File size: 14,911 Bytes
88da18c | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 | """
LLM integration module for analyzing job descriptions and tailoring CV/cover letter.
Supports OpenAI, Grok, Groq, and local Ollama via the OpenAI-compatible API format.
"""
import os
from openai import OpenAI
from typing import Dict, List, Tuple, Any, Optional
import json
import re
import requests
from bs4 import BeautifulSoup
import time
class OpenAIIntegration:
"""Class for handling LLM API interactions."""
def __init__(self, api_key: Optional[str] = None):
"""Initialize LLM integration.
Args:
api_key: API key for the configured provider. Optional for local Ollama.
"""
self.provider = (os.environ.get("LLM_PROVIDER", "ollama") or "ollama").lower()
if self.provider == "openai":
self.api_key = api_key or os.environ.get("OPENAI_API_KEY")
self.base_url = os.environ.get("OPENAI_BASE_URL")
self.model = os.environ.get("OPENAI_MODEL", "gpt-4o-mini")
if self.api_key:
if self.base_url:
self.client = OpenAI(api_key=self.api_key, base_url=self.base_url)
else:
self.client = OpenAI(api_key=self.api_key)
else:
self.client = None
elif self.provider == "grok":
# Grok uses xAI API with OpenAI-compatible endpoint
self.api_key = api_key or os.environ.get("GROK_API_KEY")
self.base_url = os.environ.get("GROK_BASE_URL", "https://api.x.ai/v1")
self.model = os.environ.get("GROK_MODEL", "grok-2")
if self.api_key:
self.client = OpenAI(api_key=self.api_key, base_url=self.base_url)
else:
self.client = None
elif self.provider == "groq":
# Groq provides OpenAI-compatible chat completions.
self.api_key = api_key or os.environ.get("GROQ_API_KEY")
self.base_url = os.environ.get("GROQ_BASE_URL", "https://api.groq.com/openai/v1")
self.model = os.environ.get("GROQ_MODEL", "llama-3.3-70b-versatile")
if self.api_key:
self.client = OpenAI(api_key=self.api_key, base_url=self.base_url)
else:
self.client = None
else:
# Default to free/local Ollama (OpenAI-compatible endpoint).
self.api_key = api_key or os.environ.get("OLLAMA_API_KEY", "ollama")
self.base_url = os.environ.get("OLLAMA_BASE_URL", "http://127.0.0.1:11434/v1")
self.model = os.environ.get("OLLAMA_MODEL", "llama3.1:8b")
self.client = OpenAI(api_key=self.api_key, base_url=self.base_url)
def is_api_key_set(self) -> bool:
"""Check if API client is available.
Returns:
Boolean indicating if API client is ready
"""
if self.provider in ("openai", "grok", "groq"):
return bool(self.api_key) and bool(self.client)
return bool(self.client)
def set_api_key(self, api_key: str) -> None:
"""Set API key for the current provider.
Args:
api_key: Provider API key
"""
self.api_key = api_key
if self.base_url:
self.client = OpenAI(api_key=api_key, base_url=self.base_url)
else:
self.client = OpenAI(api_key=api_key)
def _build_model_candidates(self) -> List[str]:
"""Build an ordered model candidate list for provider fallback."""
candidates: List[str] = [self.model]
# Optional manual fallback list from environment (comma-separated).
extra = os.environ.get("LLM_FALLBACK_MODELS", "")
if extra:
candidates.extend([m.strip() for m in extra.split(",") if m.strip()])
if self.provider == "grok":
candidates.extend([
"grok-3-mini",
"grok-3",
"grok-3-fast",
"grok-2-latest",
"grok-2",
"grok-beta",
])
elif self.provider == "groq":
candidates.extend([
"llama-3.3-70b-versatile",
"llama-3.1-8b-instant",
"mixtral-8x7b-32768",
])
elif self.provider == "openai":
candidates.extend([
"gpt-4o-mini",
"gpt-4.1-mini",
])
# Deduplicate while preserving order.
deduped: List[str] = []
seen = set()
for model in candidates:
if model and model not in seen:
deduped.append(model)
seen.add(model)
return deduped
def _chat_completion_with_fallback(self, messages: List[Dict[str, str]], temperature: float, max_tokens: int):
"""Run chat completion with model fallback on model-not-found errors."""
model_candidates = self._build_model_candidates()
last_error: Optional[Exception] = None
for model_name in model_candidates:
try:
response = self.client.chat.completions.create(
model=model_name,
messages=messages,
temperature=temperature,
max_tokens=max_tokens,
)
# Persist successful model so later requests are faster/stable.
self.model = model_name
return response
except Exception as e:
last_error = e
error_text = str(e).lower()
is_model_error = (
"model not found" in error_text
or "invalid model" in error_text
or "does not exist" in error_text
)
if is_model_error:
continue
raise
if last_error is not None:
raise ValueError(
f"All model candidates failed for provider '{self.provider}': {model_candidates}. "
f"Last error: {last_error}"
)
raise ValueError("No model candidates available for completion.")
def extract_job_description_from_url(self, url: str) -> str:
"""Extract job description from LinkedIn URL.
