mirror of
https://github.com/ada-dmitry/osint_parser_ada.git
synced 2026-09-24 01:10:20 +00:00
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.
This commit is contained in:
@@ -0,0 +1,65 @@
|
||||
"""
|
||||
Модель данных для дисциплинарной меры
|
||||
"""
|
||||
|
||||
from dataclasses import dataclass, field
|
||||
from typing import Optional, Dict
|
||||
from datetime import datetime
|
||||
|
||||
|
||||
@dataclass
|
||||
class DisciplinaryAction:
|
||||
"""Модель меры дисциплинарного воздействия"""
|
||||
|
||||
# Основная информация
|
||||
subject_name: str # ФИО аудитора или название организации
|
||||
subject_type: str # "auditor" или "organization"
|
||||
ornz: str
|
||||
action_type: str # Тип меры воздействия
|
||||
|
||||
# Детали
|
||||
violation_description: str
|
||||
decision_date: datetime
|
||||
decision_number: Optional[str] = None
|
||||
|
||||
# Дополнительная информация
|
||||
inn: Optional[str] = None
|
||||
region: Optional[str] = None
|
||||
effective_date: Optional[datetime] = None
|
||||
expiry_date: Optional[datetime] = None
|
||||
|
||||
# Решение
|
||||
decision_body: Optional[str] = None
|
||||
appeal_status: Optional[str] = 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 {
|
||||
"Субъект": self.subject_name,
|
||||
"Тип субъекта": (
|
||||
"Аудитор" if self.subject_type == "auditor" else "Организация"
|
||||
),
|
||||
"ОРНЗ": self.ornz,
|
||||
"Мера воздействия": self.action_type,
|
||||
"Описание нарушения": self.violation_description,
|
||||
"Дата решения": (
|
||||
self.decision_date.strftime("%d.%m.%Y") if self.decision_date else ""
|
||||
),
|
||||
"Номер решения": self.decision_number or "",
|
||||
"ИНН": self.inn or "",
|
||||
"Регион": self.region or "",
|
||||
"Дата вступления в силу": (
|
||||
self.effective_date.strftime("%d.%m.%Y") if self.effective_date else ""
|
||||
),
|
||||
"Дата окончания": (
|
||||
self.expiry_date.strftime("%d.%m.%Y") if self.expiry_date else ""
|
||||
),
|
||||
"Орган принявший решение": self.decision_body or "",
|
||||
"Статус обжалования": self.appeal_status or "",
|
||||
"URL источника": self.source_url or "",
|
||||
"Дата сбора данных": self.parsed_at.strftime("%d.%m.%Y %H:%M:%S"),
|
||||
}
|
||||
Reference in New Issue
Block a user