endoreg-db 0.4.5__py3-none-any.whl → 0.8.6.3__py3-none-any.whl

endoreg_db/models/media/video/pipe_1.py

@@ -0,0 +1,213 @@
+import logging
+from typing import TYPE_CHECKING, Dict, List, Optional, Tuple
+
+from django.db import transaction
+
+from endoreg_db.helpers.download_segmentation_model import download_segmentation_model
+
+# Added imports
+
+# Configure logging
+logger = logging.getLogger(__name__)  # Changed from "video_file"
+
+if TYPE_CHECKING:
+    from endoreg_db.models import VideoFile
+
+    # --- Pipeline 1 ---
+
+
+def _pipe_1(
+    video_file: "VideoFile",
+    model_name: str,
+    model_meta_version: Optional[int] = None,
+    delete_frames_after: bool = False,
+    ocr_frame_fraction: float = 0.001,
+    ocr_cap: int = 10,
+    smooth_window_size_s: int = 1,
+    binarize_threshold: float = 0.5,
+    test_run: bool = False,
+    n_test_frames: int = 10,
+) -> bool:
+    """
+    Pipeline 1: Extract frames, text, predict, create segments, optionally delete frames.
+    """
+    success = False  # Initialize success flag
+    from endoreg_db.models import AiModel, LabelVideoSegment
+
+    from ...metadata import ModelMeta, VideoPredictionMeta
+    from .video_file_segments import _convert_sequences_to_db_segments  # Added import
+
+    video_file.refresh_from_db()
+    video_file.update_video_meta()
+
+    logger.info(f"Starting Pipe 1 for video {video_file.uuid}")
+    try:
+        # 1. Heavy I/O operations outside the transaction block
+        logger.info("Pipe 1: Extracting frames...")
+        video_file.extract_frames(overwrite=False)  # Avoid overwriting if already extracted
+
+        logger.info("Pipe 1: Extracting text metadata...")
+        video_file.update_text_metadata(ocr_frame_fraction=ocr_frame_fraction, cap=ocr_cap, overwrite=False)
+        with transaction.atomic():
+            state = video_file.get_or_create_state()
+            if not state.frames_extracted:
+                logger.error("Pipe 1 failed: Frame extraction did not complete successfully.")
+                return False
+
+            # 3. Perform Initial Prediction
+            logger.info(f"Pipe 1: Performing prediction with model '{model_name}'...")
+            try:
+                ai_model_obj = AiModel.objects.get(name=model_name)
+                if model_meta_version is not None:
+                    model_meta = ai_model_obj.metadata_versions.get(version=model_meta_version)
+                else:
+                    model_meta = ai_model_obj.get_latest_version()
+            except AiModel.DoesNotExist:
+                logger.error(f"Pipe 1 failed: Model '{model_name}' not found.")
+                try:
+                    model_name = download_segmentation_model()
+                    ai_model_obj = AiModel.objects.get(name=model_name)
+                    if model_meta_version is not None:
+                        model_meta = ai_model_obj.metadata_versions.get(version=model_meta_version)
+                    else:
+                        model_meta = ai_model_obj.get_latest_version()
+                except AiModel.DoesNotExist:
+                    logger.error(f"Pipe 1 failed: Model '{model_name}' not found.")
+                    return False
+
+            except ModelMeta.DoesNotExist:
+                try:
+                    model_name = download_segmentation_model()
+                    ai_model_obj = AiModel.objects.get(name=model_name)
+                    if model_meta_version is not None:
+                        model_meta = ai_model_obj.metadata_versions.get(version=model_meta_version)
+                    else:
+                        model_meta = ai_model_obj.get_latest_version()
+                except ModelMeta.DoesNotExist:
+                    logger.error(f"Pipe 1 failed: ModelMeta version {model_meta_version} for model '{model_name}' not found.")
+                    return False
+            try:
+                sequences: Optional[Dict[str, List[Tuple[int, int]]]] = video_file.predict_video(
+                    model_meta=model_meta,
+                    smooth_window_size_s=smooth_window_size_s,
+                    binarize_threshold=binarize_threshold,
+                    test_run=test_run,
+                    n_test_frames=n_test_frames,
+                )
+            except Exception as e:
+                logger.error(f"Pipe 1 failed during prediction: {e}", exc_info=True)
+                return False
+
+            if sequences is None:
+                logger.error("Pipe 1 failed: Prediction pipeline returned None.")
+                return False
+            logger.info("Pipe 1: Prediction complete.")
+
+            # --- Set and Save State ---
+            state.initial_prediction_completed = True
+            state.save(update_fields=["initial_prediction_completed"])
+            logger.info("Pipe 1: Set initial_prediction_completed state to True.")
