import random
from functools import reduce
import pandas as pd
from fuzzywuzzy import process
from fuzzywuzzy import fuzz
import utils.gennumbers as gennumbers
# ---------------------------------------------------------------------
def fuzzy_match_ents(ents, choices, limit=2, thresh=80):
fuzz_matches_out = []
for ent in ents:
top_matches = process.extract(
ent,
set(choices),
limit=limit,
scorer=fuzz.partial_ratio)
for match, score in top_matches:
if score >= thresh:
fuzz_matches_out.append(match)
return fuzz_matches_out
def gen_rank_translation():
# generate valid chart ranks in number (1) and word (one) forms
numranks = {str(x): x for x in range(1, 101)}
numranks_str = {x: i + 1 for i, x in enumerate(gennumbers.word_nums_100())}
#
# generate ordinal ranks in number (1st) and word forms (first)
ordranks = {x: i + 1 for i, x in enumerate(gennumbers.ordinal_nums_100())}
ordranks_str = [
'first', 'second', 'third', 'fourth', 'fifth',
'sixth', 'seventh', 'eighth', 'ninth', 'tenth',
'eleventh', 'twelfth', 'thirteenth', 'fourteenth', 'fifteenth'
]
ordranks_str = {x: i + 1 for i, x in enumerate(ordranks_str)}
#
word2int = {}
for d in (numranks, numranks_str, ordranks, ordranks_str):
word2int.update(d)
#
return word2int
rank_dict = gen_rank_translation()
def respond(message, interpreter, app_data_path):
app_data = pd.read_csv(app_data_path)
parsed = interpreter.parse(message)
generic_app_response = False
if parsed['intent']['name'] == 'greet':
greetings = ['hey there', 'hi', 'howdy', 'hello']
suggestions = [
'"what rank is snapchat?"',
'"what is the number 1 free app?"',
'"what are the top ranked productivity apps?"',
'"what games are popular?"'
]
greeting = random.choice(greetings)
suggestion = random.choice(suggestions)
response = ['{}, try asking me a question about apps like {}'.format(greeting, suggestion)]
elif parsed['intent']['name'] == 'goodbye':
closings = ['cya!', 'goodbye', 'have a good day!', 'peace out']
response = [random.choice(closings)]
elif parsed['intent']['name'] == 'app_rank_search':
if len(parsed['entities']) > 0:
ents_df = pd.DataFrame(parsed['entities'])
chart_rows = ents_df['entity'] == 'chart'
genre_rows = ents_df['entity'] == 'genre'
app_rows = ents_df['entity'] == 'app'
rank_rows = ents_df['entity'].isin(['ordrank', 'numrank'])
# convert ranks to ints
chart_matches = list(ents_df.loc[chart_rows, 'value'])
genre_matches = list(ents_df.loc[genre_rows, 'value'])
app_matches = list(ents_df.loc[app_rows, 'value'])
chart_filters = fuzzy_match_ents(chart_matches, app_data['chart'])
genre_filters = fuzzy_match_ents(genre_matches, app_data['genre'])
app_filters = fuzzy_match_ents(app_matches, app_data['app'])
rank_filters = list(ents_df.loc[rank_rows, 'value'].apply(lambda x: rank_dict[x]))
df_list = []
if len(chart_filters) > 0:
chart_filtered = app_data.loc[app_data['chart'].isin(chart_filters)]
df_list.append(chart_filtered)
if len(genre_filters) > 0:
genre_filtered = app_data.loc[app_data['genre'].isin(genre_filters)]
df_list.append(genre_filtered)
if len(app_filters) > 0:
app_filtered = app_data.loc[app_data['app'].isin(app_filters)]
df_list.append(app_filtered)
if len(rank_filters) > 0:
rank_filtered = app_data.loc[app_data['rank'].isin(rank_filters)]
df_list.append(rank_filtered)
if len(df_list) > 1:
filt_inds = reduce(lambda x, y: set(x.index).intersection(set(y.index)), df_list)
elif len(df_list) == 1:
filt_inds = set(df_list[0].index)
else:
filt_inds = None
if filt_inds is not None:
query_result = app_data.iloc[list(filt_inds),]
query_result = query_result.sort_values(by=['rank']).reset_index(drop=True)
if query_result.shape[0] > 0:
response_statements = []
responses = [
'{app} is a {genre} app ranked {rank} on the {chart} chart',
'a popular {genre} app is {app}; it\'s ranked {rank} on the {chart} chart',
'number {rank} on the {chart} chart is the {genre} app {app}',
'if you\'re interested in {genre}, then check out {app}. it\'s ranked {rank} among {chart}'
]
response_choice = random.choice(responses)
for i, row in query_result.iterrows():
response_i = response_choice
response_i = response_i.format(
app=row['app'],
rank=row['rank'],
chart=row['chart'],
genre=row['genre'])
response_statements.append(response_i)
if i >= 2:
break
response = response_statements
else:
response = ["i couldn't find anything related to your app search :("]
else:
response = ["i couldn't find anything related to your app search :("]
else:
query_result = app_data.sort_values(by=['rank']).reset_index(drop=True)
response_statements = []
for i, row in query_result.iterrows():
response_i = '- {app} is a {genre} app ranked {rank} on the {chart} chart'
response_i = response_i.format(
app=row['app'],
rank=row['rank'],
chart=row['chart'],
genre=row['genre'])
response_statements.append(response_i)
if i >= 2:
break
response = response_statements
generic_app_response = True
else:
apologies = [
"i didn't understand that",
"i'm not the smartest bot",
"sorry, i didn't catch that"
]
suggestions = [
'"what rank is facebook?"',
'"what is the number 1 top grossing app?"',
'"what are the top ranked medical apps?"',
'"what apps are popular?"'
]
apology = random.choice(apologies)
suggestion = random.choice(suggestions)
response = ['{}, try asking me something like {}'.format(apology, suggestion)]
return response, generic_app_response
