File size: 18,888 Bytes
3dde54d | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 | import json
import os
import fitz
import re
from datetime import datetime, timezone
from fastapi import FastAPI, File, HTTPException, UploadFile
from pydantic import BaseModel
from CV_Parser.Extraction import extract_and_infer_profile, test_llm_connection
from groq import Groq
from prompts import INTERVIEW_EVALUATION_PROMPT, INTERVIEW_QUESTION_PROMPT
app = FastAPI(
title="AI Service",
description="CV Parsing + Candidate Profile Extraction",
version="1.1.0",
)
SECTION_HINTS = (
"profile",
"projects",
"education",
"professional experience",
"experience",
"skills & languages",
"certificates & achievements",
"courses",
)
SECTION_BOUNDARIES = (
"professional experience",
"certificates & achievements",
"skills & languages",
"education",
"projects",
"profile",
"courses",
)
def split_section_boundaries(text):
normalized = str(text or "")
for heading in sorted(SECTION_BOUNDARIES, key=len, reverse=True):
pattern = rf"(?i)(?<!\n)(?<!^)(?<!\b)(?<![\r\n])(?<!\s)({re.escape(heading)})(?!\s*$)"
normalized = re.sub(pattern, r"\n\1\n", normalized)
normalized = re.sub(
rf"(?i)(?<!\n)({re.escape(heading)})(?=\s+[A-Za-z])",
r"\1\n",
normalized,
)
normalized = re.sub(
rf"(?i)(?<=\w)({re.escape(heading)})(?=\w)",
r"\n\1\n",
normalized,
)
return normalized
def score_text_layout(text):
normalized = str(text or "").strip()
if not normalized:
return -1
lines = [line for line in normalized.splitlines() if line.strip()]
heading_hits = sum(1 for hint in SECTION_HINTS if hint in normalized.lower())
long_lines = sum(1 for line in lines if len(line) > 220)
short_lines = sum(1 for line in lines if len(line) < 18)
return (heading_hits * 20) + len(lines) - long_lines - (short_lines // 3)
def order_blocks_by_columns(page, blocks):
page_mid = page.rect.width / 2
left_column = []
right_column = []
full_order = []
for block in blocks:
if len(block) < 5:
continue
text = str(block[4]).strip()
if not text:
continue
x0, y0, x1, y1 = block[:4]
entry = (y0, x0, text)
full_order.append(entry)
if x1 <= page_mid:
left_column.append(entry)
elif x0 >= page_mid:
right_column.append(entry)
elif x0 < page_mid:
left_column.append(entry)
else:
right_column.append(entry)
left_text = "\n".join(text for _, _, text in sorted(left_column))
right_text = "\n".join(text for _, _, text in sorted(right_column))
full_text = "\n".join(text for _, _, text in sorted(full_order))
return [candidate for candidate in (left_text + ("\n" + right_text if right_text else ""), full_text) if candidate.strip()]
def extract_page_text(page):
candidates = []
plain_text = page.get_text("text", sort=True) or ""
if plain_text.strip():
candidates.append(plain_text)
blocks = page.get_text("blocks", sort=False) or []
candidates.extend(order_blocks_by_columns(page, blocks))
words = page.get_text("words", sort=True) or []
if words:
word_lines = {}
for word in words:
if len(word) < 5:
continue
text = str(word[4]).strip()
if not text:
continue
y_key = round(float(word[1]), 1)
x_key = float(word[0])
word_lines.setdefault(y_key, []).append((x_key, text))
reconstructed = []
for _, entries in sorted(word_lines.items()):
reconstructed.append(" ".join(text for _, text in sorted(entries)))
word_text = "\n".join(reconstructed)
