R/readers.R

In eteppo/finnordear: Finnish Nordea Bank Data Analysis Tools

#' Read and clean a Nordea bank statement pdf-file.
#'
#' \code{read_statement} reads and cleans up a Finnish Nordea bank statement
#' pdf-file and returns a tidy dataset of classes data.frame (base) and "tbl_df" (tibble).
#'
#' @param file A full path to a Finnish Nordea bank statement pdf-file.
#'
#' @return A dataset of balances and transactions of classes data.frame (base) and tbl_df (tibble).
#'
#' @importFrom dplyr "%>%"
#'
#' @export

read_statement <- function(file) {

    pdftools::pdf_text(file) %>%
        stringr::str_split("\n") %>%
        unlist() %>%
        stringr::str_squish() %>%
        dplyr::tibble(line = 1:length(.), text = .) %>%
        # lines with numbers with , and ending with + or -
        dplyr::filter(
            stringr::str_detect(text, ",[0-9]+\\+") |
                stringr::str_detect(text, ",[0-9]+-")
        ) %>%
        tidyr::extract(text, "amount", "([0-9\\.]+,[0-9\\.\\+-]+)", remove = FALSE) %>%
        tidyr::extract(text, "counterparty", "([a-zA-Z]+.+[a-zA-Z]+)", remove = FALSE) %>%
        tidyr::extract(text, "record_date", "(^[0-9][0-9\\.]+[0-9]+)") %>%
        dplyr::filter(!is.na(record_date)) %>%
        dplyr::mutate(type = dplyr::case_when(
            is.na(record_date) ~ "balance",
            stringr::str_detect(record_date, "\\.[0-9]+\\.") ~ "balance",
            TRUE ~ "transaction"
        )) %>%
        dplyr::filter(!(dplyr::row_number() == nrow(.) & type == "transaction")) %>%
        tidyr::separate(record_date, c("day", "month", "year"), fill = "right") %>%
        dplyr::mutate(year = dplyr::if_else(
            condition = is.na(year),
            true = year %>%
                na.omit() %>%
                unique() %>%
                as.numeric() %>%
                max() %>%
                as.character(),
            false = year
        )) %>%
        tidyr::unite(col = record_date, day, month, year, sep = "-") %>%
        dplyr::mutate(
            record_date = lubridate::dmy(record_date),
            amount = amount %>%
                stringr::str_remove_all("\\.") %>%
                stringr::str_replace(",", ".") %>%
                stringr::str_remove("\\+")
        ) %>%
        dplyr::mutate(amount = dplyr::if_else(
            stringr::str_detect(amount, "-"),
            amount %>% stringr::str_remove("-") %>% stringr::str_c("-", .),
            amount
        )) %>%
        dplyr::mutate(
            amount = as.numeric(amount),
            counterparty = dplyr::if_else(counterparty %in% c("Saldo", "Loppusaldo"), "N/A", counterparty)
        ) %>%
        dplyr::select(
            type,
            record_date,
            counterparty,
            amount,
            -line
        ) %>%
        dplyr::arrange(type, record_date)

}

#' Read and clean many Finnish Nordea bank statement pdf-files at once.
#'
#' This is just a convenient plural version of \code{read_statement}.
#'
#' @param directory A full path to a directory of Finnish Nordea bank statement pdf-files.
#' @return A dataset of transactions and balances of classes data.frame (base) and tbl_df (tibble).
#' @importFrom dplyr "%>%"
#' @export

read_statements <- function(directory) {

    files <- directory %>%
        stringr::str_c("/", base::dir(directory)) %>%
        stringr::str_subset(".pdf")

    # problem : directory can contain duplicate files
    # solution: select unique statement files by calculating MD5 hashes on parsed text
    md5_hashes <- files %>%
        purrr::map_chr(function(x) digest::digest(pdftools::pdf_text(x), algo = "md5"))

    unique_files <- files[!duplicated(md5_hashes)]

    # problem: multiple different statements contain the same transaction
    # solution: remove all duplicated rows
    ## NOTE: valid identical transactions are lost too in this approach, e.g.,
    ## multiple identical purchases on the same day
    statements <- unique_files %>%
        purrr::map_dfr(finnordear::read_statement) %>%
        dplyr::distinct_all()

    return(statements)
}