#!/usr/bin/env python
'''
Permission is hereby granted, free of charge, to any person obtaining a
copy of this software and associated documentation files (the "Software"),
to deal in the Software without restriction, including without limitation
the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the
Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
DEALINGS IN THE SOFTWARE.
'''

import argparse
import urlparse
import urllib2
import urllib
import hashlib
import base64
import hmac
import json
import sys
import os
import time
import datetime
import random
import math
import csv
import re
import codecs
import errno
import socket

__version__ = '1.13'
ALPHANUM    = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789'

FDAT_DIR    = 'fdat'
IMG_DIR     = 'img'
TT_EXT      = '.dat'
FDAT_EXT    = '.f'
FAV_EXT     = '.fav'
TWT_EXT     = '.twt'
FMAX        = 5000
EDG_EXT     = '.gml' # Graph Modeling Language
WAIT_CODES  = (400, 503, 429)
RETRY_CODES = (500, 502, 110)
SKIP_CODES  = (401, 403, 404)
SHAPES      = ['box', 'circle', 'rect', 'triangle-up', 'diamond', 'triangle-down']

from collections import namedtuple
Consumer = namedtuple('Consumer', 'key secret')
Token = namedtuple('Token', 'key secret')

def _quote(text):
  return urllib.quote(text, '-._~')

def _encode(params):
  return '&'.join(['%s=%s' % (k, v) for k, v in params])

def _parse_uri(req):
  method = req.get_method()
  if method == 'POST':
    uri = req.get_full_url()
    query = req.get_data() or ''
  else:
    url = req.get_full_url()
    if url.find('?') != -1:
      uri, query = req.get_full_url().split('?', 1)
    else:
      uri = url
      query = ''
  return method, uri, query

class Request(urllib2.Request):
  def __init__(self, url, \
    data=None, headers={}, origin_req_host=None, unverifiable=False, \
    method=None, oauth_params={}):
    urllib2.Request.__init__( \
      self, url, data, headers, origin_req_host, unverifiable)
    self.method = method
    self.oauth_params = oauth_params

  def get_method(self):
    if self.method is not None:
      return self.method
    if self.has_data():
      return 'POST'
    else:
      return 'GET'

class OAuthHandler(urllib2.BaseHandler):
  def __init__(self, consumer, token=None, timeout=None):
    self.consumer = consumer
    self.token = token
    self.timeout = timeout

  def get_signature(self, method, uri, query):
    key = '%s&' % _quote(self.consumer.secret)
    if self.token is not None:
      key += _quote(self.token.secret)
    signature_base = '&'.join((method.upper(), _quote(uri), _quote(query)))
    signature = hmac.new(str(key), signature_base, hashlib.sha1)
    return base64.b64encode(signature.digest())

  def http_request(self, req):
    if not req.has_header('Host'):
      req.add_header('Host', req.get_host())
    method, uri, query = _parse_uri(req)
    if method == 'POST':
      req.add_header('Content-type', 'application/x-www-form-urlencoded')

    query = map(lambda (k, v): (k, urllib.quote(v)), urlparse.parse_qsl(query))

    oauth_params = [
      ('oauth_consumer_key', self.consumer.key),
      ('oauth_signature_method', 'HMAC-SHA1'),
      ('oauth_timestamp', int(time.time())),
      ('oauth_nonce', ''.join([random.choice(ALPHANUM) for i in range(16)])),
      ('oauth_version', '1.0')]
    if self.token is not None:
      oauth_params.append(('oauth_token', self.token.key))
    if hasattr(req, 'oauth_params'):
      oauth_params += req.oauth_params.items()

    query += oauth_params
    query.sort()
    signature = self.get_signature(method, uri, _encode(query))

    oauth_params.append(('oauth_signature', _quote(signature)))
    oauth_params.sort()

    auth = ', '.join(['%s="%s"' % (k, v) for k, v in oauth_params])
    req.headers['Authorization'] = 'OAuth ' + auth

    req = Request(req.get_full_url(), \
      data=req.get_data(), \
      headers=req.headers, \
      origin_req_host=req.get_origin_req_host(), \
      unverifiable=req.is_unverifiable(), \
      method=method)

    req.timeout = self.timeout
    return req

  def https_request(self, req):
    return self.http_request(req)

