Spaces:
Runtime error
Runtime error
Dy
commited on
Commit
•
e7e3fea
1
Parent(s):
298dbbb
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,231 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
from bs4 import BeautifulSoup
|
3 |
+
import gradio as gr
|
4 |
+
import openai
|
5 |
+
import requests
|
6 |
+
from langchain import OpenAI, ConversationChain, LLMChain, PromptTemplate
|
7 |
+
from langchain.memory import ConversationBufferWindowMemory
|
8 |
+
from langchain.chains import LLMChain
|
9 |
+
from langchain.agents import load_tools, initialize_agent
|
10 |
+
from langchain.chat_models import ChatOpenAI
|
11 |
+
from langchain.output_parsers import CommaSeparatedListOutputParser
|
12 |
+
from langchain.prompts import PromptTemplate, ChatPromptTemplate, HumanMessagePromptTemplate
|
13 |
+
from langchain.llms import OpenAI
|
14 |
+
from collections import Counter
|
15 |
+
import pandas as pd
|
16 |
+
|
17 |
+
OPENAI_API_KEY = os.environ['OPENAI_API_KEY']
|
18 |
+
GOOGLE_MAPS_API = os.environ['GOOGLE_MAPS_API']
|
19 |
+
|
20 |
+
#### TAB 1 ####
|
21 |
+
|
22 |
+
def get_location_data(search_term, location):
|
23 |
+
# First, we get the latitude and longitude coordinates of the location
|
24 |
+
url = "https://maps.googleapis.com/maps/api/geocode/json"
|
25 |
+
params = {
|
26 |
+
"address": location,
|
27 |
+
"key": GOOGLE_MAPS_API
|
28 |
+
}
|
29 |
+
response = requests.get(url, params=params)
|
30 |
+
location_data = response.json()["results"][0]["geometry"]["location"]
|
31 |
+
|
32 |
+
# Next, we use the Places API nearbysearch endpoint to find places matching the search term
|
33 |
+
url = "https://maps.googleapis.com/maps/api/place/nearbysearch/json"
|
34 |
+
params = {
|
35 |
+
"location": f"{location_data['lat']},{location_data['lng']}",
|
36 |
+
"radius": "10000", # 10km radius
|
37 |
+
#"type": search_term,
|
38 |
+
"keyword" : search_term,
|
39 |
+
"key": GOOGLE_MAPS_API
|
40 |
+
}
|
41 |
+
response = requests.get(url, params=params)
|
42 |
+
results = response.json()["results"]
|
43 |
+
|
44 |
+
# We only want the first 5 results
|
45 |
+
results = results[:5]
|
46 |
+
|
47 |
+
# For each result, we get the place details to retrieve the description and top reviews
|
48 |
+
locations = []
|
49 |
+
for result in results:
|
50 |
+
place_id = result["place_id"]
|
51 |
+
url = "https://maps.googleapis.com/maps/api/place/details/json"
|
52 |
+
params = {
|
53 |
+
"place_id": place_id,
|
54 |
+
"fields": "name,formatted_address,formatted_phone_number,rating,review",
|
55 |
+
"key": GOOGLE_MAPS_API
|
56 |
+
}
|
57 |
+
response = requests.get(url, params=params)
|
58 |
+
place_details = response.json()["result"]
|
59 |
+
|
60 |
+
# Create a dictionary representing the location and add it to the list
|
61 |
+
location_dict = {
|
62 |
+
"name": place_details["name"],
|
63 |
+
"address": place_details["formatted_address"],
|
64 |
+
#"phone_number": place_details.get("formatted_phone_number", "N/A"),
|
65 |
+
#"rating": place_details.get("rating", "N/A"),
|
66 |
+
"reviews": []
|
67 |
+
}
|
68 |
+
|
69 |
+
# Add the top 3 reviews to the dictionary
|
70 |
+
reviews = place_details.get("reviews", [])
|
71 |
+
for review in reviews[:3]:
|
72 |
+
review_dict = {
|
73 |
+
#"author": review["author_name"],
|
74 |
+
#"rating": review["rating"],
|
75 |
+
"text": review["text"],
|
76 |
+
#"time": review["relative_time_description"]
|
77 |
+
}
|
78 |
+
location_dict["reviews"].append(review_dict)
|
79 |
+
|
80 |
+
locations.append(location_dict)
|
81 |
+
|
82 |
+
return locations
|
83 |
+
|
84 |
+
# Define the function to be used in the Gradio app
|
85 |
+
def find_competitors(product, location):
|
86 |
+
locations = get_location_data(product, location)
|
87 |
+
if len(locations) == 0:
|
88 |
+
return f"No competitors found for {product} in {location}."
