Spaces:
Sleeping
Sleeping
Update server.py
Browse files
server.py
CHANGED
|
@@ -42,6 +42,7 @@ Settings.embed_model = embed_model
|
|
| 42 |
|
| 43 |
|
| 44 |
|
|
|
|
| 45 |
DOCUMENTS_BASE_PATH = "./"
|
| 46 |
RETRIEVERS_JSON_PATH = Path("./retrievers.json")
|
| 47 |
|
|
@@ -63,6 +64,7 @@ indices: Dict[str, VectorStoreIndex] = {}
|
|
| 63 |
for source, rel_path in SOURCES.items():
|
| 64 |
full_path = os.path.join(DOCUMENTS_BASE_PATH, rel_path)
|
| 65 |
if not os.path.exists(full_path):
|
|
|
|
| 66 |
continue
|
| 67 |
|
| 68 |
for root, dirs, files in os.walk(full_path):
|
|
@@ -77,6 +79,7 @@ for source, rel_path in SOURCES.items():
|
|
| 77 |
|
| 78 |
|
| 79 |
|
|
|
|
| 80 |
|
| 81 |
|
| 82 |
|
|
@@ -86,14 +89,21 @@ mcp = FastMCP("OnBase", port=port)
|
|
| 86 |
|
| 87 |
|
| 88 |
|
|
|
|
| 89 |
@mcp.resource(
|
| 90 |
uri="info://available_retrievers",
|
| 91 |
name="AvailableRetrievers",
|
| 92 |
-
description="Lista
|
| 93 |
mime_type="application/json"
|
| 94 |
)
|
| 95 |
def get_available_retrievers() -> dict:
|
| 96 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 97 |
available = []
|
| 98 |
|
| 99 |
for source, indexes in retrievers_metadata.items():
|
|
@@ -101,16 +111,17 @@ def get_available_retrievers() -> dict:
|
|
| 101 |
available.append({
|
| 102 |
"name": index_name,
|
| 103 |
"source": source,
|
|
|
|
| 104 |
"description": metadata.get("description", ""),
|
| 105 |
-
"
|
| 106 |
})
|
| 107 |
|
| 108 |
return {
|
| 109 |
"retrievers": available,
|
| 110 |
-
"sources": list(retrievers_metadata.keys()),
|
| 111 |
"count": len(available)
|
| 112 |
}
|
| 113 |
|
|
|
|
| 114 |
@mcp.tool()
|
| 115 |
def retrieve_docs(
|
| 116 |
query: str,
|
|
@@ -118,21 +129,17 @@ def retrieve_docs(
|
|
| 118 |
top_k: int = 3
|
| 119 |
) -> dict:
|
| 120 |
"""
|
| 121 |
-
Realiza búsqueda semántica en
|
| 122 |
|
| 123 |
Parámetros:
|
| 124 |
-
query (str
|
| 125 |
-
retrievers (List[str]
|
| 126 |
-
|
| 127 |
-
top_k (int, opcional): Número máximo de resultados por retriever. Default: 3
|
| 128 |
-
|
| 129 |
-
Retorna:
|
| 130 |
-
dict
|
| 131 |
|
| 132 |
-
Ejemplo
|
| 133 |
retrieve_docs(
|
| 134 |
-
query="
|
| 135 |
-
retrievers=["
|
| 136 |
top_k=2
|
| 137 |
)
|
| 138 |
"""
|
|
@@ -148,16 +155,15 @@ def retrieve_docs(
|
|
| 148 |
retriever = indices[name].as_retriever(similarity_top_k=top_k)
|
| 149 |
nodes = retriever.retrieve(query)
|
| 150 |
|
| 151 |
-
# Obtener metadatos
|
| 152 |
-
metadata =
|
| 153 |
-
|
| 154 |
-
|
| 155 |
-
|
| 156 |
-
|
| 157 |
-
source = src
|
| 158 |
-
break
|
| 159 |
|
| 160 |
results[name] = {
|
|
|
|
| 161 |
"documents": [
|
| 162 |
{
|
| 163 |
"content": node.get_content(),
|
|
@@ -167,7 +173,10 @@ def retrieve_docs(
|
|
| 167 |
for node in nodes
|
| 168 |
],
|
| 169 |
"description": metadata.get("description", ""),
|
| 170 |
-
"source":
|
|
|
|
|
|
|
|
|
|
| 171 |
}
|
| 172 |
except Exception as e:
|
| 173 |
results[name] = {"error": str(e)}
|
|
|
|
| 42 |
|
| 43 |
|
| 44 |
|
| 45 |
+
# Configuración de paths
|
| 46 |
DOCUMENTS_BASE_PATH = "./"
|
| 47 |
RETRIEVERS_JSON_PATH = Path("./retrievers.json")
