From 4d3962e96a6eae6c888020e443be77344d80103e Mon Sep 17 00:00:00 2001
From: Flavio Poletti <flavio.poletti@hotmail.ch>
Date: Thu, 27 Jan 2022 10:34:40 +0100
Subject: [PATCH] functions to convert empty strings to NA and vice versa

---
 NAMESPACE                   |  1 +
 R/io.R                      |  3 +++
 R/utils.R                   | 39 +++++++++++++++++++++++++++++++++++++
 man/empty_strings_to_na.Rd  | 17 ++++++++++++++++
 man/na_to_empty_strings.Rd  | 14 +++++++++++++
 tests/testthat/test-utils.R | 18 +++++++++++++++++
 6 files changed, 92 insertions(+)
 create mode 100644 man/empty_strings_to_na.Rd
 create mode 100644 man/na_to_empty_strings.Rd

diff --git a/NAMESPACE b/NAMESPACE
index 2ec062a5..497c75ee 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -4,6 +4,7 @@ S3method(plot,tidygtfs)
 S3method(print,tidygtfs)
 S3method(summary,tidygtfs)
 export(cluster_stops)
+export(empty_strings_to_na)
 export(filter_feed_by_area)
 export(filter_feed_by_date)
 export(filter_feed_by_stops)
diff --git a/R/io.R b/R/io.R
index be1f3788..cecca559 100644
--- a/R/io.R
+++ b/R/io.R
@@ -62,6 +62,9 @@ write_gtfs <- function(gtfs_obj, zipfile, compression_level = 9, as_dir = FALSE)
   # convert sf tables
   gtfs_out = sf_as_tbl(gtfs_obj)
   
+  # convert NA to empty strings
+  gtfs_out <- na_to_empty_strings(gtfs_out)
+  
   # data.tables
   gtfs_out <- gtfs_out[names(gtfs_out) != "."]
   gtfs_out <- lapply(gtfs_out, as.data.table)
diff --git a/R/utils.R b/R/utils.R
index 27e326e6..575ef49e 100644
--- a/R/utils.R
+++ b/R/utils.R
@@ -6,3 +6,42 @@ feed_contains <- function(gtfs_obj, table_name) {
   exists(table_name, where = gtfs_obj) ||
     (exists(".", where = gtfs_obj) && exists(table_name, where = gtfs_obj$.))
 }
+
+#' Convert empty strings ("") to NA values in gtfs tables
+#' 
+#' @param gtfs_obj tidygtfs object
+#' 
+#' @return a gtfs_obj where all empty strings in tables have been replaced with NA
+#' 
+#' @export
+empty_strings_to_na = function(gtfs_obj) {
+  tbl_names = names(gtfs_obj)
+  tbl_names <- tbl_names[tbl_names != "."]
+  for(tbl in tbl_names) {
+    if(inherits(gtfs_obj[[tbl]], "data.frame")) {
+      gtfs_obj[[tbl]][gtfs_obj[[tbl]] == ""] <- NA
+    }
+  }
+  gtfs_obj
+}
+
+#' Convert NA values to empty strings ("")
+#'  
+#' @param gtfs_obj tidygtfs object
+na_to_empty_strings = function(gtfs_obj) {
+  lapply(gtfs_obj, function(df) {
+    if(inherits(df, "data.frame")) {
+      df2 = lapply(df, function(.col) {
+        if(is.character(.col)) {
+          .col[is.na(.col)] <- ""
+        }
+        .col
+      })
+      attributes(df2) <- attributes(df)
+      df <- df2
+    } else if(inherits(df, "list")) {
+      df <- na_to_empty_strings(df)
+    }
+    df
+  })
+}
diff --git a/man/empty_strings_to_na.Rd b/man/empty_strings_to_na.Rd
new file mode 100644
index 00000000..a2671e9c
--- /dev/null
+++ b/man/empty_strings_to_na.Rd
@@ -0,0 +1,17 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/utils.R
+\name{empty_strings_to_na}
+\alias{empty_strings_to_na}
+\title{Convert empty strings ("") to NA values in gtfs tables}
+\usage{
+empty_strings_to_na(gtfs_obj)
+}
+\arguments{
+\item{gtfs_obj}{tidygtfs object}
+}
+\value{
+a gtfs_obj where all empty strings in tables have been replaced with NA
+}
+\description{
+Convert empty strings ("") to NA values in gtfs tables
+}
diff --git a/man/na_to_empty_strings.Rd b/man/na_to_empty_strings.Rd
new file mode 100644
index 00000000..45d71716
--- /dev/null
+++ b/man/na_to_empty_strings.Rd
@@ -0,0 +1,14 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/utils.R
+\name{na_to_empty_strings}
+\alias{na_to_empty_strings}
+\title{Convert NA values to empty strings ("")}
+\usage{
+na_to_empty_strings(gtfs_obj)
+}
+\arguments{
+\item{gtfs_obj}{tidygtfs object}
+}
+\description{
+Convert NA values to empty strings ("")
+}
diff --git a/tests/testthat/test-utils.R b/tests/testthat/test-utils.R
index 3370439f..6b633ccf 100644
--- a/tests/testthat/test-utils.R
+++ b/tests/testthat/test-utils.R
@@ -100,3 +100,21 @@ test_that("gtfs_meta", { # empty test
   expect_equal(gtfs_meta, get_gtfs_meta())
 })
 
+test_that("empty_strings_to_na", {
+  gpath = system.file("extdata", "sample-feed-fixed.zip", package = "tidytransit")
+  g1 = read_gtfs(gpath)
+  g_na = empty_strings_to_na(g1)
+  g2 = na_to_empty_strings(g_na)
+
+  for(tbl in names(g1)) {
+    expect_equal(g1[[tbl]], g2[[tbl]])
+  }
+  
+  tmppath = tempfile(fileext = ".zip")
+  write_gtfs(g_na, tmppath)
+  g3 = read_gtfs(tmppath)
+   
+  for(tbl in names(g1)) {
+    expect_equal(g1[[tbl]], g3[[tbl]])
+  } 
+})