aio-sf 0.1.0b8__py3-none-any.whl → 0.1.0b10__py3-none-any.whl

aio_sf/api/collections/__init__.py

@@ -1,7 +1,7 @@
 """Salesforce Collections API module."""
 
 from .client import CollectionsAPI
-from .batch import ProgressInfo, ProgressCallback
+from .batch import ResultInfo, ResultCallback
 from .retry import ShouldRetryCallback, default_should_retry
 from .types import (
     CollectionError,
@@ -14,8 +14,8 @@ from .types import (
 
 __all__ = [
     "CollectionsAPI",
-    "ProgressInfo",
-    "ProgressCallback",
+    "ResultInfo",
+    "ResultCallback",
     "ShouldRetryCallback",
     "default_should_retry",
     "CollectionError",

aio_sf/api/collections/batch.py

@@ -2,7 +2,7 @@
 
 import asyncio
 import logging
-from typing import Any, Awaitable, Callable, List, Optional, TypedDict, Union
+from typing import Any, Awaitable, Callable, Dict, List, Optional, TypedDict, Union
 
 from .retry import (
     RecordWithAttempt,
@@ -17,19 +17,25 @@ from .types import CollectionResult
 logger = logging.getLogger(__name__)
 
 
-class ProgressInfo(TypedDict):
-    """Progress information for batch operations."""
+class ResultInfo(TypedDict):
+    """Result information provided after each batch completes."""
 
-    batch_index: int  # Current batch number (0-indexed)
-    total_batches: int  # Total number of batches
-    records_processed: int  # Number of records processed so far
-    total_records: int  # Total number of records to process
-    batch_size: int  # Size of the current batch
-    retry_count: int  # Number of records being retried in this operation
+    successes: List[CollectionResult]  # Successful results from this batch
+    errors: List[
+        CollectionResult
+    ]  # Failed results from this batch (API and HTTP errors)
+    total_records: int  # Total records being processed
+    records_completed: int  # Records finished (succeeded or failed permanently)
+    records_succeeded: int  # Records that succeeded so far
+    records_failed: int  # Records that failed permanently so far
+    records_pending: int  # Records still being retried
+    current_attempt: int  # Current retry attempt number (1-indexed)
+    current_batch_size: int  # Batch size for current attempt
+    current_concurrency: int  # Concurrency level for current attempt
 
 
-# Type alias for progress callback
-ProgressCallback = Callable[[ProgressInfo], Awaitable[None]]
+# Type alias for result callback
+ResultCallback = Callable[[ResultInfo], Awaitable[None]]
 
 
 def split_into_batches(
@@ -63,7 +69,9 @@ async def process_batches_concurrently(
     operation_func,
     max_concurrent_batches: int,
     total_records: int,
-    on_batch_complete: Optional[ProgressCallback] = None,
+    on_result: Optional[ResultCallback] = None,
+    progress_state: Optional[Dict[str, int]] = None,
+    final_results: Optional[List] = None,
     *args,
     **kwargs,
 ) -> List[Any]:
@@ -77,7 +85,8 @@ async def process_batches_concurrently(
     :param operation_func: Function to call for each batch
     :param max_concurrent_batches: Maximum number of concurrent batch operations
     :param total_records: Total number of records being processed
-    :param on_batch_complete: Optional callback invoked after each batch completes
+    :param on_result: Optional callback invoked after each batch completes with results
+    :param progress_state: Dict with progress state (to include in callback)
     :param args: Additional positional arguments for operation_func
     :param kwargs: Additional keyword arguments for operation_func
     :returns: List of results from all batches in the same order as input
@@ -87,12 +96,9 @@ async def process_batches_concurrently(
         raise ValueError("max_concurrent_batches must be greater than 0")
 
     semaphore = asyncio.Semaphore(max_concurrent_batches)
-    total_batches = len(batches)
-    callback_lock = asyncio.Lock() if on_batch_complete else None
-    records_processed = 0
+    callback_lock = asyncio.Lock() if on_result else None
 
     async def process_batch_with_semaphore(batch_index: int, batch):
-        nonlocal records_processed
         async with semaphore:
             try:
                 result = await operation_func(batch, *args, **kwargs)
@@ -103,20 +109,52 @@ async def process_batches_concurrently(
                 )
                 result = [e for _ in range(len(batch))]
 
-            # Invoke progress callback if provided
-            if on_batch_complete and callback_lock:
-                batch_size = len(batch)
+            # Invoke callback if provided, with results and progress state
+            if on_result and callback_lock and progress_state:
                 async with callback_lock:
-                    records_processed += batch_size
-                    progress_info: ProgressInfo = {
-                        "batch_index": batch_index,
-                        "total_batches": total_batches,
-                        "records_processed": records_processed,
-                        "total_records": total_records,
-                        "batch_size": batch_size,
-                        "retry_count": 0,  # Set by wrapper in process_with_retries
+                    # Split results into successes and errors
+                    successes: List[CollectionResult] = []
+                    errors: List[CollectionResult] = []
+
+                    for item in result:
+                        # Convert exceptions to CollectionResult format
+                        if isinstance(item, Exception):
+                            error_result = convert_exception_to_result(item)
+                            errors.append(error_result)
+                        elif item.get("success", False):
+                            successes.append(item)
+                        else:
+                            errors.append(item)
+
+                    # Compute current counts dynamically from final_results
+                    if final_results:
+                        records_succeeded = sum(
+                            1
+                            for r in final_results
+                            if r is not None and r.get("success", False)
+                        )
+                        records_failed = sum(
+                            1
+                            for r in final_results
+                            if r is not None and not r.get("success", False)
+                        )
+                        records_completed = records_succeeded + records_failed
+                    else:
+                        records_succeeded = records_failed = records_completed = 0
+
+                    result_info: ResultInfo = {
+                        "successes": successes,
+                        "errors": errors,
+                        "total_records": progress_state["total_records"],
+                        "records_completed": records_completed,
+                        "records_succeeded": records_succeeded,
+                        "records_failed": records_failed,
+                        "records_pending": progress_state["records_pending"],
+                        "current_attempt": progress_state["current_attempt"],
+                        "current_batch_size": progress_state["current_batch_size"],
+                        "current_concurrency": progress_state["current_concurrency"],
                     }
-                    await on_batch_complete(progress_info)
+                    await on_result(result_info)
 
             return result
 
@@ -140,7 +178,7 @@ async def process_with_retries(
     max_attempts: int,
     should_retry_callback: Optional[ShouldRetryCallback],
     max_concurrent_batches: Union[int, List[int]],
-    on_batch_complete: Optional[ProgressCallback],
+    on_result: Optional[ResultCallback],
     max_limit: int,
     *args,
     **kwargs,
@@ -154,7 +192,7 @@ async def process_with_retries(
     :param max_attempts: Maximum number of attempts per record
     :param should_retry_callback: Optional callback to determine if record should be retried
     :param max_concurrent_batches: Maximum concurrent batches (int or list of ints per attempt)
-    :param on_batch_complete: Progress callback
+    :param on_result: Callback invoked after each batch completes with results and progress
     :param max_limit: Maximum batch size limit for the operation
     :param args: Additional args for operation_func
     :param kwargs: Additional kwargs for operation_func
@@ -163,9 +201,21 @@ async def process_with_retries(
     # Initialize result array with None placeholders
     max_index = max(r.original_index for r in records_with_attempts)
     final_results: List[Optional[CollectionResult]] = [None] * (max_index + 1)
+    total_records_count = max_index + 1
+
+    # Initialize progress state
+    progress_state = {
+        "total_records": total_records_count,
+        "records_completed": 0,
+        "records_succeeded": 0,
+        "records_failed": 0,
+        "records_pending": total_records_count,
+        "current_attempt": 1,
+        "current_batch_size": 0,
+        "current_concurrency": 0,
+    }
 
     current_records = records_with_attempts
-    total_retry_count = 0
 
     while current_records:
         current_attempt = current_records[0].attempt
@@ -176,6 +226,11 @@ async def process_with_retries(
             current_attempt, max_concurrent_batches
         )
 
+        # Update progress state for current attempt
+        progress_state["current_attempt"] = current_attempt
+        progress_state["current_batch_size"] = current_batch_size
+        progress_state["current_concurrency"] = current_concurrency
+
         logger.debug(
             f"Processing {len(current_records)} records on attempt {current_attempt} "
             f"with batch_size={current_batch_size}, concurrency={current_concurrency}"
@@ -185,23 +240,15 @@ async def process_with_retries(
         records_to_process = [r.record for r in current_records]
         batches = split_into_batches(records_to_process, current_batch_size, max_limit)
 
