chore: initial sync

lib/entity_candidate.py

@@ -0,0 +1,34 @@
+"""EntityCandidate — candidato de entidad extraido por el LLM."""
+
+from dataclasses import dataclass, field
+
+
+@dataclass
+class EntityCandidate:
+    """Candidato de entidad extraido por el LLM.
+
+    Puede venir de un solo chunk o ser el resultado de mergear multiples
+    extracciones. `merged_from` rastrea los nombres originales para debugging.
+    """
+
+    name: str
+    name_normalized: str = ""
+    type_ref: str = ""
+    type_label: str = ""
+    attributes: dict = field(default_factory=dict)
+    confidence: float = 0.0
+    source_chunk_indices: list[int] = field(default_factory=list)
+    merged_from: list[str] = field(default_factory=list)
+
+    def to_dict(self) -> dict:
+        """Serializa el candidato a un diccionario."""
+        return {
+            "name": self.name,
+            "name_normalized": self.name_normalized,
+            "type_ref": self.type_ref,
+            "type_label": self.type_label,
+            "attributes": self.attributes,
+            "confidence": self.confidence,
+            "source_chunk_indices": self.source_chunk_indices,
+            "merged_from": self.merged_from,
+        }