feat: add --recursive to dvs get (SUPERSEDED — split into #203 + #204 + #205)

dvs-cli/src/main.rs

@@ -76,12 +76,16 @@ pub enum Command {
     },
     /// Retrieves the given files from dvs storage. You can use a glob or paths.
     /// If you pass a directory and a glob, the glob will be ran from that directory.
-    /// At least one path or --glob must be provided
+    /// At least one path or --glob must be provided.
     Get {
-        #[clap(required_unless_present = "glob")]
         paths: Vec<PathBuf>,
         #[clap(long, short)]
         glob: Option<String>,
+        /// Recursively include files in subdirectories for directory inputs.
+        /// Without this flag, directories return only their direct children.
+        /// Has no effect when no explicit paths are given.
+        #[clap(long, short)]
+        recursive: bool,
         /// Show what would be retrieved without making any actual changes
         #[clap(long)]
         dry_run: bool,
@@ -405,14 +409,22 @@ fn try_main() -> Result<()> {
         Command::Get {
             paths,
             glob,
+            recursive,
             dry_run,
         } => {
+            if paths.is_empty() && glob.is_none() {
+                bail!(
+                    "dvs get with no path or --glob would restore every tracked file. \
+                     If that's the intent, use `dvs get --glob '**/*'`."
+                );
+            }
             let config =
                 Config::find(&current_dir).ok_or_else(|| anyhow!("Not in a DVS repository"))??;
             let dvs_paths = DvsPaths::from_cwd(&config)?;
-            let all_paths: Vec<_> = resolve_paths_for_get(paths, glob.as_deref(), &dvs_paths)?
-                .into_iter()
-                .collect();
+            let all_paths: Vec<_> =
+                resolve_paths_for_get(paths, glob.as_deref(), &dvs_paths, recursive)?
+                    .into_iter()
+                    .collect();
             if all_paths.is_empty() {
                 return Err(anyhow!("No files to get"));
             }

dvs-rpkg/R/dvs-commands.R

@@ -80,7 +80,7 @@ dvs_status <- function(
 #' @inherit dvs_get_impl title description params
 #' @rdname dvs_get
 #' @export
-dvs_get <- function(paths = character(0), glob = NULL, dry_run = NULL) {
+dvs_get <- function(paths = character(0), glob = NULL, recursive = NULL, dry_run = NULL) {
   dvs_set_threads_impl(getOption("dvs.num_threads"))
   progress_callback <- NULL
   if (!isTRUE(dry_run)) {
@@ -90,6 +90,7 @@ dvs_get <- function(paths = character(0), glob = NULL, dry_run = NULL) {
   get_data_frame <- dvs_get_impl(
     paths = paths,
     glob = glob,
+    recursive = recursive,
     dry_run = dry_run,
     progress_callback = progress_callback
   )

dvs-rpkg/R/dvs-wrappers.R

@@ -139,26 +139,33 @@ dvs_add_impl <- function(paths = character(0), message = NULL, glob = NULL, dry_
   .val
 }
 
-# Generated from Rust fn `dvs_get` (lib.rs:436:15)
+# Generated from Rust fn `dvs_get` (lib.rs:447:15)
 #' @title Retrieve files from DVS storage into the working directory
-#' @description Fetches the specified files from DVS storage and writes them to their original paths in the working directory.
+#' @description Fetches the specified files from DVS storage and writes them to their original paths in the working directory. Calling `dvs_get()` with no arguments restores every tracked file in the repository.
 #' @param paths Character vector of file paths to retrieve from DVS storage.
 #' @param glob Optional glob pattern to select files (e.g. `"data/*.csv"`).
 #' Globs use a literal path separator: `*.csv` only matches files in the
 #' target directory and will not match `subdir/file.csv`. Use `**/*.csv` to
 #' match recursively across subdirectories.
+#' @param recursive If `TRUE`, directory inputs include all descendants;
+#' if `FALSE` or `NULL` (default), only direct children of the directory
+#' are returned. The flag only constrains descendants of paths passed
+#' explicitly — when `paths` is empty, every tracked file is returned
+#' regardless of nesting depth, and `recursive` has no effect.
 #' @param dry_run If `TRUE`, report what would be retrieved without writing files.
 #' @param progress_callback Optional handle to enable progress bar display.
 #' @keywords internal
 #' @source Generated by miniextendr from Rust fn `dvs_get`
-dvs_get_impl <- function(paths = character(0), glob = NULL, dry_run = NULL, progress_callback = NULL) {
+dvs_get_impl <- function(paths = character(0), glob = NULL, recursive = NULL, dry_run = NULL, progress_callback = NULL) {
   stopifnot(
     "'glob' must be NULL or character" = is.null(glob) || is.character(glob),
     "'glob' must be NULL or have length 1" = is.null(glob) || length(glob) == 1L,
+    "'recursive' must be NULL or logical" = is.null(recursive) || is.logical(recursive),
+    "'recursive' must be NULL or have length 1" = is.null(recursive) || length(recursive) == 1L,
     "'dry_run' must be NULL or logical" = is.null(dry_run) || is.logical(dry_run),
     "'dry_run' must be NULL or have length 1" = is.null(dry_run) || length(dry_run) == 1L
   )
-  .val <- .Call(C_dvs_get, .call = match.call(), paths, glob, dry_run, progress_callback)
+  .val <- .Call(C_dvs_get, .call = match.call(), paths, glob, recursive, dry_run, progress_callback)
   if (inherits(.val, "rust_error_value") && isTRUE(attr(.val, "__rust_error__"))) return(.miniextendr_raise_condition(.val, sys.call()))
   .val
 }
@@ -192,11 +199,15 @@ dvs_init_impl <- function(storage_path, root_dir = NULL, group = NULL, metadata_
   invisible(.val)
 }
 
-# Generated from Rust fn `dvs_status` (lib.rs:294:15)
+# Generated from Rust fn `dvs_status` (lib.rs:298:15)
 #' @title Report the sync status of DVS-managed files
 #' @description Reports the sync status of DVS-managed files. By default all statuses are shown; pass a character vector of status names (e.g. `c("current", "absent")`) to restrict output.
 #' @param paths Character vector of file or directory paths to check status for.
-#' @param recursive If `TRUE`, recursively include files in subdirectories.
+#' @param recursive If `TRUE`, directory inputs include all descendants;
+#' if `FALSE` or `NULL` (default), only direct children of the directory
+#' are returned. The flag only constrains descendants of paths passed
+#' explicitly — when `paths` is empty, every tracked file is returned
+#' regardless of nesting depth, and `recursive` has no effect.
 #' @param status Character vector of statuses to include. Valid values are
 #' `"current"`, `"absent"`, and `"unsynced"`. When empty (default), all
 #' statuses are shown.
@@ -214,7 +225,7 @@ dvs_status_impl <- function(paths = character(0), recursive = NULL, status = c("
   .val
 }
 
-# Generated from Rust fn `dvs_version` (lib.rs:494:8)
+# Generated from Rust fn `dvs_version` (lib.rs:511:8)
 #' @source Generated by miniextendr from Rust fn `dvs_version`
 #' @export
 dvs_version <- function() {
@@ -223,7 +234,7 @@ dvs_version <- function() {
   .val
 }
 
-# Generated from Rust fn `dvs_set_threads` (lib.rs:480:15)
+# Generated from Rust fn `dvs_set_threads` (lib.rs:497:15)
 #' @title Set the number of threads used by DVS parallel operations
 #' @description Controls the thread pool size for add, get, and status operations. Pass `NULL` to revert to automatic detection.
 #' @param threads Integer number of threads, or `NULL` to reset.
@@ -240,7 +251,7 @@ dvs_set_threads_impl <- function(threads = NULL) {
   invisible(.val)
 }
 
-# Generated from Rust fn `format_byte_size` (lib.rs:488:8)
+# Generated from Rust fn `format_byte_size` (lib.rs:505:8)
 #' @title Format a byte count as a human-readable size string
 #' @param size_bytes non-negative integer representing file sizes in bytes.
 #' @source Generated by miniextendr from Rust fn `format_byte_size`
@@ -256,7 +267,7 @@ format_byte_size <- function(size_bytes) {
   .val
 }
 
-# Generated from Rust fn `set_dvs_log_level` (lib.rs:523:8)
+# Generated from Rust fn `set_dvs_log_level` (lib.rs:540:8)
 #' @title Set the log level for DVS internals
 #' @description Controls which log messages from the DVS internals are routed to R's console. `error` and `warn` go to [stderr()]; `info`, `debug`, and `trace` go to stdout.
 #' @param level Character string giving the desired log level. The default

dvs-rpkg/src/rust/Cargo.toml

@@ -34,7 +34,7 @@ connections = ["miniextendr-api/connections"]
 # block, or rely on `just` recipes that vendor from the local checkout.
 [dependencies]
 miniextendr-api = { git = "https://github.com/A2-ai/miniextendr", features = ["serde", "time", "log"] }
-dvs = { git = "https://github.com/A2-ai/dvs2", branch = "main", package = "dvs" }
+dvs = { git = "https://github.com/A2-ai/dvs2", branch = "feat/get-recursive", package = "dvs" }
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
 anyhow = "1"

dvs-rpkg/src/rust/lib.rs

@@ -285,7 +285,11 @@ impl From<StatusChoice> for Status {
 /// of status names (e.g. `c("current", "absent")`) to restrict output.
 ///
 /// @param paths Character vector of file or directory paths to check status for.
-/// @param recursive If `TRUE`, recursively include files in subdirectories.
+/// @param recursive If `TRUE`, directory inputs include all descendants;
+///   if `FALSE` or `NULL` (default), only direct children of the directory
+///   are returned. The flag only constrains descendants of paths passed
+///   explicitly — when `paths` is empty, every tracked file is returned
+///   regardless of nesting depth, and `recursive` has no effect.
 /// @param status Character vector of statuses to include. Valid values are
 ///   `"current"`, `"absent"`, and `"unsynced"`. When empty (default), all
 ///   statuses are shown.
@@ -422,20 +426,28 @@ impl<'a> From<&'a FileMetadata> for FileMetadataView<'a> {
 /// Retrieve files from DVS storage into the working directory.
 ///
 /// Fetches the specified files from DVS storage and writes them
-/// to their original paths in the working directory.
+/// to their original paths in the working directory. Calling
+/// `dvs_get()` with no arguments restores every tracked file in
+/// the repository.
 ///
 /// @param paths Character vector of file paths to retrieve from DVS storage.
 /// @param glob Optional glob pattern to select files (e.g. `"data/*.csv"`).
 ///   Globs use a literal path separator: `*.csv` only matches files in the
 ///   target directory and will not match `subdir/file.csv`. Use `**/*.csv` to
 ///   match recursively across subdirectories.
+/// @param recursive If `TRUE`, directory inputs include all descendants;
+///   if `FALSE` or `NULL` (default), only direct children of the directory
+///   are returned. The flag only constrains descendants of paths passed
+///   explicitly — when `paths` is empty, every tracked file is returned
+///   regardless of nesting depth, and `recursive` has no effect.
 /// @param dry_run If `TRUE`, report what would be retrieved without writing files.
 /// @param progress_callback Optional handle to enable progress bar display.
 /// @keywords internal
 #[miniextendr(r_name = "dvs_get_impl")]
 pub(crate) fn dvs_get(
     #[miniextendr(default = "character(0)")] paths: Vec<PathBuf>,
     #[miniextendr(default = "NULL")] glob: Option<String>,
+    #[miniextendr(default = "NULL")] recursive: Option<bool>,
     #[miniextendr(default = "NULL")] dry_run: Option<bool>,
     #[miniextendr(default = "NULL")] progress_callback: Option<ExternalPtr<ProgressBarCallback>>,
 ) -> Result<DataFrame<AsSerializeRow<GetResult>>> {
@@ -444,9 +456,14 @@ pub(crate) fn dvs_get(
     let config = Config::find(&current_dir).ok_or_else(|| anyhow!("Not in a DVS repository"))??;
     let dvs_paths = DvsPaths::from_cwd(&config)?;
 
-    let all_paths: Vec<_> = resolve_paths_for_get(paths, glob.as_deref(), &dvs_paths)?
-        .into_iter()
-        .collect();
+    let all_paths: Vec<_> = resolve_paths_for_get(
+        paths,
+        glob.as_deref(),
+        &dvs_paths,
+        recursive.unwrap_or(false),
+    )?
+    .into_iter()
+    .collect();
     if all_paths.is_empty() {
         return Err(anyhow!("No files to get"));
     }

dvs-rpkg/tests/testthat/test-get.R

@@ -49,3 +49,16 @@ test_that("dvs_get on never-added path errors", {
   new_dvs_test_repo()
   expect_error(dvs_get(paths = "never-added.csv"))
 })
+
+test_that("dvs_get() with no args restores every tracked file at every depth", {
+  new_dvs_test_repo()
+  write_theoph("top.csv")
+  write_theoph_shuffled("data/raw/deep.csv", seed = 1)
+  dvs_add(paths = file.path(getwd(), c("top.csv", "data/raw/deep.csv")))
+  file.remove(c("top.csv", "data/raw/deep.csv"))
+
+  result <- dvs_get()
+  expect_s3_class(result, "tbl_df")
+  expect_setequal(result$path, c("top.csv", "data/raw/deep.csv"))
+  expect_true(all(file.exists(c("top.csv", "data/raw/deep.csv"))))
+})

dvs/src/files/status.rs

@@ -44,7 +44,7 @@ pub struct StatusFilter {
 
 /// We need to handle `.`, `./` etc but we can't canonicalize because
 /// the path might not exist and we want the path relative to the directory so no symlink resolution
-fn normalize_path(p: PathBuf) -> Option<PathBuf> {
+pub(crate) fn normalize_path(p: PathBuf) -> Option<PathBuf> {
     let mut out = PathBuf::new();
     for c in p.components() {
         match c {