chore: initial sync

lib/extraction_stats.py

@@ -0,0 +1,25 @@
+"""ExtractionStats — estadisticas del proceso de extraccion."""
+
+from dataclasses import dataclass, field
+
+
+@dataclass
+class ExtractionStats:
+    """Estadisticas del proceso de extraccion.
+
+    Util para reporting y debugging. Registra conteos antes y despues de
+    deduplicacion, tiempo de procesamiento y distribucion de tipos encontrados.
+    """
+
+    total_chunks: int = 0
+    total_chars: int = 0
+    raw_entities_count: int = 0
+    final_entities_count: int = 0
+    entities_merged: int = 0
+    raw_relations_count: int = 0
+    final_relations_count: int = 0
+    relations_merged: int = 0
+    relations_discarded: int = 0
+    entity_types_found: dict[str, int] = field(default_factory=dict)
+    relation_types_found: dict[str, int] = field(default_factory=dict)
+    processing_time_seconds: float = 0.0