ssc_codegen 0.10.1__tar.gz → 0.11.0__tar.gz

{ssc_codegen-0.10.1 → ssc_codegen-0.11.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ssc_codegen
-Version: 0.10.1
+Version: 0.11.0
 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
@@ -247,7 +247,11 @@ You can use any html source:
 
 ## See also
 
-- [Brief](docs/briefing.md) about css selectors and regular expressions.
+
+- [Brief](docs/briefing.md) about css selectors and regular expressions
+- [Explain](docs/explain.md) short document on how to understand DSL syntax
+- [LLM](llm.md) experimental prompt for generate code
+- [Explain](docs/explain.md) short note how to explain and read sscgen schema configs
 - [Quickstart](docs/quickstart.md) about css selectors and regular expressions.
 - [Tutorial](docs/tutorial.md) basic usage ssc-gen
 - [AST reference](docs/ast_reference.md) about generation code from AST

{ssc_codegen-0.10.1 → ssc_codegen-0.11.0}/README.md

@@ -210,7 +210,11 @@ You can use any html source:
 
 ## See also
 
-- [Brief](docs/briefing.md) about css selectors and regular expressions.
+
+- [Brief](docs/briefing.md) about css selectors and regular expressions
+- [Explain](docs/explain.md) short document on how to understand DSL syntax
+- [LLM](llm.md) experimental prompt for generate code
+- [Explain](docs/explain.md) short note how to explain and read sscgen schema configs
 - [Quickstart](docs/quickstart.md) about css selectors and regular expressions.
 - [Tutorial](docs/tutorial.md) basic usage ssc-gen
 - [AST reference](docs/ast_reference.md) about generation code from AST

{ssc_codegen-0.10.1 → ssc_codegen-0.11.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ssc_codegen"
-version = "0.10.1"
+version = "0.11.0"
 description = "Python-dsl code converter to html parser for web scraping "
 readme = "README.md"
 requires-python = ">=3.10"

{ssc_codegen-0.10.1 → ssc_codegen-0.11.0}/ssc_codegen/__init__.py

@@ -7,7 +7,7 @@ from ssc_codegen.schema import ItemSchema, DictSchema, ListSchema, FlatListSchem
 
 setup_logger()
 
-VERSION = "0.10.1"
+VERSION = "0.11.0"
 
 
 class __MISSING(object):
@@ -28,10 +28,12 @@ class Document(
     JsonDocument,
     BooleanDocument
 ):
+    """Special Common Document or Element marker manipulations"""
     pass
 
 
 class Nested(HTMLDocument, NestedDocument, ArrayDocument, AssertDocument):
+    """Special Common Document or Element marker for provide Nested structure parsers"""
     pass
 
 
@@ -66,7 +68,7 @@ def F() -> DocumentFilter:
     return DocumentFilter()
 
 
-def CV(value, self_cls_path: str | None = None, *, returns: bool = False) -> ClassVarDocument:
+def CV(value: int | str | float | bool | None | list, self_cls_path: str | None = None, *, returns: bool = False) -> ClassVarDocument:
     """Shortcut as ClassVarDocument(value, self_cls, parse_returns) object
 
