From 659de6a2f26b0b31478131f737ef71cb5f617c25 Mon Sep 17 00:00:00 2001
From: Stanislas Polu <spolu@dust.tt>
Date: Mon, 22 Jan 2024 09:28:14 +0100
Subject: [PATCH] core: debug upsertion failure (#3348)

* core: debug upsertion failure

* orer
---
 core/src/data_sources/data_source.rs | 26 ++++++++++++++++++++++++--
 1 file changed, 24 insertions(+), 2 deletions(-)

diff --git a/core/src/data_sources/data_source.rs b/core/src/data_sources/data_source.rs
index 5518b2f36073..6ff2b7ced231 100644
--- a/core/src/data_sources/data_source.rs
+++ b/core/src/data_sources/data_source.rs
@@ -786,12 +786,34 @@ impl DataSource {
         )?;
 
         utils::done(&format!(
-            "Created document blob: data_source_id={} document_id={} duration={}ms",
+            "Created document blob: data_source_id={} document_id={} duration={}ms blob_url={}",
             self.data_source_id,
             document_id,
-            utils::now() - now
+            utils::now() - now,
+            format!("gs://{}/{}", bucket, content_path)
         ));
 
+        match document_id {
+            "notion-95804d6b-0274-43f6-8957-5b024234e3bf" => {
+                let debug_path = format!("{}/{}/debug.json", bucket_path, document_hash);
+                Object::create(
+                    &bucket,
+                    serde_json::to_string(&text).unwrap().into_bytes(),
+                    &debug_path,
+                    "application/json",
+                )
+                .await?;
+                utils::done(&format!(
+                    "Uploaded buggy document: data_source_id={} document_id={} debug_blob_url={}",
+                    self.data_source_id,
+                    document_id,
+                    format!("gs://{}/{}", bucket, debug_path)
+                ));
+                panic!("BUGGY document `{}`", document_id);
+            }
+            _ => (),
+        };
+
         let now = utils::now();
 
         // ChunkInfo is used to store the chunk text and associated hash to avoid recomputing the