Spaces:
Running
Running
File size: 19,331 Bytes
aa8e154 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 | """
modules/interview_graph.py β Phase 2: LangGraph Multi-Agent Interview Flow
StateGraph with three nodes:
1. InterviewerAgent β asks questions via TTS, captures answers via STT,
optionally asks one follow-up
2. EvaluatorAgent β scores the answer via LLM and stores feedback
3. SummaryNode β generates final summary and saves session JSON
Entry point:
run_interview(profile: dict) -> dict (returns final_summary)
"""
import os
import json
from datetime import datetime
from typing import TypedDict
from langgraph.graph import StateGraph, END
# ββ Sibling module imports ββββββββββββββββββββββββββββββββββββββββββββββββββββ
from modules.llm import (
generate_questions,
generate_followup,
evaluate_answer,
generate_final_summary,
)
from modules.stt import transcribe
from modules.tts import speak
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# SHARED STATE SCHEMA
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
class InterviewState(TypedDict):
profile: dict # from resume_to_profile()
questions: list[str] # generated question bank
current_index: int # which question we're on
results: list[dict] # per-question results
final_summary: dict # filled by SummaryNode
status: str # "interviewing" | "evaluating" | "done"
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# NODE 1 β INTERVIEWER AGENT
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
def interviewer_agent(state: InterviewState) -> dict:
"""
1. Pick the current question from the bank.
2. Speak it via TTS.
3. Capture the candidate's answer via STT.
4. Optionally ask ONE follow-up and capture its answer.
5. Store raw Q&A data in the results list for the EvaluatorAgent.
"""
idx = state["current_index"]
questions = state["questions"]
profile = state["profile"]
question = questions[idx]
q_num = idx + 1
total = len(questions)
# ββ Speak the question ββββββββββββββββββββββββββββββββββββββββββββββββββββ
print(f"\n{'='*60}")
print(f" QUESTION {q_num}/{total}")
print(f"{'='*60}")
print(f" Q: {question}\n")
speak(f"Question {q_num}. {question}")
# ββ Capture answer via STT ββββββββββββββββββββββββββββββββββββββββββββββββ
print("[INTERVIEWER] Listening for your answer...")
try:
answer = transcribe()
except Exception as e:
print(f"[INTERVIEWER] STT error: {e}")
answer = ""
if not answer.strip():
answer = "[No answer detected]"
print(f"[INTERVIEWER] Heard: {answer[:150]}{'...' if len(answer) > 150 else ''}")
# ββ Follow-up logic (max 1 per question) ββββββββββββββββββββββββββββββββββ
followup_q = None
followup_answer = ""
if answer and not answer.startswith("["):
job_role = profile.get("job_role", "Software Engineer")
try:
followup_q = generate_followup(question, answer, job_role)
except Exception as e:
print(f"[INTERVIEWER] Follow-up generation error: {e}")
if followup_q:
print(f"\n[INTERVIEWER] Follow-up: {followup_q}")
speak(f"Follow-up question: {followup_q}")
print("[INTERVIEWER] Listening for follow-up answer...")
try:
followup_answer = transcribe()
except Exception as e:
print(f"[INTERVIEWER] Follow-up STT error: {e}")
followup_answer = ""
if not followup_answer.strip():
followup_answer = "[No follow-up answer detected]"
print(f"[INTERVIEWER] Follow-up answer: {followup_answer[:150]}{'...' if len(followup_answer) > 150 else ''}")
# ββ Build result entry (feedback will be filled by EvaluatorAgent) ββββββββ
result_entry = {
"question": question,
"answer": answer,
"followup": followup_q,
"followup_answer": followup_answer,
"feedback": {}, # placeholder β filled by evaluator
}
# Append to results list
updated_results = list(state["results"]) + [result_entry]
return {
"results": updated_results,
"status": "evaluating",
}
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# NODE 2 β EVALUATOR AGENT
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
def evaluator_agent(state: InterviewState) -> dict:
"""
1. Evaluate the latest answer using the LLM.
2. Store structured feedback in the result entry.
3. Advance current_index.
4. Set status back to 'interviewing' (routing logic decides next step).
"""
profile = state["profile"]
results = list(state["results"]) # shallow copy for mutation
idx = state["current_index"]
# The latest result is the one we just captured
latest = results[-1]
job_role = profile.get("job_role", "Software Engineer")
print(f"\n[EVALUATOR] Scoring answer for Q{idx + 1}...")