def _replace_opener():
  filename = os.path.expanduser('~')+'/.'+os.path.basename(sys.argv[0])
  if os.path.isfile(filename):
    f = open(filename, 'r')
    lines = f.readlines()
    key = lines[0].strip()
    secret = lines[1].strip()
  else:
    sys.stderr.write('''TWITTER API AUTHENTICATION SETUP
(1) Open the following link in your browser and register this script...
'>>> https://apps.twitter.com/\n''')
    sys.stderr.write('What is its consumer key? ')
    key = sys.stdin.readline().rstrip('\r\n')
    sys.stderr.write('What is its consumer secret? ')
    secret = sys.stdin.readline().rstrip('\r\n')
    lines = [key, secret]
  consumer = Consumer(key, secret)
  try:
    oauth = lines[2].strip()
    oauth_secret = lines[3].strip()
    atoken = Token(oauth, oauth_secret)
  except IndexError:
    opener = urllib2.build_opener(OAuthHandler(consumer))
    resp = opener.open(Request('https://api.twitter.com/oauth/request_token'))
    rtoken = urlparse.parse_qs(resp.read())
    rtoken = Token(rtoken['oauth_token'][0], rtoken['oauth_token_secret'][0])
    sys.stderr.write('''(2) Now, open this link and authorize the script...
'>>> https://api.twitter.com/oauth/authorize?oauth_token=%s\n''' % rtoken.key)
    sys.stderr.write('What is the PIN? ')
    verifier = sys.stdin.readline().rstrip('\r\n')
    opener = urllib2.build_opener(OAuthHandler(consumer, rtoken))
    resp = opener.open( \
      Request('https://api.twitter.com/oauth/access_token', \
      oauth_params={'oauth_verifier': verifier}))
    atoken = urlparse.parse_qs(resp.read())
    atoken = Token(atoken['oauth_token'][0], atoken['oauth_token_secret'][0])
    f = open(filename, 'w')
    f.write(key+'\n')
    f.write(secret+'\n')
    f.write(atoken.key+'\n')
    f.write(atoken.secret+'\n')
    f.close()
    sys.stderr.write('Setup complete and %s created.\n' % filename)
  opener = urllib2.build_opener(OAuthHandler(consumer, atoken))
  urllib2.install_opener(opener)

def _fetch_img(user_id, avatar):
  conn = urllib.urlopen(avatar)
  data = conn.read()
  info = conn.info().get('Content-Type').lower()
  conn.close()
  if not os.path.exists(IMG_DIR):
    os.makedirs(IMG_DIR)
  filename = IMG_DIR+'/'+str(user_id)
  if info == 'image/gif':
    filename += '.gif'
  elif info == 'image/jpeg' or info == 'image/pjpeg':
    filename += '.jpg'
  elif info == 'image/png':
    filename += '.png'
  file = open(filename, 'wb')
  file.write(data)
  file.close()
  return filename

def resolve(args):
  for sn in args.screen_name:
    try:
      if sn.isdigit():
        url = 'https://api.twitter.com/1.1/users/show.json?user_id=%s'
      else:
        url = 'https://api.twitter.com/1.1/users/show.json?screen_name=%s'
      conn = urllib2.urlopen(url % sn)
    except urllib2.HTTPError, e:
      if e.code in SKIP_CODES:
        sys.stdout.write('HTTPError %s with %s. Skipping...\n' % (e.code, sn))
        continue
      else:
        raise
    data = json.loads(conn.read())
    conn.close()
    if sn.isdigit():
      sys.stdout.write(data['screen_name']+' ')
    else:
      sys.stdout.write(data['id_str']+' ')
    sys.stdout.write('(%s friends | %s followers | %s memberships | %s tweets)\n' % (\
      data['friends_count'], data['followers_count'], data['listed_count'], data['statuses_count']))

class CursorError(urllib2.HTTPError):
  def __init__(self, code, cursor=None, res=None, headers=None):
    urllib2.HTTPError.__init__(self, None, code, None, None, None)
    self.cursor = cursor
    self.res = res
    self.headers = headers

# Results are given in groups of 5,000
def _ids(relation, param, c=None, init=None):
  res = init or []
  cursor = c or -1
  while True:
    try:
      url = 'https://api.twitter.com/1.1/%s/ids.json?%s&cursor=%s'
      conn = urllib2.urlopen(url % (relation, param, cursor))
    except urllib2.HTTPError, e:
      raise CursorError(e.code, cursor, res, e.headers)
    data = json.loads(conn.read())
    conn.close()
    res = res + data['ids']
    if data['next_cursor'] != 0:
      cursor = data['next_cursor']
    else:
      break
  return res

def _members(param, c=None, init=None):
  res = init or []
  cursor = c or -1
  while True:
    try:
      url = 'https://api.twitter.com/1.1/lists/members.json?%s&cursor=%s&include_entities=false&skip_status=true'
      conn = urllib2.urlopen(url % (param, cursor))
    except urllib2.HTTPError, e:
      raise CursorError(e.code, cursor, res)
    data = json.loads(conn.read())
    conn.close()
    for user in data['users']:
      res.append(user['id'])
    if data['next_cursor'] != 0:
      cursor = data['next_cursor']
    else:
      break
  return res

