nv-ingest-api 2025.2.23.dev20250223233317__tar.gz → 2025.2.24.dev20250224233316__tar.gz

{nv_ingest_api-2025.2.23.dev20250223233317/src/nv_ingest_api.egg-info → nv_ingest_api-2025.2.24.dev20250224233316}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: nv-ingest-api
-Version: 2025.2.23.dev20250223233317
+Version: 2025.2.24.dev20250224233316
 Summary: Python module with core document ingestion functions.
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest_api-2025.2.23.dev20250223233317 → nv_ingest_api-2025.2.24.dev20250224233316}/src/nv_ingest_api/primitives/control_message_task.py

@@ -1,10 +1,12 @@
+from uuid import UUID
+
 from pydantic import BaseModel, Field, ConfigDict
-from typing import Any, Dict
+from typing import Any, Dict, Union
 
 
 class ControlMessageTask(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
-    name: str
-    id: str
+    type: str
+    id: Union[str, UUID]
     properties: Dict[str, Any] = Field(default_factory=dict)

{nv_ingest_api-2025.2.23.dev20250223233317 → nv_ingest_api-2025.2.24.dev20250224233316}/src/nv_ingest_api/primitives/ingest_control_message.py

@@ -12,6 +12,46 @@ from nv_ingest_api.primitives.control_message_task import ControlMessageTask
 logger = logging.getLogger(__name__)
 
 
+def remove_task_by_type(ctrl_msg, task: str):
+    """
+    Remove a task from the control message by matching its type.
+
+    This function iterates over the tasks in the control message, and if it finds a task
+    whose type matches the provided task string, it removes that task (using its unique id)
+    and returns the task's properties.
+
+    Parameters
+    ----------
+    ctrl_msg : IngestControlMessage
+        The control message from which to remove the task.
+    task : str
+        The task type to remove.
+
+    Returns
+    -------
+    dict
+        The properties of the removed task.
+
+    Raises
+    ------
+    ValueError
+        If no task with the given type is found.
+    """
+    task_obj = None
+    for t in ctrl_msg.get_tasks():
+        if t.type == task:
+            task_obj = t
+            break
+
+    if task_obj is None:
+        err_msg = f"process_control_message: Task '{task}' not found in control message."
+        logger.error(err_msg)
+        raise ValueError(err_msg)
+
+    removed_task = ctrl_msg.remove_task(task_obj.id)
+    return removed_task.properties
+
+
 class IngestControlMessage:
     """
     A control message class for ingesting tasks and managing associated metadata,
@@ -53,14 +93,18 @@ class IngestControlMessage:
         """
         return task_id in self._tasks
 
-    def remove_task(self, task_id: str) -> None:
+    def remove_task(self, task_id: str) -> ControlMessageTask:
         """
         Remove a task from the control message. Logs a warning if the task does not exist.
         """
         if task_id in self._tasks:
+            _task = self._tasks[task_id]
+
             del self._tasks[task_id]
+
+            return _task
         else:
-            logger.warning(f"Attempted to remove non-existent task with id: {task_id}")
+            raise RuntimeError(f"Attempted to remove non-existent task with id: {task_id}")
 
     def config(self, config: Dict[str, Any] = None) -> Dict[str, Any]:
         """
@@ -154,7 +198,8 @@ class IngestControlMessage:
         Retrieve timestamps whose keys match the regex filter.
         """
         pattern = re.compile(regex_filter)
-        return {key: ts for key, ts in self._timestamps.items() if pattern.search(key)}
+        timestamps_snapshot = self._timestamps.copy()
+        return {key: ts for key, ts in timestamps_snapshot.items() if pattern.search(key)}
 
     def get_timestamp(self, key: str, fail_if_nonexist: bool = False) -> datetime:
         """
@@ -188,12 +233,14 @@ class IngestControlMessage:
         """
         if isinstance(timestamp, datetime):
             self._timestamps[key] = timestamp
+
         elif isinstance(timestamp, str):
             try:
                 dt = datetime.fromisoformat(timestamp)
                 self._timestamps[key] = dt
             except ValueError as e:
                 raise ValueError(f"Invalid timestamp format: {timestamp}") from e
+
         else:
             raise ValueError("timestamp must be a datetime object or ISO format string")
 

{nv_ingest_api-2025.2.23.dev20250223233317 → nv_ingest_api-2025.2.24.dev20250224233316/src/nv_ingest_api.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: nv-ingest-api
-Version: 2025.2.23.dev20250223233317
+Version: 2025.2.24.dev20250224233316
 Summary: Python module with core document ingestion functions.
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License