|
| 48 |
|
|
|
|
| 64 |
for source, rel_path in SOURCES.items():
|
| 65 |
full_path = os.path.join(DOCUMENTS_BASE_PATH, rel_path)
|
| 66 |
if not os.path.exists(full_path):
|
| 67 |
+
print(f"Advertencia: No se encontró la ruta para {source}")
|
| 68 |
continue
|
| 69 |
|
| 70 |
for root, dirs, files in os.walk(full_path):
|
|
|
|
| 79 |
|
| 80 |
|
| 81 |
|
| 82 |
+
|
| 83 |
|
| 84 |
|
| 85 |
|
|
|
|
| 89 |
|
| 90 |
|
| 91 |
|
| 92 |
+
# Resource para listar retrievers
|
| 93 |
@mcp.resource(
|
| 94 |
uri="info://available_retrievers",
|
| 95 |
name="AvailableRetrievers",
|
| 96 |
+
description="Lista todos los retrievers disponibles con sus metadatos",
|
| 97 |
mime_type="application/json"
|
| 98 |
)
|
| 99 |
def get_available_retrievers() -> dict:
|
| 100 |
+
"""
|
| 101 |
+
Devuelve lista completa de retrievers con:
|
| 102 |
+
- Nombre del índice
|
| 103 |
+
- Título del documento
|
| 104 |
+
- Descripción
|
| 105 |
+
- Fecha de actualización
|
| 106 |
+
"""
|
| 107 |
available = []
|
| 108 |
|
| 109 |
for source, indexes in retrievers_metadata.items():
|
|
|
|
| 111 |
available.append({
|
| 112 |
"name": index_name,
|
| 113 |
"source": source,
|
| 114 |
+
"title": metadata.get("title", ""),
|
| 115 |
"description": metadata.get("description", ""),
|
| 116 |
+
"last_updated": metadata.get("last_updated", "")
|
| 117 |
})
|
| 118 |
|
| 119 |
return {
|
| 120 |
"retrievers": available,
|
|
|
|
| 121 |
"count": len(available)
|
| 122 |
}
|
| 123 |
|
| 124 |
+
# Función principal de búsqueda
|
| 125 |
@mcp.tool()
|
| 126 |
def retrieve_docs(
|
| 127 |
query: str,
|
|
|
|
| 129 |
top_k: int = 3
|
| 130 |
) -> dict:
|
| 131 |
"""
|
| 132 |
+
Realiza búsqueda semántica en documentos indexados.
|
| 133 |
|
| 134 |
Parámetros:
|
| 135 |
+
query (str): Texto de búsqueda (requerido)
|
| 136 |
+
retrievers (List[str]): Nombres de retrievers a consultar (requerido)
|
| 137 |
+
top_k (int): Número de resultados por retriever (opcional, default=3)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
|
| 139 |
+
Ejemplo:
|
| 140 |
retrieve_docs(
|
| 141 |
+
query="estándares farmacéuticos",
|
| 142 |
+
retrievers=["vec_who_1"],
|
| 143 |
top_k=2
|
| 144 |
)
|
| 145 |
"""
|
|
|
|
| 155 |
retriever = indices[name].as_retriever(similarity_top_k=top_k)
|
| 156 |
nodes = retriever.retrieve(query)
|
| 157 |
|
| 158 |
+
# Obtener metadatos
|
| 159 |
+
metadata = next(
|
| 160 |
+
(meta for src in retrievers_metadata.values()
|
| 161 |
+
for idx, meta in src.items() if idx == name),
|
| 162 |
+
{}
|
| 163 |
+
)
|
|
|
|
|
|
|
| 164 |
|
| 165 |
results[name] = {
|
| 166 |
+
"title": metadata.get("title", name),
|
| 167 |
"documents": [
|
| 168 |
{
|
| 169 |
"content": node.get_content(),
|
|
|
|
| 173 |
for node in nodes
|
| 174 |
],
|
| 175 |
"description": metadata.get("description", ""),
|
| 176 |
+
"source": next(
|
| 177 |
+
(s for s, idx in retrievers_metadata.items() if name in idx),
|
| 178 |
+
"unknown"
|
| 179 |
+
)
|
| 180 |
}
|
| 181 |
except Exception as e:
|
| 182 |
results[name] = {"error": str(e)}
|