当目标包中缺少文件时触发代码

问题描述

考虑以下代码，其中我的 targets 管道中有两个文件。

options(Crayon.enabled = FALSE,tidyverse.quiet = TRUE)
library(targets)
library(tidyverse)
write_csv(tibble(x1 = 1,x2 = 1),"a.csv")
write_csv(tibble(x1 = 1,"b.csv")
tar_script({
  options(Crayon.enabled = FALSE,tidyverse.quiet = TRUE)
  library(readr)
  list(
    tarchetypes::tar_files(paths,c("a.csv","b.csv")),tar_target(data,read_csv(paths,col_types = "dd"),pattern = map(paths))
  )
})
tar_make()
#> ● run target paths_files
#> ● run branch paths_5c47d23d
#> ● run branch paths_63f6955e
#> ● run branch data_ecdaefee
#> ● run branch data_34ea7b1c
#> ● end pipeline

这适用于下游效果，即如果我修改 b.csv，那么下游的相应分支 (data) 会再次运行。但是，我将如何创建一个向上游传播的目标（即，如果 b.csv 缺失，那么我希望运行一个规则来重新创建它）。

谢谢，

解决方法

在这种情况下，管道中的目标应创建文件。草图：

# _targets.R file:
library(readr)
library(targets)
library(tibble)
options(crayon.enabled = FALSE,tidyverse.quiet = TRUE)

helper <- function(path) {
  write_csv(tibble(x1 = 1,x2 = 1),path) # 1. Write the file.
  path                                    # 2. Return the path.
}

list(
  tar_target(paths,c("a.csv","b.csv")),tar_target(files,helper(paths),pattern = map(paths),format = "file"),tar_target(data,read_csv(paths,col_types = cols()),pattern = map(files))
)