     pass `self_cls_path` argument if need use this var inner schema context:
@@ -88,5 +90,4 @@ class A(ItemShema):
 ```
 
     """
-    
     return ClassVarDocument(value, self_cls_path, parse_returns=returns)

{ssc_codegen-0.10.1 → ssc_codegen-0.11.0}/ssc_codegen/ast_/base.py

@@ -45,10 +45,16 @@ class BaseAstNode(Generic[T_MAPPING_FIELD, T_ARGUMENTS]):
         """extract all values from kwargs field wout keys"""
         return tuple(self.kwargs.values())  # type: ignore
 
-    # todo: cached property?
     @property
     def index(self) -> int:
         if self.parent:
+            # self.parent.body.index(self) add side effect if expr has same operations:
+            #      0      1      1       1
+            # D().raw().trim().trim().trim()
+            # use compare by id
+            for i, child in enumerate(self.parent.body):
+                if id(child) == id(self):
+                    return i
             return self.parent.body.index(self)
         return -1
 

ssc_codegen-0.11.0/ssc_codegen/ast_build/builder.py

@@ -0,0 +1,422 @@
+from types import ModuleType
+from typing import MutableSequence, Type, cast
+import warnings
+from typing_extensions import Self
+
+from ssc_codegen.ast_.base import BaseAstNode
+from ssc_codegen.ast_.nodes_cast import ExprJsonify, ExprNested
+from ssc_codegen.ast_.nodes_core import (
+    CodeEnd,
+    ExprCallStructClassVar,
+    ExprCallStructMethod,
+    ExprClassVar,
+    ExprDefaultValueEnd,
+    ExprDefaultValueStart,
+    ExprDefaultValueWrapper,
+    ExprNoReturn,
+    ExprReturn,
+    JsonStruct,
+    JsonStructField,
+    ModuleProgram,
+    Docstring,
+    ModuleImports,
+    CodeStart,
+    StartParseMethod,
+    StructFieldMethod,
+    StructInitMethod,
+    StructParser,
+    StructPartDocMethod,
+    StructPreValidateMethod,
+    TypeDef,
+    TypeDefField,
+)
+from ssc_codegen.ast_build.utils import (
+    generate_docstring_signature,
+    extract_schemas_from_module,
+    extract_json_structs_from_module,
+    is_literals_only_schema,
+)
+from ssc_codegen.document import BaseDocument, ClassVarDocument
+from ssc_codegen.document_utlis import (
+    convert_css_to_xpath,
+    convert_xpath_to_css,
+)
+from ssc_codegen.json_struct import Json
+from ssc_codegen.schema import BaseSchema
+from ssc_codegen.tokens import StructType, TokenType, VariableType
+
+
+class AstBuilder:
+    """USAGE:
+
+    builder = AstBuilder()
+    builder.add_header("test").add_json_types(*[]).add_struct_parsers(*[]).module
+    """
+
+    def __init__(
+        self,
+        gen_docstr: bool = True,
+        css_to_xpath: bool = False,
+        xpath_to_css: bool = False,
+    ):
+        self.module = ModuleProgram()
+        self.gen_docstr = gen_docstr
+        if css_to_xpath and xpath_to_css:
+            raise TypeError("allowed css_to_xpath=True or xpath_to_css=True")
+        self.css_to_xpath = css_to_xpath
+        self.xpath_to_css = xpath_to_css
+
+    @classmethod
+    def build_from_moduletype(
+        cls,
+        module: ModuleType,
+        gen_docstr: bool = True,
+        css_to_xpath: bool = False,
+        xpath_to_css: bool = False,
+    ) -> ModuleProgram:
+        builder = cls(
+            gen_docstr=gen_docstr,
+            css_to_xpath=css_to_xpath,
+            xpath_to_css=xpath_to_css,
+        )
+        structs = extract_schemas_from_module(module)
+        jsons = extract_json_structs_from_module(module)
+        return (
+            builder.add_header(module.__dict__.get("__doc__", None))
+            .add_json_types(*jsons)
+            .add_struct_parsers(*structs)
+            .module
+        )
+
+    @classmethod
+    def build_from_ssc_schemas(
+        cls,
+        *schemas: Type[BaseSchema],  # TODO: it not support json serialized
+        gen_docstr: bool = False,
+        css_to_xpath: bool = False,
+        xpath_to_css: bool = False,
+    ) -> ModuleProgram:
+        builder = cls(
+            gen_docstr=gen_docstr,
+            css_to_xpath=css_to_xpath,
+            xpath_to_css=xpath_to_css,
+        )
+        return (
+            builder.add_header(module_doc=None)
+            .add_struct_parsers(*schemas)
+            .module
+        )
+
+    def add_header(self, module_doc: str | None = None) -> Self:
+        if self.gen_docstr and module_doc:
+            self.module.body.append(
+                Docstring(kwargs={"value": module_doc}, parent=self.module)
+            )
+        # import node
