Skip to content

LLM-Driven Extraction of Unstructured Data — Built for API Deployments & ETL Pipeline Workflows

UN-3454 [FIX] Add CSV/JSON export to workflow execution logs modal (#1949)

ali0538fdcc28e7
diff --git a/backend/workflow_manager/execution/urls.py b/backend/workflow_manager/execution/urls.py
index a699a8340..c5bf5cc30 100644
--- a/backend/workflow_manager/execution/urls.py
+++ b/backend/workflow_manager/execution/urls.py
@@ -13,11 +13,17 @@ execution_list = ExecutionViewSet.as_view(
)
execution_detail = ExecutionViewSet.as_view({"get": "retrieve"})
execution_log_list = ExecutionLogViewSet.as_view({"get": "list"})
+execution_log_export = ExecutionLogViewSet.as_view({"get": "export"})
urlpatterns = format_suffix_patterns(
[
path("", execution_list, name="execution-list"),
path("<uuid:pk>/", execution_detail, name="execution-detail"),
path("<uuid:pk>/logs/", execution_log_list, name="execution-log"),
+ path(
+ "<uuid:pk>/logs/export/",
+ execution_log_export,
+ name="execution-log-export",
+ ),
]
)
diff --git a/backend/workflow_manager/workflow_v2/execution_log_view.py b/backend/workflow_manager/workflow_v2/execution_log_view.py
index 623b258bf..2d8575080 100644
--- a/backend/workflow_manager/workflow_v2/execution_log_view.py
+++ b/backend/workflow_manager/workflow_v2/execution_log_view.py
@@ -1,10 +1,16 @@
+import csv
+import io
+import json
import logging
from django.db.models import Q
from django.db.models.query import QuerySet
+from django.http import HttpResponse
+from django.utils import timezone
from permissions.permission import IsOwner
-from rest_framework import viewsets
+from rest_framework import status, viewsets
from rest_framework.permissions import IsAuthenticated
+from rest_framework.response import Response
from rest_framework.versioning import URLPathVersioning
from utils.pagination import CustomPagination
@@ -14,6 +20,11 @@ from workflow_manager.workflow_v2.serializers import WorkflowExecutionLogSeriali
logger = logging.getLogger(__name__)
+# Cap on synchronous export size. Above this, callers should narrow the
+# filter (file_execution_id, log_level) — async deployment-wide export
+# is intentionally a separate, future feature.
+MAX_SYNC_EXPORT_ROWS = 50_000
+
class WorkflowExecutionLogViewSet(viewsets.ModelViewSet):
versioning_class = URLPathVersioning
@@ -32,3 +43,115 @@ class WorkflowExecutionLogViewSet(viewsets.ModelViewSet):
return ExecutionLog.objects.filter(
Q(wf_execution_id=execution_id) | Q(execution_id=execution_id)
)
+
+ def export(self, request, *args, **kwargs):
+ """Export logs for a single workflow execution as CSV or JSON.
+
+ Honors the same filters as the list endpoint (file_execution_id,
+ log_level). Returns 413 if the result set exceeds the sync cap so
+ the client can prompt the user to narrow their filter.
+ """
+ # NOTE: do not name this query param `format` — DRF reserves it for
+ # content negotiation and will 404 if no renderer matches the value.
+ export_format = request.query_params.get("file_format", "json").lower()
+ if export_format not in ("json", "csv"):
+ return Response(
+ {"error": "file_format must be one of: json, csv"},
+ status=status.HTTP_400_BAD_REQUEST,
+ )
+
+ queryset = (
+ self.filter_queryset(self.get_queryset())
+ .order_by("event_time")
+ .only("id", "event_time", "data", "file_execution_id")
+ )
+
+ # Single materialization with cap+1 sentinel — avoids a separate COUNT
+ # query and a second full scan during build.
+ rows = list(queryset[: MAX_SYNC_EXPORT_ROWS + 1])
+ if len(rows) > MAX_SYNC_EXPORT_ROWS:
+ return Response(
+ {
+ "error": (
+ f"Too many logs to export (>{MAX_SYNC_EXPORT_ROWS} rows). "
+ "Narrow the filter (e.g. by file or log level) and retry."
+ ),
+ "limit": MAX_SYNC_EXPORT_ROWS,
+ },
+ status=status.HTTP_413_REQUEST_ENTITY_TOO_LARGE,
+ )
+
+ if export_format == "csv":
+ body = self._build_csv(rows)
+ content_type = "text/csv; charset=utf-8"
+ else:
+ body = self._build_json(rows)
+ content_type = "application/json"
+
+ execution_id = self.kwargs.get("pk")
+ timestamp = timezone.now().strftime("%Y%m%d_%H%M%S")
+ filename = f"execution_logs_{execution_id}_{timestamp}.{export_format}"
+
+ response = HttpResponse(body, content_type=content_type)
+ response["Content-Disposition"] = f'attachment; filename="{filename}"'
+ return response
+
+ def _normalize(self, log: ExecutionLog) -> dict:
+ """Single source of truth for per-row null-handling and dict-guard.
+
+ Logs warning when `data` is non-dict so silent blank rows in CSV
+ still leave a diagnostic trail in operator logs.
+ """
+ if isinstance(log.data, dict):
+ data = log.data
+ else:
+ if log.data is not None:
+ logger.warning(
+ "ExecutionLog %s has non-dict data of type %s; "
+ "emitting blanks in CSV export",
+ log.id,
+ type(log.data).__name__,
+ )
+ data = {}
+ return {
+ "id": str(log.id),
+ "event_time": log.event_time.isoformat() if log.event_time else None,
+ "file_execution_id": (
+ str(log.file_execution_id) if log.file_execution_id else None
+ ),
+ "level": data.get("level", ""),
+ "stage": data.get("stage", ""),
+ "log_message": data.get("log", ""),
+ "raw_data": log.data,
+ }
+
+ def _build_csv(self, rows: list[ExecutionLog]) -> str:
+ output = io.StringIO()
+ writer = csv.writer(output)
+ writer.writerow(["event_time", "level", "stage", "log", "file_execution_id"])
+ for log in rows:
+ n = self._normalize(log)
+ writer.writerow(
+ [
+ n["event_time"] or "",
+ n["level"],
+ n["stage"],
+ n["log_message"],
+ n["file_execution_id"] or "",
+ ]
+ )
+ return output.getvalue()
+
+ def _build_json(self, rows: list[ExecutionLog]) -> str:
+ # JSON path passes raw_data through verbatim (faithful to the DB)
+ # rather than projecting it to extracted level/stage/log fields.
+ entries = [
+ {
+ "id": n["id"],
+ "event_time": n["event_time"],
+ "file_execution_id": n["file_execution_id"],
+ "data": n["raw_data"],
+ }
+ for n in (self._normalize(log) for log in rows)
+ ]
+ return json.dumps(entries)
diff --git a/backend/workflow_manager/workflow_v2/urls/workflow.py b/backend/workflow_manager/workflow_v2/urls/workflow.py
index e9f404fed..f83fcb8b4 100644
--- a/backend/workflow_manager/workflow_v2/urls/workflow.py
+++ b/backend/workflow_manager/workflow_v2/urls/workflow.py
@@ -21,6 +21,7 @@ workflow_execute = WorkflowViewSet.as_view({"post": "execute", "put": "activate"
execution_entity = WorkflowExecutionViewSet.as_view({"get": "retrieve"})
execution_list = WorkflowExecutionViewSet.as_view({"get": "list"})
execution_log_list = WorkflowExecutionLogViewSet.as_view({"get": "list"})
+execution_log_export = WorkflowExecutionLogViewSet.as_view({"get": "export"})
workflow_clear_file_marker = WorkflowViewSet.as_view({"get": "clear_file_marker"})
workflow_schema = WorkflowViewSet.as_view({"get": "get_schema"})
can_update = WorkflowViewSet.as_view({"get": "can_update"})
@@ -71,6 +72,11 @@ urlpatterns = format_suffix_patterns(
execution_log_list,
name="execution-log",
),
+ path(
+ "execution/<uuid:pk>/logs/export/",
+ execution_log_export,
+ name="execution-log-export",
+ ),
path(
"schema/",
workflow_schema,
diff --git a/frontend/src/components/logging/log-modal/LogModal.css b/frontend/src/components/logging/log-modal/LogModal.css
index f363408be..3440916c0 100644
--- a/frontend/src/components/logging/log-modal/LogModal.css
+++ b/frontend/src/components/logging/log-modal/LogModal.css
@@ -21,3 +21,8 @@
border-color: #1677ff;
color: #1677ff;
}
+
+.log-modal-title .export-btn-outlined {
+ margin-left: 16px;
+ height: auto;
+}
diff --git a/frontend/src/components/logging/log-modal/LogModal.jsx b/frontend/src/components/logging/log-modal/LogModal.jsx
index d70978031..5baa6e5a1 100644
--- a/frontend/src/components/logging/log-modal/LogModal.jsx
+++ b/frontend/src/components/logging/log-modal/LogModal.jsx
@@ -1,5 +1,5 @@
-import { CopyOutlined } from "@ant-design/icons";
-import { Button, Modal, Table, Tooltip } from "antd";
+import { CopyOutlined, DownloadOutlined } from "@ant-design/icons";
+import { Button, Dropdown, Modal, Table, Tooltip } from "antd";
import PropTypes from "prop-types";
import { useEffect, useState } from "react";
@@ -32,6 +32,7 @@ function LogModal({
const [executionLogs, setExecutionLogs] = useState([]);
const [loading, setLoading] = useState(false);
+ const [exporting, setExporting] = useState(false);
const [selectedLogLevel, setSelectedLogLevel] = useState(null);
const [ordering, setOrdering] = useState(null);
const [pagination, setPagination] = useState({
@@ -84,6 +85,89 @@ function LogModal({
setPagination({ current: 1, pageSize: 10, total: 0 }); // Reset pagination
};
+ const handleExport = async (fileFormat) => {
+ if (exporting) {
+ return;
+ }
+ setExporting(true);
+ try {
+ const url = getUrl(`/execution/${executionId}/logs/export/`);
+ // `file_format` (not `format`) — `format` is reserved by DRF for
+ // content negotiation and 404s when set to "csv".
+ const params = {
+ file_format: fileFormat,
+ file_execution_id: fileId || "null",
+ };
+ // Skip log_level when nothing is selected; axios serializes `null` as
+ // the literal string "null" which falls through the level filter to
+ // an INFO-min default and silently drops DEBUG rows.
+ if (selectedLogLevel) {
+ params.log_level = selectedLogLevel;
+ }
+ const response = await axiosPrivate.get(url, {
+ params,
+ responseType: "blob",
+ });
+
+ const filename =
+ response.headers?.["content-disposition"]?.match(
+ /filename="?([^"]+)"?/,
+ )?.[1] || `execution_logs_${fileId || executionId}.${fileFormat}`;
+ const blobUrl = globalThis.URL.createObjectURL(new Blob([response.data]));
+ const link = document.createElement("a");
+ link.href = blobUrl;
+ link.download = filename;
+ document.body.appendChild(link);
+ link.click();
+ link.remove();
+ globalThis.URL.revokeObjectURL(blobUrl);
+ } catch (err) {
+ // 413 means we hit the server-side row cap. Always surface a
+ // narrow-your-filter message — even if decoding the JSON blob body
+ // fails we keep the right user-facing copy rather than falling
+ // through to a generic "Request failed" alert.
+ if (err?.response?.status === 413) {
+ let message = "Export too large — narrow your filter and retry.";
+ try {
+ const text = await err.response.data.text();
+ const parsed = JSON.parse(text);
+ if (parsed?.error) {
+ message = parsed.error;
+ }
+ } catch (parseErr) {
+ console.error("Failed to parse 413 body:", parseErr);
+ }
+ setAlertDetails({ type: "error", content: message });
+ return;
+ }
+ setAlertDetails(handleException(err));
+ } finally {
+ setExporting(false);
+ }
+ };
+
+ const exportMenuItems = [
+ {
+ key: "json",
+ label: "Download as JSON",
+ onClick: () => handleExport("json"),
+ },
+ {
+ key: "csv",
+ label: "Download as CSV",
+ onClick: () => handleExport("csv"),
+ },
+ ];
+
+ let exportTooltip;
+ if (!pagination.total) {
+ exportTooltip = "No logs to export";
+ } else if (exporting) {
+ exportTooltip = "Export in progress…";
+ } else {
+ exportTooltip = "Export logs";
+ }
+
const logDetailsColumns = [
{
title: "Event Time",
@@ -167,6 +251,27 @@ function LogModal({
onClick={() => copyToClipboard(displayId, "File Execution ID")}
/>
)}
+ <Dropdown
+ menu={{ items: exportMenuItems }}
+ trigger={["click"]}
+ disabled={!pagination.total || exporting}
+ >
+ <Tooltip title={exportTooltip}>
+ {/* span wrapper — antd disabled Buttons swallow pointer
+ events so a direct Tooltip child never receives
+ mouseenter and the disabled-state tooltip never shows. */}
+ <span>
+ <Button
+ className="export-btn-outlined"
+ icon={<DownloadOutlined />}
+ loading={exporting}
+ disabled={!pagination.total || exporting}
+ >
+ Export
+ </Button>
+ </span>
+ </Tooltip>
+ </Dropdown>
</span>
}
centered