if word_text.strip():
candidates.append(word_text)
best_text = max(candidates, key=score_text_layout, default="")
if not best_text.strip():
try:
ocr_text_page = page.get_textpage_ocr(language="eng")
best_text = page.get_text(textpage=ocr_text_page) or ""
except Exception:
best_text = plain_text
return best_text
class CVRequest(BaseModel):
clean_cv_text: str
class InterviewSessionRequest(BaseModel):
profile: dict
question_count: int = 5
class InterviewAnswerRequest(BaseModel):
profile: dict
question: str
answer: str
history: list[dict] = []
def get_groq_client():
api_key = os.environ.get("GROQ_API_KEY")
if not api_key:
raise ValueError("GROQ_API_KEY is not set")
return Groq(api_key=api_key)
def normalize_text(value):
return " ".join(str(value or "").split()).strip()
def normalize_list(values):
return [normalize_text(value) for value in values or [] if normalize_text(value)]
def sanitize_profile_for_interview(profile):
candidate = profile.get("candidate", {}) if isinstance(profile, dict) else {}
extraction = profile.get("extraction", {}) if isinstance(profile, dict) else {}
return {
"candidate": {
"fullName": normalize_text(candidate.get("fullName")),
"currentRole": normalize_text(candidate.get("currentRole")),
"suggestedRole": normalize_text(candidate.get("suggestedRole")),
"experienceYears": candidate.get("experienceYears", 0),
"experienceLevel": normalize_text(candidate.get("experienceLevel")),
"summary": normalize_text(candidate.get("summary")),
},
"extraction": {
"skills": normalize_list(extraction.get("skills")),
"highlights": normalize_list(extraction.get("highlights")),
"experience": normalize_list(extraction.get("experience")),
"projects": normalize_list(extraction.get("projects")),
"certifications": normalize_list(extraction.get("certifications")),
},
}
def fallback_interview_plan(profile, question_count=5):
candidate = profile["candidate"]
extraction = profile["extraction"]
target_role = (
candidate.get("suggestedRole")
or candidate.get("currentRole")
or "the target role"
)
focus_areas = normalize_list(
[
target_role,
*extraction.get("skills", [])[:2],
*extraction.get("projects", [])[:1],
"communication",
]
)[:4]
questions = [
{
"id": "q1",
"category": "Personal",
"prompt": f"Tell me a bit about yourself and why you want {target_role}.",
"why": "Opens the interview with role motivation and self-presentation.",
},
{
"id": "q2",
"category": "Personal",
"prompt": "What project from your CV are you most proud of?",
"why": "Checks ownership, impact, and storytelling.",
},
{
"id": "q3",
"category": "Technical",
"prompt": (
f"What skill will help you most as a {target_role}, and where did you use it?"
),
"why": "Connects declared skills to real experience.",
},
{
"id": "q4",
"category": "Technical",
"prompt": "Tell me about a technical problem you solved in a project.",
"why": "Assesses reasoning, resilience, and execution.",
},
{
"id": "q5",
"category": "Personal",
"prompt": "What skill would you like to improve next?",
"why": "Measures self-awareness and growth mindset.",
},
]
return {
"interviewerIntro": (
f"Hi {candidate.get('fullName') or 'there'}, I will simulate a live interview for "
f"{target_role}. Answer naturally, and I will give coaching feedback after each response."
