regmatches从gregexpr返回多个匹配项

Question

我想在单词之前和之后抓2-3个单词：“help”

我有一个文本块如下：

....功能和许多绿色植物，以帮助安抚神经......等等... ...在他们的毛绒毯子舒适，以帮助放松神经

这就是我做的

x <- paste("(\\S+\\s+|^)(\\S+\\s+|)(\\S+\\s+|)", treatSym[i], ".?(\\s+\\S+|)(\\s+\\S+|$)(\\s+\\S+|$)", sep="")

matching <- gregexpr(x,text)

regmatches(text, matching, invert = FALSE)

我得到了这个错误，因为我猜测长度（匹配）= 2.但是当只有一个匹配时，它运行良好。

Error in regmatches(text, matching, invert = FALSE) : 
  ‘x’ and ‘m’ must have the same length

是否有更好的解决方案可以在关键字之前和之后调出2-3个字？

Answer 1

n是长度为2的向量，给出关键字之前和之后的单词数

n <- c(2, 2)
x <- "....features and lots of greenery to help soothe the nerves...blah blah...cozy up in their plush blankets to help relax the nerves"

pat <- sprintf('(?:[a-z]+ ){%s}help(?: [a-z]+){%s}', n[1], n[2])
m <- gregexpr(pat, x, perl = TRUE)
regmatches(x, m)[[1]]
# [1] "greenery to help soothe the" "blankets to help relax the" 

作为一个功能

f <- function(string, keyword, n = c(2,2)) {
  # pat <- sprintf('(?:[a-z]+ ){%s}%s(?: [a-z]+){%s}', n[1], keyword, n[2])
  pat <- sprintf('(?:[a-z]+ ){0,%s}%s(?: [a-z]+){0,%s}', n[1], keyword, n[2])
  m <- gregexpr(pat, string, perl = TRUE)
  regmatches(string, m)[[1]]
}

f(x, 'help', c(1, 2))
# [1] "to help soothe the" "to help relax the" 

Answer 2

另一个选择是分割单词，获取help索引，并在每个help之前/之后获得2或3个单词。

library(magrittr)
library(stringi)
library(SOfun)  ### https://github.com/mrdwab/SOfun

x <- "....features and lots of greenery to help soothe the nerves...blah blah...cozy up in their plush blankets to help relax the nerves"

选项1：得到的话

### Remove ... and split words
temp <- stri_replace_all_regex(pattern = "[[:punct:]]", replacement = " ", str = x) %>%
        stri_split_fixed(pattern = " ") %>%
        unlist %>%
        .[nchar(.) > 0]

data.frame(word = temp, stringsAsFactors = FALSE) %>%
getMyRows(pattern = grep(pattern = "help", x = .$word), range = -3:3) %>%
lapply(function(ana){ana[-grep(pattern = "help", x = ana)]})

#[[1]]
#[1] "of"       "greenery" "to"       "soothe"   "the"      "nerves"  
#
#[[2]]
#[1] "plush"    "blankets" "to"       "relax"    "the"      "nerves" 

如果您想查看为每个help选择的单词，可以尝试以下操作。

选项2：创建数据框

temp <- stri_replace_all_regex(pattern = "[[:punct:]]", replacement = " ", str = x) %>%
        stri_split_fixed(pattern = " ") %>%
        unlist %>%
        .[nchar(.) > 0]

data.frame(word = temp, stringsAsFactors = FALSE) %>%
getMyRows(pattern = grep(pattern = "help", x = .$word), range = -3:3) %>%
lapply(function(ana){ana[-grep(pattern = "help", x = ana)]}) -> temp


do.call(rbind,
        lapply(temp, function(y){
                        data.frame(word = y,
                                   ind = c(-3:-1, 1:3),
                                   stringsAsFactors = FALSE)}
              )
        )

# ind indicates relative positions of the words. words with negative
# numbers are on left side of help. Words with positive numbers on right.

#       word ind
#1        of  -3
#2  greenery  -2
#3        to  -1
#4    soothe   1
#5       the   2
#6    nerves   3
#7     plush  -3
#8  blankets  -2
#9        to  -1
#10    relax   1
#11      the   2
#12   nerves   3

Answer 3

你可以使用quanteda包quanteda这样的事情。

my.string <- "....features and lots of greenery to help soothe the nerves...blah blah...cozy up in their plush blankets to help relax the nerves"

library(quanteda)
kwic(my.string, "help", window = 3, valuetype = "fixed")

                     contextPre keyword         contextPost
[text1, 11]    of greenery to [    help ] soothe the nerves
[text1, 30] plush blankets to [    help ] relax the nerves