Args:
url: LinkedIn job posting URL
Returns:
Extracted job description text
"""
if not url.startswith(('http://', 'https://')):
raise ValueError("Invalid URL format")
if 'linkedin.com' not in url:
raise ValueError("URL must be from LinkedIn")
try:
# Add headers to mimic a browser request
headers = {
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
}
# Make the request
response = requests.get(url, headers=headers)
response.raise_for_status()
# Parse the HTML
soup = BeautifulSoup(response.text, 'html.parser')
# Extract job title
job_title = ""
title_element = soup.find('h1', class_='top-card-layout__title')
if title_element:
job_title = title_element.get_text(strip=True)
# Extract company name
company = ""
company_element = soup.find('a', class_='topcard__org-name-link')
if company_element:
company = company_element.get_text(strip=True)
# Extract job description
description = ""
desc_element = soup.find('div', class_='show-more-less-html__markup')
if desc_element:
description = desc_element.get_text(strip=True)
# If we couldn't find the description in the expected place, try alternative selectors
if not description:
desc_element = soup.find('div', class_='description__text')
if desc_element:
description = desc_element.get_text(strip=True)
# Combine all information
full_description = f"Job Title: {job_title}\nCompany: {company}\n\nJob Description:\n{description}"
return full_description
except requests.RequestException as e:
raise ValueError(f"Error fetching job description: {str(e)}")
except Exception as e:
raise ValueError(f"Error parsing job description: {str(e)}")
def analyze_job_description(self, job_description_or_url: str, cv_content: str) -> Dict[str, Any]:
"""Analyze job description and suggest CV modifications.
Args:
job_description_or_url: Text of the job description or LinkedIn URL
cv_content: Current content of the CV
Returns:
Dictionary with suggested modifications for different CV sections
"""
if not self.is_api_key_set():
if self.provider == "openai":
raise ValueError("OpenAI API key is not set. Please set OPENAI_API_KEY or call set_api_key().")
elif self.provider == "grok":
raise ValueError("Grok API key is not set. Please set GROK_API_KEY or call set_api_key().")
elif self.provider == "groq":
raise ValueError("Groq API key is not set. Please set GROQ_API_KEY or call set_api_key().")
raise ValueError("LLM client is not initialized. Check local Ollama settings.")
# Check if input is a LinkedIn URL
if job_description_or_url.startswith(('http://', 'https://')) and 'linkedin.com' in job_description_or_url:
try:
job_description = self.extract_job_description_from_url(job_description_or_url)
except Exception as e:
raise ValueError(f"Error extracting job description from URL: {str(e)}")
else:
job_description = job_description_or_url
# Prepare the prompt for GPT
prompt = f"""
You are an expert CV and resume tailoring assistant. Your task is to analyze a job description
and suggest modifications to a CV to better match the job requirements.
JOB DESCRIPTION:
{job_description}
CURRENT CV CONTENT:
{cv_content}
Please analyze the job description and suggest specific modifications to the following sections of the CV:
1. Profile/Summary: Suggest a tailored professional summary that highlights relevant skills and experience.
2. Skills: Identify key skills from the job description that should be emphasized or added.
3. Experience: Suggest how to reframe or emphasize certain experiences to better match the job requirements.
Format your response as a JSON object with the following structure:
{{
"profile_summary": "Suggested profile summary text",
"skills": ["skill1", "skill2", "skill3"],
"experience_highlights": ["point1", "point2", "point3"],
"keywords_to_emphasize": ["keyword1", "keyword2", "keyword3"]
}}
"""
try:
# Call LLM API with model fallback.
response = self._chat_completion_with_fallback(
messages=[
{"role": "system", "content": "You are an expert CV tailoring assistant that provides structured JSON responses."},
{"role": "user", "content": prompt}
],
temperature=0.5,
max_tokens=1000,
)
# Extract and parse the response
result = response.choices[0].message.content
try:
# Try to parse as JSON
return json.loads(result)
except json.JSONDecodeError:
# If parsing fails, return raw response
return {"raw_response": result}
except Exception as e:
return {"error": str(e)}
def tailor_cover_letter(self, job_description: str, current_cover_letter: str, cv_content: str) -> str:
"""Generate a tailored cover letter based on job description and CV.
Args:
job_description: Text of the job description
current_cover_letter: Current content of the cover letter
cv_content: Content of the CV for reference
Returns:
Tailored cover letter text
"""
if not self.is_api_key_set():
if self.provider == "openai":
raise ValueError("OpenAI API key is not set. Please set OPENAI_API_KEY or call set_api_key().")
elif self.provider == "grok":
raise ValueError("Grok API key is not set. Please set GROK_API_KEY or call set_api_key().")
elif self.provider == "groq":
raise ValueError("Groq API key is not set. Please set GROQ_API_KEY or call set_api_key().")
raise ValueError("LLM client is not initialized. Check local Ollama settings.")
# Prepare the prompt for LLM
prompt = f"""
You are an expert cover letter writing assistant. Your task is to tailor a cover letter
to better match a specific job description, while maintaining the original structure and tone.
JOB DESCRIPTION:
{job_description}
CURRENT COVER LETTER:
{current_cover_letter}
CV CONTENT (for reference):
{cv_content}
Please rewrite the body of the cover letter to:
1. Address specific requirements mentioned in the job description
2. Highlight relevant skills and experiences from the CV
3. Demonstrate enthusiasm for the specific role and company
4. Maintain a professional tone similar to the original
5. Keep approximately the same length as the original
Return only the tailored body text of the cover letter, without greeting or closing.
"""
try:
# Call LLM API with model fallback.
response = self._chat_completion_with_fallback(
messages=[
{"role": "system", "content": "You are an expert cover letter writing assistant."},
{"role": "user", "content": prompt}
],
temperature=0.7,
max_tokens=1000,
)
# Extract the response
result = response.choices[0].message.content
return result
except Exception as e:
return f"Error generating cover letter: {str(e)}"
|