In [17]:

    

from numpy import ndarray

# making authors list
def make_authors_list_from_csv(filename):
    authors_list = []
    with open(filename+'.csv') as f:
        next(f)
        for line in f:
            split = line.split('|')
            authors = split[2].split(',')            
            authors_clean = []
            for author in authors:
                authors_clean.append(author.lstrip())
            # authors list
            for author in authors_clean:
                author= author.lstrip()
                if not(author in authors_list):
                    authors_list.append(author.lstrip())        
    authors_list = sorted(authors_list)
    authors_dict = {}
    i = 0
    for author in authors_list:
        authors_dict[author] = i
        i+=1
    return authors_list,authors_dict



# making dictionnary for indexing author   
# construction poster dictionary  
# for each poster: the list of collaborators
def make_poster_dict_from_csv(filename):
    poster_count = {}
    poster_dict = {}
    with open(filename+'.csv') as f:
        next(f)
        for line in f:
            split = line.split('|')
            posterid =int(split[0])
            authors = split[2].split(',')            
            authors_id = []
            for author in authors:
                author = author.lstrip()
                authors_id.append(authors_dict[author.lstrip()])
                if author in poster_count:
                    poster_count[author]+=1
                else:
                    poster_count[author]=1
            poster_dict[posterid] = authors_id
            
#            for author in poster_count:                
#                poster_count_list.append([author,poster_count[author]])
            
    return poster_dict,poster_count  


import os
# poster count when multiple poster
def make_json_poster_count(poster_count,filename,number_min=1):
    poster_count_list  = [[author,poster_count[author]] for author in poster_count]
    poster_count_list = sorted(poster_count_list,key=lambda x:x[1], reverse=True)
    authors = [x[0] for x in poster_count_list][0:100]
    counts = [x[1] for x in poster_count_list][0:100]
    with open(filename+".json", "w") as outfile:
        outfile.write("{\"name\":\"" + str(authors) +"\",\n\"count\":"+str(counts)+"}\n")    
    return     

# poster count when multiple poster
def make_csv_poster_count(poster_count,filename,number_min=1):
    poster_count_list  = [[author,poster_count[author]] for author in poster_count]
    poster_count_list = sorted(poster_count_list,key=lambda x:x[1], reverse=True)
    authors = [x[0] for x in poster_count_list][0:100]
    counts = [x[1] for x in poster_count_list][0:100]
    with open(filename+".csv", "w") as outfile:
        outfile.write("Authors, Counts\n")
        for x in poster_count:
            outfile.write(x[0] +', '+ str(x[1])+"\n")


    

In [18]:

    

filename = 'AbstractsCosyne2012'
authors_list,authors_dict = make_authors_list_from_csv(filename)
poster_dict,poster_count = make_poster_dict_from_csv(filename)
make_json_poster_count(poster_count,'poster_count12',2);
make_csv_poster_count(poster_count,'poster_count12',2);


    

In [19]:

    

filename = 'AbstractsCosyne2013'
authors_list,authors_dict = make_authors_list_from_csv(filename)
poster_dict,poster_count = make_poster_dict_from_csv(filename)
make_json_poster_count(poster_count,'poster_count13',2);
make_csv_poster_count(poster_count,'poster_count13',2);


    

In [25]:

    

import csv
with open('cosyne2011.csv', 'rb') as csvfile:
    spamreader = csv.reader(csvfile, delimiter="|", quotechar="")
    for row in spamreader:
        print ', '.join(row)


    

    




---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-25-2c6c7a86dd01> in <module>()
      1 import csv
      2 with open('cosyne2011.csv', 'rb') as csvfile:
----> 3     spamreader = csv.reader(csvfile, delimiter="Ê", quotechar="|")
      4     for row in spamreader:
      5         print ', '.join(row)

TypeError: "delimiter" must be an 1-character string

In [9]:

    

lt[0:5]


    

    
Out[9]:





[0, 1, 2, 3, 4]

In [ ]: