Merge pull request #68 from mindsdb/alter-kb

Alter knowledge base syntax

Author: ea-rus

mindsdb_sql_parser/__about__.py

@@ -1,6 +1,6 @@
 __title__ = 'mindsdb_sql_parser'
 __package_name__ = 'mindsdb_sql_parser'
-__version__ = '0.11.4'
+__version__ = '0.12.0'
 __description__ = "Mindsdb SQL parser"
 __email__ = "[email protected]"
 __author__ = 'MindsDB Inc'

mindsdb_sql_parser/ast/mindsdb/__init__.py

@@ -17,8 +17,11 @@
 from .drop_job import DropJob
 from .chatbot import CreateChatBot, UpdateChatBot, DropChatBot
 from .trigger import CreateTrigger, DropTrigger
-from .knowledge_base import CreateKnowledgeBase, DropKnowledgeBase, CreateKnowledgeBaseIndex, DropKnowledgeBaseIndex \
-    , EvaluateKnowledgeBase
+from .knowledge_base import (
+    CreateKnowledgeBase, DropKnowledgeBase, CreateKnowledgeBaseIndex, DropKnowledgeBaseIndex, AlterKnowledgeBase,
+    EvaluateKnowledgeBase
+)
+
 from .skills import CreateSkill, DropSkill, UpdateSkill
 
 # remove it in next release

mindsdb_sql_parser/ast/mindsdb/knowledge_base.py

@@ -37,16 +37,15 @@ def __init__(
 
     def to_tree(self, *args, level=0, **kwargs):
         ind = indent(level)
-        storage_str = f"{ind} storage={self.storage.to_string()},\n" if self.storage else ""
-        model_str = f"{ind} model={self.model.to_string()},\n" if self.model else ""
-        out_str = f"""
-        {ind}CreateKnowledgeBase(
-        {ind}    if_not_exists={self.if_not_exists},
-        {ind}    name={self.name.to_string()},
-        {ind}    from_query={self.from_query.to_tree(level=level + 1) if self.from_query else None},
-        {model_str}{storage_str}{ind}    params={self.params}
-        {ind})
-        """
+        storage_str = f"{ind}  storage={self.storage.to_string()},\n" if self.storage else ""
+        model_str = f"{ind}  model={self.model.to_string()},\n" if self.model else ""
+        out_str = f"{ind}CreateKnowledgeBase(\n" \
+        f"{ind}  if_not_exists={self.if_not_exists},\n" \
+        f"{ind}  name={self.name.to_string()},\n" \
+        f"{ind}  from_query={self.from_query.to_tree(level=level + 1) if self.from_query else None},\n" \
+        f"{ind}{model_str}{storage_str}{ind}  params={self.params}\n" \
+        f"{ind})"
+
         return out_str
 
     def get_string(self, *args, **kwargs):
@@ -80,6 +79,55 @@ def __repr__(self) -> str:
         return self.to_tree()
 
 
+class AlterKnowledgeBase(ASTNode):
+    """
+    Update the knowledge base
+    """
+    def __init__(
+        self,
+        name,
+        params=None,
+        *args,
+        **kwargs,
+    ):
+        """
+        Args:
+            name: Identifier -- name of the knowledge base
+            params: dict -- additional parameters to pass to the knowledge base. E.g., chunking strategy, etc.
+        """
+        super().__init__(*args, **kwargs)
+        self.name = name
+        self.params = params
+    def to_tree(self, *args, level=0, **kwargs):
+        ind = indent(level)
+        out_str = f"{ind}AlterKnowledgeBase(\n" \
+        f"{ind}  name={self.name.to_string()},\n" \
+        f"{ind}  params={self.params}\n" \
+        f"{ind})"
+        return out_str
+
+    def get_string(self, *args, **kwargs):
+
+        using_ar = []
+        params = self.params.copy()
+        if params:
+            using_ar += [f"{k}={repr(v)}" for k, v in params.items()]
+        if using_ar:
+            using_str = "USING " + ", ".join(using_ar)
+        else:
+            using_str = ""
+
+        out_str = (
+            f"ALTER KNOWLEDGE_BASE {self.name.to_string()} "
+            f"{using_str}"
+        )
+
+        return out_str
+
+    def __repr__(self) -> str:
+        return self.to_tree()
+
+
 class DropKnowledgeBase(ASTNode):
     """
     Delete a knowledge base

mindsdb_sql_parser/parser.py

@@ -21,7 +21,7 @@
 from mindsdb_sql_parser.ast.mindsdb.latest import Latest
 from mindsdb_sql_parser.ast.mindsdb.evaluate import Evaluate
 from mindsdb_sql_parser.ast.mindsdb.knowledge_base import CreateKnowledgeBase, DropKnowledgeBase, \
-    CreateKnowledgeBaseIndex, DropKnowledgeBaseIndex, EvaluateKnowledgeBase
+   AlterKnowledgeBase, CreateKnowledgeBaseIndex, DropKnowledgeBaseIndex, EvaluateKnowledgeBase
 from mindsdb_sql_parser.ast.mindsdb.skills import CreateSkill, DropSkill, UpdateSkill
 from mindsdb_sql_parser.exceptions import ParsingException
 from mindsdb_sql_parser.ast.mindsdb.retrain_predictor import RetrainPredictor
@@ -95,6 +95,7 @@ class MindsDBParser(Parser):
        'create_trigger',
        'drop_trigger',
        'create_kb',
+       'alter_kb',
        'drop_kb',
        'evaluate_kb',
        'create_skill',
@@ -113,9 +114,6 @@ def query(self, p):
     @_(
         'CREATE KNOWLEDGE_BASE if_not_exists_or_empty identifier USING kw_parameter_list',
         'CREATE KNOWLEDGE_BASE if_not_exists_or_empty identifier',
-        # from select
-        'CREATE KNOWLEDGE_BASE if_not_exists_or_empty identifier FROM LPAREN select RPAREN USING kw_parameter_list',
-        'CREATE KNOWLEDGE_BASE if_not_exists_or_empty identifier FROM LPAREN select RPAREN',
     )
     def create_kb(self, p):
         params = getattr(p, 'kw_parameter_list', {})
@@ -145,6 +143,19 @@ def create_kb(self, p):
             if_not_exists=if_not_exists
         )
 
