paleolimbot commented on code in PR #12817:
URL: https://github.com/apache/arrow/pull/12817#discussion_r853357707
##########
r/src/table.cpp:
##########
@@ -226,20 +222,41 @@ arrow::Status AddMetadataFromDots(SEXP lst, int
num_fields,
cpp11::writable::list metadata(2);
metadata.names() = arrow::r::data::names_metadata;
- bool has_metadata = false;
+ bool has_top_level_metadata = false;
// "top level" attributes, only relevant if the first object is not named
and a data
// frame
cpp11::strings names = Rf_getAttrib(lst, R_NamesSymbol);
if (names[0] == "" && Rf_inherits(VECTOR_ELT(lst, 0), "data.frame")) {
SEXP top_level = metadata[0] = arrow_attributes(VECTOR_ELT(lst, 0), true);
if (!Rf_isNull(top_level) && XLENGTH(top_level) > 0) {
- has_metadata = true;
+ has_top_level_metadata = true;
}
}
// recurse to get all columns metadata
- metadata[1] = CollectColumnMetadata(lst, num_fields, has_metadata);
+ cpp11::writable::list metadata_columns = CollectColumnMetadata(lst,
num_fields);
+
+ // Remove metadata for ExtensionType columns, because these have their own
mechanism for
+ // preserving R type information
+ for (R_xlen_t i = 0; i < schema->num_fields(); i++) {
+ if (schema->field(i)->type()->id() == Type::EXTENSION) {
+ metadata_columns[i] = R_NilValue;
Review Comment:
This was a really good catch! It led me down a bit of a rabbit hole but it
now works!
``` r
# remotes::install_github("apache/arrow/r#12817")
library(arrow, warn.conflicts = FALSE)
nested <- tibble::tibble(x = vctrs::new_vctr(1:5, class = "custom_vctr"))
infer_type(nested)
#> StructType
#> struct<x: <custom_vctr[0]>>
as_arrow_array(nested)
#> StructArray
#> <struct<x: <custom_vctr[0]>>>
#> -- is_valid: all not null
#> -- child 0 type: <custom_vctr[0]>
#> [
#> 1,
#> 2,
#> 3,
#> 4,
#> 5
#> ]
```
<sup>Created on 2022-04-19 by the [reprex
package](https://reprex.tidyverse.org) (v2.0.1)</sup>
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
