Why is knitr choking on a pipe that works fine otherwise? - r

Question

I've got an RMarkdown script that works fine if I run the chunks manually, either one at a time or with "Run All". But when I try to use knitr to generate HTML or a PDF, I'm getting an error: Error in select(responses, starts_with("Q1 ") & !contains("None")) %>% : could not find function "%>%"
The actual full line reads:
cols <- select(responses, starts_with("Q1 ") & !contains("None") ) %>% colnames()
I'm working with data from a survey, where a lot of questions were "select as many as apply" type questions, and there was an open ended "None of the above" option. At this point, I'm pulling out exactly the columns I want (all the Q1 responses, but not Q10 or Q11 responses, and not the open ended response) so I can use pivot_longer() and summarize the responses. It works fine in the script: I get a list of the exact column names that I want, and then count the values.
But when I try to use knitr() it balks on the %>%.
processing file: 02_Survey2020_report.Rmd
|.... | 6%
ordinary text without R code
|......... | 12%
label: setup (with options)
List of 1
$ include: logi FALSE
|............. | 19%
ordinary text without R code
|.................. | 25%
label: demographics gender
Quitting from lines 28-46 (02_Survey2020_report.Rmd)
Error in select(responses, starts_with("Q1 ") & !contains("None")) %>% :
could not find function "%>%"
Calls: <Anonymous> ... handle -> withCallingHandlers -> withVisible -> eval -> eval
Execution halted
A simplified reproducible example gets the same results. I run the following and get what I expect, a tidy table with the count times each answer was selected:
example <- data.frame("id" = c(009,008,007,006,005,004,003,002,001,010), "Q3_Red" = c("","","","Red","","","","Red","Red","Red"), "Q3_Blue" = c("","","","","","Blue","Blue","Blue","",""),
"Q3_Green" = c("","Green","Green","","","","","Green","",""), "Q3_Purple" = c("","Purple","","","Purple","","Purple","","Purple","Purple"),
"Q3_None of the above" = c(009,008,"Verbose explanation that I don't want to count." ,006,005,004,003,002,"Another verbose entry.",010)
)
cols <- select(example, starts_with("Q3") & !contains("None") ) %>% colnames()
example %>%
pivot_longer(cols = all_of(cols),
values_to = "response") %>%
filter(response != "") %>%
count(response)
But when I use ctrlshiftk to output a document, I get the same error:
processing file: 00a_reproducible_examples.Rmd
Quitting from lines 9-25 (00a_reproducible_examples.Rmd)
Error in select(example, starts_with("Q3") & !contains("None")) %>% colnames() :
could not find function "%>%"
Calls: <Anonymous> ... handle -> withCallingHandlers -> withVisible -> eval -> eval
Execution halted
Why is knitr balking at a pipe?

Answer 1

recently I run into similar problem.
Not sure if there is more sophisticated solution, but loading library in each chunk of code worked for me.
To display result of a code, w/o message regarding library loading add message = FALSE.
Example:
```{r, echo=FALSE, message = FALSE}
library(dplyr)
>>your code with dplyr<<
```