refactor: Use nanoarrow_vctr in forthcoming nanoarrow release (#36)

GeoArrow was the wrong place for this class...it's much better in
nanoarrow! See apache/arrow-nanoarrow#461 for
the implementation.

Author: paleolimbot

DESCRIPTION

@@ -14,7 +14,7 @@ Encoding: UTF-8
 Roxygen: list(markdown = TRUE)
 RoxygenNote: 7.2.3
 Imports:
-    nanoarrow (>= 0.3.0),
+    nanoarrow (>= 0.5.0),
     wk (>= 0.6.0)
 LinkingTo:
     wk

NAMESPACE

@@ -1,8 +1,5 @@
 # Generated by roxygen2: do not edit by hand
 
-S3method("[",geoarrow_vctr)
-S3method("[<-",geoarrow_vctr)
-S3method("[[<-",geoarrow_vctr)
 S3method(as.character,geoarrow_vctr)
 S3method(as_geoarrow_array,character)
 S3method(as_geoarrow_array,default)
@@ -18,8 +15,6 @@ S3method(as_geoarrow_array_stream,geoarrow_vctr)
 S3method(as_geoarrow_array_stream,nanoarrow_array_stream)
 S3method(as_nanoarrow_array,sfc)
 S3method(as_nanoarrow_array_extension,geoarrow_extension_spec)
-S3method(as_nanoarrow_array_stream,geoarrow_vctr)
-S3method(as_nanoarrow_schema,geoarrow_vctr)
 S3method(convert_array,geoarrow_vctr)
 S3method(convert_array,sfc)
 S3method(convert_array,wk_wkb)
@@ -31,7 +26,6 @@ S3method(infer_geoarrow_schema,default)
 S3method(infer_geoarrow_schema,nanoarrow_array)
 S3method(infer_geoarrow_schema,nanoarrow_array_stream)
 S3method(infer_nanoarrow_ptype_extension,geoarrow_extension_spec)
-S3method(infer_nanoarrow_schema,geoarrow_vctr)
 S3method(infer_nanoarrow_schema,sfc)
 S3method(infer_nanoarrow_schema,wk_wkb)
 S3method(infer_nanoarrow_schema,wk_wkt)
@@ -53,7 +47,6 @@ export(na_extension_wkb)
 export(na_extension_wkt)
 importFrom(nanoarrow,as_nanoarrow_array)
 importFrom(nanoarrow,as_nanoarrow_array_extension)
-importFrom(nanoarrow,as_nanoarrow_array_stream)
 importFrom(nanoarrow,as_nanoarrow_schema)
 importFrom(nanoarrow,convert_array)
 importFrom(nanoarrow,convert_array_extension)

R/pkg-arrow.R

@@ -8,6 +8,7 @@ as_arrow_array.geoarrow_vctr <- function(x, ..., type = NULL) {
   }
 }
 
+#' @importFrom nanoarrow as_nanoarrow_schema
 as_chunked_array.geoarrow_vctr <- function(x, ..., type = NULL) {
   if (is.null(type)) {
     schema <- NULL
@@ -18,7 +19,7 @@ as_chunked_array.geoarrow_vctr <- function(x, ..., type = NULL) {
   }
 
   # as_nanoarrow_array_stream() applies the indices if vctr is sliced
-  stream <- as_nanoarrow_array_stream(x, schema = schema)
+  stream <- as_geoarrow_array_stream(x, schema = schema)
   chunks <- nanoarrow::collect_array_stream(stream, validate = FALSE)
   type <- arrow::as_data_type(type)
 

R/pkg-nanoarrow.R

@@ -20,7 +20,7 @@ register_geoarrow_extension <- function() {
 #' @importFrom nanoarrow infer_nanoarrow_ptype_extension
 #' @export
 infer_nanoarrow_ptype_extension.geoarrow_extension_spec <- function(extension_spec, x, ...) {
-  new_geoarrow_vctr(list(), x, integer())
+  nanoarrow::nanoarrow_vctr(schema = x, subclass = "geoarrow_vctr")
 }
 
 #' @importFrom nanoarrow convert_array_extension

R/vctr.R

@@ -18,48 +18,7 @@ as_geoarrow_vctr <- function(x, ..., schema = NULL) {
   }
 
   stream <- as_geoarrow_array_stream(x, ..., schema = schema)
-  chunks <- nanoarrow::collect_array_stream(stream, validate = FALSE)
-  new_geoarrow_vctr(chunks, stream$get_schema())
-}
-
-new_geoarrow_vctr <- function(chunks, schema, indices = NULL) {
-  offsets <- .Call(geoarrow_c_vctr_chunk_offsets, chunks)
-  if (is.null(indices)) {
-    indices <- seq_len(offsets[length(offsets)])
-  }
-
-  structure(
-    indices,
-    schema = schema,
-    chunks = chunks,
-    offsets = offsets,
-    class = c("geoarrow_vctr", "wk_vctr")
-  )
-}
-
-#' @export
-`[.geoarrow_vctr` <- function(x, i) {
-  attrs <- attributes(x)
-  x <- NextMethod()
-
-  if (is.null(vctr_as_slice(x))) {
-    stop(
-      "Can't subset geoarrow_vctr with non-slice (e.g., only i:j indexing is supported)"
-    )
-  }
-
-  attributes(x) <- attrs
-  x
-}
-
-#' @export
-`[<-.geoarrow_vctr` <- function(x, i, value) {
-  stop("subset assignment for geoarrow_vctr is not supported")
-}
-
-#' @export
-`[[<-.geoarrow_vctr` <- function(x, i, value) {
-  stop("subset assignment for geoarrow_vctr is not supported")
+  nanoarrow::as_nanoarrow_vctr(stream, subclass = "geoarrow_vctr")
 }
 
 #' @export
@@ -102,129 +61,8 @@ as.character.geoarrow_vctr <- function(x, ...) {
   format(x, ...)
 }
 
-#' @export
-infer_nanoarrow_schema.geoarrow_vctr <- function(x, ...) {
-  attr(x, "schema", exact = TRUE)
-}
-
-# Because zero-length vctrs are R's way of communicating "type", implement
-# as_nanoarrow_schema() here so that it works in places that expect a type
-#' @importFrom nanoarrow as_nanoarrow_schema
-#' @export
-as_nanoarrow_schema.geoarrow_vctr <- function(x, ...) {
-  attr(x, "schema", exact = TRUE)
-}
-
 #' @export
 as_geoarrow_array_stream.geoarrow_vctr <- function(x, ..., schema = NULL) {
-  as_nanoarrow_array_stream.geoarrow_vctr(x, ..., schema = schema)
-}
-
-#' @importFrom nanoarrow as_nanoarrow_array_stream
-#' @export
-as_nanoarrow_array_stream.geoarrow_vctr <- function(x, ..., schema = NULL) {
-  if (!is.null(schema)) {
-    stream <- as_nanoarrow_array_stream(x, schema = NULL)
-    return(as_geoarrow_array_stream(stream, schema = schema))
-  }
-
-  slice <- vctr_as_slice(x)
-  if (is.null(slice)) {
-    stop("Can't resolve non-slice geoarrow_vctr to nanoarrow_array_stream")
-  }
-
-  x_schema <- attr(x, "schema", exact = TRUE)
-
-  # Zero-size slice can be an array stream with zero batches
-  if (slice[2] == 0) {
-    return(nanoarrow::basic_array_stream(list(), schema = x_schema))
-  }
-
-  # Full slice doesn't need slicing logic
-  offsets <- attr(x, "offsets", exact = TRUE)
-  batches <- attr(x, "chunks", exact = TRUE)
-  if (slice[1] == 1 && slice[2] == max(offsets)) {
-    return(
-      nanoarrow::basic_array_stream(
-        batches,
-        schema = x_schema,
-        validate = FALSE
-      )
-    )
-  }
-
-  # Calculate first and last slice information
-  first_index <- slice[1] - 1L
-  end_index <- first_index + slice[2]
-  last_index <- end_index - 1L
-  first_chunk_index <- vctr_resolve_chunk(first_index, offsets)
-  last_chunk_index <- vctr_resolve_chunk(last_index, offsets)
-
-  first_chunk_offset <- first_index - offsets[first_chunk_index + 1L]
-  first_chunk_length <- offsets[first_chunk_index + 2L] - first_index
-  last_chunk_offset <- 0L
-  last_chunk_length <- end_index - offsets[last_chunk_index + 1L]
-
-  # Calculate first and last slices
-  if (first_chunk_index == last_chunk_index) {
-    batch <- vctr_array_slice(
-      batches[[first_chunk_index + 1L]],
-      first_chunk_offset,
-      last_chunk_length - first_chunk_offset
-    )
-
-    return(
-      nanoarrow::basic_array_stream(
-        list(batch),
-        schema = x_schema,
-        validate = FALSE
-      )
-    )
-  }
-
-  batch1 <- vctr_array_slice(
-    batches[[first_chunk_index + 1L]],
-    first_chunk_offset,
-    first_chunk_length
-  )
-
-  batchn <- vctr_array_slice(
-    batches[[last_chunk_index + 1L]],
-    last_chunk_offset,
-    last_chunk_length
-  )
-
-  seq_mid <- seq_len(last_chunk_index - first_chunk_index - 1)
-  batch_mid <- batches[first_chunk_index + seq_mid]
-
-  nanoarrow::basic_array_stream(
-    c(
-      list(batch1),
-      batch_mid,
-      list(batchn)
-    ),
-    schema = x_schema,
-    validate = FALSE
-  )
-}
-
-
-# Utilities for vctr methods
-
-vctr_resolve_chunk <- function(x, offsets) {
-  .Call(geoarrow_c_vctr_chunk_resolve, x, offsets)
-}
-
-vctr_as_slice <- function(x) {
-  .Call(geoarrow_c_vctr_as_slice, x)
-}
-
-vctr_array_slice <- function(x, offset, length) {
-  new_offset <- x$offset + offset
-  new_length <- length
-  nanoarrow::nanoarrow_array_modify(
-    x,
-    list(offset = new_offset, length = new_length),
-    validate = FALSE
-  )
+  stream <- nanoarrow::as_nanoarrow_array_stream(x)
+  as_geoarrow_array_stream(stream, schema = schema)
 }