+    @_(
+        'ALTER KNOWLEDGE_BASE identifier USING kw_parameter_list',
+    )
+    def alter_kb(self, p):
+        params = getattr(p, 'kw_parameter_list', {})
+        name = p.identifier
+        params = {k.lower(): v for k, v in params.items()}  # case insensitive
+
+        return AlterKnowledgeBase(
+            name=name,
+            params=params
+        )
+
     @_('CREATE INDEX ON KNOWLEDGE_BASE identifier')
     def create_index(self, p):
         return CreateKnowledgeBaseIndex(name=p.identifier)

tests/test_mindsdb/test_knowledgebase.py

@@ -3,6 +3,7 @@
 from mindsdb_sql_parser.ast.mindsdb.knowledge_base import (
     CreateKnowledgeBase,
     DropKnowledgeBase,
+    AlterKnowledgeBase,
 )
 from mindsdb_sql_parser.ast import (
     Select,
@@ -18,6 +19,7 @@
 )
 from mindsdb_sql_parser.utils import to_single_line 
 
+
 class TestKB:
 
     def test_create_knowledge_base(self):
@@ -60,42 +62,6 @@ def test_create_knowledge_base(self):
         ast = parse_sql(sql)
         assert ast == expected_ast
 
-        # create from a query
-        sql = """
-            CREATE KNOWLEDGE_BASE my_knowledge_base
-            FROM (
-                SELECT id, content, embeddings, metadata
-                FROM my_table
-                JOIN my_embedding_model
-            )
-            USING
-                MODEL = mindsdb.my_embedding_model,
-                STORAGE = my_vector_database.some_table
-        """
-        ast = parse_sql(sql)
-        expected_ast = CreateKnowledgeBase(
-            name=Identifier("my_knowledge_base"),
-            if_not_exists=False,
-            model=Identifier(parts=["mindsdb", "my_embedding_model"]),
-            storage=Identifier(parts=["my_vector_database", "some_table"]),
-            from_select=Select(
-                targets=[
-                    Identifier("id"),
-                    Identifier("content"),
-                    Identifier("embeddings"),
-                    Identifier("metadata"),
-                ],
-                from_table=Join(
-                    left=Identifier("my_table"),
-                    right=Identifier("my_embedding_model"),
-                    join_type="JOIN",
-                ),
-            ),
-            params={},
-        )
-
-        assert ast == expected_ast
-
         # create without MODEL
         sql = """
             CREATE KNOWLEDGE_BASE my_knowledge_base
@@ -189,6 +155,62 @@ def test_create_knowledge_base(self):
         )
         assert ast == expected_ast
 
+    def disabled_test_create_from_select(self):
+        # create from a query
+        sql = """
+            CREATE KNOWLEDGE_BASE my_knowledge_base
+            FROM (
+                SELECT id, content, embeddings, metadata
+                FROM my_table
+                JOIN my_embedding_model
+            )
+            USING
+                MODEL = mindsdb.my_embedding_model,
+                STORAGE = my_vector_database.some_table
+        """
+        ast = parse_sql(sql)
+        expected_ast = CreateKnowledgeBase(
+            name=Identifier("my_knowledge_base"),
+            if_not_exists=False,
+            model=Identifier(parts=["mindsdb", "my_embedding_model"]),
+            storage=Identifier(parts=["my_vector_database", "some_table"]),
+            from_select=Select(
+                targets=[
+                    Identifier("id"),
+                    Identifier("content"),
+                    Identifier("embeddings"),
+                    Identifier("metadata"),
+                ],
+                from_table=Join(
+                    left=Identifier("my_table"),
+                    right=Identifier("my_embedding_model"),
+                    join_type="JOIN",
+                ),
+            ),
+            params={},
+        )
+
+        assert ast == expected_ast
+
+    def test_update_knowledge_base(self):
+        # create without select
+        sql = """
+            ALTER KNOWLEDGE_BASE my_kb
+            USING
+                reranking_model={'provider': 'openai'},
+                embedding_model={'api_key': '123'}
+        """
+        ast = parse_sql(sql)
+        expected_ast = AlterKnowledgeBase(
+            name=Identifier("my_kb"),
+            params={
+                'reranking_model': {'provider': 'openai'},
+                'embedding_model': {'api_key': '123'},
+            },
+        )
+        assert to_single_line(str(ast)) == to_single_line(str(expected_ast))
+        assert ast == expected_ast
+
     def test_drop_knowledge_base(self):
         # drop if exists
         sql = """