+        self.module.body.append(ModuleImports(parent=self.module))
+        # CodeStart hook node
+        self.module.body.append(CodeStart(parent=self.module))
+        return self
+
+    def _json_field_nodes(
+        self, jsn_obj: Type[Json], ast_jsn: JsonStruct
+    ) -> None:
+        for name, field_type in jsn_obj.get_fields().items():
+            ast_jsn.body.append(
+                JsonStructField(
+                    kwargs={"name": name, "type": field_type}, parent=ast_jsn
+                )
+            )
+
+    def add_json_types(self, *json_structs: Type[Json]) -> Self:
+        ast_jsons = []
+        for jsn_st in json_structs:
+            ast_jsn = JsonStruct(
+                kwargs={
+                    "name": jsn_st.__name__,
+                    "is_array": jsn_st.__IS_ARRAY__,
+                },
+                parent=self.module,
+            )
+            self._json_field_nodes(jsn_st, ast_jsn)
+            ast_jsons.append(ast_jsn)
+        self.module.body.extend(ast_jsons)
+        return self
+
+    def _ref_body_parent(self, node: BaseAstNode) -> None:
+        parent = node
+        for child in node.body:
+            child.parent = parent
+
+    def _literal_struct_node(self, schema: Type[BaseSchema]) -> StructParser:
+        # dont required TypeDef struct generate
+        literals = schema.__get_mro_literals__().copy()
+        st = StructParser(
+            kwargs={
+                "name": schema.__name__,
+                "struct_type": StructType.CONFIG_CLASSVARS,
+                "docstring": schema.__doc__ or "" if self.gen_docstr else "",
+            },
+            body=[f.expr() for f in literals.values()],
+            parent=self.module,
+        )
+        self._ref_body_parent(st)
+        return st
+
+    def _ref_filter_expr(
+        self, parent: BaseAstNode, parent_body: MutableSequence[BaseAstNode]
+    ) -> None:
+        for i in parent_body:
+            i.parent = parent
+            if i.body:
+                self._ref_filter_expr(i, i.body)
+
+    def _ref_method_exprs(self, node_method: BaseAstNode) -> None:
+        parent = node_method
+        for expr in node_method.body:
+            expr.parent = parent
+            if expr.kind == TokenType.EXPR_FILTER:
+                self._ref_filter_expr(expr, expr.body)
+
+    def _unwrap_default_node(
+        self,
+        stack: list[BaseAstNode],  # noqa (use sideeffect for path stack exprs)
+        ret_type: VariableType,
+    ) -> None:
+        if stack[0].kind == ExprDefaultValueWrapper.kind:
+            default_expr = stack.pop(0)
+            value = default_expr.kwargs["value"]
+            classvar_hooks = default_expr.classvar_hooks
+            default_type = ret_type
+            if value is None:
+                match ret_type:
+                    case VariableType.STRING:
+                        default_type = VariableType.OPTIONAL_STRING
+                    case VariableType.INT:
+                        default_type = VariableType.OPTIONAL_INT
+                    case VariableType.FLOAT:
+                        default_type = VariableType.OPTIONAL_FLOAT
+                    case VariableType.LIST_STRING:
+                        default_type = VariableType.OPTIONAL_LIST_STRING
+                    case VariableType.LIST_INT:
+                        default_type = VariableType.OPTIONAL_LIST_INT
+                    case VariableType.LIST_FLOAT:
+                        default_type = VariableType.OPTIONAL_LIST_FLOAT
+                    # TODO: warning for BOOL type
+                    case _:
+                        warnings.warn(
+                            f"'None' default value not allowed return type '{ret_type.name}'. ",
+                            category=SyntaxWarning,
+                        )
+                        default_type = VariableType.ANY
+            elif isinstance(value, list):
+                # todo: check if empty list passed
+                match ret_type:
+                    case VariableType.LIST_STRING:
+                        default_type = VariableType.LIST_STRING
+                    case VariableType.LIST_INT:
+                        default_type = VariableType.LIST_INT
+                    case VariableType.LIST_FLOAT:
+                        default_type = VariableType.LIST_FLOAT
+                    case _:
+                        warnings.warn(
+                            f"`empty list` default value not allowed return type `{ret_type.name}`. "
+                            f"Expected types `{(VariableType.LIST_STRING.name, VariableType.LIST_INT, VariableType.LIST_FLOAT)}`",
+                            category=SyntaxWarning,
+                        )
+                        default_type = VariableType.ANY
+            expr_default_start = ExprDefaultValueStart(
+                kwargs={"value": value}, classvar_hooks=classvar_hooks
+            )
+            expr_default_end = ExprDefaultValueEnd(
+                kwargs={"value": value},
+                ret_type=default_type,
+                classvar_hooks=classvar_hooks,
+            )
+            stack.insert(0, expr_default_start)
+            stack.append(expr_default_end)
+
+    def _struct_pop_pre_validate(
+        self,
+        node_start_parse: StartParseMethod,
+        fields: dict[str, BaseDocument],
+    ) -> StructPreValidateMethod:
+        # fmt: off
+        assert "__PRE_VALIDATE__" in fields, "check `__PRE_VALIDATE__` key before extract"
+
+        validate_field = fields.pop("__PRE_VALIDATE__")
+        method = StructPreValidateMethod()
+        exprs = validate_field.stack.copy()
+        # not allowed use default exprs, skip check
+        exprs.append(ExprNoReturn())
+        method.body.extend(exprs)
+        self._ref_method_exprs(method)
+        node_start_parse.body.append(ExprCallStructMethod(kwargs={"type": VariableType.NULL, "name": "__PRE_VALIDATE__",}))
+        return method
+        # fmt: on
+
+    def _struct_pop_split_doc(
+        self,
+        node_start_parse: StartParseMethod,
+        fields: dict[str, BaseDocument],
+    ) -> StructPartDocMethod:
+        # fmt: off
+        assert "__SPLIT_DOC__" in fields, "check `__SPLIT_DOC__` key before extract"
+
+        split_field = fields.pop("__SPLIT_DOC__")
+        method = StructPartDocMethod()
+        exprs = split_field.stack.copy()
+        # not allowed use default exprs, skip check
+        exprs.append(ExprReturn(ret_type=VariableType.LIST_DOCUMENT))
+        method.body.extend(exprs)
+        self._ref_method_exprs(method)
+        node_start_parse.body.append(ExprCallStructMethod(kwargs={"type": VariableType.LIST_DOCUMENT, "name": "__SPLIT_DOC__",}))
+        return method
+        # fmt: on
+
+    def _struct_call_literals(
+        self,
+        node_start_parse: StartParseMethod,
+        literals: dict[str, ClassVarDocument],
+    ) -> None:
+        # fmt: off
+        for literal in literals.values():
+            expr = literal.expr()
+            if expr.kwargs["parse_returns"]:
+                _, struct_name, field_name, _, _ = expr.unpack_args()
+                node_start_parse.body.append(
+                    ExprCallStructClassVar(
+                        kwargs={"struct_name": struct_name, "field_name": field_name, "type": expr.ret_type,}))
+        # fmt: on
+
+    def _struct_fields_parse(
+        self,
+        node_start_parse: StartParseMethod,
+        fields: dict[str, BaseDocument],
+    ) -> list[StructFieldMethod]:
+        st_fields = []
+        # fmt: off
+        for field_name, document in fields.items():
+            if self.css_to_xpath:
+                document = convert_css_to_xpath(document)
+            elif self.xpath_to_css:
+                document = convert_xpath_to_css(document)
+            
+            ret_type = document.stack_last_ret
+            method = StructFieldMethod(kwargs={"name": field_name}, ret_type=ret_type,)
+            exprs = document.stack.copy()
+            # in inheritance schemas, child classes use same fields are used as in the parent class
+            # avoid duplicate ExprReturn or ExprDefaultValueWrapper node
+            if (exprs[-1].kind != ExprReturn.kind and exprs[-1].kind != ExprDefaultValueEnd.kind):
+                exprs.append(ExprReturn(accept_type=ret_type, ret_type=ret_type))
+            self._unwrap_default_node(exprs, exprs[-1].ret_type)
+            method.body.extend(exprs)
+            self._ref_method_exprs(method)
+            st_fields.append(method)
+            node_start_parse.body.append(
+                ExprCallStructMethod(kwargs={"type": exprs[-1].ret_type, "name": field_name,}))
+        return st_fields
+        # fmt: on
+
+    def _node_typedef(self, struct: StructParser) -> TypeDef | None:
+        # fmt: off
+        if struct.struct_type == StructType.CONFIG_CLASSVARS:
+            return None
+        typedef = TypeDef(
+            parent=self.module,
+            kwargs={"name": struct.kwargs["name"], "struct_type": struct.struct_type},
