MAP(BLOB, BLOB) type is not supported

[mgacc0]

    library(tidyverse)
    library(duckplyr)

    duckplyr::db_exec("COPY (SELECT 42 AS i, 84 AS j) TO 'xx.parquet';")
    duckplyr::read_sql_duckdb("SELECT * FROM parquet_metadata('xx.parquet');")
#> Error in `rel_to_df()`:
#> ! rel_to_altrep: Unknown column type for altrep: MAP(BLOB, BLOB)
#> ---
#> Backtrace:
#>     ▆
#>  1. └─duckplyr::read_sql_duckdb("SELECT * FROM parquet_metadata('xx.parquet');")
#>  2.   ├─duckplyr::rel_to_df(rel, prudence = prudence)
#>  3.   └─duckplyr:::rel_to_df.duckdb_relation(rel, prudence = prudence)

[krlmlr]

Thanks for raising this. The development version of duckdb is now telling us which column is affected.

It might take a while to support the MAP() type. In the meantime:

options(conflicts.policy = list(warn = FALSE))
library(duckplyr)
#> Loading required package: dplyr
#> ✔ Overwriting dplyr methods with duckplyr methods.
#> ℹ Turn off with `duckplyr::methods_restore()`.

db_exec("COPY (SELECT 42 AS i, 84 AS j) TO 'xx.parquet';")
read_sql_duckdb("SELECT * EXCLUDE key_value_metadata FROM parquet_metadata('xx.parquet');")
#> # A duckplyr data frame: 25 variables
#>   file_name  row_group_id row_group_num_rows row_group_num_columns
#>   <chr>             <dbl>              <dbl>                 <dbl>
#> 1 xx.parquet            0                  1                     2
#> 2 xx.parquet            0                  1                     2
#> # ℹ 21 more variables: row_group_bytes <dbl>, column_id <dbl>,
#> #   file_offset <dbl>, num_values <dbl>, path_in_schema <chr>, type <chr>,
#> #   stats_min <chr>, stats_max <chr>, stats_null_count <dbl>,
#> #   stats_distinct_count <dbl>, stats_min_value <chr>, stats_max_value <chr>,
#> #   compression <chr>, encodings <chr>, index_page_offset <dbl>,
#> #   dictionary_page_offset <dbl>, data_page_offset <dbl>,
#> #   total_compressed_size <dbl>, total_uncompressed_size <dbl>, …

^{Created on 2025-04-21 with reprex v2.1.1}