#!/bin/bash

# Script to generate synthetic patients directly using modules created by the pipeline
# Usage: ./generate_patients.sh <module_name> <output_directory> <population_size>

MODULE_NAME=$1
OUTPUT_DIR=$2
POPULATION=${3:-10}

if [ -z "$MODULE_NAME" ] || [ -z "$OUTPUT_DIR" ]; then
  echo "Usage: $0 <module_name> <output_directory> [population_size]"
  echo "Example: $0 diabetes /path/to/output 20"
  exit 1
fi

# Create output directory
mkdir -p "$OUTPUT_DIR/m" "$OUTPUT_DIR/f"

# Location of module file
MODULE_PATH="/data/olamide/synthea-alldiseases/modules/${MODULE_NAME}.json"

if [ ! -f "$MODULE_PATH" ]; then
  echo "Module file not found: $MODULE_PATH"
  exit 1
fi

# Create a temporary directory for the container output
TEMP_DIR=$(mktemp -d)
echo "Created temporary directory: $TEMP_DIR"

# Run for male patients
echo "Generating male patients..."
docker run --rm \
  -v "$MODULE_PATH:/app/modules/${MODULE_NAME}.json" \
  -v "$TEMP_DIR:/app/output" \
  synthea-module-generator \
  bash -c "cd /app && ./run_synthea -p $((POPULATION/2)) -g M -m ${MODULE_NAME} Massachusetts"

# Copy male patient files to the output directory
echo "Copying male patient files..."
find "$TEMP_DIR/fhir" -name "*.json" ! -name "*hospital*" ! -name "*practitioner*" | while read file; do
  # Check if it's a patient file by looking for gender field
  if grep -q '"gender"' "$file"; then
    cp "$file" "$OUTPUT_DIR/m/"
  fi
done

# Clear the temp directory
rm -rf "$TEMP_DIR/fhir"/*

# Run for female patients
echo "Generating female patients..."
docker run --rm \
  -v "$MODULE_PATH:/app/modules/${MODULE_NAME}.json" \
  -v "$TEMP_DIR:/app/output" \
  synthea-module-generator \
  bash -c "cd /app && ./run_synthea -p $((POPULATION/2)) -g F -m ${MODULE_NAME} Massachusetts"

# Copy female patient files to the output directory
echo "Copying female patient files..."
find "$TEMP_DIR/fhir" -name "*.json" ! -name "*hospital*" ! -name "*practitioner*" | while read file; do
  # Check if it's a patient file by looking for gender field
  if grep -q '"gender"' "$file"; then
    cp "$file" "$OUTPUT_DIR/f/"
  fi
done

# Count the results
male_count=$(find "$OUTPUT_DIR/m" -type f -name "*.json" | wc -l)
female_count=$(find "$OUTPUT_DIR/f" -type f -name "*.json" | wc -l)

# Report results
echo "Patient generation complete. Results saved to $OUTPUT_DIR"
echo "Male patients: $male_count"
echo "Female patients: $female_count"

# Clean up temp directory
rm -rf "$TEMP_DIR"