Spaces:
Paused
Paused
File size: 20,390 Bytes
9f2b0ed 308d699 8e4e001 9f2b0ed 32acb92 9f2b0ed 2ae57cb 32acb92 b5d3943 9f2b0ed 2ae57cb 9f2b0ed 44441db 308d699 44441db 308d699 81341b4 330157f 9f2b0ed 330157f 9f2b0ed 330157f 308d699 330157f 2ae57cb b5d3943 6720344 b5d3943 9f2b0ed 44441db a1b807c 44441db 308d699 a1b807c 49fbb3a 308d699 a1b807c 308d699 6720344 49fbb3a cb47676 49fbb3a a1b807c 49fbb3a a1b807c 49fbb3a a1b807c 49fbb3a 308d699 44441db 49fbb3a 308d699 a1b807c 32acb92 a1b807c 32acb92 a1b807c 308d699 49fbb3a 308d699 e83166d 308d699 49fbb3a 308d699 a1b807c 44441db a1b807c 44441db 308d699 e83166d 88df9ba |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 |
import os
import uuid
import json
import logging
from flask import Blueprint, request, jsonify, send_file, url_for, current_app
from flask_login import login_required, current_user
from backend.models.database import db, Job, Application
from backend.services.interview_engine import (
generate_first_question,
generate_next_question,
edge_tts_to_file_sync,
whisper_stt,
evaluate_answer
)
# Additional imports for report generation
from backend.models.database import Application
from backend.services.report_generator import generate_llm_interview_report, create_pdf_report
from flask import abort
interview_api = Blueprint("interview_api", __name__)
@interview_api.route("/start_interview", methods=["POST"])
@login_required
def start_interview():
"""
Start a new interview. Generates the first question based on the user's
resume/profile and the selected job. Always returns a JSON payload
containing the question text and, if available, a URL to an audio
rendition of the question.
"""
try:
data = request.get_json() or {}
job_id = data.get("job_id")
# Validate the job and the user's application
job = Job.query.get_or_404(job_id)
application = Application.query.filter_by(
user_id=current_user.id,
job_id=job_id
).first()
if not application or not application.extracted_features:
return jsonify({"error": "No application/profile data found."}), 400
# Parse the candidate's profile
try:
profile = json.loads(application.extracted_features)
except Exception as e:
logging.error(f"Invalid profile JSON: {e}")
return jsonify({"error": "Invalid profile JSON"}), 500
# Generate the first question using the LLM
question = generate_first_question(profile, job)
if not question:
question = "Tell me about yourself and why you're interested in this position."
# Attempt to generate a TTS audio file for the question
audio_url = None
try:
audio_dir = "/tmp/audio"
os.makedirs(audio_dir, exist_ok=True)
filename = f"q_{uuid.uuid4().hex}.wav"
audio_path = os.path.join(audio_dir, filename)
audio_result = edge_tts_to_file_sync(question, audio_path)
if audio_result and os.path.exists(audio_path) and os.path.getsize(audio_path) > 1000:
audio_url = url_for("interview_api.get_audio", filename=filename)
logging.info(f"Audio generated successfully: {audio_url}")
else:
logging.warning("Audio generation failed or file too small")
except Exception as e:
logging.error(f"Error generating TTS audio: {e}")
audio_url = None
return jsonify({
"question": question,
"audio_url": audio_url
})
except Exception as e:
logging.error(f"Error in start_interview: {e}")
return jsonify({"error": "Internal server error"}), 500
import subprocess
@interview_api.route("/transcribe_audio", methods=["POST"])
@login_required
def transcribe_audio():
"""Transcribe uploaded .webm audio using ffmpeg conversion and Faster-Whisper"""
audio_file = request.files.get("audio")
if not audio_file:
return jsonify({"error": "No audio file received."}), 400
temp_dir = "/tmp/interview_temp"
os.makedirs(temp_dir, exist_ok=True)
original_path = os.path.join(temp_dir, f"user_audio_{uuid.uuid4().hex}.webm")
wav_path = original_path.replace(".webm", ".wav")
audio_file.save(original_path)
# Convert to WAV using ffmpeg
try:
subprocess.run(
["ffmpeg", "-y", "-i", original_path, wav_path],
stdout=subprocess.DEVNULL,
stderr=subprocess.DEVNULL
)
except Exception as e:
logging.error(f"FFmpeg conversion failed: {e}")
return jsonify({"error": "Failed to convert audio"}), 500
# Transcribe
transcript = whisper_stt(wav_path)
# Cleanup
try:
os.remove(original_path)
os.remove(wav_path)
except:
pass
if not transcript or not transcript.strip():
return jsonify({"error": "No speech detected in audio. Please try again."}), 400
return jsonify({"transcript": transcript})
