scripts/audit/_common.py

"""Shared helpers for audit scripts."""

import sys
from pathlib import Path
from datetime import datetime

import pandas as pd
import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import seaborn as sns
from scipy import stats
import warnings

warnings.filterwarnings('ignore')
sns.set_style('whitegrid')
plt.rcParams['figure.figsize'] = (14, 6)

PROJECT_ROOT = Path(__file__).resolve().parent.parent.parent
PARQUET_DIR = PROJECT_ROOT / 'output' / 'parquet'
DATA_DIR = PROJECT_ROOT / 'data'
OUTPUT_BASE = PROJECT_ROOT / 'output' / 'audit'


def setup_output_dir(dataset_name):
    out = OUTPUT_BASE / dataset_name
    out.mkdir(parents=True, exist_ok=True)
    return out


def save_fig(output_dir, name):
    path = output_dir / f'{name}.png'
    plt.savefig(path, dpi=150, bbox_inches='tight')
    print(f'  -> saved {path.relative_to(PROJECT_ROOT)}')
    plt.close()


def load_parquet(dataset_name):
    if dataset_name == 'users':
        path = PARQUET_DIR / 'users.parquet'
    elif dataset_name == 'cardS':
        path = PARQUET_DIR / 'cardS_cleaned.parquet'
    elif dataset_name == 'cardD':
        path = PARQUET_DIR / 'cardD_cleaned.parquet'
    else:
        path = PARQUET_DIR / f'{dataset_name}_cleaned.parquet'
    print(f'Loading {path.name}...')
    df = pd.read_parquet(path)
    mem = df.memory_usage(deep=True).sum()
    unit, divisor = ('GB', 1024**3) if mem > 1024**3 else ('MB', 1024**2)
    print(f'  {len(df):,} rows, {mem / divisor:.2f} {unit}')
    return df


def print_section(title):
    print()
    print('=' * 80)
    print(title)
    print('=' * 80)


def standard_temporal_analysis(df, dataset_title, output_dir):
    """Yearly bar chart + gap detection. Returns (yearly_counts, gaps)."""
    print_section('TEMPORAL ANALYSIS')

    ts_col = next((c for c in ['timestamp', 'start_time'] if c in df.columns), None)
    if not ts_col:
        print('No timestamp column found — skipping temporal analysis')
        return pd.Series(dtype=int), []

    df['timestamp'] = pd.to_datetime(df[ts_col])
    df['date'] = df['timestamp'].dt.date
    df['year'] = df['timestamp'].dt.year

    print(f'First record: {df["timestamp"].min()}')
    print(f'Last record:  {df["timestamp"].max()}')
    print(f'Time span:    {(df["timestamp"].max() - df["timestamp"].min()).days} days')
    print(f'Years:        {df["year"].min()} - {df["year"].max()}')

    yearly_counts = df['year'].value_counts().sort_index()
    print('\nYEARLY DISTRIBUTION')
    print(yearly_counts)

    plt.figure(figsize=(14, 6))
    yearly_counts.plot(kind='bar', color='steelblue')
    plt.title(f'{dataset_title} - Yearly Distribution', fontsize=14, fontweight='bold')
    plt.xlabel('Year')
    plt.ylabel('Number of Records')
    plt.xticks(rotation=45)
    plt.grid(axis='y', alpha=0.3)
    plt.tight_layout()
    save_fig(output_dir, 'yearly_distribution')

    # Gap detection
    daily_counts = df.groupby('date').size()
    all_dates = pd.date_range(start=daily_counts.index.min(),
                              end=daily_counts.index.max(), freq='D')
    missing_dates = pd.to_datetime(all_dates.difference(pd.to_datetime(daily_counts.index)))

    gaps = []
    if len(missing_dates) > 0:
        gap_start = missing_dates[0]
        gap_end = missing_dates[0]
        for i in range(1, len(missing_dates)):
            if (missing_dates[i] - missing_dates[i - 1]).days == 1:
                gap_end = missing_dates[i]
            else:
                if (gap_end - gap_start).days >= 7:
                    gaps.append((gap_start, gap_end, (gap_end - gap_start).days))
                gap_start = missing_dates[i]
                gap_end = missing_dates[i]
        if (gap_end - gap_start).days >= 7:
            gaps.append((gap_start, gap_end, (gap_end - gap_start).days))

    if gaps:
        print(f'\nFound {len(gaps)} gaps >= 7 days:')
        for start, end, days in gaps:
            print(f'  {start.date()} to {end.date()} ({days} days)')
    else:
        print('\nNo significant temporal gaps (all gaps < 7 days)')

    return yearly_counts, gaps


def standard_missing_data_analysis(df, output_dir):
    """Missing data summary + chart."""
    print_section('MISSING DATA ANALYSIS')

    missing = df.isnull().sum()
    missing_pct = (missing / len(df) * 100).round(2)
    missing_df = pd.DataFrame({
        'Missing Count': missing,
        'Missing %': missing_pct
    }).sort_values('Missing %', ascending=False)

    has_missing = missing_df[missing_df['Missing Count'] > 0]
    if len(has_missing) > 0:
        print(has_missing)

        top_missing = has_missing.head(10)
        plt.figure(figsize=(12, 6))
        top_missing['Missing %'].plot(kind='barh', color='coral')
        plt.title('Top 10 Columns by Missing Data %', fontweight='bold')
        plt.xlabel('Missing %')
        plt.gca().invert_yaxis()
        plt.tight_layout()
        save_fig(output_dir, 'missing_data')
    else:
        print('No missing data found.')

    return missing_df


def standard_audit_summary(dataset_title, df, extra_lines=None):
    """Print a summary block at the end of an audit."""
    print_section(f'AUDIT SUMMARY: {dataset_title}')
    print(f'Audit Date:    {datetime.now().strftime("%Y-%m-%d")}')
    print(f'Total Rows:    {len(df):,}')
    ts_col = next((c for c in ['timestamp', 'start_time'] if c in df.columns), None)
    if ts_col:
        print(f'Date Range:    {df[ts_col].min().date()} to {df[ts_col].max().date()}')
    uid_col = next((c for c in ['user_id_hash', 'user_id', 'username'] if c in df.columns), None)
    if uid_col:
        print(f'Unique Users:  {df[uid_col].nunique():,}')
    if extra_lines:
        for line in extra_lines:
            print(line)
    print('=' * 80)