Feat/dat 474 constraints (#67)

Author: tunglxfast

datamimic_ce/constants/attribute_constants.py

@@ -73,3 +73,5 @@
 ATTR_STRING: Final = "string"
 ATTR_BUCKET: Final = "bucket"
 ATTR_MP_PLATFORM: Final = "mpPlatform"
+ATTR_IF: Final = "if"
+ATTR_THEN: Final = "then"

datamimic_ce/constants/element_constants.py

@@ -25,3 +25,5 @@
 EL_CONDITION = "condition"
 EL_ELSE_IF = "else-if"
 EL_ELSE = "else"
+EL_CONSTRAINTS = "constraints"
+EL_RULE = "rule"

datamimic_ce/model/constraints_model.py

@@ -0,0 +1,11 @@
+# DATAMIMIC
+# Copyright (c) 2023-2024 Rapiddweller Asia Co., Ltd.
+# This software is licensed under the MIT License.
+# See LICENSE file for the full text of the license.
+# For questions and support, contact: [email protected]
+
+from pydantic import BaseModel
+
+
+class ConstraintsModel(BaseModel):
+    pass

datamimic_ce/model/rule_model.py

@@ -0,0 +1,19 @@
+# DATAMIMIC
+# Copyright (c) 2023-2024 Rapiddweller Asia Co., Ltd.
+# This software is licensed under the MIT License.
+# See LICENSE file for the full text of the license.
+# For questions and support, contact: [email protected]
+from pydantic import BaseModel, Field, field_validator
+
+from datamimic_ce.constants.attribute_constants import ATTR_IF, ATTR_THEN
+from datamimic_ce.model.model_util import ModelUtil
+
+
+class RuleModel(BaseModel):
+    if_rule: str = Field(alias=ATTR_IF)
+    then_rule: str = Field(alias=ATTR_THEN)
+
+    @field_validator("if_rule", "then_rule")
+    @classmethod
+    def validate_name(cls, value):
+        return ModelUtil.check_not_empty(value=value)

datamimic_ce/parsers/constraints_parser.py

@@ -0,0 +1,46 @@
+# DATAMIMIC
+# Copyright (c) 2023-2024 Rapiddweller Asia Co., Ltd.
+# This software is licensed under the MIT License.
+# See LICENSE file for the full text of the license.
+# For questions and support, contact: [email protected]
+
+
+from pathlib import Path
+from xml.etree.ElementTree import Element
+
+from datamimic_ce.constants.element_constants import EL_CONSTRAINTS
+from datamimic_ce.parsers.statement_parser import StatementParser
+from datamimic_ce.statements.composite_statement import CompositeStatement
+from datamimic_ce.statements.constraints_statement import ConstraintsStatement
+from datamimic_ce.utils.base_class_factory_util import BaseClassFactoryUtil
+
+
+class ConstraintsParser(StatementParser):
+    def __init__(
+        self,
+        class_factory_util: BaseClassFactoryUtil,
+        element: Element,
+        properties: dict,
+    ):
+        super().__init__(
+            element,
+            properties,
+            valid_element_tag=EL_CONSTRAINTS,
+            class_factory_util=class_factory_util,
+        )
+
+    def parse(self, descriptor_dir: Path, parent_stmt: CompositeStatement) -> ConstraintsStatement:
+        """
+        Parse element "constraints" into ConstraintsStatement
+        :return:
+        """
+        constraints_stmt = ConstraintsStatement(parent_stmt)
+        sub_stmt_list = self._class_factory_util.get_parser_util_cls()().parse_sub_elements(
+            class_factory_util=self._class_factory_util,
+            descriptor_dir=descriptor_dir,
+            element=self._element,
+            properties=self._properties,
+            parent_stmt=constraints_stmt,
+        )
+        constraints_stmt.sub_statements = sub_stmt_list
+        return constraints_stmt

datamimic_ce/parsers/generate_parser.py

@@ -10,6 +10,7 @@
 from datamimic_ce.constants.element_constants import EL_GENERATE
 from datamimic_ce.model.generate_model import GenerateModel
 from datamimic_ce.parsers.statement_parser import StatementParser
+from datamimic_ce.statements.constraints_statement import ConstraintsStatement
 from datamimic_ce.statements.generate_statement import GenerateStatement
 from datamimic_ce.statements.statement import Statement
 from datamimic_ce.utils.base_class_factory_util import BaseClassFactoryUtil
@@ -50,5 +51,20 @@ def parse(self, descriptor_dir: Path, parent_stmt: Statement, lazy_parse: bool =
             self._properties,
             gen_stmt,
         )
+
+        self._check_only_one_constraints_tag(sub_stmt_list)
+
         gen_stmt.sub_statements = sub_stmt_list
         return gen_stmt
+
+    @staticmethod
+    def _check_only_one_constraints_tag(sub_stmt_list: list[Statement]):
+        """
+        Only one 'constraints' tag per generate
+        """
+        count = 0
+        for stmt in sub_stmt_list:
+            if isinstance(stmt, ConstraintsStatement):
+                count += 1
+        if count > 1:
+            raise SyntaxError("Only once <constraints> allow in per <generate>")

datamimic_ce/parsers/parser_util.py

@@ -15,6 +15,7 @@
 from datamimic_ce.constants.element_constants import (
     EL_ARRAY,
     EL_CONDITION,
+    EL_CONSTRAINTS,
     EL_DATABASE,
     EL_ECHO,
     EL_ELEMENT,
@@ -32,12 +33,14 @@
     EL_MONGODB,
     EL_NESTED_KEY,
     EL_REFERENCE,
+    EL_RULE,
     EL_SETUP,
     EL_VARIABLE,
 )
 from datamimic_ce.logger import logger
 from datamimic_ce.parsers.array_parser import ArrayParser
 from datamimic_ce.parsers.condition_parser import ConditionParser
+from datamimic_ce.parsers.constraints_parser import ConstraintsParser
 from datamimic_ce.parsers.database_parser import DatabaseParser
 from datamimic_ce.parsers.echo_parser import EchoParser
 from datamimic_ce.parsers.element_parser import ElementParser
@@ -54,6 +57,7 @@
 from datamimic_ce.parsers.memstore_parser import MemstoreParser
 from datamimic_ce.parsers.nested_key_parser import NestedKeyParser
 from datamimic_ce.parsers.reference_parser import ReferenceParser
+from datamimic_ce.parsers.rule_parser import RuleParser
 from datamimic_ce.parsers.variable_parser import VariableParser
 from datamimic_ce.statements.array_statement import ArrayStatement
 from datamimic_ce.statements.composite_statement import CompositeStatement
@@ -109,6 +113,7 @@ def get_valid_sub_elements_set_by_tag(ele_tag: str) -> set | None:
                 EL_ELEMENT,
                 EL_ARRAY,
                 EL_CONDITION,
+                EL_CONSTRAINTS,
             },
             EL_CONDITION: {EL_IF, EL_ELSE_IF, EL_ELSE},
             EL_GENERATE: {
@@ -122,6 +127,7 @@ def get_valid_sub_elements_set_by_tag(ele_tag: str) -> set | None:
                 EL_ECHO,
                 EL_CONDITION,
                 EL_INCLUDE,
+                EL_CONSTRAINTS,
             },
             EL_INCLUDE: {EL_SETUP},
             EL_ITEM: {EL_KEY, EL_NESTED_KEY, EL_LIST, EL_ARRAY, EL_ELEMENT},
@@ -130,6 +136,7 @@ def get_valid_sub_elements_set_by_tag(ele_tag: str) -> set | None:
             EL_IF: None,
             EL_ELSE_IF: None,
             EL_ELSE: None,
+            EL_CONSTRAINTS: {EL_RULE},
         }
 
         return valid_sub_element_dict.get(ele_tag, set())
@@ -189,6 +196,10 @@ def get_parser_by_element(class_factory_util: BaseClassFactoryUtil, element: Ele
             return ElementParser(class_factory_util, element=element, properties=properties)
         elif tag == EL_GENERATOR:
             return GeneratorParser(class_factory_util, element=element, properties=properties)
+        elif tag == EL_CONSTRAINTS:
+            return ConstraintsParser(class_factory_util, element=element, properties=properties)
+        elif tag == EL_RULE:
+            return RuleParser(class_factory_util, element=element, properties=properties)
         else:
             raise ValueError(f"Cannot get parser for element <{tag}>")
 
