wos translator: year format

Author: Gerit Wagner

search_query/wos/translator.py

@@ -1,5 +1,7 @@
 #!/usr/bin/env python3
 """WOS query translator."""
+import re
+
 from search_query.constants import Fields
 from search_query.constants import Operators
 from search_query.query import Query
@@ -116,6 +118,24 @@ def _translate_search_fields(cls, query: Query) -> None:
         for child in query.children:
             cls._translate_search_fields(child)
 
+    @classmethod
+    def _format_year(cls, query: Query) -> None:
+        """Format year search fields to WOS syntax."""
+
+        if query.is_term() and query.search_field and query.search_field.value == "PY=":
+            if re.fullmatch(r"^\d{4}$", query.value):
+                pass
+
+            match = re.search(r"\d{4}", query.value)
+            if match:
+                query.value = match.group(0)
+                return
+
+            raise ValueError(f"Invalid year format: {query.value}")
+
+        for child in query.children:
+            cls._format_year(child)
+
     @classmethod
     def to_specific_syntax(cls, query: Query) -> Query:
         """Convert the query to a specific syntax."""
@@ -126,5 +146,6 @@ def to_specific_syntax(cls, query: Query) -> Query:
         cls.move_fields_to_operator(query)
         cls._remove_contradicting_search_fields(query)
         cls._remove_redundant_terms(query)
+        cls._format_year(query)
 
         return query

test/test_query_translation.py

@@ -212,3 +212,16 @@ def test_translation_pubmed_to_generic() -> None:
     expected = "AND[quantum[title], dot[title], spin[title]]"
 
     assert converted_query == expected
+
+
+def test_translation_pubmed_date_to_wos_year() -> None:
+    query_str = '("quantum"[ti]) AND (2022/05/01[dp])'
+    query = search_query.parser.parse(
+        query_str,
+        platform=PLATFORM.PUBMED.value,
+    )
+    translated_query = query.translate(PLATFORM.WOS.value)
+    converted_query = translated_query.to_string()
+
+    expected = 'TI="quantum" AND PY=2022'
+    assert converted_query == expected