+        )
+        for sc_field in struct.body:
+            if sc_field.kind not in (TokenType.STRUCT_FIELD, TokenType.CLASSVAR,):
+                continue
+            elif sc_field.kind == ExprClassVar.kind and sc_field.kwargs["parse_returns"]:
+                _value, _struct_name, field_name, _parse_returns, _is_regex = sc_field.unpack_args()
+                tdef_field = TypeDefField(
+                    parent=typedef,
+                    # cls_nested: always self-used field, set None
+                    kwargs={"name": field_name, "type": sc_field.ret_type, "cls_nested": None, "cls_nested_type": None,},
+                )
+                typedef.body.append(tdef_field)
+            elif sc_field.kind == StructFieldMethod.kind:
+                if sc_field.body[-1].ret_type == VariableType.NESTED:
+                    node = [i for i in sc_field.body if i.kind == TokenType.EXPR_NESTED][0]
+                    node = cast(ExprNested, node)
+                    cls_name = node.kwargs["schema_name"]
+                    cls_nested_type = node.kwargs["schema_type"]
+                elif sc_field.body[-1].ret_type == VariableType.JSON:
+                    node = [i for i in sc_field.body if i.kind == TokenType.TO_JSON][0]
+                    node = cast(ExprJsonify, node)
+                    cls_name = node.kwargs["json_struct_name"]
+                    # hack: for provide more consistent Typedef field gen api reuse StructType enum 
+                    # or perceive it as bad AST design
+                    cls_nested_type = StructType.LIST if node.kwargs["is_array"] else StructType.ITEM 
+                else:
+                    cls_name, cls_nested_type = None, None
+                tdef_field = TypeDefField(
+                    parent=typedef,
+                    kwargs={"name": sc_field.kwargs["name"], "type": sc_field.body[-1].ret_type, 
+                            "cls_nested": cls_name, "cls_nested_type": cls_nested_type,}
+                    )
+                typedef.body.append(tdef_field)
+        return typedef
+        # fmt: on
+
+    def _parser_struct_node(self, schema: Type[BaseSchema]) -> StructParser:
+        # fmt: off
+        docstring = ((schema.__doc__ or "") + "\n\n" + generate_docstring_signature(schema))
+        st = StructParser(
+            kwargs={"name": schema.__name__, "struct_type": schema.__SCHEMA_TYPE__, "docstring": docstring if self.gen_docstr else "",},
+            parent=self.module,
+            )
+        literals = schema.__get_mro_literals__().copy()
+
+        for literal in literals.values():
+            st.body.append(literal.expr())
+        st.body.append(StructInitMethod())
+
+        fields = schema.__get_mro_fields__().copy()
+        node_start_parse = StartParseMethod(parent=st)  # insert to END
+        if "__PRE_VALIDATE__" in fields:
+            pre_validate = self._struct_pop_pre_validate(node_start_parse, fields)
+            st.body.append(pre_validate)
+
+        if fields.get("__SPLIT_DOC__"):
+            split_doc = self._struct_pop_split_doc(node_start_parse, fields)
+            st.body.append(split_doc)
+        # literals call (if required return)
+        self._struct_call_literals(node_start_parse, literals)
+
+        fields_st = self._struct_fields_parse(node_start_parse, fields)
+        # assembly struct node
+        st.body.extend(fields_st)
+        st.body.append(node_start_parse)
+        self._ref_body_parent(st)
+        # generate typedef as annotation node
+        return st
+
+    def add_struct_parsers(self, *struct_parsers: Type[BaseSchema]) -> Self:
+        ssc_structs = []
+        ssc_typedefs = []
+        for schema in struct_parsers:
+            if is_literals_only_schema(schema):
+                if schema.__SCHEMA_TYPE__ != StructType.ITEM:
+                    raise TypeError(
+                        "Literals-only init allowed only ItemSchema"
+                    )
+                struct = self._literal_struct_node(schema)
+                ssc_structs.append(struct)
+                continue
+            else:
+                struct = self._parser_struct_node(schema)
+                ssc_structs.append(struct)
+        for struct in ssc_structs:
+            typedef = self._node_typedef(struct)
+            # literals-only schema skip
+            if not typedef:
+                continue
+            ssc_typedefs.append(typedef)
+        # assembly
