From ffc3ddfe3b4bb9c98984568ca242e63964468de0 Mon Sep 17 00:00:00 2001
From: McElwain <sean.mcelwain@outlook.com>
Date: Sat, 18 Apr 2026 16:15:39 -0500
Subject: [PATCH] feat: add training exports and dataset inspection tooling

---
 scripts/inspect_training_data.py | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)
 create mode 100644 scripts/inspect_training_data.py

diff --git a/scripts/inspect_training_data.py b/scripts/inspect_training_data.py
new file mode 100644
index 0000000..b3643b6
--- /dev/null
+++ b/scripts/inspect_training_data.py
@@ -0,0 +1,24 @@
+import json
+from pathlib import Path
+
+path = Path("/mnt/storage/document-processor/exports/document_training.jsonl")
+
+count = 0
+approved = 0
+
+with path.open() as f:
+    for line in f:
+        row = json.loads(line)
+        count += 1
+        if row["review"]["is_approved"]:
+            approved += 1
+
+        if count <= 3:
+            print("\n--- SAMPLE ---")
+            print("ID:", row["document"]["document_id"])
+            print("Merchant:", row["extracted_fields"].get("merchant_normalized"))
+            print("Total:", row["extracted_fields"].get("total"))
+            print("OCR len:", len(row["ocr_text"]))
+
+print("\nTotal docs:", count)
+print("Approved:", approved)