Search refactoring to better utilize index and improve capabilities (#…

…906)

docs/user-guide/7-Lucene-search.rst

@@ -89,7 +89,7 @@ This syntax is still not very convenient, so we have introduced shorter syntax i
 .. code-block::
 
    size:>=500000
-   size:">=500000"
+   size:>="500000"
 
 MWDB-Core supports human-readable file size so instead of specifying the number of bytes, we can refer to larger units like kB, MB and GB.
 
@@ -141,7 +141,7 @@ If you want to search for objects uploaded after certain hour:
 
 .. code-block::
 
-   upload_time:">=2020-09-28 08:00"
+   upload_time:>="2020-09-28 08:00"
 
 If you want to search for objects uploaded at certain minute:
 

mwdb/core/search/__init__.py

@@ -1,7 +1,7 @@
-from .exceptions import SQLQueryBuilderBaseException
-from .search import SQLQueryBuilder
+from .exceptions import QueryBaseException
+from .search import build_query
 
 __all__ = [
-    "SQLQueryBuilderBaseException",
-    "SQLQueryBuilder",
+    "QueryBaseException",
+    "build_query",
 ]

mwdb/core/search/exceptions.py

@@ -1,29 +1,54 @@
-class SQLQueryBuilderBaseException(Exception):
+from typing import Optional, Tuple, Type
+
+from luqum.tree import Item
+
+
+class QueryBaseException(Exception):
     """
-    Base exception for SQLQueryBuilder
+    Base exception for search engine
     """
 
 
-class UnsupportedGrammarException(SQLQueryBuilderBaseException):
+class QueryParseException(QueryBaseException):
     """
-    Raised when SQLQueryBuilder does not support given Lucene grammar
+    Raised when Lucene parser is unable to parse a query
     """
 
 
-class FieldNotQueryableException(SQLQueryBuilderBaseException):
+class UnsupportedNodeException(QueryBaseException):
+    def __init__(self, message: str, node: Item) -> None:
+        super().__init__(f"{message} ({node.pos}:{node.pos + node.size - 1})")
+
+
+class UnsupportedNodeTypeException(UnsupportedNodeException):
     """
-    Raised when field does not exists, so it can't be queried, eg. file.unexistent_field
+    Raised when query visitor does not support given Lucene grammar
     """
 
+    def __init__(self, node: Item, expected: Optional[Tuple[Type, ...]] = None) -> None:
+        message = f"{node.__class__.__name__} is not supported here"
+        if expected:
+            message += f", expected {', '.join(typ.__name__ for typ in expected)}"
+        super().__init__(message, node)
+
+
+class UnsupportedPatternValue(UnsupportedNodeException):
+    def __init__(self, node: Item) -> None:
+        super().__init__("Pattern values are not supported here", node)
+
+
+class InvalidValueException(QueryBaseException):
+    def __init__(self, value: str, expected: str) -> None:
+        super().__init__(f"Invalid value format: {value}, expected {expected}")
+
 
-class MultipleObjectsQueryException(SQLQueryBuilderBaseException):
+class FieldNotQueryableException(QueryBaseException):
     """
-    Raised when multiple object types are queried,
-    e.g. file.file_name:something AND static.cfg:something2
+    Raised when field does not exist, so it can't be queried, eg. file.unexistent_field
     """
 
 
-class ObjectNotFoundException(SQLQueryBuilderBaseException):
+class ObjectNotFoundException(QueryBaseException):
     """
     Raised when object referenced in query condition can't be found
     """