csvReader.py

# python 3.6
import sys as S
import csv as CS
import collections as C
import xml.etree.ElementTree as ET
import os  # dir paths

infile = S.argv[1]
csvfile = open(infile, 'r')
csvlist = [x.strip() for x in csvfile.readlines()]
csvfile.close()

AnnoDict = C.defaultdict(list)

for csv in csvlist:
    print (csv)
    with open(csv, 'r') as file:
        fileReader = CS.reader(file, delimiter=';', quotechar='|')
        print ('opened ' + csv)

        header = next(fileReader)  # note Python 3 & 2 next differences
        while not 'sentence id' in header:
            next(header)

        IdCol = None
        AnnoCols = {}
        for (i, col) in enumerate(header):         

            if col.endswith(' id'):  # get id column index
                print ('ID column found!')
                IdCol = i

            if col.endswith(' class'):  # get annotation column indices
                AnnoCols[i] = col[:-6]

        for row in fileReader:
            AnnotationID = row[IdCol]  # sentence id
            for anno_col in AnnoCols.keys():
                if row[anno_col].strip():
                    AnnoDict[AnnotationID].append((AnnoCols[anno_col],
                            row[anno_col]))

        print (AnnoDict)  # print the resulting annotation dictionary