import datetime

from dateutil import tz

import json

import sys

import time

import os

from granary import microformats2

from granary import twitter

reload(sys)

sys.setdefaultencoding('utf8')

if __name__ == '__main__':

# set up our timezone objects, so we can set local times on all posts

utc = tz.gettz('UTC')

est = tz.gettz('America/New_York')

# load tweet JSON from command line argument

posts = json.loads(open(sys.argv[1], 'r').read())

if not os.path.isdir('note'):

os.makedirs('note')

last_date = ''

counter = 0

for post in posts:

decoded = twitter.Twitter('token', 'secret', 'smerrill').tweet_to_activity(post)

obj = twitter.Twitter('token', 'secret', 'smerrill').tweet_to_object(post)

# first we need to deal with legacy tweets in the export, because they

# all have a published time of midnight UTC on the day of publication.

# Set them to my timezone while I'm at it.

date = datetime.datetime.strptime(decoded['published'], "%Y-%m-%d %H:%M:%S +0000").replace(tzinfo=utc).astimezone(est)

# we have the date, which is enough to build the directory path

path = "note/%s" % date.strftime('%Y/%m/%d')

if not os.path.isdir(path):

os.makedirs(path)

index = "%s/_index.md" % path

open(index,'w').close()

if last_date == date:

# this tweet has the same timestamp as the last one.

# we'll just increment by one minute. I don't recall ever

# having tweeted more than 60 times in a day.

counter += 1

date = date + datetime.timedelta(minutes=counter)

else:

counter = 0

last_date = date

# now we have a unique date and time, which we can use to

# create the output file

filename = "note/%s/%s.md" % (date.strftime('%Y/%m/%d'), date.strftime('%H%M%S'))

file = open(filename, 'w')

file.write('---\n')

file.write('author: skippy\n')

file.write("date: %s\n" % date.strftime('%Y-%m-%d %H:%M:%S'))

# now we figure out what kind of Tweet this is:

# new post, reply to someone, or retweet.

# i don't "like" anything, so i don't need to worry about

# that activity here.

if decoded['verb'] == 'share':

# we'll deal with retweets first.

file.write("repost_url: %s\n" % decoded['object']['url'])

file.write("repost_name: %s\n" % decoded['object']['author']['username'])

if decoded['verb'] == 'post':

# this is either a new post, or a reply

if 'inReplyTo' in decoded['object']:

# okay, this is a rewteet.

# but sometimes the user ID is invalid.

if 'tags' in decoded['object'] and decoded['object']['tags'][0]['url']:

reply_user = "'@%s'" % decoded['object']['tags'][0]['url'].replace('https://twitter.com/', '')

else:

reply_user = 'a former Twitter user'

file.write("reply_to_user: %s\n" % reply_user)

file.write("reply_to_url: %s\n" % decoded['object']['inReplyTo'][0]['url'])

file.write("tweet_url: %s\n" % decoded['url'])

file.write('---\n')

# we use the render_content() method here to ensure that we get

# untruncated links in the output.

file.write('%s\n' % microformats2.render_content(obj, False))

file.write("\n")

file.close()