Parsers: Elsevier, OUP, Springer orcids parsing

* Adapted tests. * ref: cern-sis/issues-scoap3#345
cern-sis · Jul 17, 2024 · f01ecf1 · f01ecf1
1 parent 7bcdcf5
commit f01ecf1
Show file tree

Hide file tree

Showing 8 changed files with 4,762 additions and 203 deletions.
diff --git a/dags/elsevier/parser.py b/dags/elsevier/parser.py
@@ -122,6 +122,7 @@ def _get_authors(self, article):
     def _get_authors_details(self, author_group):
         authors = []
         for author in author_group.findall("author"):
+            orcid = author.get("orcid")
             surname = extract_text(
                 article=author, path="surname", field_name="surname", dois=self.dois
             )
@@ -146,6 +147,8 @@ def _get_authors_details(self, author_group):
                 auth_dict["affiliations"] = affiliations
             if emails:
                 auth_dict["email"] = emails
+            if orcid:
+                auth_dict["orcid"] = orcid
             authors.append(auth_dict)
 
         if not authors:

diff --git a/dags/oup/parser.py b/dags/oup/parser.py
@@ -163,6 +163,7 @@ def _get_authors(self, article):
         )
         authors = []
         for contribution in contributions:
+            orcid = get_text_value(contribution.find("contrib-id"))
             surname = get_text_value(contribution.find("name/surname"))
             given_names = get_text_value(contribution.find("name/given-names"))
             email = get_text_value(contribution.find("email"))
@@ -183,20 +184,20 @@ def _get_authors(self, article):
                 if country:
                     country = country.capitalize()
                     _aff["country"] = country
-
                 full_affiliation.append(_aff)
 
             if not all([surname, given_names, email]) and not full_affiliation:
                 pass
             else:
-                authors.append(
-                    {
-                        "surname": surname,
-                        "given_names": given_names,
-                        "email": email,
-                        "affiliations": full_affiliation,
-                    }
-                )
+                author = {
+                    "surname": surname,
+                    "given_names": given_names,
+                    "email": email,
+                    "affiliations": full_affiliation,
+                }
+                if orcid:
+                    author.update({"orcid": orcid})
+                authors.append(author)
         return authors
 
     def _get_date(self, article):

diff --git a/dags/springer/parser.py b/dags/springer/parser.py
@@ -9,7 +9,7 @@
     CustomExtractor,
     TextExtractor,
 )
-from common.utils import construct_license
+from common.utils import construct_license, clean_text
 from structlog import get_logger
 
 
@@ -211,7 +211,7 @@ def _get_affiliations(self, author_group, contrib):
                 affiliations.append(cleaned_aff)
 
         mapped_affiliations = [
-            {"value": aff, "organization": org, **({"country": country} if country else {})}
+            {"value": clean_text(aff), "organization": clean_text(org), **({"country": country} if country else {})}
             for aff, org, country, in affiliations
         ]