add hw2 and tests for hw2

.github/workflows/pylint.yml

@@ -1,35 +1,58 @@
-name: Проверка
+name: Main test
 on: [push]
 jobs:
-  linter:
-    name: Линтер
+  linter_hw1:
+    name: Linter for HW1
     runs-on: ubuntu-latest
     steps:
     - uses: actions/checkout@v2
-    - name: Установка Python
+        python -m pip install --upgrade pip
+        pip install flake8==3.9.0 wemake-python-styleguide==0.15.3 bandit==1.7.2
+    - name: Flake8
+      run: |
+        cd hw1
+        flake8 
+  tests_for_hw1:
+    name: Tests for hw1
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v2
+        pip install pytest==6.2.5
+    - name: Pytest
+      run: |
+        cd hw1
+        pytest
+  linter_hw2:
+    name: Linter for HW2
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v2
+    - name: Python installation
       uses: actions/setup-python@v2
       with:
         python-version: 3.10.6
-    - name: Установка зависимостей
+    - name: Dependencies installtion
       run: |
         python -m pip install --upgrade pip
         pip install flake8==3.9.0 wemake-python-styleguide==0.15.3 bandit==1.7.2
     - name: Flake8
-      run: flake8 .
-  tests:
-    name: Тесты
+      run: |
+        cd hw2
+        flake8 
+  tests_for_hw2:
+    name: Tests for hw2
     runs-on: ubuntu-latest
     steps:
     - uses: actions/checkout@v2
-    - name: Установка Python
+    - name: Python installation
       uses: actions/setup-python@v2
       with:
         python-version: 3.10.6
-    - name: Установка зависимостей
+    - name: Dependencies installation
       run: |
         python -m pip install --upgrade pip
         pip install pytest==6.2.5
-        pip install numpy
     - name: Pytest
-      run: pytest
-
+      run: |
+        cd hw2
+        pytest

hw2/consts_hw2.py

@@ -0,0 +1,14 @@
+"""Provides constants for hw2."""
+
+AGE = 'age'
+AGE_MAX = 'age_max'
+AGE_MIN = 'age_min'
+AGE_AVERAGE = 'age_average'
+AGE_MEDIAN = 'age_median'
+
+LAST_LOGIN = 'last_login'
+LESS_TWO_DAYS = 'percent_active_users_last_two_days'
+LESS_WEEK = 'percent_active_users_last_week'
+LESS_MONTH = 'percent_active_users_last_month'
+LESS_HALFYEAR = 'percent_active_users_last_halfyear'
+MORE_HALFYEAR = 'percent_active_not_users_last_halfyear'

hw2/exceptions_hw2.py

@@ -0,0 +1,61 @@
+"""Provides exceptions for hw2."""
+
+
+class InputFilepathError(Exception):
+    """Error that is thrown when the file could not be read."""
+
+    def __init__(self, filepath: str) -> None:
+        """Initialize the exception.
+
+        Args:
+            filepath (str): path to file.
+        """
+        super().__init__(f'Path {filepath} does not exist.')
+
+
+class MissingFieldError(Exception):
+    """Error that is thrown when one of the users missing a field."""
+
+    def __init__(self, fld: str) -> None:
+        """Initialize the exception.
+
+        Args:
+            fld (str): field name.
+        """
+        super().__init__(f'"{fld}" field does not exist. Every user should have "{fld}" field.')
+
+
+class InvalidDateFormatError(Exception):
+    """Error that is thrown when date format in field is not correct."""
+
+    def __init__(self, got_format: str) -> None:
+        """Initialize the exception.
+
+        Args:
+            got_format (str): date format.
+        """
+        super().__init__(f'Wrong date format, expected format: 2020-12-30, but got {got_format}.')
+
+
+class InvalidJSONFormatError(Exception):
+    """Error that is thrown when JSON structure is not correct."""
+
+    def __init__(self, filepath: str) -> None:
+        """Initialize the exception.
+
+        Args:
+            filepath (str): path to file.
+        """
+        super().__init__(f'An incorrect json file structure was found at path {filepath}.')
+
+
+class EmptyJSONError(Exception):
+    """Error that is thrown when was given empty json file."""
+
+    def __init__(self, filepath: str) -> None:
+        """Initialize the exception.
+
+        Args:
+            filepath (str): path to file.
+        """
+        super().__init__(f'Empty json file was found at path {filepath}.')

hw2/hw2.py

@@ -0,0 +1,157 @@
+"""
+Module for homework_2.
+
+Напишите модуль, в котором функция process_data принимает путь к json-файлу
+с данными о клиентах сайта (пример файла в data_hw2.json) и путь к json-файлу вывода.
+
+Функция process_data записывает в этот общий json статистику:
+минимальный, максимальный, средний и медианный возраст пользователя,
+а также процент клиентов, которые были онлайн:
+менее двух дней, недели, месяца, полугода, и более полугода назад.
+"""
+import json
+import os
+from datetime import datetime
+
+import consts_hw2 as cnst
+import exceptions_hw2
+
+TWO_DAYS = 2
+WEEK = 7
+MONTH = 30
+HALFYEAR = 180
+
+ROUND_UP_TO_TWO = 2
+
+
+def _median(inp_values: list[int | float]) -> int | float:
+    inp_values = sorted(inp_values)
+    length = len(inp_values)
+    middle = length // 2
+    if length % 2 == 0:
+        median = (inp_values[middle - 1] + inp_values[middle]) / 2
+    else:
+        median = inp_values[middle]
+    return round(median, ROUND_UP_TO_TWO)
+
+
+def _average(inp_values: list[int | float]) -> int | float:
+    return round(sum(inp_values) / max(len(inp_values), 1), ROUND_UP_TO_TWO)
+
+
+def _last_login_date(user: dict) -> datetime:
+    try:
+        return datetime.strptime(user[cnst.LAST_LOGIN], '%Y-%m-%d')
+    except KeyError:
+        raise exceptions_hw2.MissingFieldError(cnst.LAST_LOGIN)
+    except ValueError:
+        raise exceptions_hw2.InvalidDateFormatError(user[cnst.LAST_LOGIN])
+
+
+def calculate_ages_stats(users: dict) -> dict:
+    """Calculate statistics of the user ages.
+
+    Args:
+        users (dict): user input from JSON file.
+
+    Raises:
+        MissingFieldError: if field does not exist.
+
+    Returns:
+        dict: statistics of user ages.
+    """
+    try:
+        ages = [users[user][cnst.AGE] for user in users]
+    except Exception:
+        raise exceptions_hw2.MissingFieldError(cnst.AGE)
+
+    ages_processed = {}
+    ages_processed[cnst.AGE_MAX] = max(ages)
+    ages_processed[cnst.AGE_MIN] = min(ages)
+    ages_processed[cnst.AGE_AVERAGE] = _average(ages)
+    ages_processed[cnst.AGE_MEDIAN] = _median(ages)
+
+    return ages_processed
+
+
+def calculate_dates_stats(users: dict) -> dict:
+    """Calculate statistics of the dates of last user logins.
+
+    Args:
+        users (dict): user input from JSON file.
+
+    Returns:
+        dict: statistics of last user login, percent of total users.
+    """
+    last_login_dates = [_last_login_date(user) for user in users.values()]
+
+    last_login_statistics = {
+        cnst.LESS_TWO_DAYS: 0,
+        cnst.LESS_WEEK: 0,
+        cnst.LESS_MONTH: 0,
+        cnst.LESS_HALFYEAR: 0,
+        cnst.MORE_HALFYEAR: 0,
+    }
+
+    current_date = datetime.now()
+
+    for last_login_date in last_login_dates:
+        days_passed = (current_date - last_login_date).days
+        match days_passed:
+            case _ if days_passed < TWO_DAYS:
+                last_login_statistics[cnst.LESS_TWO_DAYS] += 1
+            case _ if days_passed < WEEK:
+                last_login_statistics[cnst.LESS_WEEK] += 1
+            case _ if days_passed < MONTH:
+                last_login_statistics[cnst.LESS_MONTH] += 1
+            case _ if days_passed < HALFYEAR:
+                last_login_statistics[cnst.LESS_HALFYEAR] += 1
+            case _:
+                last_login_statistics[cnst.MORE_HALFYEAR] += 1
+
+    quantity_users = len(last_login_dates)
+    for period in last_login_statistics.keys():
+        percent = round(last_login_statistics[period] / quantity_users * 100, ROUND_UP_TO_TWO)
+        last_login_statistics[period] = f'{percent}%'
+
+    return last_login_statistics
+
+
+def _aggregate_data(users: dict) -> dict:
+    age_stats = calculate_ages_stats(users)
+    data_stats = calculate_dates_stats(users)
+    return {
+        **age_stats,
+        **data_stats,
+    }
+
+
+def _save_data(input_path: str, output_path: str) -> None:
+    try:
+        with open(input_path, 'r') as input_file:
+            users = json.load(input_file)
+    except ValueError:
+        raise exceptions_hw2.InvalidJSONFormatError(input_path)
+    except FileNotFoundError:
+        raise exceptions_hw2.InputFilepathError(input_path)
+    if not users:
+        raise exceptions_hw2.EmptyJSONError(input_path)
+
+    statistics = _aggregate_data(users)
+    os.makedirs(os.path.abspath(os.path.dirname(output_path)), exist_ok=True)
+    with open(output_path, 'w') as out:
+        json.dump(statistics, out, indent=4)
+
+
+def process_data(input_path: str = 'input.json', output_path: str = 'output.json'):
+    """Process data from input json and write statistics to output json file.
+
+    Args:
+        input_path (str): path to source json file. Defaults to 'input.json'.
+        output_path (str): path to destionation json file. Defaults to 'output.json'.
+    """
+    try:
+        _save_data(input_path, output_path)
+    except Exception as error:
+        with open(output_path, 'w') as out:
+            json.dump({'ERROR': str(error)}, out, indent=4)

hw2/test_hw2.py

@@ -0,0 +1,82 @@
+"""Tests for hw2."""
+import json
+import tempfile
+from typing import Any
+
+import pytest
+
+import consts_hw2 as cnst
+import hw2
+
+ZEROS = '0.0%'
+ERROR = 'ERROR'
+CORRECT_DATA = (
+    ('tests/normal.json', {
+        cnst.AGE_MAX: 143,
+        cnst.AGE_MIN: 18,
+        cnst.AGE_AVERAGE: 80.5,
+        cnst.AGE_MEDIAN: 80.5,
+        cnst.LESS_TWO_DAYS: ZEROS,
+        cnst.LESS_WEEK: ZEROS,
+        cnst.LESS_MONTH: '50.0%',
+        cnst.LESS_HALFYEAR: ZEROS,
+        cnst.MORE_HALFYEAR: '50.0%',
+    }),
+    ('tests/normal2.json', {
+        cnst.AGE_MAX: 90,
+        cnst.AGE_MIN: 4,
+        cnst.AGE_AVERAGE: 45.0,
+        cnst.AGE_MEDIAN: 38.5,
+        cnst.LESS_TWO_DAYS: ZEROS,
+        cnst.LESS_WEEK: '16.67%',
+        cnst.LESS_MONTH: '16.67%',
+        cnst.LESS_HALFYEAR: '33.33%',
+        cnst.MORE_HALFYEAR: '33.33%',
+    }),
+)
+
+INVALID_DATA = (
+    ('tests/empty.json', {
+        ERROR: 'Empty json file was found at path tests/empty.json.',
+    }),
+    ('tests/incorrect_data.json', {
+        ERROR: 'Wrong date format, expected format: 2020-12-30, but got 2012-1001.',
+    }),
+    ('tests/invalid_format.json', {
+        'ERROR': 'An incorrect json file structure was found at path tests/invalid_format.json.',
+    }),
+    ('tests/no_age.json', {
+        ERROR: '\"age\" field does not exist. Every user should have \"age\" field.',
+    }),
+    ('tests/no_last_login.json', {
+        ERROR: '\"last_login\" field does not exist. Every user should have \"last_login\" field.',
+    }),
+)
+
+
+@pytest.mark.parametrize('input_path, expected', CORRECT_DATA)
+def test_correct_data(input_path: str, expected: dict[str, Any]):
+    """Checks that the process data function correctly calculates and saves data.
+
+    Args:
+        input_path (str): path to source json file.
+        expected (dict[str, Any]): expected processed stats.
+    """
+    with tempfile.NamedTemporaryFile() as output:
+        hw2.process_data(input_path, output.name)
+        got = json.load(output)
+        assert got == expected
+
+
+@pytest.mark.parametrize('input_path, expected', INVALID_DATA)
+def test_exceptions(input_path: str, expected: type):
+    """Checks that the process data function correctly calls exceptions.
+
+    Args:
+        input_path (str): path to source json file.
+        expected (type): exception that should be called.
+    """
+    with tempfile.NamedTemporaryFile() as output:
+        hw2.process_data(input_path, output.name)
+        got = json.load(output)
+        assert got.items() == expected.items()

hw2/tests/empty.json

@@ -0,0 +1,3 @@
+{
+
+}