YiLi225

# Install: !pip install pyspellchecker
from spellchecker import SpellChecker
spell = SpellChecker()

# Text from 122001 of the wikiText data; modified to inculde typos: 
# 'commentary' -> 'commentyra', 'gimmick' -> 'gimimick'; 
# curText = doc_set[122001]
# Use the spellchecker to identify and correct the typos
correctDict = {}
for val in re.split(r'[^\w]', curText): 

YiLi225

# Replace with the correct words
from functools import reduce
reduce(lambda x, y: x.replace(*y), [curText, *list(correctDict.items())])

YiLi225

# Install: !pip install next_word_prediction
# curText = doc_set[1135957]
from next_word_prediction import GPT2
gpt2 = GPT2()

# Predict the next word after 'vegetarian'
gpt2.predict_next(curText.replace('[MASK]', ''), 5)

YiLi225

## Top 5 next words
['dishes', 'and', ',', '-', 'recipes']

## Comparing with the original text
falafel has become popular among vegetarians and vegans , 
as an alternative to meat @-@ laden street foods , 
and is now sold in packaged mixes in health @-@ food stores .
falafel 's versatility has allowed for the reformulating of recipes for meatloaf , 
sloppy joes and spaghetti and meatballs into vegetarian [dishes] .

YiLi225

from wordcloud import WordCloud, ImageColorGenerator
from nltk.corpus import stopwords
from nltk.util import ngrams
import nltk

def replace(match):
    return swMapping[match.group(0)]
## Define stopwords
curSW = stopwords.words('english')
curSW += ['unk']

YiLi225

import re
# curText = doc_set[122001]
curText = '''
six months ago we thought it would be a fun idea to release our album on election day but this is not the election to be cute . 
we felt as though rather than making a commentary we were only riding the wave of the election . this seemed less and less like what we 
intended to do and more of a gimmick . 
'''
## Replace the words 'day', 'we', 'is'
print(' '.join(re.sub(r'|'.join(map(re.escape, ['day', 'we', 'is'])), ' ', curText).split()))

YiLi225

## Query to pull data
timeSeries_query = """
SELECT 
	InvoiceDate, 
	Description,
	ROUND(SUM(Quantity * UnitPrice), 2) AS Total_Sale_Amt  
FROM 
	OnlineRetail
WHERE InvoiceDate BETWEEN '2010-12-01' AND '2011-01-01'
	AND Description IN ('CHOCOLATE HOT WATER BOTTLE', 'GREY HEART HOT WATER BOTTLE')

YiLi225

import streamlit as st
st.title("Online Retail")
## !!! date options
date = st.sidebar.selectbox(
            "Select a Invoice Date Range",
            [
              "'2010-12-01' and '2011-01-01'", 
              "'2010-12-01' and '2010-12-15'"              
            ])
## !!! product options           

YiLi225

## Query with placeholders
timeSeries_query = f"""
SELECT 
	InvoiceDate, 
	Description,
	ROUND(SUM(Quantity * UnitPrice), 2) AS Total_Sale_Amt  
FROM 
	OnlineRetail
WHERE InvoiceDate BETWEEN {date}           ## !!! Placeholder for date
	AND Description IN ('{product}')   ## !!! Placeholder for product 

YiLi225

## Read in external SQL file 
sqlFile = open(r'...\OnlineRetailPull.sql', 'r')
myQuery = sqlFile.read()
OnlineRetailData = pd.read_sql_query(myQuery, engine)