Merge pull request #87 from uw-ssec/carlos/issue75

feat: Support data transformation "verbs" in the docker container #77

CONTRIBUTING.md

@@ -126,7 +126,10 @@ docker run --rm \
   -v /path/to/data-raw/:/home/biodiversity-horizons/data-raw/ \
   -v /path/to/outputs/:/home/biodiversity-horizons/outputs/ \
   biodiversity-horizons \
-  Rscript scripts/exposure_workflow.R /home/biodiversity-horizons/data-raw multicore 4
+  exposure \
+  -d /home/biodiversity-horizons/data-raw \
+  -p multicore \
+  -w 4
 ```
 
 This command:
@@ -151,7 +154,7 @@ Example:
 ```bash
 docker run --rm \
   -v $(pwd)/data-raw/:/home/biodiversity-horizons/data-raw/ \
-  biodiversity-horizons -d ./data-raw
+  biodiversity-horizons exposure -d ./data-raw
 ```
 
 ## Pull Requests

DESCRIPTION

@@ -23,7 +23,8 @@ Imports:
     logger,
     tidyverse,
     optparse,
-    here
+    here,
+    stars
 Suggests:
     testthat (>= 3.0.0)
 Config/testthat/edition: 3

Dockerfile

@@ -33,11 +33,10 @@ COPY NAMESPACE .
 # and the image builds much faster when changing code
 RUN Rscript -e "remotes::install_local('.', dependencies=TRUE)" # Install dependencies
 COPY R ./R
+COPY utility ./utility
 RUN Rscript -e "remotes::install_local('.', dependencies=TRUE)" # install package code
 
 COPY scripts ./scripts
 
-# Run the script with "data-raw/" as path since that is where, run_container.sh will mount the data
-# We'll also use "multisession", and (availableCores()-1) workers as default
-# The user can override by passing in arguments at runtime, e.g.: "multisession" 4
+# Run the main script, which can take arguments to determine the workflow to run
 ENTRYPOINT ["Rscript", "scripts/main.R"]

docker_exposure.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+source run_container.sh
+
+if [ $# -lt 2 ]; then
+    echo "usage: $0 <data_dir> <output_dir> <extra-args>"
+    exit 1
+fi
+
+DATA_DIR=$1
+OUT_DIR=$2
+shift 2
+run_with_mounts $DATA_DIR $OUT_DIR exposure -d data-raw $*

docker_shp2rds.sh

@@ -0,0 +1,13 @@
+#!/bin/bash
+source run_container.sh
+
+if [ $# -lt 2 ]; then
+    echo "Usage: $0 <input_file> <output__file> <extra-args>"
+    exit 1
+fi
+INPUT_FILE=$1
+OUT_FILE=$2
+shift 2
+run_with_mounts $(dirname $INPUT_FILE) $(dirname $OUT_FILE) \
+shp2rds -i ./data-raw/$(basename $INPUT_FILE) \
+-o ./outputs/$(basename $OUT_FILE) $*

run_container.sh

@@ -1,30 +1,27 @@
 #!/bin/bash
-# This script runs the Docker container with the provided directories for input and output data
+# CONTAINER=biodiv # for development, change this to the name of your local Docker ccontainer build
+CONTAINER=ghcr.io/uw-ssec/biodiversity-horizons
 
-# Check if two arguments are provided
-if [ "$#" -lt 2 ]; then
-    echo "Usage: $0 <data dir> <output dir>"
-    exit 1
-fi
+function run_with_mounts() {
 
-# Assign arguments to variables
-DATA_DIR=$1
-OUT_DIR=$2
-shift 2
+    # run the container with the given directories mounted as volumes
+    # and then pass any additional arguments to the container
+    DATA_DIR=$1
+    OUT_DIR=$2
+    shift 2
 
-# Check if directories exist
-if [ ! -d "$DATA_DIR" ]; then
-    echo "Error: Directory $DATA_DIR does not exist."
-    exit 1
-fi
+    # Check if directories exist
+    if [ ! -d "$DATA_DIR" ]; then
+        echo "Error: Directory $DATA_DIR does not exist."
+        exit 1
+    fi
 
-if [ ! -d "$OUT_DIR" ]; then
-    echo "Warning: Directory $OUT_DIR does not exist, creating it ..."
-    mkdir -p "$OUT_DIR"
-fi
+    if [ ! -d "$OUT_DIR" ]; then
+        echo "Warning: Directory $OUT_DIR does not exist, creating it ..."
+        mkdir -p "$OUT_DIR"
+    fi
 
-# Run the Docker container with the provided directories mounted as volumes
-# TODO: Replace with the published Docker image name
-docker run -v "$DATA_DIR":/home/biodiversity-horizons/data-raw \
--v "$OUT_DIR":/home/biodiversity-horizons/outputs \
-ghcr.io/uw-ssec/biodiversity-horizons $* # Pass any additional arguments to the Docker container
+    docker run -v "$DATA_DIR":/home/biodiversity-horizons/data-raw \
+    -v "$OUT_DIR":/home/biodiversity-horizons/outputs \
+    $CONTAINER $* # Pass any additional arguments to the Docker container
+}

scripts/main.R

@@ -17,17 +17,39 @@ parse_extent <- function(extent) {