# first retrieve list of IDs from relation, tweets or list memberships
# then retrieve details for each 100 at a time
def init(args):
  if args.l:
    type = args.l
  else:
    type = 'followers' if args.followers else 'friends'
  if args.query:
    # retrieve handles from file
    bag = []
    users = []
    for tweet in open(args.screen_name+TWT_EXT):
      handles = re.findall(r'@([\w]+)', tweet.lower())
      if len(handles) > 0:
        if args.nomention:
            handles = [handles[0]]
        bag += handles
        users.append(handles[0])
    bag = list(set(bag))
    users = list(set(users))
  else:
    # retrieve IDs from relationship
    cursor = res = None
    while True:
      try:
        if args.l:
          bag = _members('slug='+args.l+'&owner_screen_name='+args.screen_name, cursor, res)
        else:
          url = 'https://api.twitter.com/1.1/users/show.json?screen_name=%s'
          conn = urllib2.urlopen(url % args.screen_name)
          data = json.loads(conn.read())
          conn.close()
          bag = [data['id']] + _ids(type, 'screen_name='+args.screen_name, cursor, res)
        break
      except CursorError, e:
        if e.code in WAIT_CODES:
          waiting_time = int(e.headers['x-rate-limit-reset']) - int(round(time.time())) + random.randint(10,30)
          sys.stderr.write('HTTPError %s at %s. Waiting %sm to resume...' % \
            (e.code, time.strftime('%H:%M', time.localtime()), waiting_time/60))
          time.sleep(waiting_time)
          cursor = e.cursor
          res = e.res
          sys.stderr.write('\n')
          continue
        else:
          raise
  filename = args.screen_name+TT_EXT
  if not args.force and os.path.isfile(filename):
    # ignore previously processed data found in file
    f = open(filename, 'r+')
    for item in csv.reader(f):
      try:
        bag.remove(item[1].lower() if args.query else int(item[0]))
      except:
        continue
  else:
    f = open(filename, 'w')
  # retrieve details for all IDs in the bag
  while len(bag) > 0:
    next_items = []
    if len(bag) > 100:
      while len(next_items) < 100:
        next_items = [bag.pop() for _ in xrange(100)]
    else:
      next_items = bag
      bag = []
    sys.stdout.write('Processing %i starting from %s...\n' % (len(next_items), next_items[0]))
    try:
      if isinstance(next_items[0], str):
        url = 'https://api.twitter.com/1.1/users/lookup.json?screen_name=%s&include_entities=false'
        items_str = ','.join(next_items)
      else:
        url = 'https://api.twitter.com/1.1/users/lookup.json?user_id=%s&include_entities=false'
        items_str = ','.join(map(str, next_items))
      conn = urllib2.urlopen(url % items_str)
    except (urllib2.HTTPError, socket.error) as e:
      if hasattr(e, 'code') and e.code in SKIP_CODES:
        # 404 if no lookup criteria could be satisified
        sys.stdout.write('HTTPError %s starting from %s. Skipping...\n' % (e.code, next_items[0]))
        continue
      elif (hasattr(e, 'code') and e.code in RETRY_CODES) or \
          (hasattr(e, 'errno') and e.errno in (errno.ECONNRESET, errno.EAGAIN)):
        bag += next_items
        sys.stderr.write('HTTPError %s starting from %s. Deferred...' % (e.code, next_items[0]))
        sys.stderr.flush()
        time.sleep(random.randint(10,30))
        sys.stderr.write('\n')
        continue
      elif hasattr(e, 'code') and e.code in WAIT_CODES:
        bag += next_items
        waiting_time = int(e.headers['x-rate-limit-reset']) - int(round(time.time())) + random.randint(10,30)
        sys.stderr.write('HTTPError %s at %s. Waiting %sm to resume (%s items left)...' % (e.code, time.strftime('%H:%M', time.localtime()), waiting_time/60, len(bag)))
        sys.stderr.flush()
        time.sleep(waiting_time)
        sys.stderr.write('\n')
        continue
      else:
        sys.stderr.write('\n')
        raise
    response = json.loads(conn.read())
    conn.close()
    # write details for each block to disk
    for data in response:
      user_id = data['id_str']
      name = data['screen_name']
      url = data['url'] if data['url'] is not None else ''
      avatar = data['profile_image_url'] if data['profile_image_url'] is not None else ''
      if avatar != '':
        try:
          avatar = _fetch_img(user_id, avatar)
        except:
          avatar = ''
      location = data['location'].replace(',', ' ').replace('\n', ' ').replace('\r', ' ')
      f.write(user_id+ \
        ','+name+ \
        ','+('mention' if args.query and name.lower() not in users else type)+ \
        ','+str(data['friends_count'])+ \
        ','+str(data['followers_count'])+ \
        ','+str(data['listed_count'])+ \
        ','+str(data['statuses_count'])+ \
        ','+data['created_at']+ \
        ','+url.encode('ascii', 'replace')+ \
        ','+avatar+ \
        ','+location.encode('ascii', 'replace')+'\n')

