Added Transformation Model and its routes (#39)

* Added Transformation Model and its routes

sand/__main__.py

@@ -11,7 +11,7 @@
 from sand.commands.load import load_dataset
 from sand.container import use_container
 from sand.helpers.dependency_injection import use_auto_inject
-from sand.models import Project, SemanticModel, Table, TableRow
+from sand.models import Project, SemanticModel, Table, TableRow, Transformation
 from sand.models import db as dbconn
 from sand.models import init_db
 
@@ -21,7 +21,9 @@
 def init(db):
     """Init database"""
     init_db(db)
-    dbconn.create_tables([Project, Table, TableRow, SemanticModel], safe=True)
+    dbconn.create_tables(
+        [Project, Table, TableRow, SemanticModel, Transformation], safe=True
+    )
     if Project.select().where(fn.Lower(Project.name) == "default").count() == 0:
         Project(name="Default", description="The default project").save()
 

sand/app.py

@@ -12,7 +12,7 @@
 from sand.controllers.search import search_bp
 from sand.controllers.settings import setting_bp
 from sand.controllers.table import table_bp, table_row_bp
-from sand.controllers.transform import transform_bp
+from sand.controllers.transformation import transformation_bp
 from sand.helpers.namespace import NamespaceService
 from sand.models import EntityAR, SemanticModel
 from sand.models.ontology import OntClassAR, OntPropertyAR
@@ -33,7 +33,7 @@ def get_flask_app(
             table_row_bp,
             setting_bp,
             search_bp,
-            transform_bp,
+            transformation_bp,
             generate_api(
                 SemanticModel,
                 deserializers={"data": sand_deser.deserialize_graph},

sand/controllers/transform.py → sand/controllers/transformation.py

@@ -11,21 +11,26 @@
 
 from sand.models.table import Link, Table, TableRow
 from gena.deserializer import get_dataclass_deserializer
+from gena import generate_api
+from sand.models import Transformation
 
-transform_bp = Blueprint("transform", "transform")
+transformation_bp = generate_api(Transformation)
 
 
 @dataclass
 class Context:
-    """ Context dataclass to access the row of the cell that is being transformed."""
+    """Context dataclass to access the row of the cell that is being transformed."""
+
     index: int
     row: List[Union[str, float]]
 
 
 @dataclass
 class TransformRequestPayload:
     """Request Payload dataclass to validate the request obtained from the API call"""
+
     type: Literal["map", "filter", "split", "concatenate"]
+    table_id: int
     mode: str
     datapath: Union[str, List[str]]
     code: str
@@ -47,10 +52,10 @@ class Tdata(TypedDict):
 def filter_traceback_errors() -> str:
     """Filters traceback errors, removes sensitive information
 
-        Args:
+    Args:
 
-        Returns:
-            Error String without the sensitive information.
+    Returns:
+        Error String without the sensitive information.
     """
     (exc, value, tb) = sys.exc_info()
     tb = tb.tb_next
@@ -61,17 +66,20 @@ def filter_traceback_errors() -> str:
 ItemIndex = int
 
 
-def transform_map(transform_func: Callable[[Any, Context], Any], data: Iterable[Tuple[ItemIndex, Item, Context]],
-                  tolerance: int) -> List[Tdata]:
+def transform_map(
+    transform_func: Callable[[Any, Context], Any],
+    data: Iterable[Tuple[ItemIndex, Item, Context]],
+    tolerance: int,
+) -> List[Tdata]:
     """Implements map transform, performs map operation over each cell, for a given column
 
-        Args:
-            transform_func: User defined python function defined by the user
-            data: iterable with Column data and context object
-            tolerance: contains the API request data
+    Args:
+        transform_func: User defined python function defined by the user
+        data: iterable with Column data and context object
+        tolerance: contains the API request data
 
-        Returns:
-            list of Tdata objects, data transformed after applying map transform
+    Returns:
+        list of Tdata objects, data transformed after applying map transform
     """
     transformed_data = []
     for path, value, context in data:
@@ -90,20 +98,23 @@ def transform_map(transform_func: Callable[[Any, Context], Any], data: Iterable[
     return transformed_data
 
 
-def transform_filter(transform_func: Callable[[Any, Context], Any], data: Iterable[Tuple[ItemIndex, Item, Context]],
-                     tolerance: int) -> List[Tdata]:
+def transform_filter(
+    transform_func: Callable[[Any, Context], Any],
+    data: Iterable[Tuple[ItemIndex, Item, Context]],
+    tolerance: int,
+) -> List[Tdata]:
     """Implements filter transform, performs filter operation over each cell, for a given column
 