|
89 |
+
|
90 |
+
output_str = f"Top competitors for {product} in {location}:"
|
91 |
+
for i, loc in enumerate(locations):
|
92 |
+
output_str += f"\n{i+1}. {loc['name']}"
|
93 |
+
output_str += f"\nAddress: {loc['address']}"
|
94 |
+
#output_str += f"\nPhone number: {loc['phone_number']}"
|
95 |
+
#output_str += f"\nRating: {loc['rating']}"
|
96 |
+
output_str += f"\nTop 3 reviews:"
|
97 |
+
for review in loc['reviews']:
|
98 |
+
output_str += f"\n- {review['text']}"
|
99 |
+
#output_str += f"\n Author: {review['author']}"
|
100 |
+
#output_str += f"\n Rating: {review['rating']}"
|
101 |
+
#output_str += f"\n Time: {review['time']}"
|
102 |
+
|
103 |
+
output_str2 = f"Top competitors for {product} in {location}:"
|
104 |
+
for i, loc in enumerate(locations):
|
105 |
+
output_str2 += f"\n{i+1}. {loc['name']}"
|
106 |
+
output_str2 += f"\nAddress: {loc['address']}"
|
107 |
+
|
108 |
+
#return output_str
|
109 |
+
|
110 |
+
prompt_input = '''
|
111 |
+
You are an expert management consultant that rivals the best of Mckinsey, Bain, BCG.
|
112 |
+
The client wants to sell {} in {}.
|
113 |
+
{}
|
114 |
+
Provide an analysis of the following:
|
115 |
+
- From the competition and reviews about its products and come up with creative insights to recommend the client execute as part of a differentiating business strategy.
|
116 |
+
- From there, think step by step, explain 5 strategies in bullet points of a creative and effective business plan.
|
117 |
+
- Suggest a location for the client and explain the rationale of this locatioin.
|
118 |
+
- Let us think step by step.
|
119 |
+
'''.format(product, location, output_str)
|
120 |
+
|
121 |
+
template = '''
|
122 |
+
{history}
|
123 |
+
{human_input}
|
124 |
+
'''
|
125 |
+
prompt = PromptTemplate(
|
126 |
+
input_variables=["history", "human_input"],
|
127 |
+
template=template
|
128 |
+
)
|
129 |
+
|
130 |
+
chatgpt_chain = LLMChain(
|
131 |
+
llm=ChatOpenAI(model="gpt-3.5-turbo", temperature=0.5,openai_api_key=OPENAI_API_KEY),
|
132 |
+
prompt=prompt,
|
133 |
+
verbose=True,
|
134 |
+
memory=ConversationBufferWindowMemory(k=10),
|
135 |
+
)
|
136 |
+
|
137 |
+
output = output_str2 + "\n\n" + chatgpt_chain.predict(human_input=prompt_input)
|
138 |
+
|
139 |
+
return(output)
|
140 |
+
|
141 |
+
# Create the Gradio app interface
|
142 |
+
inputs = [
|
143 |
+
gr.inputs.Textbox(label="Product to research"),
|
144 |
+
gr.inputs.Textbox(label="Location")
|
145 |
+
]
|
146 |
+
|
147 |
+
|
148 |
+
output = gr.outputs.Textbox(label="AI Analysis")
|
149 |
+
|
150 |
+
iface1 = gr.Interface(fn=find_competitors, inputs=inputs, outputs=output, title="Market Research AI",
|
151 |
+
description="Input a product and a location. The AI analyst will help you research nearby competitors, formulate a business plan to differentiate you from your competitors, and recommend a strategic location for your business.")
