[ARROW-16776] [R] dplyr::glimpse method for arrow table and datasets

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 9.0.0
• Component/s: R
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/32109
• Assignee: Neal Richardson
• Reporter: Thomas Mock
• Created: 07/Jun/22 13:37
• Updated: 11/Jan/23 11:46
• Resolved: 12/Jul/22 19:48

Description

When working with Arrow datasets/tables, I often find myself wanting to interactively print or "see" the results of a query or the first few rows of the data without having to fully collect into memory.

I can perform exploratory data analysis on large out-of-memory datasets via Arrow + dplyr but in order to print the returned values I have to collect() into memory or send to_duckdb().

• compute() - returns number of rows/columns, but no data
• collect() - returns data fully into memory, can be combined with head()
• to_duckdb() - keeps data out of memory, always returns top 10 rows and all columns, optionally increase/decrease number of printed rows

While to_duckdb() gives me the ability to do true EDA, it seems counterintuitive to need to send the arrow table over to a duckdb database just to see the glimpse()/head() equivalent.

My feature request is that there is a dplyr::glimpse() method that will lazily print the first few values of table/dataset. The expected output would be something like the below.

``` r
library(dplyr)
library(arrow)

mtcars %>% arrow::write_parquet("mtcars.parquet")
car_ds <- arrow::open_dataset("mtcars.parquet")

car_ds %>%
glimpse()

Rows: ??
Columns: 11
$ mpg <dbl> 21.0, 21.0, 22.8, 21.4, 18.7, …
$ cyl <dbl> 6, 6, 4, 6, 8, 6, 8, 4, 4, 6, …
$ disp <dbl> 160.0, 160.0, 108.0, 258.0, 36…
$ hp <dbl> 110, 110, 93, 110, 175, 105, 2…
$ drat <dbl> 3.90, 3.90, 3.85, 3.08, 3.15, …
$ wt <dbl> 2.620, 2.875, 2.320, 3.215, 3.…
$ qsec <dbl> 16.46, 17.02, 18.61, 19.44, 17…
$ vs <dbl> 0, 0, 1, 1, 0, 1, 0, 1, 1, 1, …
$ am <dbl> 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, …
$ gear <dbl> 4, 4, 4, 3, 3, 3, 3, 4, 4, 4, …
$ carb <dbl> 4, 4, 1, 1, 2, 1, 4, 2, 2, 4, …
```

Currently glimpse() will return a list output where the majority of the output is erroneous to the actual data/values.

``` r
library(dplyr)
library(arrow)

mtcars %>% arrow::write_parquet("mtcars.parquet")
car_ds <- arrow::open_dataset("mtcars.parquet")

car_ds %>% 
  glimpse()
#> Classes 'FileSystemDataset', 'Dataset', 'ArrowObject', 'R6' <FileSystemDataset>
#>   Inherits from: <Dataset>
#>   Public:
#>     .:xp:.: externalptr
#>     .class_title: function () 
#>     clone: function (deep = FALSE) 
#>     files: active binding
#>     filesystem: active binding
#>     format: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     metadata: active binding
#>     NewScan: function () 
#>     num_cols: active binding
#>     num_rows: active binding
#>     pointer: function () 
#>     print: function (...) 
#>     schema: active binding
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: active binding

car_ds %>%
  filter(cyl == 6) %>%
  glimpse()
#> List of 7
#>  $ mpg :Classes 'FileSystemDataset', 'Dataset', 'ArrowObject', 'R6' <FileSystemDataset>
#>   Inherits from: <Dataset>
#>   Public:
#>     .:xp:.: externalptr
#>     .class_title: function () 
#>     clone: function (deep = FALSE) 
#>     files: active binding
#>     filesystem: active binding
#>     format: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     metadata: active binding
#>     NewScan: function () 
#>     num_cols: active binding
#>     num_rows: active binding
#>     pointer: function () 
#>     print: function (...) 
#>     schema: active binding
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: active binding 
#>  $ cyl :List of 11
#>   ..$ mpg :Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>   ..$ cyl :Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>   ..$ disp:Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>   ..$ hp  :Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>   ..$ drat:Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>   ..$ wt  :Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>   ..$ qsec:Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>   ..$ vs  :Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>   ..$ am  :Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>   ..$ gear:Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>   ..$ carb:Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>  $ disp:Classes 'Expression', 'ArrowObject', 'R6' <Expression>
#>   Inherits from: <ArrowObject>
#>   Public:
#>     .:xp:.: externalptr
#>     cast: function (to_type, safe = TRUE, ...) 
#>     clone: function (deep = FALSE) 
#>     Equals: function (other, ...) 
#>     field_name: active binding
#>     initialize: function (xp) 
#>     invalidate: function () 
#>     pointer: function () 
#>     print: function (...) 
#>     schema: Schema, ArrowObject, R6
#>     set_pointer: function (xp) 
#>     ToString: function () 
#>     type: function (schema = self$schema) 
#>     type_id: function (schema = self$schema)  
#>  $ hp  : chr(0) 
#>  $ drat: NULL
#>  $ wt  : list()
#>  $ qsec: logi(0) 
#>  - attr(*, "class")= chr "arrow_dplyr_query"
```

^{Created on 2022-06-07 by the [reprex package](https://reprex.tidyverse.org) (v2.0.1)}

Issue Links

• Blocked: ARROW-16777 [R] printing data in Table/RecordBatch print method
• links to: GitHub Pull Request #13563