SmartCane/r_app/experiments/legacy_package_management/extract_current_versions.R

#' Version Extractor for SmartCane Project
#' 
#' This script scans your R scripts to find all used packages and extracts
#' the currently installed versions. Use this to populate the package_manager.R
#' with your actual working versions.
#' 
#' Usage:
#'   source("extract_current_versions.R")
#' 
#' Author: SmartCane Team
#' Date: 2025-06-24

# =============================================================================
# PACKAGE DISCOVERY
# =============================================================================

#' Extract packages from R scripts
extract_packages_from_scripts <- function(script_dir = ".") {
  # Find all R files
  r_files <- list.files(script_dir, pattern = "\\.(R|Rmd)$", recursive = TRUE, full.names = TRUE)
  
  packages <- c()
  
  for (file in r_files) {
    cat("Scanning:", file, "\n")
    
    tryCatch({
      content <- readLines(file, warn = FALSE)
      
      # Find library() calls
      library_matches <- regmatches(content, regexpr('library\\(["\']?([^"\'\\)]+)["\']?\\)', content))
      library_packages <- gsub('library\\(["\']?([^"\'\\)]+)["\']?\\)', '\\1', library_matches)
      library_packages <- library_packages[library_packages != ""]
      
      # Find require() calls  
      require_matches <- regmatches(content, regexpr('require\\(["\']?([^"\'\\)]+)["\']?\\)', content))
      require_packages <- gsub('require\\(["\']?([^"\'\\)]+)["\']?\\)', '\\1', require_matches)
      require_packages <- require_packages[require_packages != ""]
      
      # Find package::function calls
      namespace_matches <- regmatches(content, gregexpr('[a-zA-Z][a-zA-Z0-9.]*::', content))
      namespace_packages <- unique(unlist(lapply(namespace_matches, function(x) gsub('::', '', x))))
      namespace_packages <- namespace_packages[namespace_packages != ""]
      
      packages <- c(packages, library_packages, require_packages, namespace_packages)
      
    }, error = function(e) {
      cat("Error reading", file, ":", e$message, "\n")
    })
  }
  
  # Clean and deduplicate
  packages <- unique(packages)
  packages <- packages[!packages %in% c("", "base", "stats", "utils", "graphics", "grDevices")]
  
  return(sort(packages))
}

#' Get current version of installed packages
get_current_versions <- function(packages) {
  versions <- list()
  
  cat("\nChecking installed versions...\n")
  cat("===============================\n")
  
  for (pkg in packages) {
    if (pkg %in% rownames(installed.packages())) {
      version <- as.character(packageVersion(pkg))
      versions[[pkg]] <- version
      cat(sprintf("✓ %-20s %s\n", pkg, version))
    } else {
      cat(sprintf("✗ %-20s NOT INSTALLED\n", pkg))
    }
  }
  
  return(versions)
}

#' Generate package manager configuration
generate_package_config <- function(versions) {
  cat("\n\nGenerating REQUIRED_PACKAGES configuration...\n")
  cat("=============================================\n\n")
  
  config_lines <- c(
    "# Package requirements with your current working versions",
    "REQUIRED_PACKAGES <- list("
  )
  
  # Group packages by category
  categories <- list(
    "Core data manipulation" = c("here", "dplyr", "tidyr", "readr", "readxl", "magrittr", "lubridate", "stringr"),
    "Spatial data" = c("sf", "terra", "exactextractr", "raster", "sp", "sf", "rgdal", "rgeos"),
    "Visualization" = c("tmap", "ggplot2", "RColorBrewer", "viridis", "scales"),
    "Statistical analysis" = c("lme4", "nlme", "mgcv", "survival", "cluster"),
    "Reporting" = c("knitr", "rmarkdown", "officedown", "officer", "flextable"),
    "Tidyverse" = c("tidyverse", "purrr", "forcats", "tibble"),
    "Other packages" = c()
  )
  
  # Categorize packages
  categorized <- list()
  uncategorized <- names(versions)
  
  for (category in names(categories)) {
    cat_packages <- intersect(names(versions), categories[[category]])
    if (length(cat_packages) > 0) {
      categorized[[category]] <- cat_packages
      uncategorized <- setdiff(uncategorized, cat_packages)
    }
  }
  
  # Add uncategorized packages
  if (length(uncategorized) > 0) {
    categorized[["Other packages"]] <- uncategorized
  }
  
  # Generate config
  for (category in names(categorized)) {
    config_lines <- c(config_lines, paste0("  # ", category))
    
    packages_in_cat <- categorized[[category]]
    for (i in seq_along(packages_in_cat)) {
      pkg <- packages_in_cat[i]
      version <- versions[[pkg]]
      comma <- if (i == length(packages_in_cat) && category == names(categorized)[length(categorized)]) "" else ","
      
      # Add special comment for critical packages
      comment <- ""
      if (pkg == "tmap") comment <- "  # CRITICAL: for tm_scale_continuous() syntax"
      if (pkg == "terra") comment <- "  # CRITICAL: for raster processing"
      
      config_lines <- c(config_lines, sprintf('  "%s" = "%s"%s%s', pkg, version, comma, comment))
    }
    
    if (category != names(categorized)[length(categorized)]) {
      config_lines <- c(config_lines, "")
    }
  }
  
  config_lines <- c(config_lines, ")")
  
  # Print to console
  cat(paste(config_lines, collapse = "\n"))
  
  # Save to file
  writeLines(config_lines, "generated_package_config.R")
  cat("\n\n📁 Configuration saved to: generated_package_config.R\n")
  
  return(config_lines)
}

# =============================================================================
# MAIN EXECUTION
# =============================================================================

main <- function() {
  cat("🔍 SmartCane Package Version Extractor\n")
  cat("======================================\n\n")
  
  # Step 1: Find all packages used in scripts
  cat("Step 1: Scanning R scripts for package usage...\n")
  packages <- extract_packages_from_scripts()
  
  cat("\nFound packages:\n")
  cat(paste(packages, collapse = ", "), "\n")
  cat("\nTotal packages found:", length(packages), "\n")
  
  # Step 2: Get current versions
  cat("\nStep 2: Checking installed versions...\n")
  versions <- get_current_versions(packages)
  
  installed_count <- length(versions)
  missing_count <- length(packages) - installed_count
  
  cat(sprintf("\n📊 Summary: %d installed, %d missing\n", installed_count, missing_count))
  
  if (missing_count > 0) {
    missing_packages <- setdiff(packages, names(versions))
    cat("\n⚠️  Missing packages:\n")
    cat(paste(missing_packages, collapse = ", "), "\n")
    cat("\nYou may want to install these first, then re-run this script.\n")
  }
  
  # Step 3: Generate configuration
  if (length(versions) > 0) {
    cat("\nStep 3: Generating package manager configuration...\n")
    config <- generate_package_config(versions)
    
    cat("\n✅ Next steps:\n")
    cat("1. Review generated_package_config.R\n")
    cat("2. Copy the REQUIRED_PACKAGES list to package_manager.R\n")
    cat("3. Adjust any versions as needed\n")
    cat("4. Run package_manager.R\n")
  } else {
    cat("\n❌ No installed packages found. Install packages first.\n")
  }
}

# Run the extraction
main()