-        # Wrap progress callback to include retry count
-        wrapped_callback = None
-        if on_batch_complete:
-
-            async def progress_wrapper(progress: ProgressInfo):
-                progress["retry_count"] = total_retry_count
-                await on_batch_complete(progress)
-
-            wrapped_callback = progress_wrapper
-
-        # Process batches with current concurrency level
+        # Process batches - callback will be invoked for each batch with results
         batch_results = await process_batches_concurrently(
             batches,
             operation_func,
             current_concurrency,
             len(records_to_process),
-            wrapped_callback,
+            on_result,
+            progress_state,
+            final_results,
             *args,
             **kwargs,
         )
@@ -215,12 +262,24 @@ async def process_with_retries(
             final_results,
         )
 
+        # Update progress state based on results for next iteration
+        records_succeeded = sum(
+            1 for r in final_results if r is not None and r.get("success", False)
+        )
+        records_failed = sum(
+            1 for r in final_results if r is not None and not r.get("success", False)
+        )
+
+        progress_state["records_completed"] = records_succeeded + records_failed
+        progress_state["records_succeeded"] = records_succeeded
+        progress_state["records_failed"] = records_failed
+        progress_state["records_pending"] = len(records_to_retry)
+
         if records_to_retry:
             logger.info(
                 f"Retrying {len(records_to_retry)} failed records "
                 f"(attempt {records_to_retry[0].attempt})"
             )
-            total_retry_count += len(records_to_retry)
 
         current_records = records_to_retry
 

aio_sf/api/collections/client.py

@@ -8,7 +8,7 @@ from typing import TYPE_CHECKING
 if TYPE_CHECKING:
     from ..client import SalesforceClient
 
-from .batch import ProgressCallback, process_with_retries
+from .batch import ResultCallback, process_with_retries
 from .records import (
     detect_record_type_and_sobject,
     prepare_records,
@@ -92,7 +92,7 @@ class CollectionsAPI:
         batch_size: Union[int, List[int]] = 200,
         max_concurrent_batches: Union[int, List[int]] = 5,
         api_version: Optional[str] = None,
-        on_batch_complete: Optional[ProgressCallback] = None,
+        on_result: Optional[ResultCallback] = None,
         max_attempts: int = 1,
         should_retry: Optional[ShouldRetryCallback] = None,
     ) -> CollectionInsertResponse:
@@ -109,7 +109,7 @@ class CollectionsAPI:
         :param batch_size: Batch size (int for same size, or list of ints per attempt). Max 200.
         :param max_concurrent_batches: Maximum number of concurrent batch operations
         :param api_version: API version to use
-        :param on_batch_complete: Optional async callback invoked after each batch completes
+        :param on_result: Optional async callback invoked after each batch completes with results
         :param max_attempts: Maximum number of attempts per record (default: 1, no retries)
         :param should_retry: Optional callback to determine if a failed record should be retried
         :returns: List of results for each record, in same order as input
@@ -139,7 +139,7 @@ class CollectionsAPI:
             max_attempts,
             should_retry,
             max_concurrent_batches,
-            on_batch_complete,
+            on_result,
             self.MAX_RECORDS_INSERT,
             actual_sobject_type,
             all_or_none,
@@ -200,7 +200,7 @@ class CollectionsAPI:
         batch_size: Union[int, List[int]] = 200,
         max_concurrent_batches: Union[int, List[int]] = 5,
         api_version: Optional[str] = None,
-        on_batch_complete: Optional[ProgressCallback] = None,
+        on_result: Optional[ResultCallback] = None,
         max_attempts: int = 1,
         should_retry: Optional[ShouldRetryCallback] = None,
     ) -> CollectionUpdateResponse:
@@ -217,7 +217,7 @@ class CollectionsAPI:
         :param batch_size: Batch size (int for same size, or list of ints per attempt). Max 200.
         :param max_concurrent_batches: Maximum number of concurrent batch operations
         :param api_version: API version to use
-        :param on_batch_complete: Optional async callback invoked after each batch completes
+        :param on_result: Optional async callback invoked after each batch completes with results
         :param max_attempts: Maximum number of attempts per record (default: 1, no retries)
         :param should_retry: Optional callback to determine if a failed record should be retried
         :returns: List of results for each record, in same order as input
@@ -247,7 +247,7 @@ class CollectionsAPI:
             max_attempts,
             should_retry,
             max_concurrent_batches,
-            on_batch_complete,
+            on_result,
             self.MAX_RECORDS_UPDATE,
             actual_sobject_type,
             all_or_none,
@@ -313,7 +313,7 @@ class CollectionsAPI:
         batch_size: Union[int, List[int]] = 200,
         max_concurrent_batches: Union[int, List[int]] = 5,
         api_version: Optional[str] = None,
-        on_batch_complete: Optional[ProgressCallback] = None,
+        on_result: Optional[ResultCallback] = None,
         max_attempts: int = 1,
         should_retry: Optional[ShouldRetryCallback] = None,
     ) -> CollectionUpsertResponse:
@@ -331,7 +331,7 @@ class CollectionsAPI:
         :param batch_size: Batch size (int for same size, or list of ints per attempt). Max 200.
         :param max_concurrent_batches: Maximum number of concurrent batch operations
         :param api_version: API version to use
-        :param on_batch_complete: Optional async callback invoked after each batch completes
+        :param on_result: Optional async callback invoked after each batch completes with results
         :param max_attempts: Maximum number of attempts per record (default: 1, no retries)
         :param should_retry: Optional callback to determine if a failed record should be retried
         :returns: List of results for each record, in same order as input
@@ -361,7 +361,7 @@ class CollectionsAPI:
             max_attempts,
             should_retry,
             max_concurrent_batches,
-            on_batch_complete,
+            on_result,
             self.MAX_RECORDS_UPSERT,
             external_id_field,
             actual_sobject_type,
@@ -411,7 +411,7 @@ class CollectionsAPI:
         batch_size: Union[int, List[int]] = 200,
         max_concurrent_batches: Union[int, List[int]] = 5,
         api_version: Optional[str] = None,
-        on_batch_complete: Optional[ProgressCallback] = None,
+        on_result: Optional[ResultCallback] = None,
         max_attempts: int = 1,
         should_retry: Optional[ShouldRetryCallback] = None,
     ) -> CollectionDeleteResponse:
@@ -427,7 +427,7 @@ class CollectionsAPI:
         :param batch_size: Batch size (int for same size, or list of ints per attempt). Max 200.
         :param max_concurrent_batches: Maximum number of concurrent batch operations
         :param api_version: API version to use
-        :param on_batch_complete: Optional async callback invoked after each batch completes
+        :param on_result: Optional async callback invoked after each batch completes with results
         :param max_attempts: Maximum number of attempts per record (default: 1, no retries)
         :param should_retry: Optional callback to determine if a failed record should be retried
         :returns: List of results for each record, in same order as input
@@ -453,7 +453,7 @@ class CollectionsAPI:
             max_attempts,
             should_retry,
             max_concurrent_batches,
-            on_batch_complete,
+            on_result,
             self.MAX_RECORDS_DELETE,
             all_or_none,
             api_version,

{aio_sf-0.1.0b8.dist-info → aio_sf-0.1.0b10.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aio-sf
-Version: 0.1.0b8
+Version: 0.1.0b10
 Summary: Async Salesforce library for Python
 Project-URL: Homepage, https://github.com/callawaycloud/aio-salesforce
 Project-URL: Repository, https://github.com/callawaycloud/aio-salesforce
@@ -153,24 +153,44 @@ asyncio.run(main())
 
 ### Collections API - Batch Operations
 
-Efficiently handle bulk operations with automatic batching and concurrency:
+Bulk operations (insert, update, upsert, delete) with automatic batching and concurrency.
 
+**Basic Usage:**
 ```python
 async with SalesforceClient(auth_strategy=auth) as sf:
     records = [{"Name": f"Account {i}"} for i in range(1000)]
     
-    # Insert with automatic batching and parallel processing
-    results = await sf.collections.insert(
-        records, sobject_type="Account", 
-        batch_size=200, max_concurrent_batches=5
-    )
-    
+    results = await sf.collections.insert(records, sobject_type="Account")
     # Also: update(), upsert(), delete()
 ```
 
-**Features:** Automatic batching • Concurrent processing • Order preservation • Smart retries • Progress tracking
+**Advanced - With Retries, Concurrency Scaling, and Progress:**
+```python
+from aio_sf.api.collections import ResultInfo
+
+async def on_result(info: ResultInfo):
+    # Called after each batch completes with successes and errors split
+    print(
+        f"Batch: {len(info['successes'])} succeeded, {len(info['errors'])} failed | "
+        f"Attempt {info['current_attempt']}, "
+        f"Overall: {info['records_succeeded']} OK, {info['records_failed']} failed, "
+        f"{info['records_pending']} pending"
+    )
+    # Inspect errors (includes both API errors and HTTP failures)
+    for error in info['errors']:
+        print(f"  Error: {error['errors']}")
+
+async with SalesforceClient(auth_strategy=auth) as sf:
+    results = await sf.collections.insert(
+        records=records,
+        sobject_type="Account",
+        batch_size=[200, 100, 25],         # Shrink batch size on retry
+        max_concurrent_batches=[5, 3, 1],  # Reduce concurrency on retry
+        max_attempts=5,                    # Retry up to 5 times
+        on_result=on_result,               # Callback with results
+    )
+```
 
-See [RETRY_GUIDE.md](RETRY_GUIDE.md) for advanced retry strategies, progress tracking, and custom error handling.
 
