Merge pull request #87 from stitam/tpl

TPL is no longer available for download
ropensci · Jan 31, 2025 · 705192e · 705192e
2 parents bc00b7b + c0588c3
commit 705192e
Show file tree

Hide file tree

Showing 15 changed files with 110 additions and 53 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -32,4 +32,4 @@ Imports:
 Suggests:
     testthat,
     taxize
-RoxygenNote: 7.3.1
+RoxygenNote: 7.3.2
diff --git a/NEWS.md b/NEWS.md
@@ -1,3 +1,11 @@
+dev
+==============
+
+## DEFUNCT
+
+* TPL is no longer available so `db_download_tpl()` will fail. However, if the database was downloaded earlier, it can still be queried using `taxizedb` functions.
+
+
 taxizedb 0.3.1
 ==============
 

diff --git a/R/db_download.R b/R/db_download.R
@@ -8,6 +8,9 @@
 #' @return (character) path to the downloaded SQL database
 #' @details Downloads sql database, cleans up unneeded files, returns path
 #' to sql file
+#' @note The Plant List (TPL) is no longer accessible. If you have a copy of the
+#' sqlite database you can still use the rest of the TPL functions with it. We 
+#' suggest using the World Flora Online (WFO) database as a replacement.
 #' @seealso [tdb_cache]
 #' @examples \dontrun{
 #' # ITIS
@@ -16,7 +19,6 @@
 #'
 #' # Plantlist
 #' # db_download_tpl()
-#' # db_download_tpl(overwrite=TRUE) # overwrite - download again
 #' # src_tpl()
 #'
 #' # COL
@@ -263,29 +265,13 @@ db_download_tpl <- function(verbose = TRUE, overwrite = FALSE) {
   if (file.exists(final_file) && !overwrite) {
     mssg(verbose, "Database already exists, returning old file")
     return(final_file)
+  } else {
+    stop(
+      "The Plant List (TPL) is no longer accessible. If you have a copy of the
+      sqlite database you can still use the rest of the TPL functions with it. 
+      We suggest using the World Flora Online (WFO) database as a replacement."
+    )
   }
-  unlink(final_file, force = TRUE)
-
-  # make home dir if not already present
-  tdb_cache$mkdir()
-  # download data
-  taxize::tpl_get(db_path_dir)
-  # convert to sqlite 
-  files <- list.files(db_path_dir, full.names = TRUE)
-  df <- data.table::rbindlist(lapply(files, data.table::fread), fill = TRUE)
-  names(df) <- gsub("\\s", "_", tolower(names(df)))
-  df$scientificname <- paste(df$genus, df$species)
-  x <- DBI::dbConnect(RSQLite::SQLite(), final_file)
-  DBI::dbWriteTable(x, "tpl", df)
-  # create indices
-  RSQLite::dbExecute(x, 'CREATE UNIQUE INDEX id ON tpl (id)')
-  RSQLite::dbExecute(x, 'CREATE INDEX sciname ON tpl (scientificname)')
-  # disconnect
-  DBI::dbDisconnect(x)
-  # clean up
-  unlink(db_path_dir, recursive = TRUE)
-  # return path
-  return(final_file)
 }
 
 #' @export

diff --git a/R/db_load.R b/R/db_load.R
@@ -1,6 +1,7 @@
-#' Load taxonomic databases - NO LONGER NEEDED
+#' Load taxonomic databases
 #' 
-#' Use [db_download] then [src_taxizedb]
+#' This family of functions has been removed. Use [db_download] then 
+#' [src_taxizedb].
 #'
 #' @export
 #' @param ... ignored

diff --git a/R/taxizedb-package.R b/R/taxizedb-package.R
@@ -90,8 +90,11 @@
 #' hiers %>% select(TSN, level)
 #'
 #' # data source: The PlantList
-#' ## download tpl datababase
-#' db_download_tpl()
+#' 
+#' "The Plant List (TPL) is no longer accessible. If you have a copy of the
+#' sqlite database you can still use the rest of the TPL functions with it.
+#' We suggest using the World Flora Online (WFO) database as a replacement." 
+#' 
 #' ## connecto the tpl database
 #' src <- src_tpl()
 #' ## do queries

diff --git a/README.Rmd b/README.Rmd
@@ -44,7 +44,6 @@ When you download a database with `taxizedb` it will automatically convert it to
 
 - [NCBI Taxonomy](https://www.ncbi.nlm.nih.gov/taxonomy)
 - [ITIS](https://itis.gov/)
-- [The Plant List (TPL)](http://www.theplantlist.org/) - Note that The Plant List has been superseded by World Flora Online.
 - [World Flora Online (WFO)](http://www.worldfloraonline.org/)
 - [Catalogue of Life (COL)](https://www.catalogueoflife.org/)
 - [Global Biodiversity Information Facility (GBIF)](https://www.gbif.org/)
@@ -53,6 +52,10 @@ When you download a database with `taxizedb` it will automatically convert it to
 Get in touch [in the issues](https://github.com/ropensci/taxizedb/issues) with
 any ideas on new data sources.
 
+## Data sources - legacy support
+
+[The Plant List (TPL)](https://en.wikipedia.org/wiki/The_Plant_List) has been replaced by World Flora Online. The website seems to be down so `taxizedb` can no longer facilitate new downloads. However, already downloaded database files can still be queried using `taxizedb` functions, for reproducibility.
+
 ## Package API
 
 This package for each data sources performs the following tasks:

diff --git a/README.md b/README.md
@@ -43,8 +43,6 @@ supported:
 
 - [NCBI Taxonomy](https://www.ncbi.nlm.nih.gov/taxonomy)
 - [ITIS](https://itis.gov/)
-- [The Plant List (TPL)](http://www.theplantlist.org/) - Note that The
-  Plant List has been superseded by World Flora Online.
 - [World Flora Online (WFO)](http://www.worldfloraonline.org/)
 - [Catalogue of Life (COL)](https://www.catalogueoflife.org/)
 - [Global Biodiversity Information Facility
@@ -55,6 +53,14 @@ Get in touch [in the
 issues](https://github.com/ropensci/taxizedb/issues) with any ideas on
 new data sources.
 
