feat!: generalize script by removing organizational metadata

Remove Phase class, organizational metadata blocks, and unused project fields. Update configuration
to use 'default_grants' and simplify PI usage to fallback corresponding author determination only.

BREAKING CHANGES: - Remove 'phase' and 'project' fields from configuration - Use 'default_grants'
instead of 'default_grant' - Generate only standard Dataverse citation metadata

doi2dataset.py

@@ -109,36 +109,6 @@ class FieldType(Enum):
     COMPOUND = "compound"
     VOCABULARY = "controlledVocabulary"
 
-@dataclass
-class Phase:
-    """
-    Represents a project phase with a defined time span.
-
-    Attributes:
-        name (str): The name of the project phase.
-        start (int): The start year of the project phase.
-        end (int): The end year of the project phase.
-    """
-
-    name: str
-    start: int
-    end: int
-
-    def check_year(self, year: int) -> bool:
-        """
-        Checks whether a given year falls within the project's phase boundaries.
-
-        Args:
-            year (int): The year to check.
-
-        Returns:
-            bool: True if the year is within the phase boundaries, otherwise False.
-        """
-
-        if self.start <= year <= self.end:
-            return True
-        return False
-
 @dataclass
 class BaseMetadataField[T]:
     """
@@ -301,7 +271,7 @@ class Institution:
                 "termName": self.display_name,
                 "@type": "https://schema.org/Organization"
             }
-            return PrimitiveMetadataField("authorAffiliation", False, self.ror, expanded_value)
+            return PrimitiveMetadataField("authorAffiliation", False, self.ror, expanded_value=expanded_value)
         else:
             return PrimitiveMetadataField("authorAffiliation", False, self.display_name)
 
@@ -316,14 +286,12 @@ class Person:
         orcid (str): ORCID identifier (optional).
         email (str): Email address (optional).
         affiliation (Institution): Affiliation of the person (optional).
-        project (list[str]): List of associated projects.
     """
     family_name: str
     given_name: str
     orcid: str = ""
     email: str = ""
     affiliation: Institution | str = ""
-    project: list[str] = field(default_factory=list)
 
     def to_dict(self) -> dict[str, str | list[str] | dict[str, str]]:
             """
@@ -340,8 +308,7 @@ class Person:
                 "family_name": self.family_name,
                 "given_name": self.given_name,
                 "orcid": self.orcid,
-                "email": self.email,
-                "project": self.project
+                "email": self.email
             }
 
             if isinstance(self.affiliation, Institution):
@@ -464,12 +431,10 @@ class ConfigData:
 
     Attributes:
         dataverse (dict[str, str]): Dataverse-related configuration.
-        phase (dict[str, dict[str, int]]): Mapping of project phases.
         pis (list[dict[str, Any]]): List of principal investigator configurations.
         default_grants (list[dict[str, str]]): Default grant configurations.
     """
     dataverse: dict[str, str]
-    phase: dict[str, dict[str, int]]
     pis: list[dict[str, Any]]
     default_grants: list[dict[str, str]]
 
@@ -523,7 +488,6 @@ class Config:
 
         cls._config_data = ConfigData(
             dataverse=config_data.get('dataverse', {}),
-            phase=config_data.get('phase', {}),
             pis=config_data.get('pis', []),
             default_grants=config_data.get('default_grants', [])
         )
@@ -545,16 +509,6 @@ class Config:
             raise RuntimeError("Failed to load configuration")
         return cls._config_data
 
-    @property
-    def PHASE(self) -> dict[str, dict[str, int]]:
-        """
-        Get phase configuration.
-
-        Returns:
-            dict[str, dict[str, int]]: Mapping of phases.
-        """
-        return self.get_config().phase
-
     @property
     def PIS(self) -> list[dict[str, Any]]:
         """
@@ -833,7 +787,10 @@ class AbstractProcessor:
             else:
                 console.print(f"\n{ICONS['warning']} No abstract found in CrossRef!", style="warning")
         else:
-            console.print(f"\n{ICONS['info']} License {license.name} does not allow derivative works. Reconstructing abstract from OpenAlex!", style="info")
+            if license.name:
+                console.print(f"\n{ICONS['info']} License {license.name} does not allow derivative works. Reconstructing abstract from OpenAlex!", style="info")
+            else:
+                console.print(f"\n{ICONS['info']} Custom license does not allow derivative works. Reconstructing abstract from OpenAlex!", style="info")
 
 
         openalex_abstract = self._get_openalex_abstract(data)
@@ -1406,8 +1363,7 @@ class MetadataProcessor:
                             CompoundMetadataField("grantNumber", True, grants).to_dict()
                         ],
                         "displayName": "Citation Metadata"
-                    },
-                    "crc1430_org_v1": self._build_organization_metadata(data)
+                    }
                 },
                 "files": []
             }
@@ -1473,71 +1429,22 @@ class MetadataProcessor:
         """
         return data.get("publication_year", "")
 
-    def _build_organization_metadata(self, data: dict[str, Any]) -> dict[str, Any]:
-        """
-        Build organization metadata fields (phase, project, PI names).
 
-        Args:
-            data (dict[str, Any]): The metadata.
-
-        Returns:
-            dict[str, Any]: Organization metadata.
-        """
-        publication_year = self._get_publication_year(data)
-        if publication_year:
-            phases = self._get_phases(int(publication_year))
-        else:
-            phases = []
-
-        pis = self._get_involved_pis(data)
-        projects: list[str] = []
-        for pi in pis:
-            for project in pi.project:
-                projects.append(project)
-
-        pi_names: list[str] = []
-        for pi in pis:
-            pi_names.append(pi.format_name())
-
-        # Deduplicate projects and PI names
-        unique_projects = list(set(projects))
-        unique_pi_names = list(set(pi_names))
-
-        return {
-            "fields": [
-                ControlledVocabularyMetadataField("crc1430OrgV1Phase", True, phases).to_dict(),
-                ControlledVocabularyMetadataField("crc1430OrgV1Project", True, unique_projects).to_dict(),
-                ControlledVocabularyMetadataField("crc1430OrgV1PI", True, unique_pi_names).to_dict()
-            ]
-        }
-
-    def _get_phases(self, year: int) -> list[str]:
-        """
-        Determine the project phases matching a given publication year.
-
-        Args:
-            year (int): The publication year.
-
-        Returns:
-            list[str]: List of matching phase names.
-        """
-        config = Config()
-        matching_phases: list[str] = []
-        for phase_name, phase_info in config.PHASE.items():
-            phase = Phase(phase_name, phase_info["start"], phase_info["end"])
-            if phase.check_year(year):
-                matching_phases.append(phase.name)
-        return matching_phases
 
     def _get_involved_pis(self, data: dict[str, Any]) -> list[Person]:
         """
-        Identify involved principal investigators from the metadata.
+        Identify involved principal investigators from the metadata for use as fallback
+        corresponding authors.
+
+        This method matches authors in the publication metadata against the configured
+        PIs and returns matching PIs. It is used as a fallback when no corresponding
+        authors are explicitly declared in the publication metadata.
 
         Args:
-            data (dict[str, Any]): The metadata.
+            data (dict[str, Any]): The metadata from OpenAlex.
 
         Returns:
-            list[Person]: List of PIs.
+            list[Person]: List of matching PIs for use as corresponding authors.
         """
         involved_pis: list[Person] = []
         for authorship in data.get("authorships", []):