scrapely-python 0.1.19__tar.gz → 0.1.20__tar.gz

{scrapely_python-0.1.19/src/scrapely_python.egg-info → scrapely_python-0.1.20}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scrapely-python
-Version: 0.1.19
+Version: 0.1.20
 Summary: Scrapely Python SDK - A toolkit for building Actors on the Scrapely platform
 Author-email: Scrapely <support@scrape.ly>
 License-Expression: MIT AND Apache-2.0

{scrapely_python-0.1.19 → scrapely_python-0.1.20}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "scrapely-python"
-version = "0.1.19"
+version = "0.1.20"
 description = "Scrapely Python SDK - A toolkit for building Actors on the Scrapely platform"
 readme = "README.md"
 license = "MIT AND Apache-2.0"

{scrapely_python-0.1.19 → scrapely_python-0.1.20}/src/scrapely/__init__.py

@@ -132,7 +132,7 @@ from .errors import (
     raise_for_status,
 )
 
-__version__ = '0.1.19'
+__version__ = '0.1.20'
 __all__ = [
     # Core
     'Actor',

{scrapely_python-0.1.19 → scrapely_python-0.1.20}/src/scrapely/charging.py

@@ -246,6 +246,8 @@ class ChargingManager:
         # grows), so the local budget gates on the run's true remaining budget.
         self._platform_other_usd: Decimal = Decimal('0')
         self._last_remaining_budget_nd: Optional[int] = None
+        # True once the platform reported a remaining budget (its usage is then in the local budget).
+        self._platform_budget_seen = False
         self._charge_lock = RLock()
         self._flush_lock = asyncio.Lock()
         # Serializes budget calculation/reservation with PPE-aware dataset pushes.
@@ -311,6 +313,7 @@ class ChargingManager:
         self._last_remaining_budget_nd = None
         if nd is None or self._max_total_charge_usd is None or nd < 0:
             return
+        self._platform_budget_seen = True
         other = self._max_total_charge_usd - Decimal(nd) / Decimal(1_000_000_000) - self._total_charged_usd
         if other > self._platform_other_usd:
             self._platform_other_usd = other
@@ -1717,7 +1720,7 @@ async def push_data_with_charges(
                 else:
                     await charging_manager._void_charge(name, op_id, charged)
 
-        for _attempt in range(4):
+        for _attempt in range(6):
             if rows <= 0:
                 break
             ops = []
@@ -1741,10 +1744,25 @@ async def push_data_with_charges(
             if short_to is None:
                 break
             # Some event came back short: void this attempt in full and retry with
-            # the rows that event can pay for (a void restores the budget).
+            # the whole rows the budget covers at the FULL per-row cost. Sizing
+            # from the short event alone ignores the other events of the row, so
+            # the retry came back short again (appstore 64df0a3608d74af88: 20
+            # details -> 14, retry sized 14, app-item short, 0 saved). What this
+            # attempt was accepted is budget the void gives back; with the
+            # platform's remainingBudgetNd the local budget also knows the rest.
+            attempt_spent = sum(
+                (charging_manager._calculate_event_price(name) or Decimal('0')) * Decimal(charged)
+                for name, charged, _op, _unc in ops
+            )
             await undo(ops)
             ops = []
-            rows = short_to  # got < n * rows, so short_to < rows
+            fit = int(attempt_spent / row_cost) if row_cost > 0 else 0
+            if charging_manager._platform_budget_seen and charging_manager._max_total_charge_usd is not None and row_cost > 0:
+                fit = max(fit, int(charging_manager._remaining_budget_usd() / row_cost))
+            # At least one more single-row try: without the platform's remaining
+            # budget the leftover after the short answer is unknown, and the last
+            # affordable row may still fit.
+            rows = min(rows - 1, max(fit, 1))
 
         if rows <= 0 or not ops:
             await undo(ops)

{scrapely_python-0.1.19 → scrapely_python-0.1.20/src/scrapely_python.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scrapely-python
-Version: 0.1.19
+Version: 0.1.20
 Summary: Scrapely Python SDK - A toolkit for building Actors on the Scrapely platform
 Author-email: Scrapely <support@scrape.ly>
 License-Expression: MIT AND Apache-2.0

{scrapely_python-0.1.19 → scrapely_python-0.1.20}/tests/test_charge_limit_guard.py

@@ -585,3 +585,65 @@ async def test_platform_other_spend_never_shrinks() -> None:
     api.usage = Decimal('0.10')  # a stale or lower report
     await mgr.charge('result', 1)
     assert mgr._platform_other_usd == seen
+
+
+
+# ---- multi-event retry sizing: whole rows at the FULL per-row cost ----
+
+APP_PRICES = {'app-details': Decimal('0.00015'), 'app-item': Decimal('0.00005')}  # row = $0.0002
+
+
+class RemainingMultiApi(MultiApi):
+    """MultiApi that also reports remainingBudgetNd (server #725)."""
+
+    async def post(self, path: str, json_data: Optional[Dict[str, Any]] = None, headers: Optional[Dict[str, str]] = None, **kw: Any) -> Dict[str, Any]:
+        body = await super().post(path, json_data=json_data, headers=headers, **kw)
+        if not path.endswith('/charge/void') and self.max_usd is not None:
+            body = {**body, 'remainingBudgetNd': int((self.max_usd - self.spent) * 1_000_000_000)}
+        return body
+
+
+def _app(server_max: str, remaining: bool = False):
+    api = (RemainingMultiApi if remaining else MultiApi)(APP_PRICES, Decimal(server_max))
+    return api, _manager(api, prices=dict(APP_PRICES), max_usd='1')
+
+
+@pytest.mark.asyncio
+async def test_retry_after_one_short_event_is_sized_by_the_full_row_cost() -> None:
+    """appstore 64df0a3608d74af88: details 20 -> 14 accepted; the retry must not size from 14."""
+    api, mgr = _app('0.0021')
+    store = Store()
+    r = await push_data_with_charges(mgr, [{'n': n} for n in range(20)], {'app-details': 1, 'app-item': 1}, True, store.push)
+    assert r.charged_count == 10  # $0.0021 / $0.0002 per full row
+    assert len(store.items) == 10
+    assert api.spent == Decimal('0.0020')
+    assert sorted(api.ops.values()) == [('app-details', 10), ('app-item', 10)]
+
+
+@pytest.mark.asyncio
+async def test_retry_when_several_events_come_back_short() -> None:
+    """Both events of the row run short on the way down; it still converges on whole paid rows."""
+    api, mgr = _multi('1', server_max='0.0405')  # 2.53 rows of $0.016 (1 result + 3 pages)
+    store = Store()
+    r = await push_data_with_charges(mgr, [{'n': n} for n in range(6)], {'result': 1, 'page': 3}, True, store.push)
+    assert r.charged_count == 2
+    assert len(store.items) == 2
+    assert api.spent == Decimal('0.032')
+
+
+@pytest.mark.asyncio
+async def test_last_affordable_row_is_saved_without_remaining_budget() -> None:
+    api, mgr = _app('0.0002')  # exactly one row; the short answer hides the leftover
+    store = Store()
+    r = await push_data_with_charges(mgr, [{'n': n} for n in range(5)], {'app-details': 1, 'app-item': 1}, True, store.push)
+    assert r.charged_count == 1 and len(store.items) == 1
+    assert api.spent == Decimal('0.0002')
+
+
+@pytest.mark.asyncio
+async def test_platform_remaining_budget_sizes_the_retry_exactly() -> None:
+    api, mgr = _app('0.00215', remaining=True)  # 10.75 rows
+    store = Store()
+    r = await push_data_with_charges(mgr, [{'n': n} for n in range(20)], {'app-details': 1, 'app-item': 1}, True, store.push)
+    assert r.charged_count == 10 and len(store.items) == 10
+    assert api.spent == Decimal('0.0020')