try:
feedback = evaluate_answer(
question=latest["question"],
answer=latest["answer"],
job_role=job_role,
followup=latest.get("followup") or "",
followup_answer=latest.get("followup_answer") or "",
)
except Exception as e:
print(f"[EVALUATOR] Evaluation error: {e}")
feedback = {
"score": 5,
"score_str": "5/10",
"strength": "Answer recorded.",
"improvement": "Evaluation unavailable due to error.",
"detail": "",
"raw": "",
}
# Store feedback in the result
latest["feedback"] = feedback
results[-1] = latest
print(f"[EVALUATOR] Score: {feedback.get('score_str', '?')} | "
f"Strength: {feedback.get('strength', '')}")
print(f"[EVALUATOR] Improve: {feedback.get('improvement', '')}")
# Advance to next question
new_index = idx + 1
return {
"results": results,
"current_index": new_index,
"status": "interviewing",
}
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# NODE 3 β SUMMARY NODE
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
def summary_node(state: InterviewState) -> dict:
"""
1. Generate final interview summary via LLM.
2. Save the complete session to a JSON file.
3. Return final_summary dict.
"""
profile = state["profile"]
results = state["results"]
job_role = profile.get("job_role", "Software Engineer")
print(f"\n{'='*60}")
print(" GENERATING FINAL SUMMARY")
print(f"{'='*60}")
try:
final_summary = generate_final_summary(results, job_role)
except Exception as e:
print(f"[SUMMARY] Error generating summary: {e}")
final_summary = {
"overall_score": 5,
"overall_score_str": "5/10",
"top_strength": "Completed the interview.",
"top_area_to_improve": "Practice more.",
"weak_topics": [],
"final_tip": "Keep practicing!",
"raw": "",
}
# ββ Print summary to console βββββββββββββββββββββββββββββββββββββββββββββ
print(f"\n Overall Score : {final_summary.get('overall_score_str', '?')}")
print(f" Top Strength : {final_summary.get('top_strength', '')}")
print(f" Top Improvement : {final_summary.get('top_area_to_improve', '')}")
print(f" Weak Topics : {', '.join(final_summary.get('weak_topics', []))}")
print(f" Final Tip : {final_summary.get('final_tip', '')}")
# ββ Speak a brief closing βββββββββββββββββββββββββββββββββββββββββββββββββ
closing_text = (
f"Interview complete. Your overall score is "
f"{final_summary.get('overall_score_str', 'not available')}. "
f"{final_summary.get('final_tip', 'Great job!')}"
)
try:
speak(closing_text)
except Exception:
pass
# ββ Save session to JSON ββββββββββββββββββββββββββββββββββββββββββββββββββ
session_data = {
"timestamp": datetime.now().isoformat(),
"profile": _make_serialisable(profile),
"questions": state["questions"],
"results": _make_serialisable(results),
"final_summary": _make_serialisable(final_summary),
}
sessions_dir = os.path.join(
os.path.dirname(os.path.dirname(os.path.abspath(__file__))),
"sessions",
)
os.makedirs(sessions_dir, exist_ok=True)
filename = f"session_{datetime.now().strftime('%Y%m%d_%H%M%S')}.json"
filepath = os.path.join(sessions_dir, filename)
try:
with open(filepath, "w", encoding="utf-8") as f:
json.dump(session_data, f, indent=2, ensure_ascii=False)
print(f"\n[SUMMARY] Session saved β {filepath}")
except Exception as e:
print(f"[SUMMARY] Failed to save session JSON: {e}")
return {
"final_summary": final_summary,
"status": "done",
}
# ββ Utility: make dicts JSON-serialisable βββββββββββββββββββββββββββββββββββββ
def _make_serialisable(obj):
"""Recursively convert non-serialisable types to strings."""
if isinstance(obj, dict):
return {k: _make_serialisable(v) for k, v in obj.items()}
if isinstance(obj, (list, tuple)):
return [_make_serialisable(item) for item in obj]
if isinstance(obj, (str, int, float, bool, type(None))):
return obj
return str(obj)
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# ROUTING LOGIC
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
def _after_evaluator(state: InterviewState) -> str:
"""
Conditional edge after EvaluatorAgent:
- If more questions remain β loop back to InterviewerAgent
- Otherwise β proceed to SummaryNode
"""
if state["current_index"] < len(state["questions"]):
return "interviewer"
return "summary"
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# GRAPH CONSTRUCTION
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
def _build_graph() -> StateGraph:
"""
Build and compile the LangGraph StateGraph.
