ssc_codegen 0.7.5__tar.gz → 0.7.7__tar.gz

{ssc_codegen-0.7.5 → ssc_codegen-0.7.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ssc_codegen
-Version: 0.7.5
+Version: 0.7.7
 Summary: Python-dsl code converter to html parser for web scraping 
 Project-URL: Documentation, https://github.com/vypivshiy/selector_schema_codegen#readme
 Project-URL: Issues, https://github.com/vypivshiy/selector_schema_codegen/issues

{ssc_codegen-0.7.5 → ssc_codegen-0.7.7}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ssc_codegen"
-version = "0.7.5"
+version = "0.7.7"
 description = "Python-dsl code converter to html parser for web scraping "
 readme = "README.md"
 requires-python = ">=3.10"

{ssc_codegen-0.7.5 → ssc_codegen-0.7.7}/ssc_codegen/__init__.py

@@ -3,7 +3,7 @@ from .document import HTMLDocument, StringDocument, ArrayDocument, AssertDocumen
 from .json_struct import Json
 from .schema import ItemSchema, DictSchema, ListSchema, FlatListSchema
 
-VERSION = "0.7.5"
+VERSION = "0.7.7"
 
 class __MISSING(object):
     pass

{ssc_codegen-0.7.5 → ssc_codegen-0.7.7}/ssc_codegen/ast_build_utils.py

@@ -126,6 +126,10 @@ def extract_json_structs_from_module(module: ModuleType) -> list[Type[Json]]:
 def assert_ret_type_not_document(
     field: "BaseDocument", name: str, schema: Type[BaseSchema]
 ) -> None:
+    # always ret nil, skip
+    if name == "__PRE_VALIDATE__":
+        return
+
     if field.stack_last_ret == VariableType.DOCUMENT:
         msg = f"{schema.__name__}.{name} cannot return type {VariableType.DOCUMENT.name}"
         raise TypeError(msg)
@@ -212,16 +216,29 @@ def cast_ret_type_to_optional(ret_type: VariableType) -> VariableType:
     return ret_type
 
 
-def assert_field_document_variable_types(field: BaseDocument) -> None:
+def scan_field_stack_expr(field: BaseDocument) -> None:
     """validate correct type expressions pass.
 
-    raise TypeError if not passed
+    raise TypeError if not scan passed
     """
     var_cursor = VariableType.DOCUMENT
     for expr in field.stack:
         if var_cursor == expr.accept_type:
             var_cursor = expr.ret_type
             continue
+        # currently used in to_len() operation
+        elif expr.accept_type == VariableType.LIST_ANY:
+            if var_cursor in (
+                VariableType.LIST_STRING,
+                VariableType.LIST_DOCUMENT,
+                VariableType.LIST_FLOAT,
+                VariableType.LIST_INT,
+            ):
+                var_cursor = expr.ret_type
+                continue
+            msg = f".to_len() expected LIST_ANY type, got '{var_cursor.name}'"
+            raise TypeError(msg)
+
         # this type always first (naive, used in DEFAULT and RETURN expr)
         elif expr.accept_type == VariableType.ANY:
             var_cursor = VariableType.DOCUMENT
@@ -229,8 +246,14 @@ def assert_field_document_variable_types(field: BaseDocument) -> None:
         # end token operation, ignore
         elif expr.kind == TokenType.EXPR_DEFAULT_END:
             continue
+        # end operations
         elif var_cursor == VariableType.NESTED:
-            raise TypeError("sub_parser not allowed next instructions")
+            raise TypeError(".sub_parser() not allowed next instructions")
+        elif var_cursor == VariableType.JSON:
+            raise TypeError(".jsonify() not allowed next instructions")
+        elif var_cursor == VariableType.OPTIONAL_JSON:
+            raise TypeError(".jsonify() not allowed next instructions")
 
+        # not covered cases
         msg = f"'{expr.kind.name}' expected type '{expr.accept_type.name}', got '{var_cursor.name}'"
         raise TypeError(msg)

{ssc_codegen-0.7.5 → ssc_codegen-0.7.7}/ssc_codegen/ast_builder.py

@@ -14,7 +14,7 @@ from .ast_build_utils import (
     unwrap_default_expr,
     assert_split_doc_ret_type_is_list_document,
     cast_ret_type_to_optional,
-    assert_field_document_variable_types,
+    scan_field_stack_expr,
     extract_json_structs_from_module,
 )
 from .ast_ssc import (
@@ -60,15 +60,17 @@ def fill_variables_stack_expr(
             # set DefaultEnd ret_type as DefaultStartToken
             tmp_stack[-1].ret_type = ret_type
         elif isinstance(first_expr.value, str):  # type: ignore[attr-defined]
-            tmp_stack[-1].ret_type = VariableType.STRING
-            if ret_type != VariableType.STRING:
-                msg = f"wrong default type passed (should be a STRING or NULL, got {ret_type.name})"
-                raise TypeError(msg)
-        elif isinstance(first_expr.value, float):  # type: ignore[attr-defined]
             tmp_stack[-1].ret_type = VariableType.FLOAT
             if ret_type != VariableType.FLOAT:
                 msg = f"wrong default type passed (should be a FLOAT or NULL, got {ret_type.name})"
                 raise TypeError(msg)
+
+        elif isinstance(first_expr.value, bool):  # type: ignore[attr-defined]
+            tmp_stack[-1].ret_type = VariableType.BOOL
+            if ret_type != VariableType.BOOL:
+                msg = f"wrong default type passed (should be a BOOL or NULL, got {ret_type.name})"
+                raise TypeError(msg)
+
         elif isinstance(first_expr.value, int):  # type: ignore[attr-defined]
             tmp_stack[-1].ret_type = VariableType.INT
             if ret_type != VariableType.INT:
@@ -152,11 +154,11 @@ def build_ast_struct(
     start_parse_body: list[CallStructFunctionExpression] = []
     struct_parse_functions: list[BaseAstNode] = []
     for name, field in fields.items():
-        assert_field_document_variable_types(field)
+        scan_field_stack_expr(field)
 
         # not allowed empty stack exprs
         if len(field.stack) == 0:
-            msg = f"{schema.__name__}.{name} has not exists expressions"
+            msg = f"{schema.__name__}.{name} has empty expressions"
             raise SyntaxError(msg)
 
         if css_to_xpath:

{ssc_codegen-0.7.5 → ssc_codegen-0.7.7}/ssc_codegen/ast_ssc.py

@@ -427,12 +427,7 @@ class JoinExpression(BaseExpression):
 @dataclass(kw_only=True)
 class ArrayLengthExpression(BaseExpression):
     kind: ClassVar[TokenType] = TokenType.EXPR_LIST_LEN
-    accept_type: VariableType = (
-        VariableType.LIST_STRING
-        | VariableType.LIST_DOCUMENT
-        | VariableType.LIST_INT
-        | VariableType.LIST_FLOAT
-    )
+    accept_type: VariableType = VariableType.LIST_ANY
     ret_type: VariableType = VariableType.INT
 
 

{ssc_codegen-0.7.5 → ssc_codegen-0.7.7}/ssc_codegen/tokens.py

@@ -14,6 +14,10 @@ class VariableType(IntEnum):
     # mark accept any variable type
     # used in return and default expr
     ANY = auto()
+
+    # currently used for to_len() operation
+    LIST_ANY = auto()
+
     # nested functions
     NESTED = auto()
     INT = auto()