src/r-init.c

@@ -14,9 +14,6 @@ extern SEXP geoarrow_c_make_type(SEXP geometry_type_sexp, SEXP dimensions_sexp,
                                  SEXP coord_type_sexp);
 extern SEXP geoarrow_c_schema_init_extension(SEXP schema_xptr, SEXP type_sexp);
 extern SEXP geoarrow_c_schema_parse(SEXP schema_xptr, SEXP extension_name_sexp);
-extern SEXP geoarrow_c_vctr_chunk_offsets(SEXP array_list);
-extern SEXP geoarrow_c_vctr_chunk_resolve(SEXP indices_sexp, SEXP offsets_sexp);
-extern SEXP geoarrow_c_vctr_as_slice(SEXP indices_sexp);
 extern SEXP geoarrow_c_handle_stream(SEXP data, SEXP handler_xptr);
 extern SEXP geoarrow_c_writer_new(SEXP schema_xptr, SEXP array_out_xptr);
 
@@ -28,9 +25,6 @@ static const R_CallMethodDef CallEntries[] = {
     {"geoarrow_c_make_type", (DL_FUNC)&geoarrow_c_make_type, 3},
     {"geoarrow_c_schema_init_extension", (DL_FUNC)&geoarrow_c_schema_init_extension, 2},
     {"geoarrow_c_schema_parse", (DL_FUNC)&geoarrow_c_schema_parse, 2},
-    {"geoarrow_c_vctr_chunk_offsets", (DL_FUNC)&geoarrow_c_vctr_chunk_offsets, 1},
-    {"geoarrow_c_vctr_chunk_resolve", (DL_FUNC)&geoarrow_c_vctr_chunk_resolve, 2},
-    {"geoarrow_c_vctr_as_slice", (DL_FUNC)&geoarrow_c_vctr_as_slice, 1},
     {"geoarrow_c_handle_stream", (DL_FUNC)&geoarrow_c_handle_stream, 2},
     {"geoarrow_c_writer_new", (DL_FUNC)&geoarrow_c_writer_new, 2},
     {NULL, NULL, 0}};