# used for igraph visualization
def _palette(rng):
  import colorsys
  HSV_tuples = [(t*1.0/rng, 0.75, 1.0) for t in range(rng)]
  return map(lambda x: colorsys.hsv_to_rgb(*x), HSV_tuples)

# requires igraph
def _draw(args):
  import igraph
  g = igraph.load('_'.join(args.screen_name)+EDG_EXT)
  sys.stdout.write('%s Handles\n%s Follow Relationships\n' % (g.vcount(), g.ecount()))
  sys.stdout.write('Avg Shortest Path = %.6f\n' % g.average_path_length())
  sys.stdout.write('In-Degree Distribution mean = %.6f, sd = %.6f\n' % (g.degree_distribution(mode=igraph.IN).mean, g.degree_distribution(mode=igraph.IN).sd))
  sys.stdout.write('Clustering Coefficient = %.6f\n' % g.transitivity_undirected())
  sys.stdout.write('Degree Assortativity = %.6f\n' % g.assortativity_degree())
  width = height = int(750+g.vcount()*4.73-g.vcount()**2*1.55e-3)
  comp = g.clusters(igraph.STRONG if args.strong else igraph.WEAK)
  gc = comp.giant()
  if len(args.screen_name) > 1 and not args.strong:
    RGB_dict = dict(zip(SHAPES, _palette(len(SHAPES))))
    for v in g.vs:
      v['color'] = RGB_dict[v['shape']]
  else:
    for v in g.vs:
      v['color'] = (.8, .8, .8)
    if gc.ecount() > 0:
      gc_size = gc.vcount()
      sys.stdout.write('Fraction Handles in Giant Component = %.2f%%\n' % (100.0*gc_size/g.vcount()))
      sys.stdout.write('GC Diameter %i (unweighted)\n' % gc.diameter())
      gc_vert = comp[comp.sizes().index(gc_size)]
      cim = gc.community_infomap(edge_weights=gc.es['weight'], vertex_weights=gc.vs['lfr'])
      RGB_tuples = _palette(max(cim.membership)+1)
      for v in g.vs:
        gc_v = gc.vs.select(id_eq=v['id'])
        if len(gc_v) > 0:
          v['color'] = RGB_tuples[cim.membership[gc_v[0].index]]
      g.es['color'] = [g.vs['color'][e.target] for e in g.es]
      E = max(g.es['weight'])
      g.es['width'] = [(not args.transparent)*max(1, 10*e['weight']/E) for e in g.es]
      g.es['arrow_size'] = [max(1, 3*e['weight']/E) for e in g.es]
  ID = max(g.vs.indegree()) or 1
  g.vs['label_size'] = [max(12, 36*v.indegree()/ID) for v in g.vs]
  for v in g.vs:
    if v['type'] == 'mention':
      v['label_color'] = 'blue'
    filename = '%s/%s%s' % (FDAT_DIR, v['userid'], FDAT_EXT)
    if not os.path.isfile(filename):
      v['color'] = (0, 0, 0)
  filename = '_'.join(args.screen_name) + \
    ('_s' if args.strong else '_w') + \
    ('_t' if args.transparent else '') + \
    '.' + args.format
  igraph.plot(g, filename, layout=g.layout(args.layout), bbox=(width, height), margin=50)

def _days(created):
  t = time.localtime()
  c = time.strptime(created, "%a %b %d %H:%M:%S +0000 %Y")
  return datetime.timedelta(seconds=(time.mktime(t)-time.mktime(c))).days

# support commenting out lines in file
def _skip_hash(iterable):
  for line in iterable:
    if not line.startswith('#'):
      yield line