),
"focusAreas": focus_areas,
"questions": questions[: max(1, min(question_count, len(questions)))],
}
def parse_json_response(text):
cleaned = str(text or "").strip()
if cleaned.startswith("```json"):
cleaned = cleaned[7:]
if cleaned.startswith("```"):
cleaned = cleaned[3:]
if cleaned.endswith("```"):
cleaned = cleaned[:-3]
cleaned = cleaned.strip()
if cleaned.startswith("{") and cleaned.endswith("}"):
return json.loads(cleaned)
start = cleaned.find("{")
end = cleaned.rfind("}")
if start != -1 and end != -1 and end > start:
return json.loads(cleaned[start : end + 1])
return json.loads(cleaned)
def request_interview_plan(profile, question_count=5):
client = get_groq_client()
model = os.environ.get("GROQ_MODEL", "llama-3.1-8b-instant")
prompt_payload = json.dumps(
{
"questionCount": question_count,
"profile": profile,
},
ensure_ascii=False,
)
completion = client.chat.completions.create(
model=model,
temperature=0.4,
messages=[
{"role": "system", "content": INTERVIEW_QUESTION_PROMPT},
{"role": "user", "content": prompt_payload},
],
)
parsed = parse_json_response(completion.choices[0].message.content)
questions = parsed.get("questions") if isinstance(parsed, dict) else None
if not isinstance(questions, list) or not questions:
raise ValueError("Interview plan did not return any questions")
return {
"interviewerIntro": normalize_text(parsed.get("interviewerIntro")),
"focusAreas": normalize_list(parsed.get("focusAreas"))[:6],
"questions": [
{
"id": normalize_text(item.get("id")) or f"q{index + 1}",
"category": normalize_text(item.get("category")) or "general",
"prompt": normalize_text(item.get("prompt")),
"why": normalize_text(item.get("why")),
}
for index, item in enumerate(questions)
if normalize_text(item.get("prompt"))
][: max(1, question_count)],
}
def evaluate_answer_fallback(question, answer):
answer_text = normalize_text(answer)
word_count = len(answer_text.split())
score = 5
if word_count >= 25:
score += 2
if word_count >= 45:
score += 1
if any(token in answer_text.lower() for token in ["because", "result", "impact", "learned", "improved"]):
score += 1
score = max(3, min(score, 9))
strengths = []
improvements = []
if word_count >= 25:
strengths.append("You gave enough detail to understand your thinking.")
else:
improvements.append("Add more detail so the interviewer can judge your contribution.")
if any(token in answer_text.lower() for token in ["result", "impact", "improved", "increased", "reduced"]):
strengths.append("You hinted at outcomes, which makes the answer stronger.")
else:
improvements.append("Mention the result or impact of your work.")
if not improvements:
improvements.append("Make the structure even clearer using situation, action, and result.")
return {
"score": score,
"strengths": strengths or ["Your answer addressed the question directly."],
"improvements": improvements[:2],
"followUpQuestion": f"Can you give one specific example related to: {normalize_text(question)}",
"coachReply": "Solid start. Tighten the structure and make your impact more explicit.",
}
def request_answer_evaluation(profile, question, answer, history):
client = get_groq_client()
model = os.environ.get("GROQ_MODEL", "llama-3.1-8b-instant")
prompt_payload = json.dumps(
{
"profile": profile,
"question": normalize_text(question),
"answer": normalize_text(answer),
"history": history[-3:],
},
ensure_ascii=False,
)
completion = client.chat.completions.create(
model=model,
temperature=0.3,
messages=[
{"role": "system", "content": INTERVIEW_EVALUATION_PROMPT},
{"role": "user", "content": prompt_payload},
],
)
parsed = parse_json_response(completion.choices[0].message.content)
return {
"score": max(1, min(int(parsed.get("score", 0) or 0), 10)),
"strengths": normalize_list(parsed.get("strengths"))[:3],
"improvements": normalize_list(parsed.get("improvements"))[:3],
"followUpQuestion": normalize_text(parsed.get("followUpQuestion")),
"coachReply": normalize_text(parsed.get("coachReply")),
}
def extract_text_from_pdf(file_bytes):
pages = []
pdf = fitz.open(stream=file_bytes, filetype="pdf")
for page in pdf:
page_text = extract_page_text(page)
if len(page_text.strip()) < 40:
try:
ocr_text_page = page.get_textpage_ocr(language="eng")
ocr_text = page.get_text(textpage=ocr_text_page) or ""
if len(ocr_text.strip()) > len(page_text.strip()):
page_text = ocr_text
except Exception:
pass
if page_text.strip():
pages.append(page_text)
return "\n".join(pages)
def clean_text(text):
normalized = (
str(text or "")
.replace("\u00a0", " ")
.replace("\u2022", "- ")
.replace("\u2013", "-")
.replace("\u2014", "-")
)
normalized = split_section_boundaries(normalized)
cleaned_lines = []
for raw_line in normalized.splitlines():
line = " ".join(raw_line.split()).strip()
if line:
cleaned_lines.append(line)
return "\n".join(cleaned_lines)
def analyze_cv_bytes(file_bytes):
raw_text = extract_text_from_pdf(file_bytes)
clean_cv_text = clean_text(raw_text)
if len(clean_cv_text.strip()) < 30:
raise HTTPException(
status_code=422,
detail=(
"No readable text could be extracted from this PDF. "
"The file may be scanned or image-based and requires OCR."
