rara-tools 0.7.9__py3-none-any.whl → 0.7.11__py3-none-any.whl

rara_tools/normalizers/authorities.py

@@ -69,7 +69,6 @@ class AuthoritiesRecordNormalizer(RecordNormalizer):
             record, [Field(tag="046", indicators=EMPTY_INDICATORS, subfields=subfields_046)])
 
     def _add_viaf_url_or_isni(self, record: Record, viaf_record: VIAFRecord) -> None:
-        # TODO 024. will be used to store KRATT KATA ID. Just generate one?
         viaf_url = f"https://viaf.org/viaf/{viaf_record.viaf_id}"
 
         subfields = [Subfield("0", self.get_subfield(
@@ -109,7 +108,6 @@ class AuthoritiesRecordNormalizer(RecordNormalizer):
         100, 110, 111 - non-repeatable field, attempts to add author type, if missing.
 
         """
-        # TODO: include KRATT KATA ID to 024 and remove on delete. Increment last elastic ID?
         if not viaf_record:
             return
 
@@ -119,7 +117,9 @@ class AuthoritiesRecordNormalizer(RecordNormalizer):
         self._add_author(record, viaf_record)
 
     def _normalize_record(self, record: Record, sierraID: str,
-                          viaf_record: VIAFRecord, is_editing_existing_record: bool) -> Record:
+                          viaf_record: VIAFRecord, 
+                          is_editing_existing_record: bool,
+                          original_entity: str) -> Record:
         
         self._normalize_sierra(record, sierraID)
         self._normalize_viaf(record, viaf_record)

rara_tools/normalizers/base.py

@@ -147,21 +147,26 @@ class RecordNormalizer:
         return filter(lambda field: not self._field_in_record(field, record), fields)
     
     def _format_date(self, value: str) -> str:
-        if value is None:
+        
+        if not value:
             return ""
 
         if isinstance(value, (datetime, date)):
-            return value.strftime(YYMMDD_FORMAT)
+            return value.strftime("%Y%m%d")
+
+        val = str(value).strip()
 
         try:
-            dt = parser.parse(str(value), fuzzy=True)
-            formatted_date = dt.strftime(YYMMDD_FORMAT)
-            logger.info(f"Formatted date '{formatted_date}' from value '{value}'")
-            return formatted_date
-        except Exception as e:
-            logger.info(f"Failed to format date string '{value}': {e}")
+            dt = parser.parse(val, fuzzy=False, default=datetime(1, 1, 1))
+        except Exception:
             return ""
 
+        if len(val) == 4 and val.isdigit():
+            return dt.strftime("%Y")  # YYYY
+        if len(val) in (6, 7):  # YYYYMM or YYYY-MM
+            return dt.strftime("%Y%m")  # YYYYMM
+        return dt.strftime("%Y%m%d")  # YYYYMMDD
+
     def get_subfield(self, record: Record, tag: str, subfield: str, default: str) -> str:
         """ get record existing subfield value or assign a fallback value. """
 
@@ -355,7 +360,7 @@ class RecordNormalizer:
         return viaf_record
 
     def _normalize_record(self, record: Record, sierraID: str,
-                          viaf_record: VIAFRecord, is_editing_existing_record: bool) -> Record:
+                          viaf_record: VIAFRecord, is_editing_existing_record: bool, original_entity: str) -> Record:
         return record
 
     @property
@@ -369,7 +374,11 @@ class RecordNormalizer:
                 logger.error(f"Failed to normalize record: {e}")
                 continue
         return result
-
+    
+    @property
+    def first(self) -> Record:
+        return next(iter(self))
+    
     def __iter__(self) -> Iterator:
         viaf_id_path = "viaf.queryResult.records.record.0.recordData.VIAFCluster.viafID"
         sierra_id_path = "sierraID"
@@ -386,7 +395,7 @@ class RecordNormalizer:
             record = self._normalize_common(record, is_editing_existing_record)
 