# generate GML file readable by Gephi or igraph
def edgelist(args):
  dat = {}
  i = 0
  for sn in args.screen_name:
    for item in csv.reader(_skip_hash(open(sn+TT_EXT))):
      if not dat.has_key(item[0]):
        dat[item[0]] = [i]+item[1:11]+[sn]
        i = i + 1
        # key: user id
        # col 0: GML id
        # col 1: screen name
        # col 2: friends or followers or list name
        # col 3: friend count
        # col 4: follower count
        # col 5: listed count
        # col 6: tweets
        # col 7: date joined
        # col 8: url
        # col 9: avatar
        # col 10: location
        # col 11: handle arg
  id0 = {user_id: val[0] for user_id, val in dat.iteritems() if val[1] in args.screen_name}
  e = open('_'.join(args.screen_name)+EDG_EXT, 'w')
  e.write('graph [\n  directed 1\n')
  for user_id, val in dat.iteritems():
    if not args.ego and val[1] in args.screen_name:
      continue
    if not args.missing and not os.path.isfile(FDAT_DIR+'/'+str(user_id)+FDAT_EXT):
      continue
    e.write('''  node [
    id %s
    userid "%s"
    file "%s.dat"
    label "%s"
    image "%s"
    type "%s"
    statuses %s
    friends %s
    followers %s
    listed %s''' % (val[0], user_id, val[11], val[1], os.path.abspath(val[9]), val[2], val[6], val[3], val[4], val[5]))
    ffr = (float(val[4])/float(val[3])) if float(val[3]) > 0 else 0
    lfr = (10*float(val[5])/float(val[4])) if float(val[4]) > 0 else 0
    e.write('\n    ffr %.4f' % ffr)
    e.write('\n    lfr %.4f' % lfr)
    if len(args.screen_name) > 1:
      sid = [i for i,s in enumerate(args.screen_name) if s == val[11]][0]
      e.write('\n    shape "%s"' % SHAPES[sid])
    else:
      if ffr > 1:
        e.write('\n    shape "triangle-up"')
      else:
        e.write('\n    shape "triangle-down"')
    e.write('\n  ]\n')
  for id1, val in dat.iteritems():
    if id1 in id0.keys():
      continue
    if not args.missing and not os.path.isfile(FDAT_DIR+'/'+str(id1)+FDAT_EXT):
      continue
    if args.ego:
      for idzero in id0.values():
        d = _days(val[7])
        e.write('''  edge [
    source %s
    target %s
    weight %s
  ]
''' % (idzero, val[0], float(val[6])/d if float(d) > 0 else 0))
    filename = FDAT_DIR+'/'+str(id1)+FDAT_EXT
    if os.path.isfile(filename):
      fdat = open(filename).readlines()
      for line in fdat:
        id2 = line.strip()
        if dat.has_key(id2):
          if not args.ego and id2 in id0.keys():
            continue
          if not args.missing and not os.path.isfile(FDAT_DIR+'/'+str(id2)+FDAT_EXT):
            continue
          d = _days(dat[id2][7])
          e.write('''  edge [
    source %s
    target %s
    weight %s
  ]
''' % (val[0], dat[id2][0], float(dat[id2][6])/d if float(d) > 0 else 0))
    else:
      sys.stdout.write('Missing data for %s\n' % dat[id1][1])
  e.write(']\n')
  e.close()
  sys.stdout.write('GML file created.\n')
  try:
    if len(args.screen_name) > 1:
      sys.stdout.write('Shapes mapping: %s -> %s\n' % (args.screen_name, SHAPES[0:len(args.screen_name)]))
    _draw(args)
  except:
    sys.stderr.write('Visualization skipped.\n')
    raise

# retrieve friends
def fetch(args):
  dat = [item for item in csv.reader(_skip_hash(open(args.screen_name+TT_EXT)))]
  if not os.path.exists(FDAT_DIR):
    os.makedirs(FDAT_DIR)
  cursor = res = None
  while len(dat) > 0:
    if cursor is None:
      item = dat.pop()
    if int(item[3]) > args.count:
      sys.stdout.write('Skipping %s (%s %s)\n' % (item[1], item[3], 'friends'))
      continue
    filename = FDAT_DIR+'/'+str(item[0])+FDAT_EXT
    if args.force or not os.path.isfile(filename):
      sys.stdout.write('Processing %s...\n' % item[0])
      try:
        bag = _ids('friends', 'user_id='+str(item[0]), cursor, res)
        cursor = res = None
      except (CursorError, socket.error) as e:
        if hasattr(e, 'code') and e.code in SKIP_CODES:
          sys.stdout.write('HTTPError %s with %s. Skipping...\n' % \
            (e.code, item[0]))
          cursor = res = None
          continue
        elif (hasattr(e, 'code') and e.code in RETRY_CODES) or \
            (hasattr(e, 'errno') and e.errno in (errno.ECONNRESET, errno.EAGAIN)):
          sys.stderr.write('HTTPError %s with %s. Deferred...' % \
            (e.code, item[0]))
          sys.stderr.flush()
          dat.append(item)
          time.sleep(random.randint(10,30))
          cursor = res = None
          sys.stderr.write('\n')
          continue
        elif hasattr(e, 'code') and e.code in WAIT_CODES:
          waiting_time = int(e.headers['x-rate-limit-reset']) - int(round(time.time())) + random.randint(10,30)
          sys.stderr.write('HTTPError %s at %s. Waiting %sm to resume (%s items left)...' % (e.code, time.strftime('%H:%M', time.localtime()), waiting_time/60, len(dat)+1))
          sys.stderr.flush()
          time.sleep(waiting_time)
          cursor = e.cursor
          res = e.res
          sys.stderr.write('\n')
          continue
        else:
          sys.stderr.write('\n')
          raise
      f = open(filename, 'w')
      for item in bag:
        f.write(str(item)+'\n')
      f.close()

def _destroy(screen_name):
  filename = screen_name+FAV_EXT
  dat = [line[:20].strip() for line in open(filename, 'rU')]
  while len(dat) > 0:
    item = dat.pop()
    sys.stdout.write('Purging %s...\n' % item)
    try:
      url = 'https://api.twitter.com/1.1/favorites/destroy.json'
      data = {'id': item}
      conn = urllib2.urlopen(url, urllib.urlencode(data))
    except urllib2.HTTPError, e:
      if e.code in SKIP_CODES:
        sys.stdout.write('HTTPError %s with %s. Skipping...\n' % (e.code, item))
        continue
      elif e.code in RETRY_CODES:
        sys.stderr.write('HTTPError %s. Deferred...' % e.code)
        sys.stderr.flush()
        time.sleep(random.randint(10,30))
        dat.append(item)
        sys.stderr.write('\n')
        continue
      elif e.code in WAIT_CODES:
        waiting_time = int(e.headers['x-rate-limit-reset']) - int(round(time.time())) + random.randint(10, 30)
        sys.stderr.write('HTTPError %s at %s. Waiting %sm to resume (%s items left)...' % (e.code, time.strftime('%H:%M', time.localtime()), waiting_time/60, len(dat)+1))
        sys.stderr.flush()
        time.sleep(waiting_time + random.randint(10,30))
        dat.append(item)
        sys.stderr.write('\n')
        continue
      else:
        sys.stderr.write('\n')
        raise
    data = json.loads(conn.read())
    conn.close()

# retrieve and optionally purge favorites
def likes(args):
  if args.purge:
    sys.stderr.write('Press any key to purge likes or Ctrl-C to abort...')
    sys.stderr.flush()
    while not sys.stdin.read(1):
      pass
    _destroy(args.screen_name)
  else:
    filename = args.screen_name+FAV_EXT
    f = codecs.open(filename, 'w', encoding='utf-8')
    max_id = None
    sys.stderr.write('Fetching likes')
    sys.stderr.flush()
    while True:
      try:
        url = 'https://api.twitter.com/1.1/favorites/list.json?count=%s&screen_name=%s'
        if max_id:
          url += '&max_id=%i' % max_id
        conn = urllib2.urlopen(url % (200, args.screen_name))
      except urllib2.HTTPError, e:
        if e.code in RETRY_CODES:
          sys.stderr.write('\nHTTPError %s. Retrying' % e.code)
          sys.stderr.flush()
          time.sleep(random.randint(10,30))
          sys.stderr.write('\n')
          continue
        elif e.code in WAIT_CODES:
          waiting_time = int(e.headers['x-rate-limit-reset']) - int(round(time.time())) + random.randint(10,30)
          sys.stderr.write('\nHTTPError %s at %s. Waiting %sm to resume...' % \
            (e.code, time.strftime('%H:%M', time.localtime()), waiting_time/60))
          sys.stderr.flush()
          time.sleep(waiting_time)
          sys.stderr.write('\n')
          continue
        else:
          sys.stderr.write('\n')
          raise
      data = json.loads(conn.read())
      conn.close()
      sys.stderr.write('.')
      sys.stderr.flush()
      if len(data) == 0:
        sys.stderr.write('\n')
        return
      for tweet in data:
        if tweet['id'] == max_id:
          if len(data) == 1:
            sys.stderr.write('\n')
            return
          else:
            continue
        max_id = min(max_id, tweet['id']) or tweet['id']
        f.write('%-20s %30s %-12s @%-20s %s\n' % ( \
          tweet['id_str'], \
          tweet['created_at'], \
          tweet['user']['id_str'], \
          tweet['user']['screen_name'], \
          tweet['text'].replace('\n', ' ')))

