VisionCareerMiniapp/vision-career
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Use llm for vacancies post processing

Browse Source
This commit is contained in:
estromenko 2025-10-31 16:37:03 +03:00
parent 44ca1dcf98
commit 59eac494ba
3 changed files with 22 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
vacancies/main/bot.py
View File

@ -62,6 +62,8 @@ async def error_handler(update: object, context: ContextTypes.DEFAULT_TYPE) -> N
async def handle_document(update: Update, context: ContextTypes.DEFAULT_TYPE): async def handle_document(update: Update, context: ContextTypes.DEFAULT_TYPE):
message = await context.bot.send_message(update.effective_chat.id, "📝 Обрабатываю твой запрос. Пожалуйста, подождите...")
if not update.message.document: if not update.message.document:
await context.bot.send_message(chat_id=update.effective_chat.id, text="Не удалось прочитать информацию из файла! Попробуйте другой формат.") await context.bot.send_message(chat_id=update.effective_chat.id, text="Не удалось прочитать информацию из файла! Попробуйте другой формат.")
return return
@ -84,7 +86,7 @@ async def handle_document(update: Update, context: ContextTypes.DEFAULT_TYPE):
{'content': customer_cv.content, 'features_json': features.model_dump()}, {'content': customer_cv.content, 'features_json': features.model_dump()},
) )
await context.bot.send_message(chat_id=update.effective_chat.id, text="Отлично! Запомнил Ваше резюме.") await context.bot.editMessageText("Отлично! Запомнил Ваше резюме.", update.effective_chat.id, message.id)
application = ApplicationBuilder().token(os.environ["BOT_TOKEN"]).build() application = ApplicationBuilder().token(os.environ["BOT_TOKEN"]).build()

6
vacancies/main/management/commands/generate_recommended_vacancies.py
View File

@ -20,7 +20,11 @@ class Command(BaseCommand):
).values_list('vacancy_id', flat=True) ).values_list('vacancy_id', flat=True)
query_filter = Filter(must_not=[HasIdCondition(has_id=recommended_vacancy_ids)]) query_filter = Filter(must_not=[HasIdCondition(has_id=recommended_vacancy_ids)])
search_result_id, vacancy_content, link = search_similarities(query_filter, customer_cv.id) result = search_similarities(query_filter, customer_cv.id)
if not result:
continue
search_result_id, vacancy_content, link = result
recommendation = RecommendedVacancy.objects.create( recommendation = RecommendedVacancy.objects.create(
customer=customer_cv.customer, customer=customer_cv.customer,

19
vacancies/main/vector_store.py
View File

@ -106,7 +106,7 @@ def add_vectors(collection_name: str, _id: int, features: dict, payload: dict):
) )
def search_similarities(query_filter: Filter, cv_id: int) -> list[dict]: def search_similarities(query_filter: Filter, cv_id: int):
vectors = client.retrieve( vectors = client.retrieve(
collection_name="cvs", collection_name="cvs",
ids=[cv_id], ids=[cv_id],
@ -141,10 +141,19 @@ def search_similarities(query_filter: Filter, cv_id: int) -> list[dict]:
total = sum(feature_sims[feature] * weights.get(feature, 1) for feature in feature_sims) total = sum(feature_sims[feature] * weights.get(feature, 1) for feature in feature_sims)
scored.append({"id": vid, "score": total, "content": vacancies_content[vid]["content"], "link": vacancies_content[vid]["link"]}) scored.append({"id": vid, "score": total, "content": vacancies_content[vid]["content"], "link": vacancies_content[vid]["link"]})
scored.sort(key=lambda x: x["score"], reverse=True) prompt = f"""
for i in range(20): Среди вакансий ниже выбери одну наиболее релевантную и выведи ее индекс.
print(f"{scored[i]['content']} {scored[i]['score']}") Если среди вакансий нет подходящих, то верни -1.
return scored[0]["id"], scored[0]["content"], scored[0]["link"] В ответе выведи только число.
{scored}
"""
openai_client = ChatOpenAI(model_name="gpt-5-mini", reasoning_effort="minimal", temperature=0, seed=42, top_p=1)
response = openai_client.invoke(prompt)
index = int(response.content)
if index == -1:
return None
return scored[index]["id"], scored[index]["content"], scored[index]["link"]
def extract_features(content: str) -> VacancyFeatures: def extract_features(content: str) -> VacancyFeatures: