scrapely-python 0.1.17__py3-none-any.whl → 0.1.18__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
scrapely/__init__.py CHANGED
@@ -132,7 +132,7 @@ from .errors import (
132
132
  raise_for_status,
133
133
  )
134
134
 
135
- __version__ = '0.1.17'
135
+ __version__ = '0.1.18'
136
136
  __all__ = [
137
137
  # Core
138
138
  'Actor',
scrapely/actor.py CHANGED
@@ -28,6 +28,7 @@ from .charging import (
28
28
  ChargingManager,
29
29
  is_default_dataset_item_event,
30
30
  push_data_and_charge,
31
+ push_data_with_charges,
31
32
  set_active_charging_manager,
32
33
  )
33
34
  from .configuration import Configuration
@@ -781,6 +782,7 @@ class Actor(metaclass=_ActorMeta):
781
782
  charged_event_name: Optional[str] = None,
782
783
  *,
783
784
  event_name: Optional[str] = None,
785
+ charges: Optional[Dict[str, int]] = None,
784
786
  ) -> ChargeResult:
785
787
  """Push data to the default dataset, charging pay-per-event items.
786
788
 
@@ -796,6 +798,10 @@ class Actor(metaclass=_ActorMeta):
796
798
  (Apify Python SDK name; also accepted positionally).
797
799
  event_name: Deprecated alias of ``charged_event_name`` (JS ``eventName``),
798
800
  kept so existing callers keep working.
801
+ charges: Several events per row, charged atomically: ``{'result': 1, 'page': 3}``
802
+ charges 1 ``result`` and 3 ``page`` events for every row. A row is stored
803
+ only when all of its events were charged; rows the spending limit cannot
804
+ pay for in full are not stored. Cannot be combined with ``charged_event_name``.
799
805
 
800
806
  Returns:
801
807
  ChargeResult describing how many items were charged and pushed.
@@ -818,6 +824,16 @@ class Actor(metaclass=_ActorMeta):
818
824
 
819
825
  items = data if isinstance(data, list) else [data]
820
826
  charging_manager = cls.get_charging_manager()
827
+ if charges is not None:
828
+ if event_name is not None:
829
+ raise ValueError('Pass either charged_event_name or charges, not both')
830
+ return await push_data_with_charges(
831
+ charging_manager,
832
+ items,
833
+ charges,
834
+ True, # Actor.push_data always targets the default dataset
835
+ cls._default_dataset.push_data,
836
+ )
821
837
  return await push_data_and_charge(
822
838
  charging_manager,
823
839
  items,
scrapely/charging.py CHANGED
@@ -1585,6 +1585,166 @@ async def push_data_and_charge(
1585
1585
  )
1586
1586
 
1587
1587
 
1588
+ def _validate_row_charges(charges: Any) -> Dict[str, int]:
1589
+ if not isinstance(charges, dict) or not charges:
1590
+ raise ValueError('charges must be a non-empty dict of {event_name: count_per_row}')
1591
+ out: Dict[str, int] = {}
1592
+ for name, count in charges.items():
1593
+ if not isinstance(name, str) or not name:
1594
+ raise ValueError('charges keys must be event names')
1595
+ if is_default_dataset_item_event(name):
1596
+ raise ValueError(f"Cannot charge for synthetic event '{name}' manually")
1597
+ if not isinstance(count, int) or isinstance(count, bool) or count < 1:
1598
+ raise ValueError(f'charges[{name!r}] must be a positive integer (events per row)')
1599
+ out[name] = count
1600
+ return out
1601
+
1602
+
1603
+ async def push_data_with_charges(
1604
+ charging_manager: ChargingManager,
1605
+ items: Union[Dict[str, Any], List[Dict[str, Any]]],
1606
+ charges: Dict[str, int],
1607
+ is_default_dataset: bool,
1608
+ push_fn: Any,
1609
+ ) -> ChargeResult:
1610
+ """Push rows that each cost several events, atomically per row.
1611
+
1612
+ ``charges`` maps event name -> events per row, e.g. ``{'result': 1, 'page': 3}``.
1613
+ A row is stored only when every one of its events was charged: the full set
1614
+ is charged first, rows the budget cannot fully pay for are not stored, and
1615
+ any operation that came back short is voided and re-charged for the rows
1616
+ that are kept, so no row is ever stored partly paid and no event is billed
1617
+ for a row that was not stored. Unpriced events are ignored. On the default
1618
+ dataset the default-dataset-item event (when priced) is added per row.
1619
+ """
1620
+ charges = _validate_row_charges(charges)
1621
+ rows_in = items if isinstance(items, list) else [items]
1622
+ if not rows_in:
1623
+ return ChargeResult(chargeable_within_limit={}, charged_count=0, event_charge_limit_reached=False)
1624
+
1625
+ async with charging_manager._operation_lock:
1626
+ if charging_manager._closed:
1627
+ raise RuntimeError('ChargingManager is closed (actor is shutting down); charge rejected')
1628
+ voids_confirmed = await charging_manager._retry_pending_voids()
1629
+
1630
+ if charging_manager._pricing_info is None:
1631
+ await _gated_push(push_fn, rows_in)
1632
+ return ChargeResult(chargeable_within_limit={}, charged_count=0, event_charge_limit_reached=False)
1633
+
1634
+ per_row: Dict[str, int] = {
1635
+ name: n for name, n in charges.items()
1636
+ if (charging_manager._calculate_event_price(name) or Decimal('0')) > 0
1637
+ }
1638
+ if is_default_dataset:
1639
+ default_event = charging_manager._config.get('default_dataset_item_event_name') or \
1640
+ resolve_default_dataset_item_event_name(charging_manager._per_event_prices)
1641
+ if default_event and default_event not in per_row and \
1642
+ (charging_manager._calculate_event_price(default_event) or Decimal('0')) > 0:
1643
+ per_row[default_event] = 1
1644
+ if not per_row:
1645
+ # Every listed event is free: a plain (unpaid) push.
1646
+ return await _unpaid_push_locked(charging_manager, rows_in, is_default_dataset, push_fn)
1647
+ if not voids_confirmed and any(not is_default_dataset_item_event(n) for n in per_row):
1648
+ raise RuntimeError('An earlier charge for a failed dataset push could not be voided yet; retry the push')
1649
+
1650
+ # Rows the budget can pay for in full (no one-event overdraw: a row is all or nothing).
1651
+ row_cost = sum(
1652
+ (charging_manager._calculate_event_price(n) or Decimal('0')) * Decimal(c) for n, c in per_row.items()
1653
+ )
1654
+ rows = len(rows_in)
1655
+ if charging_manager._max_total_charge_usd is not None and row_cost > 0:
1656
+ remaining = charging_manager._max_total_charge_usd - charging_manager._total_charged_usd
1657
+ rows = min(rows, max(0, int(remaining / row_cost)))
1658
+
1659
+ ops: List[tuple] = [] # (event, charged, op_id, uncertain)
1660
+
1661
+ async def undo(entries: List[tuple]) -> None:
1662
+ for name, charged, op_id, uncertain in entries:
1663
+ if charged <= 0:
1664
+ continue
1665
+ if op_id is None:
1666
+ charging_manager._unwind_reservation(name, charged)
1667
+ elif uncertain:
1668
+ await charging_manager._cancel_uncertain_charge(name, op_id, charged)
1669
+ else:
1670
+ await charging_manager._void_charge(name, op_id, charged)
1671
+
1672
+ for _attempt in range(4):
1673
+ if rows <= 0:
1674
+ break
1675
+ ops = []
1676
+ short_to: Optional[int] = None
1677
+ for name, n in sorted(per_row.items(), key=lambda kv: (is_default_dataset_item_event(kv[0]), kv[0])):
1678
+ want = n * rows
1679
+ charging_manager._take_last_accepted_operation_id()
1680
+ try:
1681
+ r = await charging_manager._charge_exclusive(name, want)
1682
+ op_id = charging_manager._take_last_accepted_operation_id()
1683
+ ops.append((name, r.charged_count, op_id, False))
1684
+ got = r.charged_count
1685
+ except ChargeUncertainError as unconfirmed:
1686
+ # Not confirmed: the intent stays pending and is re-posted at exit,
1687
+ # so it counts as accepted (items stored, billed once).
1688
+ ops.append((name, unconfirmed.count, unconfirmed.operation_id, True))
1689
+ got = unconfirmed.count
1690
+ if got < want:
1691
+ short_to = got // n
1692
+ break
1693
+ if short_to is None:
1694
+ break
1695
+ # Some event came back short: void this attempt in full and retry with
1696
+ # the rows that event can pay for (a void restores the budget).
1697
+ await undo(ops)
1698
+ ops = []
1699
+ rows = short_to # got < n * rows, so short_to < rows
1700
+
1701
+ if rows <= 0 or not ops:
1702
+ await undo(ops)
1703
+ charging_manager._mark_charge_limit_hit()
1704
+ if is_default_dataset:
1705
+ charging_manager._note_refused_default_dataset_items(len(rows_in))
1706
+ return ChargeResult(
1707
+ chargeable_within_limit=charging_manager.compute_chargeable(),
1708
+ charged_count=0,
1709
+ event_charge_limit_reached=True,
1710
+ )
1711
+
1712
+ kept = rows_in[:rows]
1713
+ if rows < len(rows_in):
1714
+ charging_manager._mark_charge_limit_hit()
1715
+ if is_default_dataset:
1716
+ charging_manager._note_refused_default_dataset_items(len(rows_in) - rows)
1717
+ try:
1718
+ await _gated_push(push_fn, kept)
1719
+ except Exception as push_err:
1720
+ logger.error('Dataset push failed after charges accepted (rows=%s): %s; voiding the charges', rows, push_err)
1721
+ if _push_certainly_failed(push_err):
1722
+ await undo(ops)
1723
+ raise
1724
+ return ChargeResult(
1725
+ chargeable_within_limit=charging_manager.compute_chargeable(),
1726
+ charged_count=rows,
1727
+ event_charge_limit_reached=rows < len(rows_in) or charging_manager.is_charge_limit_reached(),
1728
+ )
1729
+
1730
+
1731
+ async def _unpaid_push_locked(
1732
+ charging_manager: ChargingManager, rows: List[Dict[str, Any]], is_default_dataset: bool, push_fn: Any
1733
+ ) -> ChargeResult:
1734
+ """A push whose rows carry no priced event (caller holds the operation lock)."""
1735
+ past_limit = charging_manager.is_charge_limit_reached()
1736
+ allowed = charging_manager._take_unpaid_rows(len(rows)) if is_default_dataset else len(rows)
1737
+ if allowed < len(rows):
1738
+ charging_manager._note_refused_default_dataset_items(len(rows) - allowed)
1739
+ if allowed > 0:
1740
+ await _gated_push(push_fn, rows[:allowed])
1741
+ return ChargeResult(
1742
+ chargeable_within_limit=charging_manager.compute_chargeable(),
1743
+ charged_count=0,
1744
+ event_charge_limit_reached=past_limit,
1745
+ )
1746
+
1747
+
1588
1748
  def _push_is_paid_per_item(
1589
1749
  charging_manager: ChargingManager, event_name: Optional[str], resolved_event: Optional[str]
1590
1750
  ) -> bool:
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: scrapely-python
3
- Version: 0.1.17
3
+ Version: 0.1.18
4
4
  Summary: Scrapely Python SDK - A toolkit for building Actors on the Scrapely platform
5
5
  Author-email: Scrapely <support@scrape.ly>
6
6
  License-Expression: MIT AND Apache-2.0
@@ -1,9 +1,9 @@
1
- scrapely/__init__.py,sha256=TqM8pFfiE1_j1M69_2BOtRVAy1jMK6HzvKNpmYtNnuo,5746
1
+ scrapely/__init__.py,sha256=EHGVITjobhdZu3nhkmBmECrlftUMCsnDyLAIxdnZ_7A,5746
2
2
  scrapely/_browser_locations.py,sha256=8QVVnkms2bDn3_LDYgYDR59MvFx8757l_JYnZ7xb3oI,145269
3
- scrapely/actor.py,sha256=1MXGyHql5sRfh15HMRY8vAp_TNr_yhavDALlHBcjLO0,80810
3
+ scrapely/actor.py,sha256=Lsb4_14RCGQy-EWACnqijo7Hz5xvPKSDZcu4B8_ZrHc,81685
4
4
  scrapely/browser_configuration.py,sha256=jAZapNSHLFW05B96df4wNBtcyiiU9DA3ml0yTL1FwfA,5770
5
5
  scrapely/cdp_retry.py,sha256=hdTJEAIKhApM7IOk1DrQEl8NoflSJSmqd4RbNJ2IsgI,8540
6
- scrapely/charging.py,sha256=ImnjF4RfzLbpRp3JZIl2RM-TfHp22p0Xzt0NSVs2M1c,71477
6
+ scrapely/charging.py,sha256=maRgijO8vM8mCb_S9lnvF99zNR_G8sB3LsubU1r6t6M,79326
7
7
  scrapely/charging_client.py,sha256=-4q-Ww7gu9EZXpM9QIn1Vd4vwAXmgL_WEKqaQIx3ViA,3961
8
8
  scrapely/configuration.py,sha256=Of-lKm7hoPGUQO8d08kV7fAy2co9QTfb8Of_8ds5PSA,17028
9
9
  scrapely/errors.py,sha256=DCAlVvE3jToZuPEG2HVDL7i8D8hLboomQ5JpSOcsXoQ,4107
@@ -39,10 +39,10 @@ scrapely/storage/_file_system/_key_value_store_client.py,sha256=4Ipgbh4qEqAkwQJ9
39
39
  scrapely/storage/_file_system/_storage_client.py,sha256=aaGJ4P0QAtIgEMl6yHpuobeWFD5M4rMC5HwhOOvuGuo,1794
40
40
  scrapely/storage_clients/__init__.py,sha256=cs1YX7JV2_GKBXeiiJgkQFali08b5_y_jH6Yw7iROAA,656
41
41
  scrapely/storages/__init__.py,sha256=4PAELhJNZ3exJ7Uouicn-bCQ58NafKjhLBixQ8ATQc4,173
42
- scrapely_python-0.1.17.dist-info/licenses/LICENSE,sha256=pQGY0XwjIYh-a2HfawxHq1en53bXkFkDj2Q70jKWHr4,1088
43
- scrapely_python-0.1.17.dist-info/licenses/LICENSE-APACHE,sha256=N8m0FzMq8i4UbTgXZvKEZJSD-27OF8pWZYS_atz4kTU,11355
44
- scrapely_python-0.1.17.dist-info/licenses/NOTICE,sha256=DEsNrw6UdSIqdYYqMWVFgJnyt6jUivTD7iFU6YM_e1A,1122
45
- scrapely_python-0.1.17.dist-info/METADATA,sha256=CUcmUdVmdGoDotCHQvZUa2eYIYKugeJZvsbA9tWrVRg,3487
46
- scrapely_python-0.1.17.dist-info/WHEEL,sha256=YVMoNqKzERt-wjUZwJ33xBGAwnFl-4cqbYkTtWa4itE,91
47
- scrapely_python-0.1.17.dist-info/top_level.txt,sha256=T4B9UMrs4q99QSAFyD1zdXhl40KnLhBKqrDkPhOTsmY,9
48
- scrapely_python-0.1.17.dist-info/RECORD,,
42
+ scrapely_python-0.1.18.dist-info/licenses/LICENSE,sha256=pQGY0XwjIYh-a2HfawxHq1en53bXkFkDj2Q70jKWHr4,1088
43
+ scrapely_python-0.1.18.dist-info/licenses/LICENSE-APACHE,sha256=N8m0FzMq8i4UbTgXZvKEZJSD-27OF8pWZYS_atz4kTU,11355
44
+ scrapely_python-0.1.18.dist-info/licenses/NOTICE,sha256=DEsNrw6UdSIqdYYqMWVFgJnyt6jUivTD7iFU6YM_e1A,1122
45
+ scrapely_python-0.1.18.dist-info/METADATA,sha256=3znh1pGrKvP0jNRAftMgiGSKuBG1c3g8UWJtgMYI95g,3487
46
+ scrapely_python-0.1.18.dist-info/WHEEL,sha256=YVMoNqKzERt-wjUZwJ33xBGAwnFl-4cqbYkTtWa4itE,91
47
+ scrapely_python-0.1.18.dist-info/top_level.txt,sha256=T4B9UMrs4q99QSAFyD1zdXhl40KnLhBKqrDkPhOTsmY,9
48
+ scrapely_python-0.1.18.dist-info/RECORD,,