.env.example

# ------------------------------------------------------------------------------
# Recommended Environment Variables (can be set via CLI, except for the passwords)
# ------------------------------------------------------------------------------

# Database Configuration - REQUIRED
# You can either set these individual database parameters:
LEMATERIALFETCHER_DB_USER=username
LEMATERIALFETCHER_DB_PASSWORD=password
LEMATERIALFETCHER_DB_HOST=localhost
LEMATERIALFETCHER_DB_NAME=lematerial
# Or alternatively set a complete connection string (which takes precedence):
# LEMATERIALFETCHER_DB_CONN_STR=host=localhost user=username password=password dbname=database_name sslmode=disable

# Table to store fetched data
LEMATERIALFETCHER_TABLE_NAME=raw_materials_source

# MySQL Configuration (for OQMD)
LEMATERIALFETCHER_MYSQL_HOST=localhost
LEMATERIALFETCHER_MYSQL_USER=mysql_username
LEMATERIALFETCHER_MYSQL_PASSWORD=mysql_password
LEMATERIALFETCHER_MYSQL_DATABASE=oqmd
# LEMATERIALFETCHER_MYSQL_CERT_PATH=./path/to/cert

# Transformer Configuration
LEMATERIALFETCHER_SOURCE_TABLE_NAME=raw_materials_source
LEMATERIALFETCHER_TASK_TABLE_NAME=raw_materials_task

# Destination database - If not provided, source database credentials will be used
# Uncomment and set these if destination database is different from source
# LEMATERIALFETCHER_DEST_DB_USER=dest_username
# LEMATERIALFETCHER_DEST_DB_PASSWORD=dest_password
# LEMATERIALFETCHER_DEST_DB_HOST=localhost
# LEMATERIALFETCHER_DEST_DB_NAME=dest_database
LEMATERIALFETCHER_DEST_TABLE_NAME=optimade_materials


# ------------------------------------------------------------------------------
# Optional Environment Variables
# ------------------------------------------------------------------------------

# Directory for log files
# LEMATERIALFETCHER_LOG_DIR=./logs

# Maximum retry attempts for failed requests
# LEMATERIALFETCHER_MAX_RETRIES=3

# Delay between retries in seconds
# LEMATERIALFETCHER_RETRY_DELAY=2

# Number of parallel workers
# LEMATERIALFETCHER_NUM_WORKERS=4

# Number of items to fetch per page
# LEMATERIALFETCHER_PAGE_LIMIT=20

# Starting page offset
# LEMATERIALFETCHER_PAGE_OFFSET=0

# API Configuration
# Base URL for the API endpoint - REQUIRED
# Examples for different data sources:
# Alexandria structures PBE: https://alexandria.icams.rub.de/pbesol/v1/structures
# Alexandria trajectories PBE: https://alexandria.icams.rub.de/data/pbe/geo_opt_paths/
# OQMD: https://oqmd.org/download/

# LEMATERIALFETCHER_API_BASE_URL=https://example.com/api

# Directory for downloading OQMD data
# LEMATERIALFETCHER_OQMD_DOWNLOAD_DIR=./data

# Materials Project Configuration
# For structure data:

# LEMATERIALFETCHER_MP_BUCKET_NAME=materialsproject-build
# LEMATERIALFETCHER_MP_BUCKET_PREFIX=collections

# For task data:

# LEMATERIALFETCHER_MP_BUCKET_NAME=materialsproject-parsed
# LEMATERIALFETCHER_MP_BUCKET_PREFIX=tasks_atomate2

# Transformer processing settings
# LEMATERIALFETCHER_BATCH_SIZE=500
# LEMATERIALFETCHER_OFFSET=0
# LEMATERIALFETCHER_LOG_EVERY=1000

# ------------------------------------------------------------------------------
# LeMatRho Configuration (charge density pipeline)
# ------------------------------------------------------------------------------

# AWS credentials for authenticated S3 access (LeMatRho bucket)
# AWS_ACCESS_KEY_ID=your_access_key
# AWS_SECRET_ACCESS_KEY=your_secret_key
# AWS_DEFAULT_REGION=us-east-1

# LeMatRho S3 bucket name
# LEMATERIALFETCHER_LEMATRHO_BUCKET_NAME=lemat-rho

# VASP pseudopotential directory (required for Bader/DDEC6 analysis)
# PMG_VASP_PSP_DIR=/path/to/vasp/pseudopotentials

# External tool paths (optional, auto-detected on PATH if not set)
# LEMATERIALFETCHER_BADER_PATH=/path/to/bader
# LEMATERIALFETCHER_CHARGEMOL_PATH=/path/to/chargemol
# LEMATERIALFETCHER_ATOMIC_DENSITIES_PATH=/path/to/atomic_densities

# HuggingFace (for pushing dataset after pipeline completes)
# HF_REPO_ID=your-org/lematrho-dataset
# HF_TOKEN=hf_your_token