Add options for external_data_endpoints and environment_variables

Signed-off-by: Ben Sherman <[email protected]>
nextflow-io · Oct 21, 2024 · 4594477 · 4594477
1 parent ce27ca3
commit 4594477
Show file tree

Hide file tree

Showing 2 changed files with 49 additions and 9 deletions.
diff --git a/BCO.md b/BCO.md
@@ -1,5 +1,7 @@
 # Additional BCO configuration
 
+*New in version 1.3.0*
+
 The `bco` format supports additional "pass-through" options for certain BCO fields. These fields cannot be inferred automatically from a pipeline or run, and so must be entered through the config. External systems can use these config options to inject fields automatically.
 
 The following config options are supported:
@@ -11,9 +13,13 @@ The following config options are supported:
 - `prov.formats.bco.usability_domain`
 - `prov.formats.bco.description_domain.keywords`
 - `prov.formats.bco.description_domain.xref`
+- `prov.formats.bco.execution_domain.external_data_endpoints`
+- `prov.formats.bco.execution_domain.environment_variables`
 
 These options correspond exactly to fields in the BCO JSON schema. Refer to the [BCO User Guide](https://docs.biocomputeobject.org/user_guide/) for more information about these fields.
 
+*NOTE: The `environment_variables` setting differs from the BCO standard in that it only specifies the variable names. Only the variables specified in this list will be populated in the BCO, if they are present in the execution environment.*
+
 Here is an example config based on the BCO User Guide:
 
 ```groovy
@@ -94,6 +100,23 @@ prov {
           ]
         ]
       }
+      execution_domain {
+        external_data_endpoints = [
+          [
+            "url": "protocol://domain:port/application/path",
+            "name": "generic name"
+          ],
+          [
+            "url": "ftp://data.example.com:21/",
+            "name": "access to ftp server"
+          ],
+          [
+            "url": "http://eutils.ncbi.nlm.nih.gov/entrez/eutils",
+            "name": "access to e-utils web service"
+          ]
+        ]
+        environment_variables = ["HOSTTYPE", "EDITOR"]
+      }
     }
   }
 }
@@ -116,6 +139,10 @@ prov {
         keywords = params.bco_description_domain_keywords
         xref = params.bco_description_domain_xref
       }
+      execution_domain {
+        external_data_endpoints = params.bco_execution_domain_external_data_endpoints
+        environment_variables = params.bco_execution_domain_environment_variables
+      }
     }
   }
 }
@@ -142,6 +169,12 @@ This way, the pass-through options can be provided as JSON in a [params file](ht
     ],
     "bco_description_domain_xref": [
         // ...
+    ],
+    "bco_execution_domain_external_data_endpoints": [
+        // ...
+    ],
+    "bco_execution_domain_environment_variables": [
+        // ...
     ]
 }
 ```
diff --git a/plugins/nf-prov/src/main/nextflow/prov/BcoRenderer.groovy b/plugins/nf-prov/src/main/nextflow/prov/BcoRenderer.groovy
@@ -23,6 +23,7 @@ import java.time.format.DateTimeFormatter
 import groovy.json.JsonOutput
 import groovy.transform.CompileStatic
 import nextflow.Session
+import nextflow.SysEnv
 import nextflow.processor.TaskRun
 import nextflow.script.WorkflowMetadata
 import nextflow.util.CacheHelper
@@ -68,13 +69,15 @@ class BcoRenderer implements Renderer {
         final params = session.config.params as Map
 
         final config = session.config
-        final review            = config.navigate('prov.formats.bco.provenance_domain.review', []) as List<Map<String,?>>
-        final derived_from      = config.navigate('prov.formats.bco.provenance_domain.derived_from') as String
-        final obsolete_after    = config.navigate('prov.formats.bco.provenance_domain.obsolete_after') as String
-        final embargo           = config.navigate('prov.formats.bco.provenance_domain.embargo') as Map<String,String>
-        final usability         = config.navigate('prov.formats.bco.usability_domain', []) as List<String>
-        final keywords          = config.navigate('prov.formats.bco.description_domain.keywords', []) as List<String>
-        final xref              = config.navigate('prov.formats.bco.description_domain.xref', []) as List<Map<String,?>>
+        final review                  = config.navigate('prov.formats.bco.provenance_domain.review', []) as List<Map<String,?>>
+        final derived_from            = config.navigate('prov.formats.bco.provenance_domain.derived_from') as String
+        final obsolete_after          = config.navigate('prov.formats.bco.provenance_domain.obsolete_after') as String
+        final embargo                 = config.navigate('prov.formats.bco.provenance_domain.embargo') as Map<String,String>
+        final usability               = config.navigate('prov.formats.bco.usability_domain', []) as List<String>
+        final keywords                = config.navigate('prov.formats.bco.description_domain.keywords', []) as List<String>
+        final xref                    = config.navigate('prov.formats.bco.description_domain.xref', []) as List<Map<String,?>>
+        final external_data_endpoints = config.navigate('prov.formats.bco.execution_domain.external_data_endpoints', []) as List<Map<String,String>>
+        final environment_variables   = config.navigate('prov.formats.bco.execution_domain.environment_variables', []) as List<String>
 
         // create BCO manifest
         final bco = [
@@ -125,8 +128,12 @@ class BcoRenderer implements Renderer {
                         ]
                     ]
                 ],
-                "external_data_endpoints": [],
-                "environment_variables": [:]
+                "external_data_endpoints": external_data_endpoints,
+                "environment_variables": environment_variables.inject([:]) { acc, name ->
+                    if( SysEnv.containsKey(name) )
+                        acc.put(name, SysEnv.get(name))
+                    acc
+                }
             ],
             "parametric_domain": params.toConfigObject().flatten().collect( (k, v) -> [
                 "param": k,