R/wr.data.R

In finlaycampbell/WhatsAppR: Visualizing WhatsApp Conversation Data

##' Reading WhatsApp conversation data
##'
##' @author Finlay Campbell <f.campbell15@@imperial.ac.uk>
##'
##' @return Returns a data frame with formatted WhatsApp data
##' @export
##'
##' @param file the directory of the WhatsApp .txt file
##' @param OS indicating if the operating system is "android" or "iOS"
##' @param CheckForErrors will check for errors in reading of input data, mostly due to quoted text

wr.data <- function(file,OS,CheckForErrors=FALSE){

    input <- readLines(file)

    data <- data.frame(date=as.Date(substr(input,1,10),"%d/%m/%Y"))

    if(OS=="android"){
        find.index <- 2
        start_name <- 21
        data$time <- chron::times(paste(substr(input,13,17),":00",sep=""))
    }

    if(OS=="iOS"){
        find.index <- 4
        start_name <- 24
        data$time <- chron::times(substr(input,13,20))
    }

    if(OS=="jason"){
        find.index <- 4
        start_name <- 22
        data$time <- chron::times(substr(input,12,19))
    }


    data$person <- sapply(input,function(line) substr(line,start_name,gregexpr(":", line)[[1]][find.index]-1))

    data$text <- sapply(input,function(line) substr(line,gregexpr(":", line)[[1]][find.index]+2,nchar(line)))

    data$words <- sapply(data$text,function(line) length(strsplit(line," ")[[1]]))

    data$fulltime <- strptime(paste(data$date,data$time),format='%Y-%m-%d %H:%M')

    data <- data[order(data$fulltime),]

    #Remove 'people' that are found less than ten times (error when you have pasted wa conversations in chat)
    if(CheckForErrors){
        people <- names(sort(table(data$person)[table(data$person)>10],decreasing=TRUE))
        data$person <- factor(data$person,levels=people)
        errors <- !data$person %in% people
        if(sum(errors) > 0.05*nrow(data)) stop("More than 10% of messages can't be read, error in input format")
        data <- data[!errors,]
    } else {
        people <- names(sort(table(data$person),decreasing=TRUE))
        data$person <- factor(data$person,levels=people)
    }

    return(data)

}