Slater-Victoroff

import os

CREDENTIALS = ['AWSAccessKeyId', 'AWSSecretKey']


def credentials(filename=None):
	"""
	Generally made for standard AWS export format. Turns conf file into python dict
 
	If no filename is given, will attempt to read environment variables instead, then global variables

Slater-Victoroff

@BotoSave('bucketname')
def method():
	return {<key>: <data>}

Slater-Victoroff

import json

from boto.s3.connection import S3Connection, OrdinaryCallingFormat
from boto.s3.key import Key


class BotoSave(object):
 
	def __init__(self, bucket, filename=None):
		"""

Slater-Victoroff

class MissingKeyException(Exception):
	"""
	If we can't figure out what the key is of an item being saved, we just say so.
 
	Should probably make a better error message that says what we can figure out and what they can do about it.
	"""
 
	def __init__(self, data_structure):
		self.ds = data_structure
 

Slater-Victoroff

import requests
import json
import urllib

URLS = {
	'css': 'http://cssminifier.com/raw',
	'js': 'http://javascript-minifier.com/raw'
}

def new_filepath(filepath):

Slater-Victoroff

def get_voter_links(outfile="voter_info.txt"):
    start_urls = ("http://usavoters.directory/complete.php?id=%s" % i for i in xrange(128555, 214545))
    with open(outfile, 'a') as sink:
        for url in start_urls:
            document = etree.HTML(requests.get(url).content)
            link_selector = CSSSelector('tr>td>a')
            person_links = link_selector(document)[14:]
            sink.write('\n'.join(link.get('href') for link in person_links if link.get('href')))
            print url

Slater-Victoroff

id	title	description	google_product_category	product_type	price	sale_price	sale_price_effective_date	link	mobile_link	image_link	additional_image_link	brand	gtin	mpn	identifier_exists	condition	availability	availability_date	item_group_id	color	material	pattern	size	size_type	size_system	gender	age_group	tax	shipping	shipping_weight	shipping_label	multipack	is_bundle	adult	adwords_redirect	adwords_grouping	adwords_labels	custom_label_0	custom_label_1	custom_label_2	custom_label_3	custom_label_4	excluded_destination	expiration_date	promotion_id	display_ads_id	display_ads_link	display_ads_similar_id	display_ads_title	display_ads_value	shipping_length	shipping_width	shipping_height
sku152300450	Washable-Crepe Straight-Leg Pants, Petite, Size: PS (6/8), BLACK - Eileen Fisher	Eileen Fisher Washable-Crepe Straight-Leg Pants, Petite Details From Eileen Fisher, bi-stretch crepe pants with elegant day-to-night texture and remarkable fit memory, great for travel. 29" approx. inseam. Regular rise; yoked waist contours t

Slater-Victoroff

{"http://images.neimanmarcus.com/product_assets/T/8/A/P/X/NMT8APX_mz.jpg": [0.0, 0.0, 0.7503045797348022, 0.0, 4.050353527069092, 0.0, 0.0, 3.970454454421997, 2.56343936920166, 0.0, 0.0, 0.0, 0.0, 0.8763924241065979, 2.9059646129608154, 1.7854936122894287, 0.0, 0.0, 0.0, 3.603835105895996, 0.0, 0.0, 0.0, 0.0, 2.9097509384155273, 0.6788361072540283, 0.0, 0.0, 0.0, 0.0, 0.9168252348899841, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.21072697639465332, 0.0, 2.585055112838745, 0.0, 0.18340826034545898, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 4.772193431854248, 0.0, 0.0, 0.0, 0.0, 1.1300804615020752, 0.0, 0.0, 0.0, 1.8126261234283447, 0.0, 0.0, 2.4351160526275635, 1.0090645551681519, 6.7620415687561035, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.6218217611312866, 0.0, 3.785642385482788, 0.0, 0.0, 0.0, 0.4094317555427551, 0.0, 1.046784520149231, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.5131956338882446, 0.0, 2.341181516647339, 3.7282121181488037, 2.7647757530212402, 0.0, 0.0, 0.2037421613931656, 0.0, 0.0, 2.629900932

Slater-Victoroff

import json
from collections import OrderedDict

import numpy as np
from scipy.spatial.distance import cdist
from indicoio.custom import vectorize
from nouns import NOUNS

FEATURES = json.load(open("features.json"), object_pairs_hook=OrderedDict)

Slater-Victoroff

def firstNonRepeated(s):
    return ([letter for i, letter in enumerate(s) if letter not in (s[:i] + s[i+1:])] + [""])[0]

def first_non_repeated1(s):
    for i, letter in enumerate(s):
	if letter not in (s[:i] + s[i+1:]):
	    return letter
    return ""

def first_non_repeated1(s):