   return(ext(extent))
 }
 
+safe_parse_opts <- function(opt_parser, args) {
+  # Function to safely parse the options. Shows the help if there's an error.
+  opt <- tryCatch(
+    {
+      opt <- parse_args(opt_parser, args = args)
+      opt
+    },
+    error = function(e) {
+      cat("Error parsing arguments:", e$message, "\n")
+      parse_args(opt_parser, args = c("--help"))
+      FALSE
+    }
+  )
+  return(opt)
+}
+
 
 run_tif2rds <- function(args) {
   source("utility/format_conversion_util.R")
   option_list <- list(
-    make_option(c("-i", "--input"), type = "character",
-                help = "Path to the input .tif file"),
-    make_option(c("-o", "--output"), type = "character",
-                help = "Path to save the output .rds file"),
-    make_option(c("-y", "--year_range"), type = "character",
-                default = "1850:2014",
-                help = "Year range as a sequence (e.g., '1850:2014')")
+    make_option(c("-i", "--input"),
+      type = "character",
+      help = "Path to the input .tif file"
+    ),
+    make_option(c("-o", "--output"),
+      type = "character",
+      help = "Path to save the output .rds file"
+    ),
+    make_option(c("-y", "--year_range"),
+      type = "character",
+      default = "1850:2014",
+      help = "Year range as a sequence (e.g., '1850:2014')"
+    )
   )
   opt <- safe_parse_opts(OptionParser(option_list = option_list), args[-1])
   check_not_null(opt$input, "input")
@@ -39,11 +61,12 @@ run_tif2rds <- function(args) {
   cat("Output:", opt$output, "\n")
   cat("Year range:", opt$year_range, "\n")
 
-  climate_data <- prepare_climate_data_from_tif(input_file = opt$input,
-                                                output_file = opt$output,
-                                                year_range = year_range)
+  climate_data <- prepare_climate_data_from_tif(
+    input_file = opt$input,
+    output_file = opt$output,
+    year_range = year_range
+  )
   print("File converted successfully!")
-
 }
 
 
@@ -86,7 +109,7 @@ run_shp2rds <- function(args) {
     make_option(c("-w", "--workers"),
       type = "numeric",
       help = "Number of workers to use. Default is availableCores()-1.",
-      default = availableCores()-1
+      default = availableCores() - 1
     )
   )
 
@@ -107,33 +130,22 @@ run_shp2rds <- function(args) {
   cat("Parallel:", opt$parallel, "\n")
   cat("Workers:", opt$workers, "\n")
 
-  grid <- create_grid(extent_vals = extent,
-                      resolution = opt$resolution,
-                      crs = opt$crs)
-  range_data <- prepare_range_data_from_shp_file(input_file_path = opt$input,
-                                                 grid = grid,
-                                                 realm = opt$realm,
-                                                 use_parallel = opt$parallel,
-                                                 number_of_workers = opt$workers,
-                                                 rds_output_file_path = opt$output)
+  grid <- create_grid(
+    extent_vals = extent,
+    resolution = opt$resolution,
+    crs = opt$crs
+  )
+  range_data <- prepare_range_data_from_shp_file(
+    input_file_path = opt$input,
+    grid = grid,
+    realm = opt$realm,
+    use_parallel = opt$parallel,
+    number_of_workers = opt$workers,
+    rds_output_file_path = opt$output
+  )
   print("File converted successfully!")
 }
 
-safe_parse_opts <- function(opt_parser, args) {
-  # Function to safely parse the options. Shows the help if there's an error.
-  opt <- tryCatch(
-    {
-      opt <- parse_args(opt_parser, args = args)
-      opt
-    },
-    error = function(e) {
-      cat("Error parsing arguments:", e$message, "\n")
-      opt <- parse_args(opt_parser, args = c("--help"))
-      FALSE
-    }
-  )
-  return(opt)
-}
 
 run_exposure <- function(args) {
   source("scripts/exposure_workflow.R")
@@ -164,6 +176,7 @@ run_exposure <- function(args) {
   exposure_time_workflow(opt$data_path, opt$plan_type, opt$workers)
 }
 
+# Main function
 args <- commandArgs(trailingOnly = TRUE)
 if (length(args) == 0) {
   stop("No command provided. Use 'shp2rds', 'tif2rds' or 'exposure'.")