# ----------------------------------------------------------------------------
# Interview report download
#
# Recruiters can download a PDF summarising a candidate's interview performance.
# This route performs several checks: it verifies that the current user has
# recruiter or admin privileges, ensures that the requested application exists
# and belongs to one of the recruiter's jobs, generates a textual report via
# the ``generate_llm_interview_report`` helper, converts it into a PDF, and
# finally sends the PDF as a file attachment. The heavy lifting is
# encapsulated in ``services/report_generator.py`` to keep this route
# lightweight.
@interview_api.route('/download_report/<int:application_id>', methods=['GET'])
@login_required
def download_report(application_id: int):
"""Generate and return a PDF report for a candidate's interview.
The ``application_id`` corresponds to the ID of the Application record
representing a candidate's job application. Only recruiters (or admins)
associated with the job are permitted to access this report.
"""
# Fetch the application or return 404 if not found
application = Application.query.get_or_404(application_id)
# Authorisation: ensure the current user is a recruiter or admin
if current_user.role not in ('recruiter', 'admin'):
# 403 Forbidden if the user lacks permissions
return abort(403)
# Further check that the recruiter owns the job unless admin
job = getattr(application, 'job', None)
if job is None:
return abort(404)
if current_user.role != 'admin' and job.recruiter_id != current_user.id:
return abort(403)
try:
# Generate the textual report using the helper function. At this
# stage, interview answers and evaluations are not stored server‑side,
# so the report focuses on the candidate's application data and
# computed skill match. Should answer/score data be persisted in
# future iterations, ``generate_llm_interview_report`` can be
# extended accordingly without touching this route.
report_text = generate_llm_interview_report(application)
# Convert the text to a PDF. The helper returns a BytesIO buffer
# ready for sending via Flask's ``send_file``. Matplotlib is used
# under the hood to avoid heavy dependencies like reportlab.
pdf_buffer = create_pdf_report(report_text)
pdf_buffer.seek(0)
filename = f"{application.name.replace(' ', '_')}_interview_report.pdf"
return send_file(
pdf_buffer,
download_name=filename,
as_attachment=True,
mimetype='application/pdf'
)
except Exception as exc:
# Log the error for debugging; return a 500 to the client
logging.error(f"Error generating report for application {application_id}: {exc}")
return jsonify({"error": "Failed to generate report"}), 500
@interview_api.route("/process_answer", methods=["POST"])
@login_required
def process_answer():
"""
Process a user's answer and return a follow‑up question along with an
evaluation. Always responds with JSON.
"""
try:
data = request.get_json() or {}
answer = data.get("answer", "").strip()
question_idx = data.get("questionIndex", 0)
# ``job_id`` is required to determine how many total questions are
# expected for this interview. Without it we fall back to a
# three‑question interview.
job_id = data.get("job_id")
if not answer:
return jsonify({"error": "No answer provided."}), 400
# Get the current question for evaluation context
current_question = data.get("current_question", "Tell me about yourself")
# Evaluate the answer
evaluation_result = evaluate_answer(current_question, answer)
# 🔥 Save Q&A in interview_log for report
try:
application = Application.query.filter_by(
user_id=current_user.id,
job_id=job_id
).first()
if application:
log_data = []
if application.interview_log:
try:
log_data = json.loads(application.interview_log)
except Exception:
log_data = []
log_data.append({
"question": current_question,
"answer": answer,
"evaluation": evaluation_result
})
application.interview_log = json.dumps(log_data, ensure_ascii=False)
db.session.commit()
except Exception as log_err:
logging.error(f"Error saving interview log: {log_err}")
# Determine the number of questions configured for this job
total_questions = 4
if job_id is not None:
try:
job = Job.query.get(int(job_id))
if job and job.num_questions and job.num_questions > 0:
total_questions = job.num_questions
except Exception:
# If lookup fails, keep default
pass
# Check completion. ``question_idx`` is zero‑based; the last index
# corresponds to ``total_questions - 1``. When the current index
# reaches or exceeds this value, the interview is complete.
is_complete = question_idx >= (total_questions - 1)
next_question_text = None
audio_url = None
if not is_complete:
next_idx = question_idx + 1
# Determine which question to ask next. If next_idx is the last
# question (i.e. equals total_questions - 1), use the final
# question. Otherwise, select a follow‑up question from the
# bank based on ``next_idx - 1`` (because index 0 is for the
# first follow‑up). If out of range, cycle through the list.
if next_idx == (total_questions - 1):
next_question_text = (
"What are your salary expectations? Are you looking for a full-time or part-time role, "
"and do you prefer remote or on-site work?"
)
else:
# 🔥 Use Qdrant-powered next question
try:
# You need profile + job for Qdrant context
job = Job.query.get(int(job_id)) if job_id else None
application = Application.query.filter_by(
user_id=current_user.id,
job_id=job_id
).first()
profile = {}
if application and application.extracted_features:
profile = json.loads(application.extracted_features)
conversation_history = data.get("conversation_history", [])
next_question_text = generate_next_question(
profile,
job,
conversation_history,
answer
)
except Exception as e:
logging.error(f"Error generating next question from Qdrant: {e}")
next_question_text = "Could you elaborate more on your last point?"
# Try to generate audio for the next question
try:
audio_dir = "/tmp/audio"
os.makedirs(audio_dir, exist_ok=True)
filename = f"q_{uuid.uuid4().hex}.wav"
audio_path = os.path.join(audio_dir, filename)
audio_result = edge_tts_to_file_sync(next_question_text, audio_path)
if audio_result and os.path.exists(audio_path) and os.path.getsize(audio_path) > 1000:
audio_url = url_for("interview_api.get_audio", filename=filename)
logging.info(f"Next question audio generated: {audio_url}")
except Exception as e:
logging.error(f"Error generating next question audio: {e}")
audio_url = None
return jsonify({
"success": True,
"next_question": next_question_text,
"audio_url": audio_url,
"evaluation": evaluation_result,
"is_complete": is_complete,
"redirect_url": url_for("interview_api.interview_complete") if is_complete else None
})
except Exception as e:
logging.error(f"Error in process_answer: {e}")
return jsonify({"error": "Error processing answer. Please try again."}), 500
@interview_api.route("/audio/<string:filename>", methods=["GET"])
@login_required
def get_audio(filename: str):
"""Serve previously generated TTS audio from the /tmp/audio directory."""
try:
# Sanitize filename to prevent directory traversal
safe_name = os.path.basename(filename)
if not safe_name.endswith('.wav'):
return jsonify({"error": "Invalid audio file format."}), 400
audio_path = os.path.join("/tmp/audio", safe_name)
if not os.path.exists(audio_path):
logging.warning(f"Audio file not found: {audio_path}")
return jsonify({"error": "Audio file not found."}), 404
if os.path.getsize(audio_path) == 0:
logging.warning(f"Audio file is empty: {audio_path}")
return jsonify({"error": "Audio file is empty."}), 404
return send_file(
audio_path,
mimetype="audio/wav",
as_attachment=False,
conditional=True # Enable range requests for better audio streaming
)
except Exception as e:
logging.error(f"Error serving audio file {filename}: {e}")
return jsonify({"error": "Error serving audio file."}), 500
from flask import render_template
@interview_api.route("/interview/complete", methods=["GET"])
@login_required
def interview_complete():
"""
Final interview completion page. After the last question has been
answered, redirect here to show the candidate a brief summary of
their overall performance. The summary consists of a percentage
score and a high‑level label (e.g. "Excellent", "Good"). These
values are derived from the candidate's application data and
interview evaluations.