|
152 |
+
|
153 |
+
#### TAB 2 ####
|
154 |
+
|
155 |
+
template2 = '''
|
156 |
+
{history}
|
157 |
+
{human_input}
|
158 |
+
'''
|
159 |
+
prompt2 = PromptTemplate(
|
160 |
+
input_variables=["history", "human_input"],
|
161 |
+
template=template2
|
162 |
+
)
|
163 |
+
|
164 |
+
chatgpt_chain = LLMChain(
|
165 |
+
llm=ChatOpenAI(model="gpt-3.5-turbo", temperature=0.5,openai_api_key=OPENAI_API_KEY),
|
166 |
+
prompt=prompt2,
|
167 |
+
verbose=True,
|
168 |
+
memory=ConversationBufferWindowMemory(k=10),
|
169 |
+
)
|
170 |
+
|
171 |
+
# Scrape the URL
|
172 |
+
def scrape(url):
|
173 |
+
response = requests.get(url)
|
174 |
+
soup = BeautifulSoup(response.text, "html.parser")
|
175 |
+
|
176 |
+
# Remove script and style elements
|
177 |
+
for script in soup(["script", "style"]):
|
178 |
+
script.extract()
|
179 |
+
|
180 |
+
return soup.get_text()
|
181 |
+
|
182 |
+
# Extract keywords
|
183 |
+
def extract_keywords(prompt_input, num_keywords):
|
184 |
+
|
185 |
+
output= chatgpt_chain.predict(human_input=prompt_input)
|
186 |
+
output_parser = CommaSeparatedListOutputParser()
|
187 |
+
ret_list = output_parser.parse(output)
|
188 |
+
|
189 |
+
return ret_list
|
190 |
+
|
191 |
+
# Define the function to be used in Gradio
|
192 |
+
def keywords_from_url(url, num_keywords):
|
193 |
+
url_text = scrape(url)
|
194 |
+
prompt_input2 = '''
|
195 |
+
You are an expert SEO optimized, consultant and manager.
|
196 |
+
|
197 |
+
Here is the text from a website:
|
198 |
+
|
199 |
+
{}
|
200 |
+
|
201 |
+
From the text above, extract {} SEO keyphrase that are highly valueble in terms of SEO purpose.
|
202 |
+
|
203 |
+
Your response should be a list of comma separated values, eg: `foo, bar, baz
|
204 |
+
'''.format(url_text, num_keywords)
|
205 |
+
|
206 |
+
keywords = extract_keywords(prompt_input2, num_keywords)
|
207 |
+
|
208 |
+
df = pd.DataFrame(keywords, columns=["Keyword"])
|
209 |
+
df.index.name = "Rank"
|
210 |
+
df.index += 1
|
211 |
+
df.to_csv('keywords.csv')
|
212 |
+
|
213 |
+
return "keywords.csv"
|
214 |
+
|
215 |
+
|
216 |
+
|
217 |
+
iface2 = gr.Interface(
|
218 |
+
fn=keywords_from_url,
|
219 |
+
inputs=[gr.inputs.Textbox(label="URL"), gr.inputs.Slider(minimum=1, maximum=50, step=1, default=10, label="Number of SEO Keywords")],
|
220 |
+
outputs=gr.outputs.File(label="Download CSV File"),
|
221 |
+
title="SEO Keyword Extractor",
|
222 |
+
description="Enter a URL and the number of keywords you want to extract from that page. The output will be a CSV file containing the SEO keywords."
|
223 |
+
)
|
224 |
+
|
225 |
+
|
226 |
+
|
227 |
+
#tab1 = gr.Tab("AI Market Research", inputs=iface1.inputs, outputs=iface1.outputs)
|
228 |
+
#tab2 = gr.Tab("SEO Keyword Extractor", inputs=iface2.inputs, outputs=iface2.outputs)
|
229 |
+
|
230 |
+
demo = gr.TabbedInterface([iface2, iface1], ["SEO Keyword Extractor", "AI Market Researcher"])
|
231 |
+
demo.launch()
|