fix(): better resolution of relevant vacancies

.gitignore

@@ -3,6 +3,9 @@ config.toml
 docker-compose.yml
 .idea
 firebase.json
+dumps
+full_skills_unique.json
+filtered_vacancies.csv
 
 # Byte-compiled / optimized / DLL files
 __pycache__/

src/dataset/dump_data.py

@@ -0,0 +1,43 @@
+#!/usr/bin/env python3
+import subprocess
+from pathlib import Path
+from template_project.web_api.configuration import load_configuration
+
+
+def main() -> None:
+    config_path = Path("config.toml")
+    configuration = load_configuration(config_path)
+    
+    db_url = str(configuration.database.url.get_value())
+    db_url = db_url.replace("postgresql+psycopg://", "postgresql://")
+    
+    output_dir = Path("dumps")
+    output_dir.mkdir(exist_ok=True)
+    
+    output_file = output_dir / "data_dump.sql"
+    
+    print("Создание дампа таблиц vacancy, vacancy_embedding, key_skills...")
+    
+    subprocess.run(
+        [
+            "pg_dump",
+            db_url,
+            "--table=vacancy",
+            "--table=vacancy_embedding",
+            "--table=key_skills",
+            "--data-only",
+            "--column-inserts",
+            f"--file={output_file}",
+        ],
+        check=True,
+    )
+    
+    print(f"\nДамп создан: {output_file}")
+    print(f"Размер файла: {output_file.stat().st_size / 1024 / 1024:.2f} MB")
+    print("\nДля импорта на прод сервере выполните:")
+    print(f"  psql <PROD_DB_URL> -f {output_file}")
+
+
+if __name__ == "__main__":
+    main()
+

src/dataset/dump_data.sh

@@ -0,0 +1,17 @@
+#!/bin/bash
+
+DB_URL="${DATABASE_URL:-postgresql://user:password@localhost:5432/dbname}"
+
+echo "Создание дампа таблиц vacancy, vacancy_embedding, key_skills..."
+
+pg_dump "$DB_URL" \
+  --table=vacancy \
+  --table=vacancy_embedding \
+  --table=key_skills \
+  --data-only \
+  --column-inserts \
+  --file=dump_data.sql
+
+echo "Дамп создан: dump_data.sql"
+echo "Размер файла: $(du -h dump_data.sql | cut -f1)"
+

src/dataset/load_dump.sh

@@ -0,0 +1,16 @@
+#!/bin/bash
+
+DB_URL="${DATABASE_URL:-postgresql://user:password@localhost:5432/dbname}"
+DUMP_FILE="${1:-dump_data.sql}"
+
+if [ ! -f "$DUMP_FILE" ]; then
+    echo "Ошибка: файл $DUMP_FILE не найден"
+    exit 1
+fi
+
+echo "Импорт дампа из $DUMP_FILE в БД..."
+
+psql "$DB_URL" -f "$DUMP_FILE"
+
+echo "Импорт завершен!"
+

src/dataset/load_vacancies.py

@@ -41,7 +41,7 @@ async def main() -> None:
     ml_container = make_ml_ioc(ml_configuration)
 
     csv_path = Path("filtered_vacancies.csv")
-    max_records = 1000
+    max_records = 100_000
 
     try:
         async with backend_container() as backend_request_container, ml_container() as ml_request_container:

src/template_project/adapters/data_gateways/tables.py

@@ -22,7 +22,7 @@ from sqlalchemy.orm import registry
 
 from template_project.application.access_token.entity import AccessToken
 from template_project.application.auth_identity.entity import AuthIdentity, AuthMethod
-from template_project.application.common.enums import EducationGrade
+from template_project.application.common.enums import EducationGrade, ExperienceType
 from template_project.application.notification_device.entity import NotificationDevice
 from template_project.application.resume.entity import (
     Resume,

src/template_project/adapters/data_gateways/vacancy.py

@@ -16,11 +16,13 @@ class DefaultVacancyDataGateway(VacancyDataGateway):
 
     @override
     async def get_suitable(self, vector: list[float]) -> Sequence[SuitableVacancy]:
+        distance_expr = vacancy_embedding_table.c.vector.cosine_distance(vector)
+        similarity_expr = 1 - distance_expr
         statement = (
-            select(Vacancy, label("resume_similarity", vacancy_embedding_table.c.vector.cosine_distance(vector)))
+            select(Vacancy, label("resume_similarity", similarity_expr))
             .join(VacancyEmbedding, vacancy_embedding_table.c.vacancy_id == vacancy_table.c.id)
-            .where(vacancy_embedding_table.c.vector.cosine_distance(vector) > 0.5)
-            .order_by(vacancy_embedding_table.c.vector.cosine_distance(vector).asc())
+            .where(similarity_expr >= 0.5)
+            .order_by(distance_expr.asc())
             .limit(100)
         )
         result = await self._session.execute(statement)

src/template_project/adapters/generators/resume_prediction.py

@@ -22,11 +22,11 @@ class DefaultResumePredictionGenerator(ResumePredictionGenerator):
             key_skills=resume.key_skills,
             suitable_vacancies=[
                 SuitableVacancyDs(
-                    vacancy_id=str(suituble_vacancy.vacancy.id),
-                    from_salary=suituble_vacancy.vacancy.from_salary,
-                    to_salary=suituble_vacancy.vacancy.to_salary,
-                    key_skills=suituble_vacancy.vacancy.key_skills,
-                    resume_similarity=suituble_vacancy.resume_similarity,
+                    vacancy_id=str(suitable_vacancy.vacancy.id),
+                    from_salary=suitable_vacancy.vacancy.from_salary,
+                    to_salary=suitable_vacancy.vacancy.to_salary,
+                    key_skills=suitable_vacancy.vacancy.key_skills,
+                    resume_similarity=suitable_vacancy.resume_similarity,
                 )
                 for suitable_vacancy in suitable_vacancies
             ],

uv.lock

@@ -2425,15 +2425,28 @@ backend = [
     { name = "sqlalchemy" },
 ]
 dev = [
+    { name = "aioboto3" },
     { name = "alembic" },
+    { name = "argon2-cffi" },
     { name = "bandit" },
     { name = "codespell" },
     { name = "coverage" },
+    { name = "cryptography" },
     { name = "dirty-equals" },
+    { name = "firebase-admin" },
+    { name = "httpx" },
     { name = "mypy" },
+    { name = "pgvector" },
+    { name = "prometheus-fastapi-instrumentator" },
+    { name = "psycopg", extra = ["binary"] },
     { name = "pytest" },
     { name = "pytest-asyncio" },
+    { name = "python-multipart" },
     { name = "ruff" },
+    { name = "sentence-transformers" },
+    { name = "sqlalchemy" },
+    { name = "torch", version = "2.2.2", source = { registry = "https://download.pytorch.org/whl/cpu" }, marker = "(python_full_version < '3.13' and platform_machine == 'aarch64' and platform_python_implementation == 'CPython' and sys_platform == 'linux') or (python_full_version < '3.13' and sys_platform == 'darwin')" },
+    { name = "torch", version = "2.2.2+cpu", source = { registry = "https://download.pytorch.org/whl/cpu" }, marker = "(python_full_version >= '3.13' and sys_platform == 'darwin') or (python_full_version >= '3.13' and sys_platform == 'linux') or (platform_machine != 'aarch64' and sys_platform == 'linux') or (platform_python_implementation != 'CPython' and sys_platform == 'linux') or (sys_platform != 'darwin' and sys_platform != 'linux')" },
     { name = "types-cachetools" },
 ]
 linters = [
@@ -2487,15 +2500,27 @@ backend = [
     { name = "sqlalchemy", specifier = "==2.0.44" },
 ]
 dev = [
+    { name = "aioboto3", specifier = "==15.5.0" },
     { name = "alembic", specifier = "==1.17.0" },
+    { name = "argon2-cffi", specifier = "==23.1.0" },
     { name = "bandit", specifier = "==1.8.6" },
     { name = "codespell", specifier = "==2.4.1" },
     { name = "coverage", specifier = "==7.11.0" },
+    { name = "cryptography", specifier = "==46.0.3" },
     { name = "dirty-equals", specifier = ">=0.11" },
+    { name = "firebase-admin", specifier = ">=7.1.0" },
+    { name = "httpx", specifier = "==0.28.1" },
     { name = "mypy", specifier = "==1.18.1" },
+    { name = "pgvector", specifier = ">=0.4.1" },
+    { name = "prometheus-fastapi-instrumentator", specifier = ">=7.1.0" },
+    { name = "psycopg", extras = ["binary"], specifier = ">=3.2.12" },
     { name = "pytest", specifier = "==8.4.0" },
     { name = "pytest-asyncio", specifier = "==1.2.0" },
+    { name = "python-multipart", specifier = ">=0.0.20" },
     { name = "ruff", specifier = "==0.12.11" },
+    { name = "sentence-transformers", specifier = ">=5.1.2" },
+    { name = "sqlalchemy", specifier = "==2.0.44" },
+    { name = "torch", index = "https://download.pytorch.org/whl/cpu" },
     { name = "types-cachetools", specifier = "==6.2.0.20250827" },
 ]
 linters = [