.env.example

# Environment Configuration for Bank Statement Processor
# Copy this file to .env and customize for your environment

# =============================================================================
# Privacy and Telemetry Configuration
# =============================================================================
# Disable telemetry and tracking in all dependencies and third-party packages
# These environment variables ensure no usage data, analytics, or crash reports
# are sent to external services by any Python packages used in this application.

# Universal opt-out flags (recognized by many tools)
DO_NOT_TRACK=1
TELEMETRY_DISABLED=true

# Python package manager telemetry
PIP_NO_INPUT=1
PIP_DISABLE_PIP_VERSION_CHECK=1

# Pandas telemetry (defensive - not currently implemented but may be in future)
PANDAS_TELEMETRY_ENABLED=false

# Pytest plugins telemetry
PYTEST_DISABLE_PLUGIN_AUTOLOAD=1

# Locust telemetry (load testing tool)
LOCUST_DISABLE_TELEMETRY=1

# General analytics opt-out
ANALYTICS_DISABLED=true
NO_ANALYTICS=1

# Sentry/error reporting opt-out
SENTRY_DSN=

# Segment analytics opt-out
SEGMENT_WRITE_KEY=

# =============================================================================
# Logging Configuration
# =============================================================================
# Log level controls verbosity of application logging
# Valid values: DEBUG, INFO, WARNING, ERROR, CRITICAL
LOG_LEVEL=INFO

# =============================================================================
# Table Extraction Configuration
# =============================================================================
# Top Y-coordinate for table extraction area (in PDF points)
# Default: 300
TABLE_TOP_Y=300

# Bottom Y-coordinate for table extraction area (in PDF points)
# Default: 720
TABLE_BOTTOM_Y=720

# Enable dynamic boundary detection (experimental)
# When true, attempts to automatically detect table boundaries
# Default: false (recommended for production)
ENABLE_DYNAMIC_BOUNDARY=false

# =============================================================================
# Processing Configuration
# =============================================================================
# Sort transactions chronologically by date
# When true, transactions are sorted from oldest to newest
# Default: true
SORT_BY_DATE=true

# =============================================================================
# Output Configuration
# =============================================================================
# Output formats to generate
# Comma-separated list of formats: csv, json, excel
# All formats available in FREE tier
# Examples:
#   - csv                  (CSV only)
#   - csv,json             (CSV and JSON)
#   - csv,excel            (CSV and Excel)
#   - csv,json,excel       (all formats - default)
OUTPUT_FORMATS=csv,json,excel

# Column totals configuration
# Comma-separated list of column patterns to calculate totals for
# The processor will sum values in columns whose names contain these patterns (case-insensitive)
# Examples:
#   - debit,credit         (default - sums Debit and Credit columns)
#   - amount               (sums any column with "amount" in the name)
#   - debit,credit,balance (sums multiple column types)
# Set to empty string to disable totals: TOTALS_COLUMNS=
TOTALS_COLUMNS=debit,credit

# Generate monthly summary JSON file (aggregates transactions by month)
# Available in FREE tier - enabled by default
# Set to false to disable
# Default: true
GENERATE_MONTHLY_SUMMARY=true

# Generate expense analysis JSON file (recurring charges, anomalies, statistics)
# Default: true
GENERATE_EXPENSE_ANALYSIS=true

# =============================================================================
# Column Configuration (Advanced)
# =============================================================================
# Custom table column definitions in JSON format
# Defines column names and their X-coordinate boundaries [x_min, x_max]
# Example:
# TABLE_COLUMNS={"Date": [26, 78], "Details": [78, 255], "Debit €": [255, 313], "Credit €": [313, 369], "Balance €": [369, 450]}
#
# Leave commented to use default configuration
# TABLE_COLUMNS=

# =============================================================================
# Directory Configuration
# =============================================================================
# Project root directory (base directory for all relative paths)
# All relative paths (INPUT_DIR, OUTPUT_DIR, LOGS_DIR) are resolved relative to this
# If INPUT_DIR/OUTPUT_DIR/LOGS_DIR are absolute paths, PROJECT_ROOT is ignored
# Default: current working directory
# Examples:
#   PROJECT_ROOT=/data/bank-app          (all dirs under /data/bank-app/)
#   PROJECT_ROOT=/home/user/documents    (all dirs under /home/user/documents/)
# PROJECT_ROOT=

# Input directory containing PDF bank statements
# Can be absolute or relative (relative to PROJECT_ROOT if set, otherwise CWD)
# Default: input (relative to PROJECT_ROOT or CWD)
# Examples:
#   INPUT_DIR=input                      (./input or $PROJECT_ROOT/input)
#   INPUT_DIR=/mnt/statements/input      (absolute path, ignores PROJECT_ROOT)
INPUT_DIR=input

# Output directory for processed files
# Can be absolute or relative (relative to PROJECT_ROOT if set, otherwise CWD)
# Default: output (relative to PROJECT_ROOT or CWD)
OUTPUT_DIR=output

# Directory for processing activity logs (for audit trail)
# Can be absolute or relative (relative to PROJECT_ROOT if set, otherwise CWD)
# Default: logs (relative to PROJECT_ROOT or CWD)
LOGS_DIR=logs

# =============================================================================
# Template Configuration
# =============================================================================
# Directory containing built-in bank statement templates
# Default: ./templates (auto-detected)
# BANK_TEMPLATES_DIR=./templates

# Directory for custom user-defined templates (optional)
# Custom templates override built-in templates with same ID
# Default: none (not set)
# CUSTOM_TEMPLATES_DIR=./custom_templates

# Force a specific template by ID (bypasses auto-detection)
# Useful for testing or when auto-detection fails
# Example: FORCE_TEMPLATE=revolut
# Default: none (auto-detection enabled)
# FORCE_TEMPLATE=

# Set default template when no template is detected
# Default: first enabled template (alphabetical order)
# DEFAULT_TEMPLATE=default

# =============================================================================
# Licensing Configuration
# =============================================================================
# Path to PAID tier license file (JSON format)
# Default: ./license.json
# LICENSE_PATH=./license.json

# =============================================================================
# Docker Configuration
# =============================================================================
# Exit container after processing completes
# Set to true for one-time batch processing
# Set to false to keep container running for debugging
EXIT_AFTER_PROCESSING=true

# =============================================================================
# Data Retention & GDPR Compliance
# =============================================================================
# Number of days to retain output files (0 = no automatic cleanup)
# Files older than this will be deleted by cleanup-old-data command
# Recommended: 90 days for financial data
# Default: 0 (manual cleanup only)
DATA_RETENTION_DAYS=0

# Automatically cleanup output files on exit (after processing completes)
# When true, removes all output files after processing
# Useful for sensitive environments where data should not persist
# Default: false
AUTO_CLEANUP_ON_EXIT=false

# =============================================================================
# Build Configuration (Optional)
# =============================================================================
# These variables are used during Docker image builds for metadata
# If not set, they will default to empty strings (no warning)
# Typically set automatically by Makefile or CI/CD
# VERSION=1.0.1
# BUILD_DATE=2024-01-28T12:00:00Z
# VCS_REF=abc1234