+        self.module.body.extend(ssc_typedefs)
+        self.module.body.extend(ssc_structs)
+        self.module.body.append(CodeEnd(parent=self.module))
+        return self

ssc_codegen-0.11.0/ssc_codegen/ast_build/main.py

@@ -0,0 +1,91 @@
+from pathlib import Path
+from typing import Type
+
+from ssc_codegen.ast_ import ModuleProgram
+
+from ssc_codegen.ast_build.builder import AstBuilder
+from ssc_codegen.ast_build.utils import (
+    exec_module_code,
+    extract_schemas_from_module,
+)
+
+from ssc_codegen.schema import BaseSchema
+from ssc_codegen.static_checker import run_analyze_schema
+import logging
+
+LOGGER = logging.getLogger("ssc-gen")
+
+
+def build_ast_module_parser(
+    path: str | Path,
+    *,
+    css_to_xpath: bool = False,
+    xpath_to_css: bool = False,
+    gen_docstring: bool = True,
+) -> ModuleProgram:
+    """build ast from python sscgen config file
+
+    WARNING!!!
+        DO NOT PASS MODULES FROM UNKNOWN SOURCE/INPUT FOR SECURITY REASONS.
+
+        THIS FUNCTION COMPILE AND EXEC PYTHON CODE in runtime WOUT CHECKS
+    """
+    if css_to_xpath and xpath_to_css:
+        raise AttributeError(
+            "Should be chosen one variant (css_to_xpath OR xpath_to_css)"
+        )
+    py_module = exec_module_code(path)
+    # check code configs before build AST
+    schemas = extract_schemas_from_module(py_module)
+    count_errors = 0
+    for schema in schemas:
+        errors = run_analyze_schema(schema)
+        if errors > 0:
+            msg = f"{schema.__name__} founded errors: {errors}"
+            LOGGER.error(msg)
+            count_errors += errors
+
+    if count_errors > 0:
+        msg = f"{path} errors count: {count_errors}"
+        LOGGER.error(msg)
+        raise SyntaxError("")
+
+    ast_module = AstBuilder.build_from_moduletype(
+        py_module,
+        css_to_xpath=css_to_xpath,
+        xpath_to_css=xpath_to_css,
+        gen_docstr=gen_docstring,
+    )
+    return ast_module
+    # static check
+
+
+def build_ast_schemas(
+    *schemas: Type[BaseSchema],
+    css_to_xpath: bool = False,
+    xpath_to_css: bool = False,
+    gen_docstring: bool = False,
+) -> ModuleProgram:
+    if css_to_xpath and xpath_to_css:
+        raise AttributeError(
+            "Should be chosen one variant (css_to_xpath OR xpath_to_css)"
+        )
+    count_errors = 0
+    for schema in schemas:
+        errors = run_analyze_schema(schema)
+        if errors > 0:
+            msg = f"{schema.__name__} founded errors: {errors}"
+            LOGGER.error(msg)
+            count_errors += errors
+
+    if count_errors > 0:
+        msg = f"scehmas errors count: {count_errors}"
+        LOGGER.error(msg)
+        raise SyntaxError
+    ast_module = AstBuilder.build_from_ssc_schemas(
+        *schemas,
+        css_to_xpath=css_to_xpath,
+        xpath_to_css=xpath_to_css,
+        gen_docstr=gen_docstring,
+    )
+    return ast_module

{ssc_codegen-0.10.1 → ssc_codegen-0.11.0}/ssc_codegen/compiler.py

@@ -5,12 +5,8 @@ import types
 from os import PathLike
 from typing import Any, Type
 
-from ssc_codegen.ast_.nodes_core import ModuleImports, ModuleProgram
-from ssc_codegen.ast_build import build_ast_module_parser
-from ssc_codegen.ast_build.main import (
-    build_ast_struct_parser,
-    build_ast_typedef,
-)
+from ssc_codegen.ast_build.builder import AstBuilder
+from ssc_codegen.ast_build.main import build_ast_module_parser
 from ssc_codegen.converters.py_base import BasePyCodeConverter
 from ssc_codegen.schema import BaseSchema
 
@@ -46,22 +42,8 @@ def compile(
         print(module.Demo(document.text).parse())
         ```
     """
-
-    module = ModuleProgram()
-    module.body.append(ModuleImports())
-
-    structs = []
-
-    # TODO: add json structs support
-    for s in schemas:
-        struct = build_ast_struct_parser(s, module, gen_docstring=False)
-        structs.append(struct)
-
-    t_defs = build_ast_typedef(module, *structs)
-    module.body.extend(t_defs)
-    module.body.extend(structs)
-
-    code_parts = converter.convert_program(module)
+    ast_module = AstBuilder.build_from_ssc_schemas(*schemas)
+    code_parts = converter.convert_program(ast_module)
     module = types.ModuleType("_ssc_single_eval")
     code = "\n".join(code_parts)
     exec(code, module.__dict__)
@@ -99,7 +81,7 @@ class Compiler:
         return instance
 
     @property
-    def cache(self):
+    def cache(self) -> dict[str, "Compiler"]:
         return self._cache
 
     def get_class(self, class_name: str) -> Any: