datajunction-server 0.0.1a79.dev0__tar.gz → 0.0.1a79.dev2__tar.gz

{datajunction_server-0.0.1a79.dev0 → datajunction_server-0.0.1a79.dev2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: datajunction-server
-Version: 0.0.1a79.dev0
+Version: 0.0.1a79.dev2
 Summary: DataJunction server library for running to a DataJunction server
 Project-URL: repository, https://github.com/DataJunction/dj
 Author-email: DataJunction Authors <yian.shang@gmail.com>

datajunction_server-0.0.1a79.dev2/datajunction_server/__about__.py

@@ -0,0 +1,4 @@
+"""
+Version for Hatch
+"""
+__version__ = "0.0.1a79.dev2"

{datajunction_server-0.0.1a79.dev0 → datajunction_server-0.0.1a79.dev2}/datajunction_server/construction/build_v2.py

@@ -297,7 +297,7 @@ def build_preaggregate_query(
                         ).type,
                     )
             for proj in temp_select.projection:
-                proj.set_semantic_entity(metric.name + SEPARATOR + measure.name)  # type: ignore
+                proj.set_semantic_entity(parent_node.name + SEPARATOR + measure.name)  # type: ignore
                 proj.set_semantic_type(SemanticType.MEASURE)  # type: ignore
             final_query.select.projection.extend(temp_select.projection)
     return final_query

{datajunction_server-0.0.1a79.dev0 → datajunction_server-0.0.1a79.dev2}/datajunction_server/sql/decompose.py

@@ -1,4 +1,5 @@
 """Used for extracting measures form metric definitions."""
+import hashlib
 from functools import lru_cache
 
 from pydantic import BaseModel
@@ -69,6 +70,7 @@ class MeasureExtractor:
 
         # Outputs from decomposition
         self._measures: list[Measure] = []
+        self._measures_tracker: set[str] = set()
         self._query_ast = query_ast
         self._extracted = False
 
@@ -103,17 +105,22 @@ class MeasureExtractor:
                         col.name = ast.Name(col.name.name)
                 self._query_ast.select.from_.relations[0].primary.set_alias(None)  # type: ignore
 
-            for idx, func in enumerate(self._query_ast.find_all(ast.Function)):
+            for func in self._query_ast.find_all(ast.Function):
                 dj_function = func.function()
                 handler = self.handlers.get(dj_function)
                 if handler and dj_function.is_aggregation:
-                    if func_measures := handler(func, idx):  # pragma: no cover
+                    if func_measures := handler(func):  # pragma: no cover
                         MeasureExtractor.update_ast(func, func_measures)
-                    self._measures.extend(sorted(func_measures, key=lambda m: m.name))
+
+                    for measure in sorted(func_measures, key=lambda m: m.name):
+                        if measure.name not in self._measures_tracker:
+                            self._measures_tracker.add(measure.name)
+                            self._measures.append(measure)
+
             self._extracted = True
         return self._measures, self._query_ast
 
-    def _simple_associative_agg(self, func, idx) -> list[Measure]:
+    def _simple_associative_agg(self, func) -> list[Measure]:
         """
         Handles measures decomposition for a single-argument associative aggregation function.
         Examples: SUM, MAX, MIN, COUNT
@@ -122,10 +129,13 @@ class MeasureExtractor:
         measure_name = "_".join(
             [str(col) for col in arg.find_all(ast.Column)] + [func.name.name.lower()],
         )
+        expression = f"{func.quantifier} {arg}" if func.quantifier else str(arg)
+        short_hash = hashlib.md5(expression.encode("utf-8")).hexdigest()[:8]
+
         return [
             Measure(
-                name=f"{measure_name}_{idx}",
-                expression=f"{func.quantifier} {arg}" if func.quantifier else str(arg),
+                name=f"{measure_name}_{short_hash}",
+                expression=expression,
                 aggregation=func.name.name.upper(),
                 rule=AggregationRule(
                     type=Aggregability.FULL
@@ -135,7 +145,7 @@ class MeasureExtractor:
             ),
         ]
 
-    def _avg(self, func, idx) -> list[Measure]:
+    def _avg(self, func) -> list[Measure]:
         """
         Handles measures decomposition for AVG (it requires both the SUM and COUNT
         of the selected measure).
@@ -145,10 +155,12 @@ class MeasureExtractor:
             [str(col) for col in arg.find_all(ast.Column)]
             + [dj_functions.Sum.__name__.lower()],
         )
+        expression = str(arg)
+        short_hash = hashlib.md5(expression.encode("utf-8")).hexdigest()[:8]
         return [
             Measure(
-                name=f"{measure_name}_{idx}",
-                expression=str(arg),
+                name=f"{measure_name}_{short_hash}",
+                expression=expression,
                 aggregation=dj_functions.Sum.__name__.upper(),
                 rule=AggregationRule(
                     type=Aggregability.FULL

{datajunction_server-0.0.1a79.dev0 → datajunction_server-0.0.1a79.dev2}/tests/api/graphql/find_nodes_test.py

@@ -637,7 +637,7 @@ async def test_find_metric(
                         {
                             "aggregation": "SUM",
                             "expression": "completed_repairs",
-                            "name": "completed_repairs_sum_0",
+                            "name": "completed_repairs_sum_81105666",
                             "rule": {
                                 "type": "FULL",
                             },
@@ -645,7 +645,7 @@ async def test_find_metric(
                         {
                             "aggregation": "SUM",
                             "expression": "total_repairs_dispatched",
-                            "name": "total_repairs_dispatched_sum_1",
+                            "name": "total_repairs_dispatched_sum_01dc2341",
                             "rule": {
                                 "type": "FULL",
                             },
@@ -653,7 +653,7 @@ async def test_find_metric(
                         {
                             "aggregation": "SUM",
                             "expression": "total_amount_in_region",
-                            "name": "total_amount_in_region_sum_2",
+                            "name": "total_amount_in_region_sum_1c94ab45",
                             "rule": {
                                 "type": "FULL",
                             },
@@ -661,23 +661,23 @@ async def test_find_metric(
                         {
                             "aggregation": "SUM",
                             "expression": "na.total_amount_nationwide",
-                            "name": "na.total_amount_nationwide_sum_3",
+                            "name": "na.total_amount_nationwide_sum_fed946fe",
                             "rule": {
                                 "type": "FULL",
                             },
                         },
                     ],
-                    "derivedQuery": "SELECT  (SUM(completed_repairs_sum_0) * 1.0 / "
-                    "SUM(total_repairs_dispatched_sum_1)) * "
-                    "(SUM(total_amount_in_region_sum_2) * 1.0 / "
-                    "SUM(na.total_amount_nationwide_sum_3)) * 100 \n"
+                    "derivedQuery": "SELECT  (SUM(completed_repairs_sum_81105666) * 1.0 / "
+                    "SUM(total_repairs_dispatched_sum_01dc2341)) * "
+                    "(SUM(total_amount_in_region_sum_1c94ab45) * 1.0 / "
+                    "SUM(na.total_amount_nationwide_sum_fed946fe)) * 100 \n"
                     " FROM default.regional_level_agg CROSS JOIN "
                     "default.national_level_agg na\n"
                     "\n",
-                    "derivedExpression": "(SUM(completed_repairs_sum_0) * 1.0 / "
-                    "SUM(total_repairs_dispatched_sum_1)) * "
-                    "(SUM(total_amount_in_region_sum_2) * 1.0 / "
-                    "SUM(na.total_amount_nationwide_sum_3)) * 100",
+                    "derivedExpression": "(SUM(completed_repairs_sum_81105666) * 1.0 / "
+                    "SUM(total_repairs_dispatched_sum_01dc2341)) * "
+                    "(SUM(total_amount_in_region_sum_1c94ab45) * 1.0 / "
+                    "SUM(na.total_amount_nationwide_sum_fed946fe)) * 100",
                 },
             },
             "name": "default.regional_repair_efficiency",

