Fix document status classification across uploads and workspaces

application/single_app/config.py

@@ -101,7 +101,7 @@
 EXECUTOR_TYPE = 'thread'
 EXECUTOR_MAX_WORKERS = 30
 SESSION_TYPE = 'filesystem'
-VERSION = "0.261.353"
+VERSION = "0.261.354"
 IS_DEVELOPMENT = is_development_env_enabled()
 
 # Opt-out for deployments where App Service Easy Auth is active but the platform

application/single_app/functions_document_queries.py

@@ -3,6 +3,8 @@
 
 from datetime import datetime, timedelta, timezone
 
+from functions_document_status import is_document_processing_failure
+
 
 DOCUMENT_PLACE_FILTERS = frozenset({
     "all", "recent", "shared", "processing", "errors", "untagged",
@@ -11,8 +13,7 @@
 
 
 def _document_processing_state(document_item):
-    status_text = str(document_item.get("status") or "").lower()
-    if "error" in status_text or "failed" in status_text:
+    if is_document_processing_failure(document_item.get("status")):
         return "error"
 
     percentage = document_item.get("percentage_complete")

application/single_app/functions_document_status.py

@@ -0,0 +1,32 @@
+# functions_document_status.py
+"""Document terminal-state predicates, independent of application bootstrap.
+
+Status details can contain filenames, sheet names, or retry diagnostics. Only
+lifecycle phrases at the start of a status can declare a terminal outcome.
+Keep the browser predicates aligned through document_processing_status_cases.json.
+"""
+
+import re
+
+
+_FAILURE = re.compile(
+    r"^(?:(?:error|failed)(?:$|[\s:!]|\.(?:$|\s))"
+    r"|(?:processing|document processing|approval|content screening) failed(?:$|[\s:;!]|\.(?:$|\s))"
+    r"|video:\s*(?:blob upload|auth|upload|indexing) failed(?:$|[\s:!]|\.(?:$|\s)))",
+    re.IGNORECASE,
+)
+_COMPLETE = re.compile(
+    r"^(?:(?:ready|complete|completed|processed|finished)[.!]?$"
+    r"|processing complete(?:$|[.!]$|\s+[-(]))",
+    re.IGNORECASE,
+)
+
+
+def is_document_processing_failure(status):
+    """Recognize terminal failures, not error words in processing details."""
+    return isinstance(status, str) and _FAILURE.match(status.strip()) is not None
+
+
+def is_document_processing_complete(status):
+    """Recognize final completion, not completion of an intermediate stage."""
+    return isinstance(status, str) and _COMPLETE.match(status.strip()) is not None

application/single_app/functions_documents.py

@@ -49,6 +49,7 @@
 )
 from config import *
 from functions_appinsights import log_event
+from functions_document_status import is_document_processing_complete, is_document_processing_failure
 from functions_document_provenance import (
     ORIGIN_FIELD_NAMES,
     ORIGIN_KIND_FIELD,
@@ -3508,10 +3509,10 @@ def calculate_processing_percentage(doc_metadata):
     total_chunks_saved = doc_metadata.get('current_file_chunk', 0)
 
     # --- Final States ---
-    if "processing complete" in status or current_pct == 100:
+    if is_document_processing_complete(status) or current_pct == 100:
         # Ensure it stays 100 if it ever reached it
         return 100
-    if "error" in status or "failed" in status:
+    if is_document_processing_failure(status):
         # Keep the last known percentage on error/failure
         return current_pct
 
@@ -3694,7 +3695,7 @@ def update_document(**kwargs):
         marker = existing_document.get(SCREENING_FIELD)
         if marker is not None and not document_is_available(existing_document):
             requested_status = str(kwargs.get("status") or "").lower()
-            if "complete" in requested_status:
+            if is_document_processing_complete(requested_status):
                 kwargs["status"] = "Content review required" if marker.get("review_required") else "Content screening pending"
             if kwargs.get("percentage_complete") == 100:
                 kwargs["percentage_complete"] = 99
@@ -3763,14 +3764,14 @@ def update_document(**kwargs):
             elif isinstance(status_lower, dict):
                 status_lower = json.dumps(status_lower).lower()
 
-            if "processing complete" in status_lower:
+            if is_document_processing_complete(status_lower):
                 new_percentage = 100
-            elif "error" in status_lower or "failed" in status_lower:
+            elif is_document_processing_failure(status_lower):
                  pass # Percentage already calculated by helper based on 'failed' status
 
             # Ensure percentage doesn't decrease (unless reset on failure or hitting 100)
             # Compare against original_percentage fetched *before* any updates in this call
-            if new_percentage < original_percentage and new_percentage != 0 and "failed" not in status_lower and "error" not in status_lower:
+            if new_percentage < original_percentage and new_percentage != 0 and not is_document_processing_failure(status_lower):
                  existing_document['percentage_complete'] = original_percentage
             else:
                  existing_document['percentage_complete'] = new_percentage
@@ -6881,7 +6882,7 @@ def detect_doc_type(document_id, user_id=None):
 
 def _update_document_for_job(*, operation_guard=None, safe_errors=False, **arguments):
     status = arguments.get("status")
-    if safe_errors and isinstance(status, str) and any(word in status.lower() for word in ("error", "failed")):
+    if safe_errors and is_document_processing_failure(status):
         arguments["status"] = "Document processing failed. Contact an administrator if the problem persists."
     if operation_guard is None:
         return update_document(**arguments)

application/single_app/route_backend_chats.py

@@ -13,6 +13,7 @@
     guard_model_callable,
 )
 from content_screening.contracts import DocumentHeldError, ScreeningConflictError, ScreeningError, document_is_available
+from functions_document_status import is_document_processing_complete, is_document_processing_failure
 from agent_execution_context import AgentExecutionCancelled, DelegationBudget
 from action_credential_contracts import ActionCredentialsRequired
 from agent_delegation_runtime import AgentExecution, delegation_citations, delegation_usage, prepare_agent_execution
@@ -1533,7 +1534,7 @@ def _is_search_ready_chat_upload_workspace_document(document_item):
         return False
 
     status = str(document_item.get('status') or '').strip().lower()
-    if 'error' in status or 'failed' in status:
+    if is_document_processing_failure(status):
         return False
 
     try:
@@ -1548,7 +1549,7 @@ def _is_search_ready_chat_upload_workspace_document(document_item):
     except (TypeError, ValueError):
         percentage_complete = 0
 
-    return percentage_complete >= 100 or 'processing complete' in status
+    return percentage_complete >= 100 or is_document_processing_complete(status)
 
 
 def _get_chat_upload_workspace_document_scope(document_item):

application/single_app/static/js/chat/chat-documents.js

@@ -182,10 +182,10 @@ function isConversationTaskDocumentReady(taskDocument = {}) {
   if (window.ContentScreening?.isHeld(taskDocument)) return false;
   const statusText = String(taskDocument.status || '').trim().toLowerCase();
   const percentageComplete = Number(taskDocument.percentage_complete || taskDocument.percentageComplete || 0);
-  if (statusText.includes('error') || statusText.includes('failed')) {
+  if (window.DocumentStatus.isFailure(statusText)) {
     return false;
   }
-  return percentageComplete >= 100 || statusText.includes('processing complete') || statusText.includes('complete');
+  return percentageComplete >= 100 || window.DocumentStatus.isComplete(statusText);
 }
 
 function normalizeConversationTaskDocument(documentInfo = {}, fallbackConversationId = null) {

application/single_app/static/js/chat/chat-messages.js

@@ -132,24 +132,21 @@ function getChatWorkspaceProgressValue(value) {
 function isChatWorkspaceDocumentComplete(doc) {
   const statusText = String(doc?.status || '').toLowerCase();
   return getChatWorkspaceProgressValue(doc?.percentage_complete) >= 100
-    || statusText.includes('processing complete')
-    || statusText.includes('complete')
-    || statusText.includes('error')
-    || statusText.includes('failed');
+    || window.DocumentStatus.isComplete(statusText)
+    || window.DocumentStatus.isFailure(statusText);
 }
 
 function isChatWorkspaceDocumentFailure(doc) {
   const statusText = String(doc?.status || '').toLowerCase();
-  return statusText.includes('error') || statusText.includes('failed');
+  return window.DocumentStatus.isFailure(statusText);
 }
 
 function isChatWorkspaceDocumentSuccessComplete(doc) {
   const statusText = String(doc?.status || '').toLowerCase();
   return !isChatWorkspaceDocumentFailure(doc)
     && (
       getChatWorkspaceProgressValue(doc?.percentage_complete) >= 100
-      || statusText.includes('processing complete')
-      || statusText.includes('complete')
+      || window.DocumentStatus.isComplete(statusText)
     );
 }
 
@@ -231,9 +228,9 @@ function buildChatWorkspaceAttachmentHtml(attachment) {
   const statusText = String(attachment?.status || 'Queued for workspace processing').trim() || 'Queued for workspace processing';
   const detailsId = getChatWorkspaceProgressDetailsId(workspaceDocumentId);
   const statusLower = statusText.toLowerCase();
-  const progressClass = statusLower.includes('error') || statusLower.includes('failed')
+  const progressClass = window.DocumentStatus.isFailure(statusLower)
     ? 'bg-danger'
-    : (pct >= 100 || statusLower.includes('complete') ? 'bg-success' : 'progress-bar-striped progress-bar-animated bg-info');
+    : (pct >= 100 || window.DocumentStatus.isComplete(statusLower) ? 'bg-success' : 'progress-bar-striped progress-bar-animated bg-info');
 
   return `
     <div class="chat-workspace-upload-progress mt-2 p-2 border rounded bg-body-tertiary"
@@ -302,8 +299,8 @@ function updateChatWorkspaceProgressContainer(container, doc) {
   const pct = getChatWorkspaceProgressValue(doc?.percentage_complete);
   const statusText = String(doc?.status || 'Processing workspace document...').trim() || 'Processing workspace document...';
   const statusLower = statusText.toLowerCase();
-  const hasFailure = statusLower.includes('error') || statusLower.includes('failed');
-  const hasCompleted = !hasFailure && (pct >= 100 || statusLower.includes('complete'));
+  const hasFailure = window.DocumentStatus.isFailure(statusLower);
+  const hasCompleted = !hasFailure && (pct >= 100 || window.DocumentStatus.isComplete(statusLower));
   const progressBar = container.querySelector('.progress-bar');
   const statusElement = container.querySelector('.chat-workspace-upload-status-text');
   const progressElement = container.querySelector('.progress');

application/single_app/static/js/document-status.js

@@ -0,0 +1,16 @@
+// document-status.js
+// Kept in parity with Python and React through document_processing_status_cases.json.
+
+window.DocumentStatus = (() => {
+    const failure = /^(?:(?:error|failed)(?:$|[\s:!]|\.(?:$|\s))|(?:processing|document processing|approval|content screening) failed(?:$|[\s:;!]|\.(?:$|\s))|video:\s*(?:blob upload|auth|upload|indexing) failed(?:$|[\s:!]|\.(?:$|\s)))/i;
+    const complete = /^(?:(?:ready|complete|completed|processed|finished)[.!]?$|processing complete(?:$|[.!]$|\s+[-(]))/i;
+
+    return Object.freeze({
+        isFailure(status) {
+            return typeof status === "string" && failure.test(status.trim());
+        },
+        isComplete(status) {
+            return typeof status === "string" && complete.test(status.trim());
+        },
+    });
+})();

application/single_app/static/js/public/public_workspace.js

@@ -1071,9 +1071,8 @@ function getPublicDocumentProcessingState(doc) {
   const pctString = String((doc.percentage_complete ?? doc.percentage) || '0');
   const pct = /^\d+(\.\d+)?$/.test(pctString) ? parseFloat(pctString) : 0;
   const docStatus = doc.status || '';
-  const normalizedStatus = docStatus.toLowerCase();
-  const hasError = normalizedStatus.includes('error') || normalizedStatus.includes('failed');
-  const isComplete = pct >= 100 || normalizedStatus.includes('complete') || hasError;
+  const hasError = window.DocumentStatus.isFailure(docStatus);
+  const isComplete = pct >= 100 || window.DocumentStatus.isComplete(docStatus) || hasError;
   return { pct, docStatus, hasError, isComplete };
 }
 
@@ -1667,8 +1666,8 @@ function renderPublicDocumentRow(doc) {
   const pctString = String((doc.percentage_complete ?? doc.percentage) || "0");
   const pct = /^\d+(\.\d+)?$/.test(pctString) ? parseFloat(pctString) : 0;
   const docStatus = doc.status || "";
-  const isComplete = pct >= 100 || docStatus.toLowerCase().includes("complete") || docStatus.toLowerCase().includes("error");
-  const hasError = docStatus.toLowerCase().includes("error") || docStatus.toLowerCase().includes("failed");
+  const hasError = window.DocumentStatus.isFailure(docStatus);
+  const isComplete = pct >= 100 || window.DocumentStatus.isComplete(docStatus) || hasError;
   const isPendingGeneratedArtifact = isPendingGeneratedArtifactDocument(doc);
 
   let firstTdHtml = "";
@@ -1896,8 +1895,8 @@ function pollPublicDocumentStatus(documentId) {
       const pctString = String((doc.percentage_complete ?? doc.percentage) || "0");
       const pct = /^\d+(\.\d+)?$/.test(pctString) ? parseFloat(pctString) : 0;
       const docStatus = doc.status || "";
-      const isComplete = pct >= 100 || docStatus.toLowerCase().includes("complete") || docStatus.toLowerCase().includes("error");
-      const hasError = docStatus.toLowerCase().includes("error") || docStatus.toLowerCase().includes("failed");
+      const hasError = window.DocumentStatus.isFailure(docStatus);
+      const isComplete = pct >= 100 || window.DocumentStatus.isComplete(docStatus) || hasError;
 
       if (!isComplete && statusRow) {
         // Update progress bar and status text if still processing
@@ -3374,8 +3373,8 @@ function buildPublicFolderDocumentsTable(docs) {
     const pctString = String((doc.percentage_complete ?? doc.percentage) || '0');
     const pct = /^\d+(\.\d+)?$/.test(pctString) ? parseFloat(pctString) : 0;
     const docStatus = doc.status || '';
-    const isComplete = pct >= 100 || docStatus.toLowerCase().includes('complete') || docStatus.toLowerCase().includes('error');
-    const hasError = docStatus.toLowerCase().includes('error') || docStatus.toLowerCase().includes('failed');
+    const hasError = window.DocumentStatus.isFailure(docStatus);
+    const isComplete = pct >= 100 || window.DocumentStatus.isComplete(docStatus) || hasError;
     const isSelected = publicSelectedDocuments.has(doc.id);
     let firstColHtml = `<input type="checkbox" class="form-check-input document-checkbox${publicSelectionMode ? '' : ' d-none'}" data-document-id="${doc.id}"${isSelected ? ' checked' : ''}>`;
     let actionsHtml = '';

application/single_app/static/js/workspace/workspace-documents.js

@@ -318,9 +318,8 @@ function getDocumentProcessingState(doc) {
     const pctString = String(doc?.percentage_complete ?? "");
     const pct = /^\d+(\.\d+)?$/.test(pctString) ? parseFloat(pctString) : 0;
     const docStatus = doc?.status || "";
-    const normalizedStatus = docStatus.toLowerCase();
-    const hasError = normalizedStatus.includes("error");
-    const isComplete = pct >= 100 || normalizedStatus.includes("complete") || hasError;
+    const hasError = window.DocumentStatus.isFailure(docStatus);
+    const isComplete = pct >= 100 || window.DocumentStatus.isComplete(docStatus) || hasError;
 
     return { pct, docStatus, hasError, isComplete };
 }
@@ -1789,8 +1788,8 @@ function renderDocumentRow(doc) {
     const pctString = String(doc.percentage_complete);
     const pct = /^\d+(\.\d+)?$/.test(pctString) ? parseFloat(pctString) : 0;
     const docStatus = doc.status || "";
-    const isComplete = pct >= 100 || docStatus.toLowerCase().includes("complete") || docStatus.toLowerCase().includes("error");
-    const hasError = docStatus.toLowerCase().includes("error");
+    const hasError = window.DocumentStatus.isFailure(docStatus);
+    const isComplete = pct >= 100 || window.DocumentStatus.isComplete(docStatus) || hasError;
 
     const docRow = document.createElement("tr");
     docRow.id = `doc-row-${docId}`;
@@ -2239,8 +2238,8 @@ function pollDocumentStatus(documentId) {
                  const pctString = String(doc.percentage_complete);
                  const pct = /^\d+(\.\d+)?$/.test(pctString) ? parseFloat(pctString) : 0;
                  const docStatus = doc.status || "";
-                 const isComplete = pct >= 100 || docStatus.toLowerCase().includes("complete") || docStatus.toLowerCase().includes("error");
-                 const hasError = docStatus.toLowerCase().includes("error");
+                 const hasError = window.DocumentStatus.isFailure(docStatus);
+                 const isComplete = pct >= 100 || window.DocumentStatus.isComplete(docStatus) || hasError;
 
                 if (!isComplete && statusRow) {
                      // Update progress bar and status text if still processing

application/single_app/static/js/workspace/workspace-tags.js

@@ -648,10 +648,8 @@ function buildFolderDocumentsTable(docs) {
         const pctStr = String(doc.percentage_complete);
         const pct = /^\d+(\.\d+)?$/.test(pctStr) ? parseFloat(pctStr) : 0;
         const docStatus = doc.status || '';
-        const isComplete = pct >= 100
-            || docStatus.toLowerCase().includes('complete')
-            || docStatus.toLowerCase().includes('error');
-        const hasError = docStatus.toLowerCase().includes('error');
+        const hasError = window.DocumentStatus.isFailure(docStatus);
+        const isComplete = pct >= 100 || window.DocumentStatus.isComplete(docStatus) || hasError;
 
         const currentUserId = window.current_user_id;
         const isOwner = doc.user_id === currentUserId;

application/single_app/templates/base.html

@@ -709,6 +709,7 @@ <h5 class="modal-title workflow-alert-title" id="workflowAlertModalLabel">Workfl
          but it's okay to load it afterwards) -->
   <script src="/static/js/chat/bootstrap.bundle.min.js"></script>
   <script src="{{ url_for('static', filename='js/toast.js') }}?v={{ config['VERSION'] }}"></script>
+  <script src="{{ url_for('static', filename='js/document-status.js') }}?v={{ config['VERSION'] }}"></script>
 
 
   <!-- Then DataTables (depends on jQuery) -->