Question

我使用data.table融合函数和patterns参数编写了一个R脚本。为了获得模式，我使用了一个变量。如果我在函数之外的脚本中使用它，一切都按预期工作。如果我在函数内的脚本中使用它，则会出现错误，表示找不到函数中调用的对象。但是，如果我在全局环境中定义它，该函数可以工作！

我想知道是否有人知道为什么需要在全局env中定义变量而不是函数env？

require(data.table)

DT <- structure(list(family_id = 1:5, age_mother = c(30L, 27L, 26L, 32L, 29L), 
                 dob_child1 = c("26/11/1998", "22/06/1996", "11/07/2002", "10/10/2004", "05/12/2000"), 
                 dob_child2 = c("29/01/2000", NA, "05/04/2004", "27/08/2009", "28/02/2005"), 
                 dob_child3 = c(NA,  NA, "02/09/2007", "21/07/2012", NA), 
                 gender_child1 = c(1L, 2L, 2L, 1L, 2L), 
                 gender_child2 = c(2L, NA, 2L, 1L, 1L), 
                 gender_child3 = c(NA, NA, 1L, 1L, NA)), 
            .Names = c("family_id", "age_mother", "dob_child1", "dob_child2", "dob_child3", "gender_child1", "gender_child2", "gender_child3"), 
            row.names = c(NA, -5L), 
            class = c("data.table", "data.frame"))

transpose_col <- "(^dob)|(^gender)"

# Works without problem.
melt(DT,
 measure = patterns(unlist(strsplit(gsub("\\(|\\)", "", transpose_col), "\\|"))), 
 value.name = unlist(strsplit(gsub("\\(|\\)|\\^", "", transpose_col), "\\|")))

# Bug:
in_local_env <-  function(DT) {
    transpose_col2 <- transpose_col
    melt(DT,
         measure = patterns(unlist(strsplit(gsub("\\(|\\)", "", transpose_col2), "\\|"))), 
         value.name = unlist(strsplit(gsub("\\(|\\)|\\^", "", transpose_col2), "\\|")))
}
in_local_env(DT) # generate an error.

in_global_env <- function(DT) {
    transpose_col3 <<- transpose_col
    transpost_col4 <- transpose_col
    melt(DT,
         measure = patterns(unlist(strsplit(gsub("\\(|\\)", "", transpose_col3), "\\|"))), 
         value.name = unlist(strsplit(gsub("\\(|\\)|\\^", "", transpost_col4), "\\|")))
}
in_global_env(DT) # Works as expected.

如果我在in_local_env中调试函数并逐行执行，那么在unlist(strsplit(gsub("\\(|\\)", "", transpose_col), "\\|"))执行问题时，熔化将无法正常工作。

我使用data.table 1.9.6

融合[data.table]参数模式和环境

0 个答案: