Question

我正在尝试按多个条件分割字符串，并存储每个分割的分割条件。

我一直在尝试使用stringr::str_split包，但不能向该函数传递多个拆分条件。

例如，如果我有以下字符串：

data = "Julie (title) : This is the text Julie has: said. Extra sentence one. Extra sentence 2 and so on. Rt Hon Ellen: This is the text Ellen has said in response to Julie. TITLE OF SECTION Julie: More words from Julie."

和拆分条件：

names = c("Julie:", "Ellen:")

我想要这样的输出：

data.frame(Names = c("Julie:", "Ellen:","Julie:"),
           text = c(" This is the text Julie has: said. Extra sentence one. Extra sentence 2 and so on. ", "This is the text Ellen has said in response to Julie.","More words from Julie."))

Answer 1

我已经在coatless的答案中看到了您的评论，并创建了一个样本数据，该数据可能反映了您所说的内容。一种方法如下。我首先创建一个数据框。我使用unnest_tokens()分割了每个句子的字符串。然后，我使用separate()拆分句子。最后，我用人的名字代替了NA。我希望这会对您有所帮助。

library(tidyverse)
library(tidytext)
library(zoo)

so <- tibble(text = "Ana: I went to school today. I learned text mining. Bob: That is great! Ana: I know what to do: practice.")

unnest_tokens(so, output = sentence,
              input = text,
              token = "sentences") %>% 
separate(col = sentence, into = c("person", "sentence"), sep = ": ",
         extra = "merge", fill = "left") %>% 
mutate(person = na.locf(person))

# A tibble: 4 x 2
#  person sentence                    
#  <chr>  <chr>                       
#1 ana    i went to school today.     
#2 ana    i learned text mining.      
#3 bob    that is great!              
#4 ana    i know what to do: practice.

Answer 2

低效的Base R解决方案：

# Store a vector of the names: 

text_names <- c("Julie", "Ellen")

# Create a dataframe of the patterns: 

pattern_search <- data.frame(name_search = c(paste0(text_names, ":"),

                                             paste0(text_names, " :"),

                                             paste0(text_names, ".* :")),

                             stringsAsFactors = F)

# Split the text into sentences: 

split_text <- data.frame(sentences = trimws(unlist(strsplit(df$Text, "[.]")), "both"), stringsAsFactors = F)

# Extract the names, store them in a vector:

names_in_order <- gsub("[[:punct:]]|\\s+.*",

                       "",

                       regmatches(grep(paste0(pattern_search$name_search, collapse = "|"),

                           split_text$sentences, value = T),

                      regexpr(paste0(pattern_search$name_search, collapse =  "|"),

                              grep(paste0(pattern_search$name_search, collapse = "|"),

                                   split_text$sentences, value = T))))

# Store a logical vector denoting which elements the names should go: 

split_text$who_said_this <- grepl(paste0(pattern_search$name_search, collapse = "|"),

                                  split_text$sentences)

# Replace all occurences of TRUE with the elements of the vector of names: 

split_text$who_said_this[which(split_text$who_said_this == TRUE)] <- names_in_order

# Replace FALSE with NA values: 

split_text$who_said_this[which(split_text$who_said_this == "FALSE")] <- NA

# Store a vector that's values denote the number of times dialogue changes between the names:

split_text$speech_group_no <- ave(split_text$who_said_this,

                                  split_text$who_said_this,

                                  FUN = seq.int)

# Apply a function to fill NA values with the non-NA value above it: 

split_text <- data.frame(lapply(split_text, function(x){na.omit(x)[cumsum(!is.na(x))]}),

                         stringsAsFactors = F)

# Row-wise concatenate the dataframe by group: 

split_text <- aggregate(list(sentences = c(split_text$sentences)), 

                 list(speech_group_no = paste0(split_text$who_said_this, " - ", split_text$speech_group_no)),

                 paste0,

                 sep = ". ")

# Flatten list vector into a character vector and clean up punctuation: 

split_text$sentences <- gsub(" [,] ", " ", sapply(split_text$sentences, toString))

# Order the dialogue:

split_text <- split_text[match(split_text$speech_group_no,

                 paste(names_in_order, ave(names_in_order, names_in_order, FUN = seq.int), sep = " - ")),]

数据：

df <-  structure(
list(Text = "Julie (title) : This is the text Julie has: said. Extra sentence one. Extra sentence 2 and so on. Rt Hon Ellen: This is the text Ellen has said in response to Julie. TITLE OF SECTION Julie: More words from Julie."),
class = "data.frame",
row.names = c(NA,-1L)
  )

R通过多个分割条件分割字符串并存储分割条件

2 个答案: