#367 support resource_type and source_type options

r-geoflow · Oct 9, 2023 · 21ff716 · 21ff716
1 parent a768cce
commit 21ff716
Show file tree

Hide file tree

Showing 5 changed files with 145 additions and 3 deletions.
diff --git a/R/geoflow_handler_entity.R b/R/geoflow_handler_entity.R
@@ -113,7 +113,13 @@ register_entity_handlers <- function(){
       id = "zenodo",
       def = "Handle metadata entities built from a Zenodo source",
       packages = list("zen4R"),
-      fun = source(system.file("metadata/entity", "entity_handler_zenodo.R", package = "geoflow"))$value
+      fun = source(system.file("metadata/entity", "entity_handler_zenodo.R", package = "geoflow"))$value,
+      available_options = list(
+        resource_type = list(def = "Type resource. By default 'deposit', use this if you are the owner of the records you fetch.
+                             Alternative option value is 'record' that can be used for third-party record handling (as anonymous user", default = "deposit"),
+        source_type = list(def = "Type of source for handling Zenodo input deposits/records. Possible values are ['query': an ElasticSearch query, 
+                           'doi': one or more DOIs]", default = 'query')
+      )
     )
   )
   .geoflow$entity_handlers <- handlers

diff --git a/inst/extdata/workflows/config_metadata_zenodo_with_options.json b/inst/extdata/workflows/config_metadata_zenodo_with_options.json
@@ -0,0 +1,37 @@
+{
+  "profile": {
+    "id": "my-workflow",
+    "name": "My workflow",
+    "project": "Test geoflow project",
+    "organization": "My organization",
+    "logos": [
+      "https://via.placeholder.com/300x150.png/09f/fff?text=geoflow"
+      ],
+    "mode": "entity"
+  },
+  "metadata": {
+    "entities": [
+      {
+        "handler": "zenodo",
+        "source": ["10.5281/zenodo.2547036"],
+        "options": {
+          "resource_type": "record",
+          "source_type": "doi"
+        }
+      }
+    ],
+    "contacts" : []
+  },
+  "software": [
+     {
+  		"id": "my-zenodo",
+  	  "type": "input",
+  	  "software_type": "zenodo",
+  	  "parameters": {
+    		"url": "https://zenodo.org/api",
+    		"logger": "DEBUG"
+  	   }
+  	 }
+    ],
+  "actions": []
+}
diff --git a/inst/extdata/workflows/workflows.csv b/inst/extdata/workflows/workflows.csv
@@ -15,4 +15,5 @@ config_metadata_gsheets_sdi_geoserver_geotiff_view.json,Load metadata entities/c
 config_metadata_gsheets_sdi_geoserver_geotiff_directory.json,Load metadata entities/contacts from Google Spreadsheets and publish multiple GeoTIFF based layers on GeoServer for a single entity
 config_metadata_ocs.json,Load metadata entities/contacts/dictionary from CSV files hosted on a cloud supporting OCS API (eg. nextcloud/owncloud)
 config_metadata_dataverse.json,Load metadata entities/contacts from Dataverse records
-config_metadata_zenodo.json,Load metadata entities/contacts from Zenodo deposits/records
+config_metadata_zenodo.json,Load metadata entities/contacts from Zenodo deposits/records
+config_metadata_zenodo_with_options.json,Load metadata entities/contacts from Zenodo deposits/records (with options declared to fetch from public records and by DOI)
diff --git a/inst/metadata/entity/entity_handler_zenodo.R b/inst/metadata/entity/entity_handler_zenodo.R
@@ -11,7 +11,56 @@ handle_entities_zenodo <- function(handler, source, config, handle = TRUE){
   }
   ZENODO_CONFIG <- config$software$input$zenodo_config
 
-  results <- ZENODO$getDepositions(q = source)
+  resource_type <- handler$getOption("resource_type") #for this handler default is "deposit"
+  source_type <- handler$getOption("source_type") #for this handler default is "query"
+
+  #fetch results depending on resource and source types
+  config$logger.info(sprintf("Zenodo entity handler: resource type '%s'", resource_type))
+  config$logger.info(sprintf("Zenodo entity handler: source type '%s'", source_type))
+  results <- switch(resource_type,
+    "deposit" = {
+      switch(source_type,
+        "query" = {
+          ZENODO$getDepositions(q = source)
+        },
+        "doi" = {
+          recs = lapply(source, function(doi){
+            rec = ZENODO$getDepositionByDOI(doi)
+            if(is.null(rec)) rec = ZENODO$getDepositionByConceptDOI(doi)
+            return(rec)
+          })
+          recs = recs[!sapply(recs, is.null)]
+          recs
+        },{
+          config$logger.warn(sprintf("Unknown source type '%s' in Zenodo entity handler", source_type))
+          list()
+        }
+      )
+     },
+    "record" = {
+     switch(source_type,
+      "query" = {
+         ZENODO$getRecords(q = source) 
+       },
+       "doi" = {
+         recs = lapply(source, function(doi){
+           rec = ZENODO$getRecordByDOI(doi)
+           if(is.null(rec)) rec = ZENODO$getRecordByConceptDOI(doi)
+           return(rec)
+         })
+         recs = recs[!sapply(recs, is.null)]
+         recs
+       },{
+         config$logger.warn(sprintf("Unknown source type '%s' in Zenodo entity handler", source_type))
+         list()
+       }
+     )  
+    },{
+      config$logger.warn(sprintf("Unknown resource type '%s' in Zenodo entity handler. Valid values ['deposit','record']", resource_type))
+      list()
+    }
+  )
+  if(length(results)==1) if(is(results, "ZenodoRecord")) results = list(results)
   if(length(results)==0) return(list())
 
   entities <- lapply(1:length(results), function(i){
@@ -27,6 +76,8 @@ handle_entities_zenodo <- function(handler, source, config, handle = TRUE){
     if(length(identifiers)>0){
       identifier = identifiers[[1]]$identifier
       entity$setIdentifier("id", substr(identifier, 5,nchar(identifier)))
+    }else{
+      entity$setIdentifier("id", gsub("/", "_", result$doi))
     }
 
     #entity Date

diff --git a/tests/testthat/test_config_metadata_zenodo_with_options.R b/tests/testthat/test_config_metadata_zenodo_with_options.R
@@ -0,0 +1,47 @@
+# test_config_metadata_zenodo_with_options.R
+# Author: Emmanuel Blondel <emmanuel.blondel1@gmail.com>
+#
+# Description: Integration tests for config_metadata_zenodo_with_options.json workflow
+#=======================
+require(geoflow, quietly = TRUE)
+require(testthat)
+
+#init
+test_that("init",{
+  testthat::skip_on_cran()
+  cfg_file = system.file("extdata/workflows/config_metadata_zenodo_with_options.json", package = "geoflow")
+  CFG <- geoflow::initWorkflow(cfg_file)
+  expect_is(CFG$metadata$content, "list")
+  expect_equal(length(CFG$metadata$content), 1L)
+  expect_equal(names(CFG$metadata$content), "entities")
+  expect_equal(length(CFG$metadata$content$entities), 1L)
+  expect_equal(length(CFG$getEntities()), 1L)
+  expect_equal(length(CFG$actions), 0L)
+  expect_equal(length(CFG$software), 2L)
+  expect_equal(names(CFG$software), c("input", "output"))
+  expect_equal(length(CFG$software$input), 2L)
+  expect_equal(length(CFG$software$output), 0L)
+  expect_equal(names(CFG$software$input), c("zenodo", "zenodo_config"))
+})
+
+#debug
+test_that("debug",{
+  testthat::skip_on_cran()
+  cfg_file = system.file("extdata/workflows/config_metadata_zenodo_with_options.json", package = "geoflow")
+  DEBUG <- geoflow::debugWorkflow(cfg_file, entityIndex = 1, dir = ".")
+  expect_equal(names(DEBUG), c("config", "entity"))
+  expect_is(DEBUG$config, "list")
+  expect_is(DEBUG$entity, "geoflow_entity")
+  expect_equal(DEBUG$entity$identifiers$doi, "10.5281/zenodo.3378733")
+})
+
+#execute
+test_that("execute",{
+  #testthat::skip_on_cran()
+  #cfg_file = system.file("extdata/workflows/config_metadata_zenodo_with_options.json", package = "geoflow")
+  #execution
+  #EXEC <- geoflow::executeWorkflow(cfg_file, dir = ".")
+  #expect_true(dir.exists(EXEC))
+  #expect_true(file.exists(file.path(EXEC, "job.json")))
+  #expect_true(file.exists(file.path(EXEC, "job-logs.txt")))
+})