NLP主要会議のGoogle Calendar生成スクリプト(http://www.cs.rochester.edu/~tetreaul/conferences.html)

2015-07-31-121816.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

"""
NLP 主要会議のGoogle Calendar日程作成用のscript
http://www.cs.rochester.edu/~tetreaul/conferences.html
"""

import pandas as pd
import dateutil as du
import re
path = '/path/to/file.csv'

schedule = pd.read_csv(path, sep='\t')

for col in schedule:
    reg_name = col.strip().lower()
    schedule[reg_name] = schedule[col]
    if reg_name != col:
        schedule.drop(col, axis=1, inplace=True)

schedule = schedule.fillna('')
data = []
for idx, conf_schedule in schedule.iterrows():
    conf_name = conf_schedule['conference'].strip().strip('*')
    du_parser = du.parser
    dformat = '%m/%d/%Y'

    if conf_schedule['submission date']:
        date = conf_schedule['submission date']
        date = date.strip().strip('*')
        try:
            dt = du_parser.parse(date)
        except ValueError as e:
            pass
        else:
            conf_info = {
                'Subject': '{}:Submission Deadline'.format(conf_name),
                'Start Date': dt.strftime(dformat),
                'All Day Event': 'True',
                'Location': conf_schedule['location'],
            }
            data.append(conf_info)

    if conf_schedule['notification date']:
        try:
            date = conf_schedule['notification date']
            dt = du_parser.parse(date)
        except ValueError as e:
            pass
        else:
            date = date.strip().strip('*')
            conf_name = re.sub('(.\+)', '', conf_name)
            conf_info = {
                'Subject': '{}:Notification'.format(conf_name),
                'Start Date': dt.strftime(dformat),
                'All Day Event': 'True',
                'Location': conf_schedule['location'],
            }
            data.append(conf_info)

    if conf_schedule['conference date']:
        conf_name = re.sub(r'\s*\(.+\)\s*', '', conf_name)
        date_string = conf_schedule['conference date']
        try:
            dts = du_parser.parse(re.sub('-\d.', '', date_string))
            dte = du_parser.parse(re.sub('\d.-', '', date_string))
        except ValueError as e:
            pass
        else:
            conf_info = {
                'Subject': '{}:Conference'.format(conf_name),
                'Start Date': dts.strftime(dformat),
                'End Date': dte.strftime(dformat),
                'All Day Event': 'True',
                'Location': conf_schedule['location'],
            }
            data.append(conf_info)

df = pd.DataFrame().from_dict(data)
df = df.set_index('Subject')
print df.to_csv()