# retrieve tweets
def tweets(args):
  f = codecs.open(args.screen_name+TWT_EXT, 'w', encoding='utf-8')
  max_id = None
  sys.stderr.write('Fetching tweets')
  sys.stderr.flush()
  while True:
    try:
      if args.query:
        url = 'https://api.twitter.com/1.1/search/tweets.json?q=%s&result_type=recent&count=100&tweet_mode=extended'
      elif args.l:
        url = 'https://api.twitter.com/1.1/lists/statuses.json?slug='+args.l+'&owner_screen_name=%s&count=100&tweet_mode=extended'
      else:
        url = 'https://api.twitter.com/1.1/statuses/user_timeline.json?screen_name=%s&count=200&tweet_mode=extended'
      if max_id:
        url += '&max_id=%i' % max_id
      conn = urllib2.urlopen(url % args.screen_name)
    except urllib2.HTTPError, e:
      if e.code in RETRY_CODES:
        sys.stderr.write('\nHTTPError %s. Retrying' % e.code)
        sys.stderr.flush()
        continue
      elif e.code in WAIT_CODES:
        waiting_time = int(e.headers['x-rate-limit-reset']) - int(round(time.time())) + random.randint(10,30)
        sys.stderr.write('\nHTTPError %s at %s. Waiting %sm to resume...' % \
          (e.code, time.strftime('%H:%M', time.localtime()), waiting_time/60))
        sys.stderr.flush()
        time.sleep(waiting_time)
        sys.stderr.write('\n')
        continue
      else:
        sys.stderr.write('\n')
        raise
    data = json.loads(conn.read())
    if args.query:
      data = data['statuses']
    conn.close()
    sys.stderr.write('.')
    sys.stderr.flush()
    if len(data) == 0:
      sys.stderr.write('\n')
      return
    for tweet in data:
      if tweet['id'] == max_id:
        if len(data) == 1:
          sys.stderr.write('\n')
          return
        else:
          continue
      max_id = min(max_id, tweet['id']) or tweet['id']
      if args.query or args.l:
        f.write('%30s @%-20s %s\n' % (tweet['created_at'], \
          tweet['user']['screen_name'], tweet['full_text'].replace('\n', ' ')))
      else:
        f.write('%30s %s\n' % (tweet['created_at'], tweet['full_text'].replace('\n', ' ')))

# find out remaining calls left by endpoint
class StatsAction(argparse.Action):
  def __call__(self, parse, namespace, values, option_string=None):
    _replace_opener()
    url = 'https://api.twitter.com/1.1/application/rate_limit_status.json?resources=users,friends,statuses,search,favorites'
    conn = urllib2.urlopen(url)
    data = json.loads(conn.read())
    conn.close()
    sys.stdout.write('init (%s), ' % \
      data['resources']['users']['/users/lookup']['remaining'])
    sys.stdout.write('fetch (%s), ' % \
      data['resources']['friends']['/friends/ids']['remaining'])
    sys.stdout.write('tweets (%s, -q %s), ' % ( \
      data['resources']['statuses']['/statuses/user_timeline']['remaining'], \
      data['resources']['search']['/search/tweets']['remaining']))
    sys.stdout.write('resolve (%s), ' % \
      data['resources']['users']['/users/show/:id']['remaining'])
    sys.stdout.write('likes (%s)\n' % \
      data['resources']['favorites']['/favorites/list']['remaining'])
    sys.exit(0)

class QuoteAction(argparse.Action):
  def __call__(self, parse, namespace, values, option_string=None):
    if isinstance(values, list):
      setattr(namespace, self.dest, map(lambda v: urllib.quote(v), values))
    else:
      setattr(namespace, self.dest, urllib.quote(values))

def main():
  parser = argparse.ArgumentParser(description='Twitter Collection Tool')
  sp = parser.add_subparsers(dest='cmd', title='sub-commands')

  sp_resolve = sp.add_parser('resolve', \
    help='retrieve user_id for screen_name or vice versa')
  sp_resolve.add_argument(dest='screen_name', nargs='+', action=QuoteAction, \
    help='Twitter screen name')
  sp_resolve.set_defaults(func=resolve)