             normalized_record = self._normalize_record(
-                record, sierra_id, viaf_record, is_editing_existing_record)
+                record, sierra_id, viaf_record, is_editing_existing_record, original_entity=entity)
 
             normalized_record.fields.sort(key=lambda field: field.tag)
 

rara_tools/normalizers/bibs.py

@@ -1,4 +1,5 @@
-from pymarc import (Field, Indicators, Subfield, Record)
+from pymarc import (Field, Subfield, Record)
+from typing import List, Optional
 
 from rara_tools.constants import EMPTY_INDICATORS
 from rara_tools.normalizers.viaf import VIAFRecord
@@ -34,10 +35,57 @@ class BibRecordNormalizer(RecordNormalizer):
         ]
 
         self._add_fields_to_record(record, fields)
+        
+    def _include_name_variations(self, record: Record, viaf_record: VIAFRecord) -> None:
+        """ Include name variations from VIAF record as 400|t fields """
+        
+        if not viaf_record or not viaf_record.name_variations:
+            return
+        
+        existing_name_variations = record.get_fields("400")
+        existing_variations = [sf.value for field in existing_name_variations for sf in field.get_subfields("t")]
+        
+        fields = []
+        
+        for variation in viaf_record.name_variations:
+            if variation not in existing_variations:
+                fields.append(
+                    Field(
+                        tag="400",
+                        indicators=EMPTY_INDICATORS,
+                        subfields=[
+                            Subfield("t", variation)
+                        ]
+                    )
+                )
+        
+        self._add_fields_to_record(record, fields)
+        
+    def _add_author(self, record: Record, viaf_record: Optional[VIAFRecord], original_entity: str) -> Optional[Field]:
+        if record.get("100") or record.get("110") or record.get("111"):
+            return record
 
-    def _normalize_viaf(self, record: Record, viaf_record: VIAFRecord) -> None:
+        type_map = {
+            "Personal": "100",
+            "Corporate": "110",
+            "Collective": "111"
+        }
+        
+        tag = type_map.get(getattr(viaf_record, "name_type", None), "100")
+        title = getattr(viaf_record, "name", None) or original_entity
 
+        fields = [Field(tag=tag, indicators=EMPTY_INDICATORS, subfields=[Subfield("t", title)])]
+        
+        self._add_fields_to_record(record, fields)
+        
+        if viaf_record:
+            self._include_name_variations(record, viaf_record)
+
+    def _normalize_viaf(self, record: Record, viaf_record: VIAFRecord, original_entity: str) -> None:
+        
         if not viaf_record:
+            # viaf record not found, include original entity as 100|t
+            self._add_author(record, viaf_record=None, original_entity=original_entity)
             return record
 
         viaf_id = viaf_record.viaf_id
@@ -52,12 +100,12 @@ class BibRecordNormalizer(RecordNormalizer):
         ]
 
         self._add_fields_to_record(record, fields)
-        self._add_author(record, viaf_record)
+        self._add_author(record, viaf_record, original_entity=original_entity)
 
     def _normalize_record(self, record: Record, sierraID: str,
-                          viaf_record: VIAFRecord, is_editing_existing_record: bool) -> Record:
+                          viaf_record: VIAFRecord, is_editing_existing_record: bool, original_entity: str) -> Record:
 
         self._normalize_sierra(record)
-        self._normalize_viaf(record, viaf_record)
+        self._normalize_viaf(record, viaf_record, original_entity=original_entity)
 
         return record

{rara_tools-0.7.9.dist-info → rara_tools-0.7.11.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rara-tools
-Version: 0.7.9
+Version: 0.7.11
 Summary: Tools to support Kata's work.
 Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.10