-        Args:
-            transform_func: User defined python function defined by the user
-            data: iterable with Column data and context object
-            tolerance: contains the API request data
+    Args:
+        transform_func: User defined python function defined by the user
+        data: iterable with Column data and context object
+        tolerance: contains the API request data
 
-        Returns:
-            list of Tdata objects, data transformed after applying filter transform
+    Returns:
+        list of Tdata objects, data transformed after applying filter transform
 
-        Raises:
-            BadRequest: An error occurred when the transform_func on execution, does not return a boolean
+    Raises:
+        BadRequest: An error occurred when the transform_func on execution, does not return a boolean
     """
     transformed_data = []
     for path, value, context in data:
@@ -124,20 +135,23 @@ def transform_filter(transform_func: Callable[[Any, Context], Any], data: Iterab
     return transformed_data
 
 
-def transform_split(transform_func: Callable[[Any, Context], Any], data: Iterable[Tuple[ItemIndex, Item, Context]],
-                    tolerance: int) -> List[Tdata]:
+def transform_split(
+    transform_func: Callable[[Any, Context], Any],
+    data: Iterable[Tuple[ItemIndex, Item, Context]],
+    tolerance: int,
+) -> List[Tdata]:
     """Implements split transform, performs split operation over each cell, for a given column
 
-        Args:
-            transform_func: User defined python function defined by the user
-            data: iterable with Column data and context object
-            tolerance: contains the API request data
+    Args:
+        transform_func: User defined python function defined by the user
+        data: iterable with Column data and context object
+        tolerance: contains the API request data
 
-        Returns:
-            list of Tdata objects, data transformed after applying split transform
+    Returns:
+        list of Tdata objects, data transformed after applying split transform
 
-        Raises:
-            BadRequest: An error occurred when transform_func on execution, does not return a list
+    Raises:
+        BadRequest: An error occurred when transform_func on execution, does not return a list
     """
     transformed_data = []
     for path, value, context in data:
@@ -158,17 +172,20 @@ def transform_split(transform_func: Callable[[Any, Context], Any], data: Iterabl
     return transformed_data
 
 
-def transform_concatenate(transform_func: Callable[[Any, Context], Any],
-                          data: Iterable[Tuple[ItemIndex, Item, Context]], tolerance: int) -> List:
+def transform_concatenate(
+    transform_func: Callable[[Any, Context], Any],
+    data: Iterable[Tuple[ItemIndex, Item, Context]],
+    tolerance: int,
+) -> List:
     """Implements concatenate transform, performs concatenate operation over each cell, for a given column
 
-        Args:
-            transform_func: User defined python function defined by the user
-            data: iterable with Column data and context object
-            tolerance: contains the API request data
+    Args:
+        transform_func: User defined python function defined by the user
+        data: iterable with Column data and context object
+        tolerance: contains the API request data
 
-        Returns:
-            list of Tdata objects, data transformed after applying concatenate transform
+    Returns:
+        list of Tdata objects, data transformed after applying concatenate transform
     """
     transformed_data = []
     for path, value, context in data:
@@ -190,30 +207,30 @@ def transform_concatenate(transform_func: Callable[[Any, Context], Any],
 def custom_getitem_guard(obj: Any, index: int) -> Any:
     """Implements __getitem__ restrictedpython policy and wraps _getitem_ function
 
-        Args:
-            obj: object that has __getitem__ implementation in python
-            index: index of the element that can be accessed from obj
+    Args:
+        obj: object that has __getitem__ implementation in python
+        index: index of the element that can be accessed from obj
 
-        Returns:
-            object element at position index
+    Returns:
+        object element at position index
     """
     return obj[index]
 
 
 def compile_function(code: str) -> Callable:
     """Executes code in string in a restricted mode using restrictedpython
 
-        Args:
-            code: object that has __getitem__ implementation in python
+    Args:
+        code: object that has __getitem__ implementation in python
 
-        Returns:
-            Callable function that wraps the code as a function body
+    Returns:
+        Callable function that wraps the code as a function body
 
