mutate() %>% compute() fails. Error in .transformer(): ! value must be a string or scalar SQL, not the number.

[erydit]

Hello.
An error started to occur in ‘dbplyr’ version 2.4.0.

I am trying to collect and make some calculations in SQLite database:

DB_CON <- dbConnect(drv = RSQLite::SQLite(), dbname = "BandsTS.db")
BANDS_tbl <- tbl(DB_CON, "bands_ts_tbl") %>%
  mutate( ARVI2 = -0.18 + 1.17 * ((B08 - B04) / (B08 + B04)) ) %>%
compute()

That code (and also show_query() instead of compute() ) throws an error: ```r Error in `.transformer()`: ! `value` must be a string or scalar SQL, not the number -0.18. Run `rlang::last_trace()` to see where the error occurred. > rlang::last_trace() Error in `.transformer()`: ! `value` must be a string or scalar SQL, not the number -0.18. --- Backtrace: ▆ 1. ├─dplyr::show_query(BANDS_tbl) 2. ├─dbplyr:::show_query.tbl_lazy(BANDS_tbl) 3. │ └─dbplyr::remote_query(x, cte = cte, sql_options = sql_options) 4. │ └─dbplyr::db_sql_render(remote_con(x), x, cte = cte, sql_options = sql_options) 5. │ ├─dbplyr::db_sql_render(con, sql, sql_options = sql_options) 6. │ └─dbplyr:::db_sql_render.DBIConnection(con, sql, sql_options = sql_options) 7. │ ├─dbplyr::sql_render(sql, con = con, ..., sql_options = sql_options) 8. │ └─dbplyr:::sql_render.tbl_lazy(sql, con = con, ..., sql_options = sql_options) 9. │ ├─dbplyr::sql_render(...) 10. │ └─dbplyr:::sql_render.lazy_query(...) 11. │ ├─dbplyr::sql_build(query, con = con, sql_options = sql_options) 12. │ └─dbplyr:::sql_build.lazy_select_query(query, con = con, sql_options = sql_options) 13. │ ├─dbplyr::sql_build(op$x, con, sql_options = sql_options) 14. │ └─dbplyr:::sql_build.lazy_select_query(op$x, con, sql_options = sql_options) 15. │ └─dbplyr:::get_select_sql(...) 16. │ └─dbplyr:::translate_select_sql(con, select) 17. │ └─dbplyr::translate_sql_(...) 18. │ └─base::lapply(...) 19. │ └─dbplyr (local) FUN(X[[i]], ...) 20. │ ├─dbplyr::escape(eval_tidy(x, mask), con = con) 21. │ └─rlang::eval_tidy(x, mask) 22. ├─-0.18 + 1.17 * ((B08 - B04)/(B08 + B04)) 23. │ └─dbplyr:::escape_infix_expr(enexpr(x), x, escape_unary_minus = TRUE) 24. │ └─dbplyr:::glue_sql2(sql_current_con(), "({x})") 25. │ ├─dbplyr::sql(...) 26. │ │ └─dbplyr:::c_character(...) 27. │ └─glue::glue(...) 28. │ └─glue::glue_data(...) 29. └─glue (local) ``("x") 30. ├─.transformer(expr, env) %||% .null 31. └─dbplyr (local) .transformer(text = expr, envir = env) ```

[hadley]

Actual reprex:

library(tidyverse)
library(DBI)

db <- dbConnect(RSQLite::SQLite(), ":memory:")
dbWriteTable(conn = db, name = "tbl", value = data.frame(x = 1:10))

db_tbl <- tbl(db, "tbl")

db_tbl %>% mutate(z = -1 + x)
#> Error in `.transformer()`:
#> ! `value` must be a string or scalar SQL, not the number -1.

^{Created on 2023-12-19 with reprex v2.0.2.9000}

Looks like problem is happening in escape_infix_expr().

[hadley]

More minimal reprex:

library(dbplyr)
library(dplyr, warn.conflicts = FALSE)

lf <- lazy_frame(x = 1)
lf %>% mutate(z = -1 + x)

[hadley]

Even more minimal reprex:

library(dbplyr)
test_translate_sql(-1 + x, con = simulate_dbi())