{datajunction_server-0.0.1a79.dev0 → datajunction_server-0.0.1a79.dev2}/tests/api/metrics_test.py

@@ -414,7 +414,7 @@ async def test_read_metrics(module__client_with_roads: AsyncClient) -> None:
         {
             "aggregation": "SUM",
             "expression": "if(discount > 0.0, 1, 0)",
-            "name": "discount_sum_0",
+            "name": "discount_sum_62846f49",
             "rule": {
                 "level": None,
                 "type": "full",
@@ -423,7 +423,7 @@ async def test_read_metrics(module__client_with_roads: AsyncClient) -> None:
         {
             "aggregation": "COUNT",
             "expression": "*",
-            "name": "count_1",
+            "name": "count_3389dae3",
             "rule": {
                 "level": None,
                 "type": "full",
@@ -431,11 +431,12 @@ async def test_read_metrics(module__client_with_roads: AsyncClient) -> None:
         },
     ]
     assert data["derived_query"] == (
-        "SELECT  CAST(sum(discount_sum_0) AS DOUBLE) / SUM(count_1) AS "
+        "SELECT  CAST(sum(discount_sum_62846f49) AS DOUBLE) / SUM(count_3389dae3) AS "
         "default_DOT_discounted_orders_rate \n FROM default.repair_orders_fact"
     )
     assert data["derived_expression"] == (
-        "CAST(sum(discount_sum_0) AS DOUBLE) / SUM(count_1) AS default_DOT_discounted_orders_rate"
+        "CAST(sum(discount_sum_62846f49) AS DOUBLE) / SUM(count_3389dae3) "
+        "AS default_DOT_discounted_orders_rate"
     )
 
 

{datajunction_server-0.0.1a79.dev0 → datajunction_server-0.0.1a79.dev2}/tests/api/sql_v2_test.py

@@ -572,7 +572,7 @@ async def test_measures_sql_with_filters__v2(  # pylint: disable=too-many-argume
             )
             SELECT
               default_DOT_repair_orders_fact_built.default_DOT_dispatcher_DOT_dispatcher_id,
-              SUM(price * discount) AS price_discount_sum_0
+              SUM(price * discount) AS price_discount_sum_017d55a8
             FROM default_DOT_repair_orders_fact_built
             GROUP BY default_DOT_repair_orders_fact_built.default_DOT_dispatcher_DOT_dispatcher_id
             """,
@@ -586,10 +586,10 @@ async def test_measures_sql_with_filters__v2(  # pylint: disable=too-many-argume
                     "type": "int",
                 },
                 {
-                    "column": "price_discount_sum_0",
-                    "name": "price_discount_sum_0",
+                    "column": "price_discount_sum_017d55a8",
+                    "name": "price_discount_sum_017d55a8",
                     "node": "default.total_repair_order_discounts",
-                    "semantic_entity": "default.total_repair_order_discounts.price_discount_sum_0",
+                    "semantic_entity": "default.total_repair_order_discounts.price_discount_sum_017d55a8",
                     "semantic_type": "measure",
                     "type": "double",
                 },
@@ -637,7 +637,7 @@ async def test_measures_sql_with_filters__v2(  # pylint: disable=too-many-argume
             )
             SELECT
               default_DOT_repair_orders_fact_built.default_DOT_dispatcher_DOT_dispatcher_id,
-              SUM(price * discount) AS price_discount_sum_0,
+              SUM(price * discount) AS price_discount_sum_017d55a8,
               COUNT(1) AS count
             FROM default_DOT_repair_orders_fact_built
             GROUP BY  default_DOT_repair_orders_fact_built.default_DOT_dispatcher_DOT_dispatcher_id
@@ -652,8 +652,8 @@ async def test_measures_sql_with_filters__v2(  # pylint: disable=too-many-argume
                     "type": "int",
                 },
                 {
-                    "column": "price_discount_sum_0",
-                    "name": "price_discount_sum_0",
+                    "column": "price_discount_sum_017d55a8",
+                    "name": "price_discount_sum_017d55a8",
                     "node": mock.ANY,
                     "semantic_entity": mock.ANY,
                     "semantic_type": "measure",
@@ -763,8 +763,8 @@ async def test_measures_sql_with_filters__v2(  # pylint: disable=too-many-argume
               default_DOT_repair_orders_fact_built.default_DOT_dispatcher_DOT_company_name,
               default_DOT_repair_orders_fact_built.default_DOT_hard_hat_DOT_last_name,
               COUNT(1) AS count,
-              SUM(CAST(time_to_dispatch AS INT)) AS time_to_dispatch_sum_0,
-              SUM(total_repair_cost) AS total_repair_cost_sum_0
+              SUM(CAST(time_to_dispatch AS INT)) AS time_to_dispatch_sum_bf99afd6,
+              SUM(total_repair_cost) AS total_repair_cost_sum_9bdaf803
             FROM default_DOT_repair_orders_fact_built
             GROUP BY
               default_DOT_repair_orders_fact_built.default_DOT_us_state_DOT_state_name,
@@ -805,18 +805,18 @@ async def test_measures_sql_with_filters__v2(  # pylint: disable=too-many-argume
                     "type": "bigint",
                 },
                 {
-                    "column": "time_to_dispatch_sum_0",
-                    "name": "time_to_dispatch_sum_0",
+                    "column": "time_to_dispatch_sum_bf99afd6",
+                    "name": "time_to_dispatch_sum_bf99afd6",
                     "node": "default.avg_time_to_dispatch",
-                    "semantic_entity": "default.avg_time_to_dispatch.time_to_dispatch_sum_0",
+                    "semantic_entity": "default.avg_time_to_dispatch.time_to_dispatch_sum_bf99afd6",
                     "semantic_type": "measure",
                     "type": "bigint",
                 },
                 {
-                    "column": "total_repair_cost_sum_0",
-                    "name": "total_repair_cost_sum_0",
+                    "column": "total_repair_cost_sum_9bdaf803",
+                    "name": "total_repair_cost_sum_9bdaf803",
                     "node": "default.total_repair_cost",
-                    "semantic_entity": "default.total_repair_cost.total_repair_cost_sum_0",
+                    "semantic_entity": "default.total_repair_cost.total_repair_cost_sum_9bdaf803",
                     "semantic_type": "measure",
                     "type": "double",
                 },
@@ -892,7 +892,7 @@ async def test_measures_sql_with_filters__v2(  # pylint: disable=too-many-argume
               default_DOT_repair_orders_fact_built.default_DOT_dispatcher_DOT_company_name,
               default_DOT_repair_orders_fact_built.default_DOT_hard_hat_DOT_last_name,
               COUNT(1) AS count,
-              SUM(CAST(time_to_dispatch AS INT)) AS time_to_dispatch_sum_0
+              SUM(CAST(time_to_dispatch AS INT)) AS time_to_dispatch_sum_bf99afd6
             FROM default_DOT_repair_orders_fact_built
             GROUP BY
               default_DOT_repair_orders_fact_built.default_DOT_dispatcher_DOT_company_name,
@@ -972,7 +972,7 @@ async def test_measures_sql_with_filters__v2(  # pylint: disable=too-many-argume
               default_DOT_repair_orders_fact_built.default_DOT_dispatcher_DOT_company_name,
               default_DOT_repair_orders_fact_built.default_DOT_hard_hat_DOT_last_name,
               COUNT(1) AS count,
-              SUM(CAST(time_to_dispatch AS INT)) AS time_to_dispatch_sum_0
+              SUM(CAST(time_to_dispatch AS INT)) AS time_to_dispatch_sum_bf99afd6
             FROM default_DOT_repair_orders_fact_built
             GROUP BY
               default_DOT_repair_orders_fact_built.default_DOT_dispatcher_DOT_company_name,