The calculation mirrors the logic used in the PDF report
generation: the skills match ratio contributes 40% of the final
score while the average of the per‑question evaluation ratings
contributes 60%. If no evaluation data is available, a default
average of 0.5 is used. The resulting number is expressed as a
percentage (e.g. "75%") and mapped to a descriptive label.
"""
score = None
feedback_summary = None
try:
# Attempt to locate the most recent application with interview data
# for the current user. Because the completion route does not
# receive a job ID, we fall back to the latest application that
# contains an interview_log. If none exists, the summary will
# remain empty and the template will render placeholders.
application = (
Application.query
.filter_by(user_id=current_user.id)
.filter(Application.interview_log.isnot(None))
.order_by(Application.id.desc())
.first()
)
if application:
# Parse candidate and job skills from stored JSON. If either
# field is missing or malformed, fall back to empty lists.
try:
candidate_features = json.loads(application.extracted_features) if application.extracted_features else {}
except Exception:
candidate_features = {}
candidate_skills = candidate_features.get('skills', []) or []
job_skills = []
try:
job_skills = json.loads(application.job.skills) if application.job and application.job.skills else []
except Exception:
job_skills = []
# Compute the skills match ratio. Normalise skills to lower
# case and strip whitespace for comparison. Avoid division
# by zero if the job has no listed skills.
candidate_set = {s.strip().lower() for s in candidate_skills}
job_set = {s.strip().lower() for s in job_skills}
common = candidate_set & job_set
ratio = (len(common) / len(job_set)) if job_set else 0.0
# Extract per‑question evaluations from the interview log. The
# interview_log stores a list of dictionaries with keys
# "question", "answer" and "evaluation". Each evaluation is
# expected to include a "score" field containing text such
# as "Poor", "Medium", "Good" or "Excellent". Convert
# these descriptors into numeric values in the range [0.2, 1.0]
# similar to the logic used in report generation.
qa_scores = []
try:
if application.interview_log:
try:
log_data = json.loads(application.interview_log)
except Exception:
log_data = []
for entry in log_data:
score_text = str(entry.get('evaluation', {}).get('score', '')).lower()
# Map textual scores to numerical values
if ('excellent' in score_text) or ('5' in score_text) or ('10' in score_text):
qa_scores.append(1.0)
elif ('good' in score_text) or ('4' in score_text) or ('8' in score_text) or ('9' in score_text):
qa_scores.append(0.8)
elif ('satisfactory' in score_text) or ('medium' in score_text) or ('3' in score_text) or ('6' in score_text) or ('7' in score_text):
qa_scores.append(0.6)
elif ('needs improvement' in score_text) or ('poor' in score_text) or ('2' in score_text):
qa_scores.append(0.4)
else:
qa_scores.append(0.2)
except Exception:
qa_scores = []
# Average the QA scores. If no scores were recorded (e.g. if
# the interview_log is empty or malformed), assume a neutral
# average of 0.5 to avoid penalising the candidate for missing
# data.
qa_average = (sum(qa_scores) / len(qa_scores)) if qa_scores else 0.5
# Weight skills match (40%) and QA average (60%) to derive
# the final overall score. Convert to a percentage for
# display.
overall = (ratio * 0.4) + (qa_average * 0.6)
percentage = overall * 100.0
# Assign a descriptive label based on the overall score.
if overall >= 0.8:
label = 'Excellent'
elif overall >= 0.65:
label = 'Good'
elif overall >= 0.45:
label = 'Satisfactory'
else:
label = 'Needs Improvement'
# Format the score as a whole‑number percentage. For example
# 0.753 becomes "75%". Note that rounding is applied.
score = f"{percentage:.0f}%"
feedback_summary = label
except Exception as calc_err:
# If any error occurs during calculation, fall back to None values.
logging.error(f"Error computing overall interview score: {calc_err}")
return render_template(
"closing.html",
score=score,
feedback_summary=feedback_summary
) |