Add more version info on build, assign class traits.build (#126)

* Ensure call for SHA works when not in a git repo
* Add traits.build version into main metadata (for #124)
* Add tests for util_get_SHA and adding `resource_type` to main metadata
* Assign output as class traits.build

---------

Co-authored-by: yangsophieee <[email protected]>

R/get.R

@@ -47,5 +47,8 @@ util_get_version <- function(path =  "config/metadata.yml") {
 #' @return 40-digit SHA character string for the latest commit to the repository
 #' @export
 util_get_SHA <- function(path = ".") {
-  git2r::sha(git2r::last_commit(git2r::repository(path)))
+  sha <- tryCatch({
+      git2r::sha(git2r::last_commit(git2r::repository(path)))
+    }, error = function(cond) {NA})
+  sha
 }

R/process.R

@@ -249,10 +249,28 @@ dataset_process <- function(filename_data_raw,
       traits %>% dplyr::filter(!(!is.na(.data$error) & (.data$error == "Missing value")))
   }
 
-  # Todo - resource_metadata
-  # - Add contributors
+  # Update metadata
+  metadata <- resource_metadata
+
+  if (is.null(metadata[["related_identifiers"]][1])) {
+    metadata[["related_identifiers"]] <- list()
+  }
+
+  metadata[["related_identifiers"]] <-
+    util_append_to_list(
+      metadata[["related_identifiers"]],
+      list(
+        related_identifier_type = "url",
+        identifier = "https://github.com/traitecoevo/traits.build",
+        relation_type = "isCompiledBy",
+        resource_type = "software",
+        version = as.character(packageVersion("traits.build"))
+      )
+    )
+
 
   # Combine for final output
+  ret <- 
   list(
     traits = traits %>% dplyr::filter(is.na(.data$error)) %>% dplyr::select(-dplyr::all_of(c("error", "unit_in"))),
     locations = locations,
@@ -271,9 +289,13 @@ dataset_process <- function(filename_data_raw,
     sources = sources,
     definitions = definitions,
     schema = schema,
-    metadata = resource_metadata,
+    metadata = metadata,
     build_info = list(session_info = utils::sessionInfo())
   )
+
+  class(ret) <- c("list", "traits.build")
+
+  ret
 }
 
 #' Build dataset
@@ -1858,6 +1880,9 @@ build_combine <- function(..., d = list(...)) {
                       session_info = utils::sessionInfo()
                       )
               )
+
+  class(ret) <- c("list", "traits.build")
+
   ret
 }
 

R/testdata.R

@@ -304,7 +304,7 @@ dataset_test_worker <-
     }
 
     expect_list <- function(data, info) {
-      expect_true(class(data) == "list", info = sprintf("%s - is not a list", info))
+      expect_true("list" %in% class(data), info = sprintf("%s - is not a list", info))
     }
 
     expect_list_names_valid <- function(data, info, label) {

tests/testthat/functions.R

@@ -54,7 +54,7 @@ expect_not_NA <- function(object, info = NULL, label = NULL) {
 
 
 expect_list <- function(data, info) {
-  expect_true(class(data) == "list", info = info)
+  expect_true("list" %in% class(data), info = info)
 }
 
 

tests/testthat/test-process.R

@@ -24,6 +24,7 @@ test_that("`dataset_process` is working", {
   expect_no_error(austraits_names <- schema$austraits$elements %>% names())
   expect_no_error(x <- dataset_process(test_data, test_config, schema, resource_metadata, unit_conversions))
   expect_type(x, "list")
+  expect_equal(class(x), c("list", "traits.build"))
   expect_length(x, 13)
   expect_named(x, austraits_names)
   expect_equal(nrow(x$excluded_data), 0)

tests/testthat/test-setup.R

@@ -594,7 +594,9 @@ test_that("`build_setup_pipeline` is working", {
   expect_true(file.copy("data/Test_2022/test-metadata.yml", "data/Test_2022/metadata.yml", overwrite = TRUE))
 
   expect_no_error(zip::unzip("config/testgit.zip"))
-  expect_no_error(sha <- git2r::sha(git2r::last_commit()))
+  # Expect no error if not within a git repo
+  expect_no_error(sha <- util_get_SHA("../../.."))
+  expect_no_error(sha <- util_get_SHA())
   # Expect error if path or method is wrong
   expect_error(build_setup_pipeline(path = "Datas"))
   expect_error(build_setup_pipeline(method = "grrrr"))
@@ -666,8 +668,29 @@ test_that("`build_setup_pipeline` is working", {
   expect_equal(austraits$build_info$git_SHA, sha)
   expect_equal(austraits$build_info$git_SHA, "6c73238d8d048781d9a4f5239a03813be313f0dd")
 
+  # Check output lists have required parts
+  ## Todo add mode here
+
+  ## sources
+
+  ## metadata
+
+  ## schema
+  expect_equal(austraits$schema, get_schema())
+
+  ## Compiled by traits.build
+  traits.build_tag <- last(austraits$metadata$related_identifiers)
+  expected_output <- list(
+    related_identifier_type = "url",
+    identifier = "https://github.com/traitecoevo/traits.build",
+    relation_type = "isCompiledBy",
+    resource_type = "software",
+    version = as.character(packageVersion("traits.build"))
+  )
+  expect_equal(traits.build_tag, expected_output)
   #expect_length(austraits_raw$taxa, 14) #not valid test with new `dataset_update_taxonomy setup`
   #expect_length(austraits$taxa, 14) #not valid test with new `dataset_update_taxonomy setup`
+
   expect_equal(nrow(austraits$taxa), nrow(austraits_raw$taxa))
 
   # Compare products from three methods, except `build_info`