r - 按非结构化文本列分组以透视数据

Question

我有一些文本数据，如下所示:

   ID                      text
1                              
2                              
3                              
4  HD                 some text
5  LP            some more text
6  AN            even more text
7                              
8                              
9                              
10 HD       some different text
11 SN  some more different text
12 AN  even more different text

每个部分是一个文档，由一些空格分隔。文档以 ID 作为 HD 开始，以 ID 和 AN 结束。我正在尝试做两件事，最终我想 pivot_wider 数据并将列作为 ID 然后每一行都是一个文档。我运行以下命令:

widerText <- textData %>% 
  pivot_wider(names_from = ID, values_from = text)


finalText <- widerText %>% 
  unnest(HD, LP, AN, SN, PP, LO, AN)

它没有保留正确的结构，文本会混淆。因此，我想在运行 pivot_wider 之前创建一个分组变量。

每个文档都以 HD 开头并以 AN 结尾，因此我想创建具有以下输出的内容:

预期输出:

   ID                      text   grp
1                                  0
2                                  0
3                                  0
4  HD                 some text    1
5  LP            some more text    1
6  AN            even more text    1
7                                  0
8                                  0
9                                  0
10 HD       some different text    2
11 SN  some more different text    2
12 AN  even more different text    2

数据:

textData <- data.frame(
  ID = c(
    " ", " ", " ", "HD", "LP", "AN",
    " ", " ", " ", "HD", "SN", "AN",
    " ", " ", " ", "HD", "PP", "AN",
    " ", " ", " ", "HD", "LO", "AN"

    ),
  text = c(
    " ", " ", " ", "some text", "some more text", "even more text",
    " ", " ", " ", "some different text", "some more different text", "even more different text",
    " ", " ", " ", "some additional text", "some more additional text", "even more additional text",
    " ", " ", " ", "some extra text", "some more extra text", "even more extra text"
    )
)

Answer 1

您可以在 textData$ID == "HD" 上使用 cumsum 获取组，并将空组设置为 0 使用ifelse.

textData$grp <- ifelse(textData$ID==" ", 0, cumsum(textData$ID == "HD"))
textData
#   ID                      text grp
#1                                 0
#2                                 0
#3                                 0
#4  HD                 some text   1
#5  LP            some more text   1
#6  AN            even more text   1
#7                                 0
#8                                 0
#9                                 0
#10 HD       some different text   2
#11 SN  some more different text   2
#12 AN  even more different text   2
#13                                0
#14                                0
#15                                0
#16 HD      some additional text   3
#17 PP some more additional text   3
#18 AN even more additional text   3
#19                                0
#20                                0
#21                                0
#22 HD           some extra text   4
#23 LO      some more extra text   4
#24 AN      even more extra text   4