Files
ada 12d5c6eaff feat: Add data models for disciplinary actions, organizations, and training centers
- Implemented DisciplinaryAction model for tracking disciplinary measures.
- Created Organization model to represent auditing organizations with relevant details.
- Developed TrainingCenter model for educational and methodological centers.
- Added methods to convert models to dictionaries for easy export.

feat: Implement parsers for auditors and organizations

- Developed AuditorsParser to scrape and parse auditor registry data.
- Created OrganizationsParser for scraping auditing organizations' registry.
- Implemented GenericRegistryParser for handling generic tabular data.

chore: Set up base parser functionality

- Established BaseParser class with common methods for making requests and parsing HTML.
- Added pagination handling and detailed page parsing capabilities.

feat: Implement Excel export functionality

- Created ExcelExporter class for exporting data to Excel files with formatting options.
- Added support for exporting multiple sheets and organization data specifically.

chore: Add logging utilities

- Implemented setup_logger function for consistent logging across the application.
- Configured logging to output to both console and file.

chore: Update requirements and scripts

- Added necessary dependencies for parsing, Excel handling, and logging.
- Created run.sh and run.fish scripts for easy execution of the parser.
2025-10-25 20:31:45 +03:00

78 lines
2.9 KiB
Python

"""
Модель данных для аудитора
"""
from dataclasses import dataclass, field
from typing import Optional, List, Dict
from datetime import datetime
@dataclass
class Auditor:
"""Модель аудитора или индивидуального аудитора"""
# Основная информация
full_name: str
ornz: str # Основной регистрационный номер записи
certificate_number: str
region: str
status: Optional[str] = None
# Дополнительная информация
inn: Optional[str] = None
snils: Optional[str] = None
qualification: Optional[str] = None
# Организация (для аудиторов)
organization_name: Optional[str] = None
organization_inn: Optional[str] = None
# Даты
certificate_issue_date: Optional[datetime] = None
membership_start_date: Optional[datetime] = None
membership_end_date: Optional[datetime] = None
# Дополнительные данные
education: Optional[str] = None
experience_years: Optional[int] = None
specializations: List[str] = field(default_factory=list)
# Метаинформация
source_url: Optional[str] = None
parsed_at: datetime = field(default_factory=datetime.now)
def to_dict(self) -> Dict:
"""Преобразование в словарь для экспорта"""
return {
"ФИО": self.full_name,
"ОРНЗ": self.ornz,
"Номер аттестата": self.certificate_number,
"Регион": self.region,
"Статус": self.status or "",
"ИНН": self.inn or "",
"СНИЛС": self.snils or "",
"Квалификация": self.qualification or "",
"Организация": self.organization_name or "",
"ИНН организации": self.organization_inn or "",
"Дата выдачи аттестата": (
self.certificate_issue_date.strftime("%d.%m.%Y")
if self.certificate_issue_date
else ""
),
"Дата начала членства": (
self.membership_start_date.strftime("%d.%m.%Y")
if self.membership_start_date
else ""
),
"Дата окончания членства": (
self.membership_end_date.strftime("%d.%m.%Y")
if self.membership_end_date
else ""
),
"Образование": self.education or "",
"Стаж (лет)": self.experience_years or "",
"Специализации": ", ".join(self.specializations),
"URL источника": self.source_url or "",
"Дата сбора данных": self.parsed_at.strftime("%d.%m.%Y %H:%M:%S"),
}