[ARROW-12603] [R] open_dataset ignoring provided schema when using select

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 4.0.0
• Fix Version/s: 4.0.1
• Component/s: R
• Environment: R version 4.0.5 (2021-03-31) ; Platform: x86_64-pc-linux-gnu (64-bit)
• External issue URL: https://github.com/apache/arrow/issues/28357
• Assignee: Unassigned
• Reporter: Eu Jing Chua
• Created: 29/Apr/21 16:42
• Updated: 11/Jan/23 08:27
• Resolved: 03/May/21 16:19

Description

While the following snippet works with arrow 3.0.0, it fails after updating to arrow 4.0.0.

An example CSV that can be used to replicate this can be found here

.
├── data
│   └── 2021-04-25-Karlen-pypm.csv
└── test.R

library(arrow)
library(tidyverse)

sch <- schema(forecast_date=string(),
 target=string(),
 target_end_date=string(),
 location=string(),
 type=string(),
 quantile=string(),
 value=string())

ds = open_dataset("data", format = "csv", schema = sch)

ds %>% select(target) %>% collect()

The error is:
Error: Invalid: In CSV column #3: CSV conversion error to int64: invalid value 'US'

However, it should be noted that these all run well and return a data frame with the right schema.

ds %>% collect()
ds %>% select(target, location) %>% collect()

Issue Links

• duplicates: ARROW-12500 [C++][Dataset] Consolidate similar tests for file formats