mirror of
https://github.com/ada-dmitry/osint_parser_ada.git
synced 2026-09-24 01:10:20 +00:00
ada
12d5c6eaff
- 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.
60 lines
2.2 KiB
Python
60 lines
2.2 KiB
Python
"""
|
|
Модель данных для квалификационного аттестата
|
|
"""
|
|
|
|
from dataclasses import dataclass, field
|
|
from typing import Optional, Dict
|
|
from datetime import datetime
|
|
|
|
|
|
@dataclass
|
|
class Certificate:
|
|
"""Модель квалификационного аттестата аудитора"""
|
|
|
|
# Основная информация
|
|
certificate_number: str
|
|
auditor_full_name: str
|
|
issue_date: datetime
|
|
status: str
|
|
|
|
# Дополнительная информация
|
|
qualification_type: Optional[str] = None
|
|
issuer: Optional[str] = None
|
|
validity_period: Optional[str] = None
|
|
|
|
# Данные аудитора
|
|
auditor_inn: Optional[str] = None
|
|
auditor_snils: Optional[str] = None
|
|
|
|
# Причина аннулирования (для аннулированных)
|
|
cancellation_reason: Optional[str] = None
|
|
cancellation_date: Optional[datetime] = None
|
|
|
|
# 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 {
|
|
"Номер аттестата": self.certificate_number,
|
|
"ФИО аудитора": self.auditor_full_name,
|
|
"Дата выдачи": (
|
|
self.issue_date.strftime("%d.%m.%Y") if self.issue_date else ""
|
|
),
|
|
"Статус": self.status,
|
|
"Тип квалификации": self.qualification_type or "",
|
|
"Выдан": self.issuer or "",
|
|
"Срок действия": self.validity_period or "",
|
|
"ИНН аудитора": self.auditor_inn or "",
|
|
"СНИЛС аудитора": self.auditor_snils or "",
|
|
"Причина аннулирования": self.cancellation_reason or "",
|
|
"Дата аннулирования": (
|
|
self.cancellation_date.strftime("%d.%m.%Y")
|
|
if self.cancellation_date
|
|
else ""
|
|
),
|
|
"URL источника": self.source_url or "",
|
|
"Дата сбора данных": self.parsed_at.strftime("%d.%m.%Y %H:%M:%S"),
|
|
}
|