FDO-Json now delivers access info about PDFs

Author: timconrad

app/mardi_fdo_server.py

@@ -91,47 +91,60 @@ def to_fdo(qid: str, entity: Dict[str, Any]) -> Dict[str, Any]:
 
 
 def to_fdo_publication(qid: str, entity: Dict[str, Any]) -> Dict[str, Any]:
-
     fdo_id = f"{FDO_IRI}{qid}"
-    access_id = f"{FDO_ACCESS_IRI}{qid}"
+    access_json_id = f"{FDO_ACCESS_IRI}{qid}-jsonld"
+    access_pdf_id = f"{FDO_ACCESS_IRI}{qid}-pdf"
 
     created = entity.get("created")
     modified = entity.get("modified")
     created_or_modified = created if created else modified if modified else ""
 
+    profile, pdf_url = build_scholarly_article_profile(qid, entity)
+
+    access_list = [{"@id": access_json_id}]
+    if pdf_url:
+        access_list.append({"@id": access_pdf_id})
+
+    access_records = [
+        {
+            "@id": access_json_id,
+            "accessURL": f"{fdo_id}?format=jsonld",
+            "mediaType": "application/ld+json",
+        }
+    ]
+    if pdf_url:
+        access_records.append(
+            {
+                "@id": access_pdf_id,
+                "accessURL": pdf_url,
+                "mediaType": "application/pdf",
+            }
+        )
+
     return {
         "@context": [
             "https://w3id.org/fdo/context/v1",
             {
                 "schema": "https://schema.org/",
                 "prov": "http://www.w3.org/ns/prov#",
-                "fdo": "https://w3id.org/fdo/vocabulary/"
-            }
+                "fdo": "https://w3id.org/fdo/vocabulary/",
+            },
         ],
-
         "@id": fdo_id,
         "@type": "DigitalObject",
-
         "kernel": {
             "@id": fdo_id,
             "digitalObjectType": "https://types.mardi4nfdi.de/ScholarlyArticle/v1",
             "created": created_or_modified,
             "modified": modified or "",
-            "access": [{"@id": access_id}]
-        },
-
-        "profile": build_scholarly_article_profile(qid, entity),
-
-        "access": {
-            "@id": access_id,
-            "accessURL": f"{fdo_id}?format=jsonld",
-            "mediaType": "application/ld+json"
+            "access": access_list,
         },
-
+        "profile": profile,
+        "access": access_records,
         "provenance": {
             "prov:generatedAtTime": modified or "",
-            "prov:wasAttributedTo": "MaRDI Knowledge Graph"
-        }
+            "prov:wasAttributedTo": "MaRDI Knowledge Graph",
+        },
     }
 
 

fdo_schemas/publication.py

@@ -14,11 +14,14 @@
 
 
 def build_scholarly_article_profile(qid: str, entity: Dict[str, Any]) -> Dict[str, Any]:
+    claims = entity.get("claims", {})
+
+    arxiv_id = extract_string_claim(claims, "P21")
+    pdf_url = f"https://arxiv.org/pdf/{arxiv_id}.pdf" if arxiv_id else None
 
+    # build the profile as before
     label = entity.get("labels", {}).get("en", {}).get("value", qid)
     description = entity.get("descriptions", {}).get("en", {}).get("value", "")
-    claims = entity.get("claims", {})
-
     author_ids = extract_item_ids(claims, "P16")
     citation_ids = extract_item_ids(claims, "P223")
     container_ids = extract_item_ids(claims, "P1433")
@@ -27,7 +30,6 @@ def build_scholarly_article_profile(qid: str, entity: Dict[str, Any]) -> Dict[st
     license_ids = extract_item_ids(claims, "P275")
     language_ids = extract_item_ids(claims, "P407")
     keyword_ids = extract_item_ids(claims, "1450")
-
     publication_date = extract_time_claim(claims, "P28") or ""
     doi_value = extract_string_claim(claims, "P27") or ""
     page_range = extract_string_claim(claims, "P304")
@@ -40,8 +42,6 @@ def build_scholarly_article_profile(qid: str, entity: Dict[str, Any]) -> Dict[st
     profile = {
         "@context": "https://schema.org",
         "@type": "ScholarlyArticle",
-
-        # Schema.org identity stays in the Wikibase namespace
         "@id": f"{ENTITY_IRI}{qid}",
         "name": label,
         "headline": label,
@@ -66,7 +66,7 @@ def build_scholarly_article_profile(qid: str, entity: Dict[str, Any]) -> Dict[st
             "@type": "PropertyValue",
             "propertyID": "doi",
             "value": doi_value,
-            "url": f"https://doi.org/{doi_value}"
+            "url": f"https://doi.org/{doi_value}",
         }
         profile["sameAs"] = [f"https://doi.org/{doi_value}"]
 
@@ -86,4 +86,14 @@ def build_scholarly_article_profile(qid: str, entity: Dict[str, Any]) -> Dict[st
     if citation_ids:
         profile["citation"] = schema_refs_from_ids(citation_ids)
 
-    return profile
+    if pdf_url:
+        profile["encoding"] = [
+            {
+                "@type": "MediaObject",
+                "contentUrl": pdf_url,
+                "encodingFormat": "application/pdf"
+            }
+        ]
+
+    return profile, pdf_url
+