 ## Exporter
 

{aio_sf-0.1.0b8.dist-info → aio_sf-0.1.0b10.dist-info}/RECORD

@@ -11,9 +11,9 @@ aio_sf/api/auth/static_token.py,sha256=xAXpLkzRh6hDEiaKeR9N3Tk-0PuAAZKKU9Ivs0W81
 aio_sf/api/bulk_v2/__init__.py,sha256=TxNM9dFmRX5k57Wj_JnposeEqqd9xcJx78wI-d6VH0o,315
 aio_sf/api/bulk_v2/client.py,sha256=t2vl8bIIvSrbg5T1mFDm7O4vq0GyDihfkVHamGYnZ7I,6817
 aio_sf/api/bulk_v2/types.py,sha256=18TN_VMisKJVCvo39q36moYXQhGLRUs73CKWIxBeAcs,1471
-aio_sf/api/collections/__init__.py,sha256=kE9NTBr-iBgQy-ZQAkJD92m7wYk7FQQ-K0gcKnvUoIg,685
-aio_sf/api/collections/batch.py,sha256=rA3mCtKYIKgKMhcbXjvkwN1wDzFFgGS7C0L8KpfTiaI,11758
-aio_sf/api/collections/client.py,sha256=IhIn6D-PpunSeztueK4Rs7AIHp57vhNyH3CuvcXnR6c,19098
+aio_sf/api/collections/__init__.py,sha256=Loy9PVf7VsFqs9J2s5n6DO1txy6H5xdHU422CnuMkaM,677
+aio_sf/api/collections/batch.py,sha256=mRhWBNIvSfgw9tk9PP1DzUmgTdEwB-bThKN8195luH8,14755
+aio_sf/api/collections/client.py,sha256=gR1z3Q5K6iZvcYzJJ6n0azoEdNYp7UdefHMV8jb2RZY,19044
 aio_sf/api/collections/records.py,sha256=UwI4mjoMkJPBCYbsNUmj5HTBjbGdcaMHT6UiJ4EmcUA,5371
 aio_sf/api/collections/retry.py,sha256=RXTpMhWyeYsXNiojgEYIVnGtg8S0iYMxgfWgX3APahE,4669
 aio_sf/api/collections/types.py,sha256=vWjVVySjV8eRFgdGHiP8QiyaZZ5PtxMl0PyFEZ6yNtY,762
@@ -26,7 +26,7 @@ aio_sf/api/query/types.py,sha256=Wfk75kJpNDCGpTHonCbzjWvayy8guA3eyZp3hE7nBt0,845
 aio_sf/exporter/__init__.py,sha256=waTegrvw_SvJzREAWD4twSDldSL-HfvhLTLLT1o765o,771
 aio_sf/exporter/bulk_export.py,sha256=2GtiwXChf7dq7dByGLPDhIJJg-yq9eyoE57H4Ekqaus,13169
 aio_sf/exporter/parquet_writer.py,sha256=jGiLooxyaqciSDUbXj5F_4uWoR_YrQaB-PrDfRuXR3Y,14495
-aio_sf-0.1.0b8.dist-info/METADATA,sha256=cLcujx9tn_Ce3FM_S97Q_srpu5B4cTM-dpMpOrpxpZg,7517
-aio_sf-0.1.0b8.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-aio_sf-0.1.0b8.dist-info/licenses/LICENSE,sha256=gu0Cbpiqs-vX7YgJJhGI1jH1mHup3dZMrZc-gmpEG60,1071
-aio_sf-0.1.0b8.dist-info/RECORD,,
+aio_sf-0.1.0b10.dist-info/METADATA,sha256=jC8DcA0rPFavqImBvSZhqfWNGwVEboglPqmhbXu_J9Y,8282
+aio_sf-0.1.0b10.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+aio_sf-0.1.0b10.dist-info/licenses/LICENSE,sha256=gu0Cbpiqs-vX7YgJJhGI1jH1mHup3dZMrZc-gmpEG60,1071
+aio_sf-0.1.0b10.dist-info/RECORD,,