{rara_tools-0.7.9.dist-info → rara_tools-0.7.11.dist-info}/RECORD

@@ -20,9 +20,9 @@ rara_tools/core_formatters/formatted_keyword.py,sha256=hhi6wh4ErFionjBqYsEeKGbf1
 rara_tools/core_formatters/formatted_meta.py,sha256=r0RPG4eM-REPIR1DrIJnvYPQtQrzkgdvX9tvhNWjQ0Y,5250
 rara_tools/core_formatters/formatted_object.py,sha256=7a499ZmcZXOqtlwxDi6FWHWF5a6HdCsduS22wV3uHIE,5656
 rara_tools/normalizers/__init__.py,sha256=_NqpS5w710DhaURytHq9JpEt8HgYpSPfRDcOtOymJgE,193
-rara_tools/normalizers/authorities.py,sha256=wUnusRRcgIbAVoOrxiT-D0dAbIlpPph69i2DvCU7lCE,4815
-rara_tools/normalizers/base.py,sha256=W68goAtez5TgF-7SUou77eirkepatBUSL1Aghp8Gj84,15211
-rara_tools/normalizers/bibs.py,sha256=mxRSYq6adCwrCeJgWXJbLGKrG48Dz93vAt8Is2HzMak,2030
+rara_tools/normalizers/authorities.py,sha256=0R1YpZCVtwKj1fgzLvPYfw-R4v_hBocB9dHwMF2JR7c,4720
+rara_tools/normalizers/base.py,sha256=wiza01rldAhdH5rszE0z7ehE8EHSeKnNb-RmetFiZYE,15388
+rara_tools/normalizers/bibs.py,sha256=7H-spiqj6x8Xk4JQkBY5GK7q5pKjYUmXTRFf0RgzC-w,4040
 rara_tools/normalizers/reader.py,sha256=GYCkAtnsNx135w5lD-_MqCZzdHQHHPDF-pDxYj839Vo,1595
 rara_tools/normalizers/viaf.py,sha256=C-NfbvL83ZcHVB9ICMw43wAMYKTqDTHU3ZT2mXKec00,24288
 rara_tools/parsers/marc_parsers/base_parser.py,sha256=Kdw4aivJf2FkWgIK7pJtHtVXF_G1pjHVQ7IcFItSqy8,1649
@@ -39,8 +39,8 @@ rara_tools/parsers/marc_records/title_record.py,sha256=XrtJ4gj7wzSaGxNaPtPuawmqq
 rara_tools/parsers/tools/entity_normalizers.py,sha256=VyCy_NowCLpOsL0luQ55IW-Qi-J5oBH0Ofzr7HRFBhM,8949
 rara_tools/parsers/tools/marc_converter.py,sha256=LgSHe-7n7aiDrw2bnsB53r3fXTRFjZXTwBYfTpL0pfs,415
 rara_tools/parsers/tools/russian_transliterator.py,sha256=5ZU66iTqAhr7pmfVqXPAI_cidF43VqqmuN4d7H4_JuA,9770
-rara_tools-0.7.9.dist-info/licenses/LICENSE.md,sha256=hkZVnIZll7e_KNEQzeY94Y9tlzVL8iVZBTMBvDykksU,35142
-rara_tools-0.7.9.dist-info/METADATA,sha256=4AzmBZEA0nqC1i1ypTLMDgz8wSzQvsCqKPsKBNaGEi8,4079
-rara_tools-0.7.9.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-rara_tools-0.7.9.dist-info/top_level.txt,sha256=JwfB5b8BAtW5OFKRln2AQ_WElTRyIBM4nO0FKN1cupY,11
-rara_tools-0.7.9.dist-info/RECORD,,
+rara_tools-0.7.11.dist-info/licenses/LICENSE.md,sha256=hkZVnIZll7e_KNEQzeY94Y9tlzVL8iVZBTMBvDykksU,35142
+rara_tools-0.7.11.dist-info/METADATA,sha256=OvVJG6looJ5yhNhMJ-jUV74t1u6aymlzajNqMokUa54,4080
+rara_tools-0.7.11.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+rara_tools-0.7.11.dist-info/top_level.txt,sha256=JwfB5b8BAtW5OFKRln2AQ_WElTRyIBM4nO0FKN1cupY,11
+rara_tools-0.7.11.dist-info/RECORD,,