fix metadata

api/core/app/app_config/entities.py

@@ -140,12 +140,12 @@ SupportedComparisonOperator = Literal[
     # for string or array
     "contains",
     "not contains",
-    "starts with",
-    "ends with",
+    "start with",
+    "end with",
     "is",
     "is not",
     "empty",
-    "is not empty",
+    "not empty",
     # for number
     "=",
     "≠",

api/core/rag/retrieval/dataset_retrieval.py

@@ -796,7 +796,7 @@ class DatasetRetrieval:
             )
             if automatic_metadata_filters:
                 for filter in automatic_metadata_filters:
-                    self._process_metadata_filter_func(
+                    document_query = self._process_metadata_filter_func(
                         filter.get("condition"), filter.get("metadata_name"), filter.get("value"), document_query
                     )
         elif metadata_filtering_mode == "manual":
@@ -805,7 +805,7 @@ class DatasetRetrieval:
                 expected_value = condition.value
                 if isinstance(expected_value, str):
                     expected_value = self._replace_metadata_filter_value(expected_value, inputs)
-                self._process_metadata_filter_func(
+                document_query = self._process_metadata_filter_func(
                     condition.comparison_operator, metadata_name, expected_value, document_query
                 )
         else:
@@ -883,31 +883,38 @@ class DatasetRetrieval:
     def _process_metadata_filter_func(self, condition: str, metadata_name: str, value: str, query):
         match condition:
             case "contains":
-                query = query.filter(Document.doc_metadata[metadata_name].like(f"%{value}%"))
+                query = query.filter(DatasetDocument.doc_metadata[metadata_name].like(f'"%{value}%"'))
             case "not contains":
-                query = query.filter(Document.doc_metadata[metadata_name].notlike(f"%{value}%"))
+                query = query.filter(DatasetDocument.doc_metadata[metadata_name].notlike(f'"%{value}%"'))
             case "start with":
-                query = query.filter(Document.doc_metadata[metadata_name].like(f"{value}%"))
+                query = query.filter(DatasetDocument.doc_metadata[metadata_name].like(f'"{value}%"'))
             case "end with":
-                query = query.filter(Document.doc_metadata[metadata_name].like(f"%{value}"))
-            case "is", "=":
-                query = query.filter(Document.doc_metadata[metadata_name] == value)
-            case "is not", "≠":
-                query = query.filter(Document.doc_metadata[metadata_name] != value)
+                query = query.filter(DatasetDocument.doc_metadata[metadata_name].like(f'"%{value}"'))
+            case "is" | "=":
+                if isinstance(value, str):
+                    query = query.filter(DatasetDocument.doc_metadata[metadata_name] == f'"{value}"')
+                else:
+                    query = query.filter(DatasetDocument.doc_metadata[metadata_name] == value)
+            case "is not" | "≠":
+                if isinstance(value, str):
+                    query = query.filter(DatasetDocument.doc_metadata[metadata_name] != f'"{value}"')
+                else:
+                    query = query.filter(DatasetDocument.doc_metadata[metadata_name] != value)
             case "is empty":
-                query = query.filter(Document.doc_metadata[metadata_name].is_(None))
+                query = query.filter(DatasetDocument.doc_metadata[metadata_name].is_(None))
             case "is not empty":
-                query = query.filter(Document.doc_metadata[metadata_name].isnot(None))
-            case "before", "<":
-                query = query.filter(Document.doc_metadata[metadata_name] < value)
-            case "after", ">":
-                query = query.filter(Document.doc_metadata[metadata_name] > value)
-            case "≤", ">=":
-                query = query.filter(Document.doc_metadata[metadata_name] <= value)
-            case "≥", ">=":
-                query = query.filter(Document.doc_metadata[metadata_name] >= value)
+                query = query.filter(DatasetDocument.doc_metadata[metadata_name].isnot(None))
+            case "before" | "<":
+                query = query.filter(DatasetDocument.doc_metadata[metadata_name] < value)
+            case "after" | ">":
+                query = query.filter(DatasetDocument.doc_metadata[metadata_name] > value)
+            case "≤" | ">=":
+                query = query.filter(DatasetDocument.doc_metadata[metadata_name] <= value)
+            case "≥" | ">=":
+                query = query.filter(DatasetDocument.doc_metadata[metadata_name] >= value)
             case _:
                 pass
+        return query
 
     def _fetch_model_config(
         self, tenant_id: str, model: ModelConfig

api/core/workflow/nodes/knowledge_retrieval/entities.py

@@ -79,12 +79,12 @@ SupportedComparisonOperator = Literal[
     # for string or array
     "contains",
     "not contains",
-    "starts with",
-    "ends with",
+    "start with",
+    "end with",
     "is",
     "is not",
     "empty",
-    "is not empty",
+    "not empty",
     # for number
     "=",
     "≠",

api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py

@@ -296,7 +296,7 @@ class KnowledgeRetrievalNode(LLMNode):
             automatic_metadata_filters = self._automatic_metadata_filter_func(dataset_ids, query, node_data)
             if automatic_metadata_filters:
                 for filter in automatic_metadata_filters:
-                    self._process_metadata_filter_func(
+                    document_query = self._process_metadata_filter_func(
                         filter.get("condition"), filter.get("metadata_name"), filter.get("value"), document_query
                     )
         elif node_data.metadata_filtering_mode == "manual":
@@ -305,7 +305,7 @@ class KnowledgeRetrievalNode(LLMNode):
                 expected_value = condition.value
                 if isinstance(expected_value, str):
                     expected_value = self.graph_runtime_state.variable_pool.convert_template(expected_value).text
-                self._process_metadata_filter_func(
+                document_query = self._process_metadata_filter_func(
                     condition.comparison_operator, metadata_name, expected_value, document_query
                 )
         else:
@@ -383,17 +383,23 @@ class KnowledgeRetrievalNode(LLMNode):
     def _process_metadata_filter_func(self, condition: str, metadata_name: str, value: str, query):
         match condition:
             case "contains":
-                query = query.filter(Document.doc_metadata[metadata_name].like(f"%{value}%"))
+                query = query.filter(Document.doc_metadata[metadata_name].like(f'"%{value}%"'))
             case "not contains":
-                query = query.filter(Document.doc_metadata[metadata_name].notlike(f"%{value}%"))
+                query = query.filter(Document.doc_metadata[metadata_name].notlike(f'"%{value}%"'))
             case "start with":
-                query = query.filter(Document.doc_metadata[metadata_name].like(f"{value}%"))
+                query = query.filter(Document.doc_metadata[metadata_name].like(f'"{value}%"'))
             case "end with":
-                query = query.filter(Document.doc_metadata[metadata_name].like(f"%{value}"))
+                query = query.filter(Document.doc_metadata[metadata_name].like(f'"%{value}"'))
             case "=" | "is":
-                query = query.filter(Document.doc_metadata[metadata_name] == value)
+                if isinstance(value, str):
+                    query = query.filter(Document.doc_metadata[metadata_name] == f'"{value}"')
+                else:
+                    query = query.filter(Document.doc_metadata[metadata_name] == value)
             case "is not" | "≠":
-                query = query.filter(Document.doc_metadata[metadata_name] != value)
+                if isinstance(value, str):
+                    query = query.filter(Document.doc_metadata[metadata_name] != f'"{value}"')
+                else:
+                    query = query.filter(Document.doc_metadata[metadata_name] != value)
             case "is empty":
                 query = query.filter(Document.doc_metadata[metadata_name].is_(None))
             case "is not empty":
@@ -408,7 +414,7 @@ class KnowledgeRetrievalNode(LLMNode):
                 query = query.filter(Document.doc_metadata[metadata_name] >= value)
             case _:
                 pass
-
+        return query
     @classmethod
     def _extract_variable_selector_to_variable_mapping(
         cls,