Update app.py
Browse files
app.py
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
-
import requests
|
2 |
import gradio as gr
|
|
|
3 |
import os
|
4 |
|
5 |
# Load API keys securely from environment variables
|
@@ -23,15 +23,57 @@ class EmailAgent:
|
|
23 |
self.company_info = None
|
24 |
self.role_description = None
|
25 |
|
26 |
-
#
|
27 |
-
def
|
28 |
-
print("Reasoning:
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
35 |
|
36 |
# Action: Fetch LinkedIn data via Proxycurl (acting based on reasoning)
|
37 |
def fetch_linkedin_data(self):
|
@@ -79,13 +121,6 @@ class EmailAgent:
|
|
79 |
print(f"Error: Unable to fetch company info via Firecrawl. Using default info.")
|
80 |
self.company_info = "A leading company in its field."
|
81 |
|
82 |
-
# Reflection: Check if we have enough data to generate the email
|
83 |
-
def reflect_on_data(self):
|
84 |
-
print("Reflection: Do we have enough data?")
|
85 |
-
if not self.bio or not self.skills or not self.company_info:
|
86 |
-
print("Warning: Some critical information is missing. Proceeding with default values.")
|
87 |
-
return True
|
88 |
-
|
89 |
# Final Action: Generate the email using Groq Cloud LLM based on gathered data
|
90 |
def generate_email(self):
|
91 |
print("Action: Generating the email with the gathered information.")
|
@@ -133,14 +168,13 @@ class EmailAgent:
|
|
133 |
|
134 |
# Main loop following ReAct pattern
|
135 |
def run(self):
|
136 |
-
self.
|
|
|
|
|
137 |
self.fetch_linkedin_data() # Fetch LinkedIn data
|
138 |
self.fetch_company_info_with_firecrawl() # Fetch company data using Firecrawl
|
139 |
-
|
140 |
-
|
141 |
-
return self.generate_email() # Final action: generate email
|
142 |
-
else:
|
143 |
-
return "Error: Not enough data to generate the email."
|
144 |
|
145 |
# Define the Gradio interface and the main app logic
|
146 |
def gradio_ui():
|
|
|
|
|
1 |
import gradio as gr
|
2 |
+
import requests
|
3 |
import os
|
4 |
|
5 |
# Load API keys securely from environment variables
|
|
|
23 |
self.company_info = None
|
24 |
self.role_description = None
|
25 |
|
26 |
+
# Use the LLM to reason and reflect on the provided data
|
27 |
+
def reason_with_llm(self):
|
28 |
+
print("Reasoning: Using LLM to reason about available data...")
|
29 |
+
|
30 |
+
# LLM reasoning prompt that evaluates the current data and reflects on next actions
|
31 |
+
reasoning_prompt = f"""
|
32 |
+
You are a reasoning agent tasked with generating a job application email. Here's what we have:
|
33 |
+
|
34 |
+
1. Candidate's LinkedIn profile URL: {self.linkedin_url}
|
35 |
+
2. Company Name: {self.company_name}
|
36 |
+
3. Role: {self.role}
|
37 |
+
4. Word Limit: {self.word_limit}
|
38 |
+
5. Candidate's Name: {self.user_name}
|
39 |
+
6. Candidate's Email: {self.email}
|
40 |
+
7. Candidate's Phone: {self.phone}
|
41 |
+
8. Candidate's LinkedIn: {self.linkedin}
|
42 |
+
|
43 |
+
Candidate's Bio: {self.bio}
|
44 |
+
Candidate's Skills: {', '.join(self.skills)}
|
45 |
+
Candidate's Experiences: {', '.join([exp['title'] for exp in self.experiences])}
|
46 |
+
|
47 |
+
Company Information: {self.company_info}
|
48 |
+
Role Description: {self.role_description}
|
49 |
+
|
50 |
+
Evaluate the completeness of the data. If some key data is missing, determine whether we should:
|
51 |
+
- Scrape for more data (e.g., company info, role descriptions).
|
52 |
+
- Proceed with the available information and generate the email using default logic.
|
53 |
+
|
54 |
+
Reflect on whether we need more data or if the current information is sufficient to proceed.
|
55 |
+
"""
|
56 |
+
|
57 |
+
# Send this reasoning prompt to the LLM
|
58 |
+
url = "https://api.groq.com/openai/v1/chat/completions"
|
59 |
+
headers = {
|
60 |
+
"Authorization": f"Bearer {groq_api_key}",
|
61 |
+
"Content-Type": "application/json",
|
62 |
+
}
|
63 |
+
|
64 |
+
data = {
|
65 |
+
"messages": [{"role": "user", "content": reasoning_prompt}],
|
66 |
+
"model": "llama3-8b-8192"
|
67 |
+
}
|
68 |
+
|
69 |
+
response = requests.post(url, headers=headers, json=data)
|
70 |
+
if response.status_code == 200:
|
71 |
+
reasoning_output = response.json()["choices"][0]["message"]["content"].strip()
|
72 |
+
print("LLM Reasoning Output:", reasoning_output)
|
73 |
+
return reasoning_output
|
74 |
+
else:
|
75 |
+
print(f"Error: {response.status_code}, {response.text}")
|
76 |
+
return "Error: Unable to complete reasoning."
|
77 |
|
78 |
# Action: Fetch LinkedIn data via Proxycurl (acting based on reasoning)
|
79 |
def fetch_linkedin_data(self):
|
|
|
121 |
print(f"Error: Unable to fetch company info via Firecrawl. Using default info.")
|
122 |
self.company_info = "A leading company in its field."
|
123 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
124 |
# Final Action: Generate the email using Groq Cloud LLM based on gathered data
|
125 |
def generate_email(self):
|
126 |
print("Action: Generating the email with the gathered information.")
|
|
|
168 |
|
169 |
# Main loop following ReAct pattern
|
170 |
def run(self):
|
171 |
+
reasoning_output = self.reason_with_llm() # LLM performs reasoning and reflection
|
172 |
+
print("LLM Reflection:", reasoning_output)
|
173 |
+
|
174 |
self.fetch_linkedin_data() # Fetch LinkedIn data
|
175 |
self.fetch_company_info_with_firecrawl() # Fetch company data using Firecrawl
|
176 |
+
|
177 |
+
return self.generate_email() # Final action: generate email
|
|
|
|
|
|
|
178 |
|
179 |
# Define the Gradio interface and the main app logic
|
180 |
def gradio_ui():
|