Flow:
START β interviewer β evaluator βββ¬βββ interviewer (loop)
ββββ summary β END
"""
graph = StateGraph(InterviewState)
# Add nodes
graph.add_node("interviewer", interviewer_agent)
graph.add_node("evaluator", evaluator_agent)
graph.add_node("summary", summary_node)
# Set entry point
graph.set_entry_point("interviewer")
# Edges
graph.add_edge("interviewer", "evaluator") # always goes to evaluator
graph.add_conditional_edges( # evaluator β loop or finish
"evaluator",
_after_evaluator,
{
"interviewer": "interviewer",
"summary": "summary",
},
)
graph.add_edge("summary", END) # summary β done
return graph.compile()
# Compile once at module level
_compiled_graph = _build_graph()
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# PUBLIC ENTRY POINT
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
def run_interview(profile: dict, num_questions: int = 2) -> dict:
"""
Run a full multi-agent mock interview.
Args:
profile: dict from resume_to_profile() containing name, job_role,
experience, skills, resume_text, etc.
num_questions: number of middle questions to generate (bookend questions
are added automatically).
Returns:
final_summary dict with overall_score, top_strength,
top_area_to_improve, weak_topics, final_tip, etc.
"""
name = profile.get("name", "Candidate")
job_role = profile.get("job_role", "Software Engineer")
experience = profile.get("experience", "Fresher")
skills = profile.get("skills", "")
resume_text = profile.get("resume_text", "")
print(f"\n{'='*60}")
print(f" AI MOCK INTERVIEW β LangGraph Multi-Agent Flow")
print(f"{'='*60}")
print(f" Candidate : {name}")
print(f" Role : {job_role}")
print(f" Experience: {experience}")
print(f"{'='*60}\n")
# ββ Generate question bank ββββββββββββββββββββββββββββββββββββββββββββββββ
print("[SETUP] Generating interview questions...")
questions = generate_questions(
name=name,
job_role=job_role,
experience=experience,
skills=skills,
resume_text=resume_text,
num_questions=num_questions,
)
print(f"[SETUP] {len(questions)} questions ready.\n")
for i, q in enumerate(questions, 1):
print(f" {i}. {q}")
print()
# ββ Greet candidate βββββββββββββββββββββββββββββββββββββββββββββββββββββββ
greeting = (
f"Welcome {name}. This is your mock interview for the {job_role} role. "
f"I will ask you {len(questions)} questions. Let's begin."
)
speak(greeting)
# ββ Build initial state βββββββββββββββββββββββββββββββββββββββββββββββββββ
initial_state: InterviewState = {
"profile": profile,
"questions": questions,
"current_index": 0,
"results": [],
"final_summary": {},
"status": "interviewing",
}
# ββ Run the graph βββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
final_state = _compiled_graph.invoke(initial_state)
print(f"\n{'='*60}")
print(" INTERVIEW SESSION COMPLETE")
print(f"{'='*60}\n")
return final_state["final_summary"]
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# CLI β quick test entry
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
if __name__ == "__main__":
import sys
sys.path.insert(0, os.path.join(os.path.dirname(os.path.abspath(__file__)), ".."))
from modules.resume_parser import resume_to_profile
print("\n" + "=" * 60)
print(" LANGGRAPH INTERVIEW β QUICK START")
print("=" * 60)
print(" [1] Upload resume (PDF/DOCX/TXT/MD)")
print(" [2] Manual profile")
print("=" * 60)
choice = input(" Choice: ").strip()
if choice == "1":
path = input(" Resume path: ").strip().strip('"')
print(" Parsing resume...")
profile = resume_to_profile(path)
print(f" β Parsed: {profile['name']} | {profile['job_role']}")
else:
profile = {
"name": input(" Name : ").strip(),
"job_role": input(" Job Role : ").strip(),
"experience": input(" Experience: ").strip(),
"skills": input(" Skills : ").strip(),
"resume_text": "",
}
num_q = input(" Number of questions (default 2): ").strip()
num_q = int(num_q) if num_q.isdigit() else 2
summary = run_interview(profile, num_questions=num_q)
print("\n[DONE] Final summary returned:")
print(json.dumps(summary, indent=2))
|