+
+            logger.info(f"Pipe 1: Sequences returned from prediction: {sequences}")
+            if not sequences:
+                logger.warning("Pipe 1: Prediction returned empty sequences dictionary. No LabelVideoSegments will be created.")
+
+            # 4. Create LabelVideoSegments
+            logger.info("Pipe 1: Creating LabelVideoSegments from predictions...")
+            try:
+                video_prediction_meta = VideoPredictionMeta.objects.get(video_file=video_file, model_meta=model_meta)
+                logger.info(f"Pipe 1: Calling _convert_sequences_to_db_segments for video {video_file.uuid} with prediction meta {video_prediction_meta.pk}")
+                _convert_sequences_to_db_segments(
+                    video=video_file,
+                    sequences=sequences,
+                    video_prediction_meta=video_prediction_meta,
+                )
+                video_file.sequences = sequences
+                video_file.save(update_fields=["sequences"])
+                state.lvs_created = True
+                state.save(update_fields=["lvs_created"])
+                logger.info("Pipe 1: Set lvs_created state to True.")
+                logger.info("Pipe 1: LabelVideoSegment creation complete.")
+                lvs_count_after = LabelVideoSegment.objects.filter(video_file=video_file).count()
+                logger.info(f"Pipe 1: Found {lvs_count_after} LabelVideoSegments after conversion attempt.")
+            except VideoPredictionMeta.DoesNotExist:
+                logger.error("Pipe 1 failed: Could not find VideoPredictionMeta after prediction.")
+                raise
+
+        logger.info(f"Pipe 1 completed successfully for video {video_file.uuid}")
+        success = True  # Set success flag
+        return True
+
+    except Exception as e:
+        logger.error(f"Pipe 1 failed for video {video_file.uuid}: {e}", exc_info=True)
+        return False
+    finally:
+        # 5. Optionally delete frames
+        if delete_frames_after and success:  # Check success flag
+            logger.info("Pipe 1: Deleting frames after processing...")
+            try:
+                video_file.delete_frames()
+                logger.info("Pipe 1: Frame deletion complete.")
+            except Exception as e:
+                logger.error(f"Pipe 1 failed during frame deletion: {e}", exc_info=True)
+        else:
+            logger.info("Pipe 1: Frame deletion skipped.")
+
+
+# --- Test after Pipe 1 ---
+def _test_after_pipe_1(video_file: "VideoFile", start_frame: int = 0, end_frame: int = 100) -> bool:
+    """
+    Simulates human annotation validation after Pipe 1.
+    Creates 'outside' segments and marks sensitive meta as verified.
+    """
+    from ...label import Label, LabelVideoSegment
+
+    logger.info(f"Starting _test_after_pipe_1 for video {video_file.uuid}")
+    try:
+        # 1. Create 'outside' LabelVideoSegments
+        try:
+            outside_label = Label.objects.get(name__iexact="outside")
+            logger.info(f"Creating 'outside' annotation segment [{start_frame}-{end_frame}]")
+            # Create a segment - assuming custom_create handles saving
+            outside_segment = LabelVideoSegment.objects.create(  # Assign to variable
+                video_file=video_file,
+                label=outside_label,
+                start_frame_number=start_frame,
+                end_frame_number=end_frame,
+                prediction_meta=None,
+            )
+            # Ensure the segment has a state and mark it as validated
+            segment_state, created = outside_segment.get_or_create_state()  # Unpack the tuple
+            segment_state.is_validated = True
+            segment_state.save()
+            logger.info(f"Marked 'outside' segment {outside_segment.pk} as validated. Created: {created}")
+
+        except Label.DoesNotExist:
+            logger.error("_test_after_pipe_1 failed: 'outside' Label not found.")
+            return False
+        except Exception as e:
+            logger.error(f"_test_after_pipe_1 failed during segment creation: {e}", exc_info=True)
+            return False
+
+        # 2. Set Sensitive Metadata state to verified
+        if video_file.sensitive_meta:
+            sm = video_file.sensitive_meta
+            st_state = sm.state
+            assert st_state, "SensitiveMeta state is None. Cannot set to verified."
+            logger.info("Setting sensitive meta state to verified.")
+
+            # Example: using a boolean field
+            video_file.sensitive_meta.state.dob_verified = True
+            video_file.sensitive_meta.state.names_verified = True
+            video_file.sensitive_meta.state.save()  # Save the SensitiveMeta instance
+            logger.info("Sensitive meta state updated.")
+
+        else:
+            logger.warning("_test_after_pipe_1: No sensitive meta found to verify.")