),
)
result = extract_and_infer_profile(clean_cv_text)
if result["status"] == "error":
raise HTTPException(status_code=500, detail=result["message"])
result["data"]["metadata"]["rawTextLength"] = len(raw_text)
result["data"]["metadata"]["cleanTextLength"] = len(clean_cv_text)
return clean_cv_text, result
@app.get("/")
async def root():
return {"message": "AI Service API is running successfully"}
@app.get("/health/llm")
async def health_llm():
checked_at = datetime.now(timezone.utc).isoformat()
try:
result = test_llm_connection()
return {
"status": "success",
"service": "llm",
"checkedAt": checked_at,
**result,
}
except Exception as error:
raise HTTPException(
status_code=503,
detail={
"status": "error",
"service": "llm",
"checkedAt": checked_at,
"message": str(error),
},
)
@app.post("/parse-cv")
async def parse_cv(file: UploadFile = File(...)):
try:
content = await file.read()
raw_text = extract_text_from_pdf(content)
clean_cv_text = clean_text(raw_text)
return {
"status": "success",
"clean_cv_text": clean_cv_text,
}
except HTTPException:
raise
except Exception as error:
raise HTTPException(status_code=500, detail=str(error))
@app.post("/extract-profile")
async def extract_profile(request: CVRequest):
if not request.clean_cv_text.strip():
raise HTTPException(
status_code=400,
detail="clean_cv_text cannot be empty",
)
result = extract_and_infer_profile(request.clean_cv_text)
if result["status"] == "error":
raise HTTPException(status_code=500, detail=result["message"])
return result
@app.post("/analyze-cv")
async def analyze_cv(file: UploadFile = File(...)):
try:
content = await file.read()
clean_cv_text, result = analyze_cv_bytes(content)
return {
"status": "success",
"message": "CV parsed and analyzed successfully",
"clean_cv_text": clean_cv_text,
"data": result["data"],
}
except HTTPException:
raise
except Exception as error:
raise HTTPException(status_code=500, detail=str(error))
@app.post("/interview/session/start")
async def start_interview_session(request: InterviewSessionRequest):
profile = sanitize_profile_for_interview(request.profile)
question_count = max(1, min(int(request.question_count or 5), 7))
try:
try:
plan = request_interview_plan(profile, question_count=question_count)
source = "llm"
except Exception:
plan = fallback_interview_plan(profile, question_count=question_count)
source = "fallback"
return {
"status": "success",
"message": "Interview session created successfully",
"data": {
**plan,
"questionCount": len(plan["questions"]),
"source": source,
},
}
except Exception as error:
raise HTTPException(status_code=500, detail=str(error))
@app.post("/interview/session/answer")
async def evaluate_interview_answer(request: InterviewAnswerRequest):
question = normalize_text(request.question)
answer = normalize_text(request.answer)
if not question:
raise HTTPException(status_code=400, detail="question cannot be empty")
if not answer:
raise HTTPException(status_code=400, detail="answer cannot be empty")
profile = sanitize_profile_for_interview(request.profile)
history = request.history if isinstance(request.history, list) else []
try:
try:
evaluation = request_answer_evaluation(profile, question, answer, history)
source = "llm"
except Exception:
evaluation = evaluate_answer_fallback(question, answer)
source = "fallback"
return {
"status": "success",
"message": "Interview answer evaluated successfully",
"data": {
**evaluation,
"source": source,
},
}
except Exception as error:
raise HTTPException(status_code=500, detail=str(error))
|