+## Data sources - legacy support
+
+[The Plant List (TPL)](https://en.wikipedia.org/wiki/The_Plant_List) has
+been replaced by World Flora Online. The website seems to be down so
+`taxizedb` can no longer facilitate new downloads. However, already
+downloaded database files can still be queried using `taxizedb`
+functions, for reproducibility.
+
 ## Package API
 
 This package for each data sources performs the following tasks:

diff --git a/man/db_download.Rd b/man/db_download.Rd
diff --git a/man/db_load-defunct.Rd b/man/db_load-defunct.Rd
diff --git a/man/taxizedb-package.Rd b/man/taxizedb-package.Rd
diff --git a/tests/testthat/test-children.R b/tests/testthat/test-children.R
@@ -64,11 +64,17 @@ test_that("missing values are consistent with taxize", {
 })
 
 test_that('children fails well', {
-  # tpl not supported
-  expect_error(
-    children("Poa", db = "tpl"),
-    "TPL database is not supported"
-  )
+
+  # TPL is no longer accessible. We can only test locally.
+  file <- file.path(tdb_cache$cache_path_get(), 'plantlist.sqlite')
+  if (file.exists(file)) {
+    # tpl not supported
+    expect_error(
+      children("Poa", db = "tpl"),
+      "TPL database is not supported"
+    )
+  }
+
   # abc does not exist
   expect_error(
     children("Poa annua", db = "abc"),

diff --git a/tests/testthat/test-classification.R b/tests/testthat/test-classification.R
@@ -118,11 +118,17 @@ test_that('classification handles mixed inputs', {
 })
 
 test_that('classification fails well', {
-  # tpl not supported
-  expect_error(
-    classification("Poa annua", db = "tpl"),
-    "TPL database is not supported"
-  )
+
+  # TPL is no longer accessible. We can only test locally.
+  file <- file.path(tdb_cache$cache_path_get(), 'plantlist.sqlite')
+  if (file.exists(file)) {
+    # tpl not supported
+    expect_error(
+      classification("Poa annua", db = "tpl"),
+      "TPL database is not supported"
+    )
+  }
+
   # abc does not exist
   expect_error(
     classification("Poa annua", db = "abc"),

diff --git a/tests/testthat/test-db_download.R b/tests/testthat/test-db_download.R
@@ -11,3 +11,20 @@ test_that("db_download - are functions", {
   expect_named(formals(db_download_itis), c("verbose", "overwrite"))
   expect_named(formals(db_download_tpl), c("verbose", "overwrite"))
 })
+
+test_that("db_download_tpl()", {
+  file <- file.path(tdb_cache$cache_path_get(), 'plantlist.sqlite')
+  if (file.exists(file)) {
+    # existing file can be imported
+    suppressMessages(expect_no_error(db_download_tpl()))
+    expect_equal(
+      capture_messages(db_download_tpl()),
+      "Database already exists, returning old file\n"
+    )
+    # existing file cannot be updated
+    expect_error(db_download_tpl(overwrite = TRUE))
+  } else {
+    # new file cannot be downloaded
+    expect_error(db_download_tpl())
+  }
+})
diff --git a/tests/testthat/test-downstream.R b/tests/testthat/test-downstream.R
@@ -36,12 +36,19 @@ test_that("downstream works with GBIF", {
   expect_named(x[[1]], c("id", "name", "rank"))
 })
 
+# TPL is no longer accessible. We can only test locally.
+test_that('downstream fails well', { 
+  file <- file.path(tdb_cache$cache_path_get(), 'plantlist.sqlite')
+  if (file.exists(file)) {
+    # tpl not supported
+    expect_error(
+      downstream("Poa", db = "tpl"),
+      "TPL database is not supported"
+    )
+  }
+})
+
 test_that('downstream fails well', {
-  # tpl not supported
-  expect_error(
-    downstream("Poa", db = "tpl"),
-    "TPL database is not supported"
-  )
   # abc does not exist
   expect_error(
     downstream("Poa annua", db = "abc"),

diff --git a/tests/testthat/test-realthing.R b/tests/testthat/test-realthing.R
@@ -9,7 +9,6 @@ test_that("We can actually download, load, and query all databases", {
   ## These should load from cache if we already have the download!
   gbif <- db_download_gbif()
   itis <- db_download_itis()
-  tpl <- db_download_tpl()
   col <- db_download_col()
   ncbi <- db_download_ncbi()
   wfo <- db_download_wfo()
@@ -25,9 +24,15 @@ test_that("We can actually download, load, and query all databases", {
   ## Load the actual data:
   gbif_db <- src_gbif()
   itis_db <- src_itis()
-  tpl_db <- src_tpl()
   col_db <- src_col()
   ncbi_db <- src_ncbi()
   wfo_db <- src_wfo()
   wikitaxa_db <- src_wikidata()
+
+  # TPL is no longer accessible. We can only test locally.
+  file <- file.path(tdb_cache$cache_path_get(), 'plantlist.sqlite')
+  if (file.exists(file)) {
+    tpl <- db_download_tpl()
+    tpl_db <- src_tpl()
+  }
 })