From 995fd77aa4a3715616604e82f36a691872f3fda9 Mon Sep 17 00:00:00 2001
From: Matija Obreza <matija.obreza@croptrust.org>
Date: Wed, 4 Sep 2019 11:29:41 +0200
Subject: [PATCH] Added functions to check taxonomic names and geo data using
 Genesys Validator

---
 DESCRIPTION   |   5 ++-
 NAMESPACE     |   5 +++
 R/validator.R | 109 ++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 117 insertions(+), 2 deletions(-)
 create mode 100644 R/validator.R

diff --git a/DESCRIPTION b/DESCRIPTION
index e22da90..bd18fe1 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -1,5 +1,5 @@
 Package: genesysr
-Version: 0.9.3
+Version: 0.9.4
 Title: Genesys PGR Client
 Description: Access data on plant genetic resources from genebanks around the world published on Genesys (<https://www.genesys-pgr.org>).
   Your use of data is subject to terms and conditions available at <https://www.genesys-pgr.org/content/legal/terms>.
@@ -10,7 +10,8 @@ Maintainer: Matija Obreza <matija.obreza@croptrust.org>
 Depends: R (>= 3.1.0)
 Imports:
   httr,
-  jsonlite
+  jsonlite,
+  dplyr
 License: Apache License 2.0
 RoxygenNote: 6.1.1
 URL: https://gitlab.croptrust.org/genesys-pgr/genesysr
diff --git a/NAMESPACE b/NAMESPACE
index f04dcac..0436cbf 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -3,6 +3,9 @@
 export(api1_url)
 export(api_url)
 export(authorization)
+export(check_country)
+export(check_landorsea)
+export(check_taxonomy)
 export(client_login)
 export(download_mcpd)
 export(download_pdci)
@@ -18,3 +21,5 @@ export(setup_production)
 export(setup_sandbox)
 export(user_login)
 importFrom(utils,browseURL)
+importFrom(utils,read.csv)
+importFrom(utils,write.table)
diff --git a/R/validator.R b/R/validator.R
new file mode 100644
index 0000000..2b3f42b
--- /dev/null
+++ b/R/validator.R
@@ -0,0 +1,109 @@
+
+#' Check MCPD taxonomic data (GENUS, SPECIES, SPAUTHOR, SUBTAXA, SUBTAUTHOR)
+#' using https://validator.genesys-pgr.org.
+#' 
+#' Duplicate input rows are removed using dplyr::distinct() and results are
+#' returned for unique rows.
+#' 
+#' @param mcpd Accession passport data in MCPD format
+#' @param toCurrentTaxa Should obsoleted names be reported?
+#' 
+#' @examples
+#' \dontrun{
+#'   taxaCheck <- genesysr::check_taxonomy(mcpd)
+#' }
+#' 
+#' @return Results from valdator
+#' @export
+#' @importFrom utils read.csv write.table
+check_taxonomy <- function(mcpd, toCurrentTaxa = FALSE) {
+  DT <- dplyr::distinct(mcpd)
+  # print(DT)
+
+  CSV <- tempfile(pattern = "file", tmpdir = tempdir(), fileext = "")
+  write.table(DT, file = CSV, row.names = FALSE, dec=".", sep = "\t", quote = FALSE, na = "")
+  
+  TMP <- tempfile(pattern = "file", tmpdir = tempdir(), fileext = "")
+
+  # Requires toCurrentTaxa
+  response <- httr::POST("https://validator.genesys-pgr.org/process", body = list(
+    toCurrentTaxa = toCurrentTaxa,
+    separator = "\t", decimalMark = ".", 
+    encoding = "UTF-8", csvText = readChar(CSV, file.info(CSV)$size)
+  ), encode = "multipart", httr::accept("text/csv"), httr::write_disk(TMP)) # , httr::verbose())
+  
+  R <- read.csv(TMP, fileEncoding = "UTF-8", sep = "\t")
+  file.remove(CSV, TMP)
+  invisible(R)
+}
+
+#' Run Land-or-Sea check on MCPD data using https://validator.genesys-pgr.org.
+#' Uploads only rows where DECLATITUDE and DECLONGITUDE are provided. 
+#' In practice it is better to use `check_country` if ORIGCTY data exists.
+#' 
+#' @param mcpd Accession passport data in MCPD format
+#' 
+#' @examples
+#' \dontrun{
+#'   waterCheck <- genesysr::check_landorsea(mcpd)
+#' }
+#' 
+#' @return Results from valdator
+#' @export
+#' @importFrom utils read.csv write.table
+check_landorsea <- function(mcpd) {
+  GEO <- dplyr::filter(mcpd, ! is.na(mcpd$DECLATITUDE) & ! is.na(mcpd$DECLONGITUDE))
+  # print(GEO)
+
+  CSV <- tempfile(pattern = "landorsea", tmpdir = tempdir(), fileext = "")
+  write.table(GEO, file = CSV, row.names = FALSE, dec=".", sep = "\t", quote = FALSE, na = "")
+  readChar(CSV, file.info(CSV)$size)
+
+  TMP <- tempfile(pattern = "landorsea", tmpdir = tempdir(), fileext = "")
+  
+  response <- httr::POST("https://validator.genesys-pgr.org/process", body = list(
+    validateType = "landorsea",
+    separator = "\t", decimalMark = ".", 
+    encoding = "UTF-8", csvText = readChar(CSV, file.info(CSV)$size)
+  ), encode = "multipart", httr::accept("text/csv"), httr::write_disk(TMP)) # , httr::verbose())
+  
+  R <- read.csv(TMP, fileEncoding = "UTF-8", sep = "\t")
+  # print(R)
+  file.remove(CSV, TMP)
+  invisible(R)
+}
+
+#' Run Land-or-Sea check on MCPD data. Uploads only rows where ORIGCTY,
+#' DECLATITUDE and DECLONGITUDE are provided.
+#' 
+#' @param mcpd Accession passport data in MCPD format
+#' 
+#' @examples
+#' \dontrun{
+#'   geoCheck <- genesysr::check_country(mcpd)
+#' }
+#' 
+#' @return Results from valdator
+#' @export
+#' @importFrom utils read.csv write.table
+check_country <- function(mcpd) {
+  GEO <- dplyr::filter(mcpd, is.character(mcpd$ORIGCTY) & ! is.na(mcpd$DECLATITUDE) & ! is.na(mcpd$DECLONGITUDE))
+  # print(GEO)
+
+  CSV <- tempfile(pattern = "landorsea", tmpdir = tempdir(), fileext = "")
+  write.table(GEO, file = CSV, row.names = FALSE, dec=".", sep = "\t", quote = FALSE, na = "")
+  readChar(CSV, file.info(CSV)$size)
+
+  TMP <- tempfile(pattern = "landorsea", tmpdir = tempdir(), fileext = "")
+  
+  response <- httr::POST("https://validator.genesys-pgr.org/process", body = list(
+    validateType = "country",
+    separator = "\t", decimalMark = ".", 
+    encoding = "UTF-8", csvText = readChar(CSV, file.info(CSV)$size)
+  ), encode = "multipart", httr::accept("text/csv"), httr::write_disk(TMP)) # , httr::verbose())
+  
+  R <- read.csv(TMP, fileEncoding = "UTF-8", sep = "\t")
+  # print(R)
+  file.remove(CSV, TMP)
+  invisible(R)
+}
-- 
GitLab