Implement targets backend for bakepipe

.gitignore

@@ -4,6 +4,10 @@ sample-project/analysis_results.rds
 sample-project/report.txt
 sample-project/.bakepipe.state
 
+# Targets backend files
+_targets.R
+_targets/
+
 # R files
 .Rproj.user
 .Rhistory

DESCRIPTION

@@ -14,7 +14,7 @@ License: MIT + file LICENSE
 Encoding: UTF-8
 Roxygen: list(markdown = TRUE)
 RoxygenNote: 7.3.2
-Imports: callr, fs
+Imports: callr, fs, targets
 Suggests: testthat (>= 3.0.0)
 Config/testthat/edition: 3
 URL: https://vangberg.github.io/bakepipe/, https://github.com/vangberg/bakepipe

NAMESPACE

@@ -4,10 +4,9 @@ export(clean)
 export(external_in)
 export(file_in)
 export(file_out)
+export(generate_targets_file)
 export(run)
 export(status)
 importFrom(callr,r)
 importFrom(fs,path_rel)
 importFrom(stats,setNames)
-importFrom(utils,read.csv)
-importFrom(utils,write.csv)

R/clean.R

@@ -4,7 +4,8 @@
 #' intermediate files. This provides a complete clean of generated artifacts.
 #' The pipeline can be regenerated by running run() again.
 #'
-#' @param verbose Logical. If TRUE (default), prints progress messages to console.
+#' @param verbose Logical. If TRUE (default), prints progress messages
+#'   to console.
 #' @return Character vector of file paths that were actually removed
 #' @export
 #' @examples
@@ -13,94 +14,50 @@
 #' dir.create(temp_dir)
 #' sample_proj <- system.file("extdata", "sample-project", package = "bakepipe")
 #' file.copy(sample_proj, temp_dir, recursive = TRUE)
-#' 
+#'
 #' # Change to the sample project directory
 #' old_wd <- getwd()
 #' setwd(file.path(temp_dir, "sample-project"))
-#' 
+#'
 #' # Run the pipeline first to create output files
-#' run(verbose = FALSE)
-#' 
+#' run()
+#'
 #' # Now clean up the generated files
 #' removed_files <- clean()
 #' print(removed_files)
-#' 
+#'
 #' # Restore working directory and clean up
 #' setwd(old_wd)
 #' unlink(temp_dir, recursive = TRUE)
 clean <- function(verbose = TRUE) {
-  # Parse all scripts to get dependencies
-  dependencies <- parse()
-
-  # If no scripts found, return empty vector
-  if (length(dependencies$scripts) == 0) {
-    if (verbose) {
-      message("\n[CLEAN] \033[1;36mBakepipe Clean\033[0m")
-      message("\033[33m   No output files found to clean\033[0m\n")
+  deps <- parse()
+
+  # Collect all output files
+  files <- unique(unlist(lapply(deps$scripts, function(s) s$outputs)))
+
+  # Add targets artifacts
+  files <- c(files, "_targets.R")
+
+  # Track removed files
+  removed_files <- character()
+  removed <- 0
+  for (f in files) {
+    if (file.exists(f)) {
+      file.remove(f)
+      removed_files <- c(removed_files, f)
+      removed <- removed + 1
     }
-    return(character(0))
   }
 
-  # Get all outputs from the top-level outputs list
-  all_outputs <- dependencies$outputs
-
-  # Only try to remove files that actually exist
-  existing_files <- all_outputs[file.exists(all_outputs)]
-
-  # Print header
-  if (verbose) {
-    message("\n[CLEAN] \033[1;36mBakepipe Clean\033[0m")
-
-    if (length(existing_files) == 0) {
-      message("\033[32m[OK] No output files to clean - all clean!\033[0m\n")
-      return(character(0))
-    }
-
-    message(paste0("\033[33m   Found ", length(existing_files), " output file",
-               if(length(existing_files) > 1) "s" else "", " to remove\033[0m\n"))
-  } else if (length(existing_files) == 0) {
-    return(character(0))
+  # Remove _targets directory
+  if (dir.exists("_targets")) {
+    unlink("_targets", recursive = TRUE)
+    removed <- removed + 1
   }
 