@@ -238,9 +249,9 @@ def parse_sub_elements(
                     | GeneratorParser,
                 ):
                     stmt = parser.parse()
-                elif isinstance(parser, KeyParser):
+                elif isinstance(parser, KeyParser | RuleParser):
                     stmt = parser.parse(descriptor_dir=descriptor_dir, parent_stmt=parent_stmt)
-                elif isinstance(parser, ConditionParser):
+                elif isinstance(parser, ConditionParser | ConstraintsParser):
                     stmt = parser.parse(
                         descriptor_dir=descriptor_dir, parent_stmt=cast(CompositeStatement, parent_stmt)
                     )

datamimic_ce/parsers/rule_parser.py

@@ -0,0 +1,36 @@
+# DATAMIMIC
+# Copyright (c) 2023-2024 Rapiddweller Asia Co., Ltd.
+# This software is licensed under the MIT License.
+# See LICENSE file for the full text of the license.
+# For questions and support, contact: [email protected]
+from pathlib import Path
+from xml.etree.ElementTree import Element
+
+from datamimic_ce.constants.element_constants import EL_RULE
+from datamimic_ce.model.rule_model import RuleModel
+from datamimic_ce.parsers.statement_parser import StatementParser
+from datamimic_ce.statements.rule_statement import RuleStatement
+from datamimic_ce.statements.statement import Statement
+from datamimic_ce.utils.base_class_factory_util import BaseClassFactoryUtil
+
+
+class RuleParser(StatementParser):
+    """
+    Parse element "rule" to RuleStatement
+    """
+
+    def __init__(
+        self,
+        class_factory_util: BaseClassFactoryUtil,
+        element: Element,
+        properties: dict,
+    ):
+        super().__init__(element, properties, valid_element_tag=EL_RULE, class_factory_util=class_factory_util)
+
+    def parse(self, descriptor_dir: Path, parent_stmt: Statement) -> RuleStatement:
+        """
+        Parse element "xml-attribute" to XmlAttributeStatement
+        :return:
+        """
+
+        return RuleStatement(self.validate_attributes(RuleModel))

datamimic_ce/statements/constraints_statement.py

@@ -0,0 +1,11 @@
+# DATAMIMIC
+# Copyright (c) 2023-2024 Rapiddweller Asia Co., Ltd.
+# This software is licensed under the MIT License.
+# See LICENSE file for the full text of the license.
+# For questions and support, contact: [email protected]
+from datamimic_ce.statements.composite_statement import CompositeStatement
+
+
+class ConstraintsStatement(CompositeStatement):
+    def __init__(self, parent_stmt: CompositeStatement):
+        super().__init__(None, parent_stmt=parent_stmt)

datamimic_ce/statements/rule_statement.py

@@ -0,0 +1,22 @@
+# DATAMIMIC
+# Copyright (c) 2023-2024 Rapiddweller Asia Co., Ltd.
+# This software is licensed under the MIT License.
+# See LICENSE file for the full text of the license.
+# For questions and support, contact: [email protected]
+from datamimic_ce.model.rule_model import RuleModel
+from datamimic_ce.statements.statement import Statement
+
+
+class RuleStatement(Statement):
+    def __init__(self, model: RuleModel):
+        super().__init__(None, None)
+        self._if_rule = model.if_rule
+        self._then_rule = model.then_rule
+
+    @property
+    def if_rule(self):
+        return self._if_rule
+
+    @property
+    def then_rule(self):
+        return self._then_rule

datamimic_ce/tasks/constraints_task.py

@@ -0,0 +1,60 @@
+# DATAMIMIC
+# Copyright (c) 2023-2024 Rapiddweller Asia Co., Ltd.
+# This software is licensed under the MIT License.
+# See LICENSE file for the full text of the license.
+# For questions and support, contact: [email protected]
+import copy
+import itertools
+
+from datamimic_ce.contexts.context import SAFE_GLOBALS, DotableDict
+from datamimic_ce.data_sources.data_source_pagination import DataSourcePagination
+from datamimic_ce.statements.constraints_statement import ConstraintsStatement
+from datamimic_ce.statements.rule_statement import RuleStatement
+from datamimic_ce.tasks.task import Task
+
+
+class ConstraintsTask(Task):
+    def __init__(self, statement: ConstraintsStatement):
+        self._statement = statement
+
+    @property
+    def statement(self) -> ConstraintsStatement:
+        return self._statement
+
+    def execute(self, source_data, pagination: DataSourcePagination | None = None, cyclic: bool | None = False) -> list:
+        filter_data = list(source_data)
+        # If source is empty, return empty list
+        if len(filter_data) == 0:
+            return []
+
+        for i in range(len(filter_data) - 1, -1, -1):  # Iterate from last to first
+            data_dict = copy.deepcopy(filter_data[i])
+
+            for key, value in data_dict.items():
+                if isinstance(value, dict):
+                    data_dict[key] = DotableDict(value)
+
+            for child_stmt in self.statement.sub_statements:
+                if isinstance(child_stmt, RuleStatement):
+                    if_condition = eval(child_stmt.if_rule, SAFE_GLOBALS, data_dict)
+                    if isinstance(if_condition, bool) and if_condition:
+                        else_condition = eval(child_stmt.then_rule, SAFE_GLOBALS, data_dict)
+                        if isinstance(else_condition, bool) and else_condition is False:
+                            del filter_data[i]  # remove data that not meet then_rule
+                            break
+        # If filtered data is empty, return empty list
+        if len(filter_data) == 0:
+            return []
+
+        if pagination is None:
+            start_idx = 0
+            end_idx = len(filter_data)
+        else:
+            start_idx = pagination.skip
+            end_idx = pagination.skip + pagination.limit
+        # Get cyclic data from filtered data source
+        if cyclic:
+            iterator = itertools.cycle(filter_data)
+            return [copy.deepcopy(ele) for ele in itertools.islice(iterator, start_idx, end_idx)]
+        else:
+            return list(itertools.islice(filter_data, start_idx, end_idx))

datamimic_ce/tasks/nested_key_task.py

@@ -15,6 +15,7 @@
 from datamimic_ce.data_sources.data_source_registry import DataSourceRegistry
 from datamimic_ce.logger import logger
 from datamimic_ce.statements.nested_key_statement import NestedKeyStatement
+from datamimic_ce.tasks.constraints_task import ConstraintsTask
 from datamimic_ce.tasks.element_task import ElementTask
 from datamimic_ce.tasks.task import Task
 from datamimic_ce.utils.base_class_factory_util import BaseClassFactoryUtil
@@ -187,6 +188,9 @@ def _try_execute_sub_tasks(self, ctx: GenIterContext) -> dict:
                 try:
                     if isinstance(sub_task, ElementTask):
                         attributes.update(sub_task.generate_xml_attribute(ctx))
+                    elif isinstance(sub_task, ConstraintsTask):
+                        # do not execute ConstraintsTask here, ConstraintsTask is for filter source data
+                        pass
                     else:
                         sub_task.execute(ctx)
                 except StopIteration:
@@ -316,7 +320,8 @@ def _modify_nestedkey_data_list(self, parent_context: GenIterContext, value: lis
         :return:
         """
         result = []
-
+        # filter source data by constraints
+        value = self._filter_source_by_constraints_task(parent_context=parent_context, source_data=value)
         # Determine len of nestedkey
         count = self._determine_nestedkey_length(context=parent_context)
         value_len = len(value)
@@ -380,3 +385,17 @@ def _post_convert(self, value):
         for converter in self._converter_list:
             value = converter.convert(value)
         return value
+
+    def _filter_source_by_constraints_task(self, parent_context: GenIterContext, source_data: list) -> list:
+        """
+        Execute ConstraintsTask to filter source data
+        """
+        result = source_data
+        if self._sub_tasks:
+            for sub_task in self._sub_tasks:
+                if isinstance(sub_task, ConstraintsTask):
+                    nestedkey_len = self._determine_nestedkey_length(context=parent_context)
+                    temp_pagination = DataSourcePagination(skip=0, limit=nestedkey_len) if nestedkey_len else None
+                    result = sub_task.execute(source_data, pagination=temp_pagination, cyclic=self.statement.cyclic)
+                    break
+        return result