gitea-admin/gh-christianlouis-docuelevate
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: enhance processing status tracking and UI display for file details

Browse Source
This commit is contained in:
Christian Krakau-Louis
2026-02-12 01:07:22 +01:00
parent 02e1445e01
commit 83c65f3c40
7 changed files with 287 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files
app/utils/step_manager.py
+52 -6
View File
@@ -156,6 +156,9 @@ def get_file_step_status(db: Session, file_id: int) -> Dict[str, Dict]:
def get_file_overall_status(db: Session, file_id: int) -> Dict:
"""
Get the overall processing status for a file based on its steps.
Only considers "real" processing steps that represent user-facing status.
Ignores diagnostic/internal steps like poll_task, upload_file, etc.
Args:
db: Database session
@@ -172,7 +175,28 @@ def get_file_overall_status(db: Session, file_id: int) -> Dict:
"in_progress_steps": 2
}
"""
steps = db.query(FileProcessingStep).filter(FileProcessingStep.file_id == file_id).all()
# Define which steps are "real" status-determining steps
# Only high-level logical steps, not implementation sub-steps
REAL_MAIN_STEPS = {
"create_file_record",
"check_text",
"extract_text",
"process_with_azure_document_intelligence",
"extract_metadata_with_gpt",
"embed_metadata_into_pdf",
"finalize_document_storage",
"send_to_all_destinations",
}
all_steps = db.query(FileProcessingStep).filter(FileProcessingStep.file_id == file_id).all()
# Filter to only real steps
steps = [
s for s in all_steps
if s.step_name in REAL_MAIN_STEPS
or s.step_name.startswith("queue_")
or s.step_name.startswith("upload_to_")
]
if not steps:
return {
@@ -217,6 +241,9 @@ def get_step_summary(db: Session, file_id: int) -> Dict:
"""
Get a summary of main steps vs upload steps with status counts.
Only counts "real" processing steps that represent user-facing status.
Ignores diagnostic/internal steps like poll_task, upload_file, etc.
Args:
db: Database session
file_id: ID of the file
@@ -230,13 +257,24 @@ def get_step_summary(db: Session, file_id: int) -> Dict:
"total_upload_tasks": 6
}
"""
# Define which steps are "real" status-determining steps
# Only high-level logical steps, not implementation sub-steps
REAL_MAIN_STEPS = {
"create_file_record",
"check_text",
"extract_text",
"process_with_azure_document_intelligence",
"extract_metadata_with_gpt",
"embed_metadata_into_pdf",
"finalize_document_storage",
"send_to_all_destinations",
}
steps = db.query(FileProcessingStep).filter(FileProcessingStep.file_id == file_id).all()
main_counts = {"queued": 0, "in_progress": 0, "success": 0, "failure": 0, "skipped": 0}
upload_counts = {"queued": 0, "in_progress": 0, "success": 0, "failure": 0, "skipped": 0}
upload_prefixes = ["upload_to_", "queue_"]
main_steps_count = 0
upload_steps_count = 0
@@ -246,14 +284,21 @@ def get_step_summary(db: Session, file_id: int) -> Dict:
if status == "pending":
status = "queued"
# Check if it's an upload task
is_upload = any(step.step_name.startswith(prefix) for prefix in upload_prefixes)
# Check if it's an upload task (only count actual upload_to_* steps, not queue_* steps)
is_upload = step.step_name.startswith("upload_to_")
# Only count "real" steps
is_real_step = step.step_name in REAL_MAIN_STEPS or is_upload or step.step_name.startswith("queue_")
if not is_real_step:
# Skip diagnostic/internal steps like poll_task, upload_file, set_custom_fields, etc.
continue
if is_upload:
if status in upload_counts:
upload_counts[status] += 1
upload_steps_count += 1
elif step.step_name in MAIN_PROCESSING_STEPS:
elif step.step_name in REAL_MAIN_STEPS:
if status in main_counts:
main_counts[status] += 1
main_steps_count += 1
@@ -264,3 +309,4 @@ def get_step_summary(db: Session, file_id: int) -> Dict:
"total_main_steps": main_steps_count,
"total_upload_tasks": upload_steps_count,
}