Propagating metadata to output works now.

Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>

4 files changed, 3 insertions, 34 deletions

diff --git a/bh20seqanalyzer/main.py b/bh20seqanalyzer/main.py
index 63ff067..193a268 100644
--- a/bh20seqanalyzer/main.py
+++ b/bh20seqanalyzer/main.py
@@ -29,7 +29,7 @@ def validate_upload(api, collection, validated_project,
     else:
         try:
             metadata_content = ruamel.yaml.round_trip_load(col.open("metadata.yaml"))
-            metadata_content["id"] = "keep:%s/metadata.yaml" % collection["portable_data_hash"]
+            metadata_content["id"] = "http://arvados.org/keep:%s/metadata.yaml" % collection["portable_data_hash"]
             add_lc_filename(metadata_content, metadata_content["id"])
             valid = qc_metadata(metadata_content) and valid
         except Exception as e:
@@ -146,7 +146,7 @@ def start_pangenome_analysis(api,
             "class": "File",
             "location": "keep:%s/metadata.yaml" % v["portable_data_hash"]
         })
-        inputobj["subjects"].append("keep:%s/sequence.fasta" % v["portable_data_hash"])
+        inputobj["subjects"].append("http://arvados.org/keep:%s/sequence.fasta" % v["portable_data_hash"])
     run_workflow(api, analysis_project, pangenome_workflow_uuid, "Pangenome analysis", inputobj)
 
 
diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
index a072bd7..8a22db1 100644
--- a/bh20sequploader/bh20seq-schema.yml
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -169,11 +169,6 @@ $graph:
     virus: virusSchema?
     technology: technologySchema
     submitter: submitterSchema
-    submission:
-      type: string
-      jsonldPredicate:
-        _id: "@id"
-        #_type: "@id"
     id:
       doc: The subject (eg the fasta/fastq file) that the metadata describes
       type: string?
diff --git a/bh20sequploader/qc_metadata.py b/bh20sequploader/qc_metadata.py
index 38edcaa..e477f21 100644
--- a/bh20sequploader/qc_metadata.py
+++ b/bh20sequploader/qc_metadata.py
@@ -5,21 +5,10 @@ import pkg_resources
 import logging
 import traceback
 
-class CustomFetcher(schema_salad.ref_resolver.DefaultFetcher):
-    def check_exists(sup, url):
-        if url.startswith("keep:"):
-            return True
-        else:
-            return super().check_exists(url)
-
-    def supported_schemes(self):  # type: () -> List[str]
-        return ["file", "http", "https", "mailto", "keep"]
-
-
 def qc_metadata(metadatafile):
     schema_resource = pkg_resources.resource_stream(__name__, "bh20seq-schema.yml")
     cache = {"https://raw.githubusercontent.com/arvados/bh20-seq-resource/master/bh20sequploader/bh20seq-schema.yml": schema_resource.read().decode("utf-8")}
-    (loader,
+    (document_loader,
      avsc_names,
      schema_metadata,
      metaschema_loader) = schema_salad.schema.load_schema("https://raw.githubusercontent.com/arvados/bh20-seq-resource/master/bh20sequploader/bh20seq-schema.yml", cache=cache)
@@ -28,19 +17,6 @@ def qc_metadata(metadatafile):
         print(avsc_names)
         return False
 
-    document_loader = schema_salad.ref_resolver.Loader(
-        loader.ctx,
-        schemagraph=loader.graph,
-        foreign_properties=loader.foreign_properties,
-        idx=loader.idx,
-        cache=loader.cache,
-        fetcher_constructor=CustomFetcher,
-        skip_schemas=loader.skip_schemas,
-        url_fields=loader.url_fields,
-        allow_attachments=loader.allow_attachments,
-        session=loader.session,
-        )
-
     try:
         doc, metadata = schema_salad.schema.load_and_validate(document_loader, avsc_names, metadatafile, True)
         return True
diff --git a/example/metadata.yaml b/example/metadata.yaml
index a2f6e57..c780921 100644
--- a/example/metadata.yaml
+++ b/example/metadata.yaml
@@ -1,5 +1,3 @@
-submission: publicSequenceResource
-
 host:
     host_id: XX1
     host_species: string