Merge pull request #15 from TheJacksonLaboratory/task/update-publicat…

…ion-info-schema

Task/update publication info schema

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "geneweaver-core"
-version = "0.9.0a7"
+version = "0.9.0a8"
 description = "The core of the Jax-Geneweaver Python library"
 authors = ["Jax Computational Sciences <[email protected]>"]
 readme = "README.md"
@@ -17,12 +17,12 @@ pydantic = { extras = ["dotenv"], version = "^1.10" }
 openpyxl = "^3.1.2"
 numpy = ">=1.22,<2"
 pandas = ">=1.5,<3"
+requests = "^2.31.0"
 
 [tool.poetry.group.dev.dependencies]
 pylint = "^2.15.4"
 pytest = "^7.1.3"
 pydocstyle = "^6.1.1"
-requests = "^2.28.1"
 pytest-cov = "^4.0.0"
 geneweaver-testing = "^0.1.0"
 

src/geneweaver/core/publication/pubmed.py

@@ -43,6 +43,8 @@ def get_publication(pubmed_id: str) -> PublicationInfo:
 
     publication_fields = extract_fields(publication_xml)
 
+    publication_fields["pubmed_id"] = pubmed_id
+
     return PublicationInfo(**publication_fields)
 
 
@@ -173,6 +175,8 @@ def extract_authors(publication_xml: ElementTree.XML) -> dict:
         if not authors_are_complete(author_list_node):
             authors.append("et al.")
 
+        authors = ", ".join(authors)
+
         add_to_dict_if_not_none(publication_authors, "authors", authors)
 
     return publication_authors

src/geneweaver/core/schema/publication.py

@@ -8,15 +8,15 @@
 class PublicationInfo(BaseModel):
     """Publication upload schema (no ID)."""
 
+    pubmed_id: int
     authors: str
     title: str
-    abstract: str
+    abstract: str = ""
     journal: Optional[str] = None
     volume: Optional[str] = None
-    pages: str
-    month: str
-    year: int
-    pubmed_id: int
+    pages: Optional[str] = None
+    month: Optional[str] = None
+    year: Optional[int] = None
 
 
 class Publication(PublicationInfo):

tests/unit/publication/pubmed/test_extract_authors.py

@@ -16,13 +16,13 @@
         "<root><AuthorList CompleteYN='Y'><Author>"
         "<ForeName>John</ForeName><LastName>Doe</LastName>"
         "</Author></AuthorList></root>",
-        {"authors": ["John Doe"]},
+        {"authors": "John Doe"},
     ),
     (
         "<root><AuthorList CompleteYN='N'><Author>"
         "<ForeName>John</ForeName><LastName>Doe</LastName>"
         "</Author></AuthorList></root>",
-        {"authors": ["John Doe", "et al."]},
+        {"authors": "John Doe, et al."},
     ),
     ("<root></root>", {}),
     (
@@ -31,7 +31,7 @@
         "</Author><Author>"
         "<ForeName>Jane</ForeName><LastName>Doe</LastName>"
         "</Author></AuthorList></root>",
-        {"authors": ["John Doe", "Jane Doe"]},
+        {"authors": "John Doe, Jane Doe"},
     ),
 ]