-        Raises:
-            BadRequest: An error occurred when the code has compilation error
+    Raises:
+        BadRequest: An error occurred when the code has compilation error
     """
     loc = {}
-    safe_globals.update({'_getitem_': custom_getitem_guard})
+    safe_globals.update({"_getitem_": custom_getitem_guard})
     compiled_result = compile_restricted_function("value,context", code, "<function>")
 
     if compiled_result.errors:
@@ -224,23 +241,26 @@ def compile_function(code: str) -> Callable:
     return loc["<function>"]
 
 
-@transform_bp.route(
-    f"/{transform_bp.name}/<table_id>/transformations", methods=["POST"]
-)
-def transform(table_id: int):
-    table = Table.get_by_id(table_id)
-    table_rows: List[TableRow] = list(
-        TableRow.select().where(TableRow.table == table).order_by(TableRow.index)
-    )
-
+@transformation_bp.route(f"/{transformation_bp.name}/test", methods=["POST"])
+def transform():
     if isinstance(request.json["datapath"], str):
         request.json["datapath"] = [request.json["datapath"]]
 
     request_data = transform_request_deserializer(request.json)
+    table = Table.get_by_id(request_data.table_id)
+    table_rows: List[TableRow] = list(
+        TableRow.select().where(TableRow.table == table).order_by(TableRow.index)
+    )
     transform_func = compile_function(request_data.code)
     col_index_list = [table.columns.index(column) for column in request_data.datapath]
-    data = ((table_row.index, [table_row.row[col_index] for col_index in col_index_list],
-             Context(index=table_row.index, row=table_row.row)) for table_row in table_rows[:request_data.rows])
+    data = (
+        (
+            table_row.index,
+            [table_row.row[col_index] for col_index in col_index_list],
+            Context(index=table_row.index, row=table_row.row),
+        )
+        for table_row in table_rows[: request_data.rows]
+    )
 
     transformed_data = None
 
@@ -256,7 +276,9 @@ def transform(table_id: int):
             raise BadRequest(
                 "For transform type map the outputpath should be a single column"
             )
-        transformed_data = transform_filter(transform_func, data, request_data.tolerance)
+        transformed_data = transform_filter(
+            transform_func, data, request_data.tolerance
+        )
 
     elif request_data.type == "split":
         if request_data.outputpath is None:
@@ -266,6 +288,8 @@ def transform(table_id: int):
         transformed_data = transform_split(transform_func, data, request_data.tolerance)
 
     elif request_data.type == "concatenate":
-        transformed_data = transform_concatenate(transform_func, data, request_data.tolerance)
+        transformed_data = transform_concatenate(
+            transform_func, data, request_data.tolerance
+        )
 
     return jsonify(transformed_data)

sand/models/__init__.py

@@ -3,5 +3,6 @@
 from sand.models.project import Project
 from sand.models.semantic_model import SemanticModel
 from sand.models.table import Table, TableRow, Link, ContextPage
+from sand.models.transformation import Transformation
 
-all_tables = [Project, SemanticModel, Table, TableRow]
+all_tables = [Project, SemanticModel, Table, TableRow, Transformation]

sand/models/transformation.py

@@ -0,0 +1,45 @@
+from __future__ import annotations
+from typing import Literal, Union, List
+from peewee import (
+    CharField,
+    ForeignKeyField,
+    TextField,
+    IntegerField,
+    BooleanField,
+)
+from playhouse.sqlite_ext import JSONField
+
+from sand.models.base import BaseModel
+from sand.models.table import Table
+
+
+class Transformation(BaseModel):
+    name = CharField()
+    table = ForeignKeyField(Table, backref="transformations", on_delete="CASCADE")
+    mode = CharField()
+    datapath: Union[List[str], str] = JSONField()  # type: ignore
+    outputpath: List[str] = JSONField()  # type: ignore
+    type = CharField()
+    code = TextField()
+    on_error: Literal[
+        "set_to_blank", "store_error", "keep_original", "abort"
+    ] = CharField()
+    is_draft = BooleanField()
+    order = IntegerField()
+    insert_after = ForeignKeyField("self", null=True, on_delete="SET NULL")
+
+    def to_dict(self):
+        return {
+            "id": self.id,
+            "name": self.name,
+            "table": self.table_id,
+            "type": self.type,
+            "mode": self.mode,
+            "datapath": self.datapath,
+            "outputpath": self.outputpath,
+            "code": self.on_error,
+            "on_error": self.on_error,
+            "is_draft": self.is_draft,
+            "order": self.order,
+            "insert_after": self.insert_after,
+        }