+
+        logger.info(f"_test_after_pipe_1 completed successfully for video {video_file.uuid}")
+        return True
+
+    except Exception as e:
+        logger.error(f"_test_after_pipe_1 failed for video {video_file.uuid}: {e}", exc_info=True)
+        return False

endoreg_db/models/media/video/pipe_2.py

@@ -0,0 +1,105 @@
+from itertools import filterfalse
+import logging
+from typing import TYPE_CHECKING
+from django.db import transaction
+
+
+# Configure logging
+logger = logging.getLogger(__name__) # Changed from "video_file"
+
+if TYPE_CHECKING:
+    from endoreg_db.models import VideoFile, VideoState
+
+def _pipe_2(video_file:"VideoFile") -> bool:
+    """
+    Process the given video file through pipeline 2 operations which include frame extraction,
+    anonymization of the video, and deletion of sensitive meta data.
+    Heavy I/O operations are performed outside the main atomic transaction for DB updates.
+
+    Parameters:
+        video_file (VideoFile): An instance of VideoFile representing the video to process.
+
+    Returns:
+        bool: True if all operations complete successfully; otherwise, False.
+    """
+    logger.info("Starting Pipe 2 for video %s", video_file.uuid)
+    try:
+        # --- Part 1: Frame Extraction ---
+        # Determine if frames are needed (short transaction for state read)
+        with transaction.atomic():
+            state: "VideoState" = video_file.get_or_create_state()
+            frames_needed = not state.frames_extracted
+
+        if frames_needed:
+            logger.info("Pipe 2: Frames not extracted. Extracting outside main DB transaction...")
+            if not video_file.extract_frames(overwrite=False):  # Heavy I/O work
+                logger.error("Pipe 2 failed: Frame extraction method returned False.")
+                return False
+
+            # Verify extraction and update state (short transaction)
+            with transaction.atomic():
+                video_file.refresh_from_db()
+                if not video_file.state or not video_file.state.frames_extracted:
+                    logger.error("Pipe 2 failed: Frame extraction did not update state successfully.")
+                    return False
+                logger.info("Pipe 2: Frame extraction complete.")
+        else:
+            logger.info("Pipe 2: Frames already extracted.")
+
+        # --- Part 2: Video Anonymization ---
+        # Determine if anonymization is needed (short transaction for state read)
+        with transaction.atomic():
+            state: "VideoState" = video_file.get_or_create_state()
+            anonymization_needed = not state.anonymized
+            if anonymization_needed:
+                state.sensitive_meta_processed = False
+
+        if anonymization_needed:
+            logger.info("Pipe 2: Video not anonymized. Anonymizing outside main DB transaction...")
+            anonymize_success = video_file.anonymize(delete_original_raw=True)  # Heavy I/O work
+            if not anonymize_success:
+                logger.error("Pipe 2 failed: Anonymization process failed (returned False).")
+                return False
+
+            # Verify anonymization and update state (short transaction)
+            with transaction.atomic():
+                video_file.refresh_from_db()
+                if not video_file.state or not video_file.state.anonymized:
+                    logger.error("Pipe 2 Error: State.anonymized is False even after anonymize() call.")
+                    return False
+                logger.info("Pipe 2: Anonymization complete.")
+        else:
+            logger.info("Pipe 2: Video already anonymized.")
+        
+
+        # --- Part 3: Final DB operations (now in its own atomic transaction) ---
+        with transaction.atomic():
+            video_file.refresh_from_db() # Ensure we have the latest video_file state for these ops
+
+            # Set sensitive_meta_processed True atomically
+            state: "VideoState" = video_file.get_or_create_state()
+            state.sensitive_meta_processed = True
+
+            # Delete Sensitive Meta Object
+            if video_file.sensitive_meta:
+                logger.info("Pipe 2: Deleting sensitive meta object...")
+                try:
+                    sm_pk = video_file.sensitive_meta.pk
+                    video_file.sensitive_meta.delete()
+                    video_file.sensitive_meta = None # Important after SET_NULL
+                    video_file.save(update_fields=['sensitive_meta']) # Persist the null relation
+                    logger.info("Pipe 2: Deleted sensitive meta object (PK: %s).", sm_pk)
+                except Exception as e:
+                    logger.error("Pipe 2: Failed to delete sensitive meta object: %s", e, exc_info=True)
+                    raise  # Reraise to ensure this transaction rolls back
+            else:
+                logger.info("Pipe 2: No sensitive meta object found to delete.")
+
+            logger.info(f"Pipe 2 completed successfully for video {video_file.uuid}")
+            return True
+
+    except Exception as e:
+        # This will catch exceptions from I/O operations if they raise,
+        # or from the final transaction block, or any other unhandled error.
+        logger.error(f"Pipe 2 failed for video {video_file.uuid} with unhandled exception: {e}", exc_info=True)
+        return False