dictFull.update({'nom':0,'nom-pr':0,'adj':0,'verb':0,'adv':0,'flex':0,'loc':0, 'stub':0})
fFinal.write('|-\n')
write_line(dictFull,fFinal,lineNumber)
fFinal.close()
def usage():
sys.stderr.write("""Options available are\n
-h --help
-v --verbose Enter verbose mode
-i --input Specify an input directory
-o --output Specify an output filename
-d (nothing changing)\n""")
def main(argv):
global _verbose
global _debug
_verbose = 0
_debug = 0
outf = 'langsTableCol'
try:
opts, args = getopt.getopt(argv, "hvi:o:d", ["help", "verbose", "input=", "output="])
except getopt.GetoptError:
sys.stderr.write("Illegal argument\n")
usage()
sys.exit(2)
for opt, arg in opts:
if opt in ("-h", "--help"):
usage()
sys.exit(0)
elif opt == '-d':
_debug = 1
#TODO: put debug condition
elif opt in ("-v", "--verbose"):
_verbose = 1
#TODO: put verbose condition
elif opt in ("-o", "--output"):
#TODO: verify that arg is a directory path
outf = arg
elif opt in ("-i", "--input"):
#TODO: verify that arg is a directory path
inputdir = arg
fpickle = open('dictOcc','r')
typeOcc=pickle.load(fpickle)
fpickle.close()
fpickle = open('listOcc','r')
langsOcc=pickle.load(fpickle)
fpickle.close()
fpickle = open('dictOcc.old','r')
typeOccOld=pickle.load(fpickle)
fpickle.close()
fpickle = open('listOcc.old','r')
langsOccOld=pickle.load(fpickle)
fpickle.close()
#Sort the result
lLangsSorted = []
for lang in langsOcc.keys():
lLangsSorted.append((langsOcc[lang], lang))
lLangsSorted.sort(reverse=True)
#Give a sum per type for all languages
sumPerType = {'nom':0,'nom-pr':0,'adj':0,'verb':0,'adv':0,'flex':0,'loc':0, 'stub':0}
for lang in typeOcc.values():
#DEBUG
#print lang
#END DEBUG
for i in sumPerType.keys() :
sumPerType[i] = sumPerType[i] + lang[i]
print "Total : nom : {nom:d} , nom-pr: {nom-pr:d} , adj: {adj:d} , verb: {verb:d} , adv: {adv:d} , flex: {flex:d} , loc: {loc:d} , stub: {stub:d}".format(**sumPerType)
#Give total for all words
print "New total : ", sum(langsOcc.values()), " Old total : ", sum(langsOccOld.values())," diff : ", sum(langsOcc.values())-sum(langsOccOld.values())
#Format it to be directly pastable in the wiktionary table
print "\n| || bgcolor='yellow' | Total général ||", sum(langsOcc.values()), "||", sum(langsOccOld.values()),"||", sum(langsOcc.values())-sum(langsOccOld.values()), "||", sumPerType["nom"], "||", sumPerType["nom-pr"], "||", sumPerType["adj"], "||", sumPerType["verb"], "||", sumPerType["adv"], "||", sum(langsOcc.values())-sumPerType["flex"], "|| {flex:d} || {loc:d} || {stub:d}\n".format(**sumPerType)
# Print the result
format_and_write_both(typeOcc,lLangsSorted,typeOccOld,langsOccOld,outf)
if _verbose:
print 'Output written in "%s"'% outf
if __name__ == '__main__':
main(sys.argv[1:])