-  # Calculate max filename width for alignment
-  max_width <- max(nchar(existing_files))
-
-  # Remove the files with progress indicators
-  removed_files <- character(0)
-  failed_files <- character(0)
-
-  for (file_path in existing_files) {
-    if (file.remove(file_path)) {
-      removed_files <- c(removed_files, file_path)
-      if (verbose) {
-        message(sprintf("\033[31m[RM]  %s\033[0m", file_path))
-      }
-    } else {
-      failed_files <- c(failed_files, file_path)
-      if (verbose) {
-        message(sprintf("\033[33m[!]  %-*s \033[2m(failed to remove)\033[0m", max_width, file_path))
-      }
-    }
-  }
-
-  # Print summary
-  if (verbose) {
-    message("\n\033[1;36m[SUMMARY]\033[0m")
-
-    if (length(removed_files) > 0) {
-      message(sprintf("\033[32m   Removed %d file%s\033[0m", 
-                  length(removed_files),
-                  if(length(removed_files) > 1) "s" else ""))
-    }
-
-    if (length(failed_files) > 0) {
-      message(sprintf("\033[33m   Failed to remove %d file%s\033[0m", 
-                  length(failed_files),
-                  if(length(failed_files) > 1) "s" else ""))
-    }
-
-    message("")
+  if (verbose && removed > 0) {
+    message(sprintf("✓ removed %d artifact%s", removed, if (removed != 1) "s" else ""))
   }
 