  sp_init = sp.add_parser('init', \
    help='retrieve friends data for screen_name')
  sp_init.add_argument('-o', '--followers', action='store_true', \
    help='retrieve followers (default: friends)')
  sp_init.add_argument('-q', '--query', action='store_true', \
    help='extract handles from query (default: screen_name)')
  sp_init.add_argument('-n', '--nomention', action='store_true', \
    help='ignore mentions from query (default: false)')
  sp_init.add_argument('-m', '--members', dest='l', \
    help='extract member handles from list named L')
  sp_init.add_argument('-f', '--force', action='store_true', \
    help='ignore existing %s file (default: False)' % TT_EXT)
  sp_init.add_argument(dest='screen_name', action=QuoteAction, \
    help='Twitter screen name')
  sp_init.set_defaults(func=init)

  sp_fetch = sp.add_parser('fetch', \
    help='retrieve friends of handles in %s file' % TT_EXT)
  sp_fetch.add_argument('-f', '--force', action='store_true', \
    help='ignore existing %s files (default: False)' % FDAT_EXT)
  sp_fetch.add_argument('-c', dest='count', type=int, default=FMAX, \
    help='skip if friends above count (default: %(FMAX)i)' % globals())
  sp_fetch.add_argument(dest='screen_name', action=QuoteAction, \
    help='Twitter screen name')
  sp_fetch.set_defaults(func=fetch)

  sp_tweets = sp.add_parser('tweets', \
    help='retrieve tweets')
  sp_tweets.add_argument('-q', '--query', action='store_true', \
    help='argument is a query (default: screen_name)')
  sp_tweets.add_argument('-m', '--members', dest='l', \
    help='extract tweets from list named L')
  sp_tweets.add_argument(dest='screen_name', action=QuoteAction, \
    help='Twitter screen name')
  sp_tweets.set_defaults(func=tweets)

  sp_likes = sp.add_parser('likes', \
    help='retrieve likes')
  sp_likes.add_argument('-p', '--purge', action='store_true', \
    help='destroy likes (default: False)')
  sp_likes.add_argument(dest='screen_name', action=QuoteAction, \
    help='Twitter screen name')
  sp_likes.set_defaults(func=likes)

  sp_edgelist = sp.add_parser('edgelist', \
    help='generate graph in GML format', \
    description='Vertices have the following attributes: twitter user_id, .dat source file, label, link to image file, type, statuses, friends, followers, listed, ffr and lfr. Vertices with a friends-to-followers ratio > 1 have a triangle-up shape, otherwise triangle-down. Label size is proportional to the in-degree. Edges have a width based on their weight attribute. Weight corresponds to the number of tweets per day since the account was created. Vertices included in the giant component are colored according to their membership to a particular community. Otherwise, they are colored in grey. Community finding is based on infomap and applied to members of the giant component. Black is used for vertices with no edges, with more than %s friends or set to private.' % FMAX)
  sp_edgelist.add_argument('-f', '--format', default='png', \
    choices=('png', 'pdf', 'ps'), \
    help='graph output format (default: png)')
  sp_edgelist.add_argument('-e', '--ego', action='store_true', \
    help='include screen_name (default: False)')
  sp_edgelist.add_argument('-l', '--layout', default='kk', \
    choices=('circle', 'fr', 'kk'), \
    help='igraph layout (default: kk)')
  sp_edgelist.add_argument('-s', '--strong', action='store_true', \
    help='use strong ties to isolate clusters (default: weak)')
  sp_edgelist.add_argument('-t', '--transparent', action='store_true', \
    help='hide edges in graph (default: visible)')
  sp_edgelist.add_argument(dest='screen_name', nargs='+', action=QuoteAction, \
    help='Twitter screen name')
  sp_edgelist.add_argument('-m', '--missing', action='store_true', \
    help='include missing handles (default: False)')
  sp_edgelist.set_defaults(func=edgelist)

  parser.add_argument('-s', '--stats', action=StatsAction, nargs=0, \
    help='show Twitter throttling stats and exit')
  parser.add_argument('-v', '--version', action='version', \
    version='%(prog)s v'+'%(__version__)s' % globals())

  try:
    args = parser.parse_args()
    _replace_opener()
    args.func(args)
  except KeyboardInterrupt:
    sys.stderr.write('\n')
    return 2
  except Exception, err:
    sys.stderr.write(str(err)+'\n')
    return 1
  else:
    sys.stderr.write('Done.\n')
    return 0

if __name__ == '__main__':
  sys.exit(main())