-  removed_files
-}
+  invisible(removed_files)
+}

R/generate_targets_file.R

@@ -0,0 +1,189 @@
+#' Generate _targets.R file from bakepipe scripts
+#'
+#' Parses all R scripts in the project and generates a _targets.R file
+#' that defines targets for the targets package. This allows bakepipe
+#' to use targets as a backend for pipeline execution.
+#'
+#' @return Invisibly returns the path to the generated _targets.R file
+#' @importFrom callr r
+#' @export
+generate_targets_file <- function() {
+  # Parse all scripts to get dependencies
+  parsed <- parse()
+
+  # Get project root
+  project_root <- root()
+  targets_file <- file.path(project_root, "_targets.R")
+
+  # Start building the targets file content
+  lines <- character(0)
+
+  # Add library calls
+  lines <- c(lines, "library(targets)")
+  lines <- c(lines, "")
+
+  # Add the list of targets
+  lines <- c(lines, "list(")
+
+  # Generate targets for each script and its dependencies
+  target_lines <- character(0)
+
+  # Track all files that need file targets
+  all_file_targets <- character(0)
+
+  # First pass: add all external input file targets
+  for (script_name in names(parsed$scripts)) {
+    script_info <- parsed$scripts[[script_name]]
+
+    for (ext_file in script_info$externals) {
+      ext_target_name <- path_to_target_name(ext_file, "")
+      if (!(ext_target_name %in% all_file_targets)) {
+        target_lines <- c(
+          target_lines,
+          sprintf(
+            '  tar_target(%s, "%s", format = "file"),',
+            ext_target_name,
+            ext_file
+          )
+        )
+        all_file_targets <- c(all_file_targets, ext_target_name)
+      }
+    }
+  }
+
+  # Second pass: generate one target per script
+  for (script_name in names(parsed$scripts)) {
+    script_info <- parsed$scripts[[script_name]]
+
+    # Generate target name from script path
+    output_target_name <- path_to_target_name(script_name, "output")
+
+    # Collect dependencies
+    deps <- character(0)
+
+    # Track the script itself so changes are detected
+    script_target_name <- path_to_target_name(script_name, "script")
+    target_lines <- c(
+      target_lines,
+      sprintf(
+        '  tar_target(%s, "%s", format = "file"),',
+        script_target_name,
+        script_name
+      )
+    )
+    deps <- c(deps, script_target_name)
+
+    # Add external input dependencies
+    for (ext_file in script_info$externals) {
+      ext_target_name <- path_to_target_name(ext_file, "")
+      deps <- c(deps, ext_target_name)
+    }
+
+    # Add input file dependencies (from other scripts)
+    for (input_file in script_info$inputs) {
+      # Search through all scripts to find which one produces this input
+      for (producer_script in names(parsed$scripts)) {
+        if (input_file %in% parsed$scripts[[producer_script]]$outputs) {
+          # Found the producer script - depend on its output target
+          input_target_name <- path_to_target_name(producer_script, "output")
+          deps <- c(deps, input_target_name)
+          break
+        }
+      }
+    }
+
+    # Build the script target
+    target_start <- c(
+      sprintf("  tar_target("),
+      sprintf("    %s,", output_target_name),
+      sprintf("    {")
+    )
+
+    # Add dependencies
+    dep_lines <- character(0)
+    for (dep in deps) {
+      dep_lines <- c(dep_lines, sprintf("      %s", dep))
+    }
+
+    # Add script execution via callr for isolation
+    exec_lines <- c(
+      sprintf("      callr::r("),
+      sprintf("        func = function(script_path) {"),
+      sprintf("          source(script_path, local = TRUE)"),
+      sprintf("        },"),
+      sprintf('        args = list(script_path = "%s")', script_name),
+      sprintf("      )")
+    )
+
+    # Add output vector if script has outputs
+    output_lines <- character(0)
+    if (length(script_info$outputs) > 0) {
+      output_files_r <- paste0('"', script_info$outputs, '"', collapse = ", ")
+      output_lines <- c(output_lines, sprintf("      c(%s)", output_files_r))
+    } else {
+      # No outputs - return empty character vector
+      output_lines <- c(output_lines, sprintf("      character(0)"))
+    }
+
+    target_end <- c(
+      sprintf("    },"),
+      sprintf("    format = \"file\""),
+      sprintf("  ),")
+    )
+
+    # Combine all parts
+    full_target <- c(target_start, dep_lines, exec_lines, output_lines, target_end)
+    target_lines <- c(target_lines, full_target)
+
+    all_file_targets <- c(all_file_targets, output_target_name)
+  }
+
+  # Remove trailing comma from last target
+  if (length(target_lines) > 0) {
+    last_line <- target_lines[length(target_lines)]
+    target_lines[length(target_lines)] <- sub(",$", "", last_line)
+  }
+
+  # Combine everything
+  lines <- c(lines, target_lines)
+  lines <- c(lines, ")")
+
+  # Write to file
+  writeLines(lines, targets_file)
+
+  invisible(targets_file)
+}
+
+#' Convert file path to valid R target name
+#'
+#' Converts a file path to a valid R identifier for use as a target name.
+#' Special characters are replaced with underscores.
+#'
+#' @param path File path
+#' @param prefix Prefix to add (e.g., "script", "run", or "")
+#' @return Valid R identifier
+#' @keywords internal
+path_to_target_name <- function(path, prefix = "") {
+  # Convert to lowercase and replace special chars with underscores
+  name <- tolower(path)
+
+  # Replace directory separators, dots (except in extensions), hyphens
+  # and other special characters with underscores
+  name <- gsub("[/\\.-]", "_", name)
+
+  # Remove any remaining non-alphanumeric characters
+  name <- gsub("[^a-z0-9_]", "_", name)
+
+  # Remove consecutive underscores
+  name <- gsub("_+", "_", name)
+
+  # Remove leading/trailing underscores
+  name <- gsub("^_|_$", "", name)
+
+  # Add prefix if provided
+  if (nzchar(prefix)) {
+    name <- paste(prefix, name, sep = "_")
+  }
+
+  name
+}