From 41a906b0ca13c17689568eac1b6d7037dd87d234 Mon Sep 17 00:00:00 2001
From: Stanislas Polu <spolu@dust.tt>
Date: Tue, 10 Dec 2024 12:02:50 +0100
Subject: [PATCH] core: log parents[0] / node_id mismatch (#9241)

* core: log parents that are not self

* KWSearch invariant log

* remove unused

* fix log

* more compact log

* remove old log

* clean-up imports

* better logging
---
 core/bin/core_api.rs                 | 55 ++++++++++++++++++++++++++++
 core/src/data_sources/data_source.rs | 18 +--------
 2 files changed, 56 insertions(+), 17 deletions(-)

diff --git a/core/bin/core_api.rs b/core/bin/core_api.rs
index 081e9c8eeb29..7881b73de5dd 100644
--- a/core/bin/core_api.rs
+++ b/core/bin/core_api.rs
@@ -1478,6 +1478,17 @@ async fn data_sources_documents_update_parents(
 ) -> (StatusCode, Json<APIResponse>) {
     let project = project::Project::new_from_id(project_id);
 
+    if payload.parents.get(0) != Some(&document_id) {
+        info!(
+            data_source_id = data_source_id,
+            node_id = document_id,
+            parents = ?payload.parents,
+            node_type = "document",
+            operation = "update_parents",
+            "[KWSEARCH] invariant_first_parent_self"
+        );
+    }
+
     match state
         .store
         .load_data_source(&project, &data_source_id)
@@ -1625,6 +1636,17 @@ async fn data_sources_documents_upsert(
         None => false,
     };
 
+    if payload.parents.get(0) != Some(&payload.document_id) {
+        info!(
+            data_source_id = data_source_id,
+            node_id = payload.document_id,
+            parents = ?payload.parents,
+            node_type = "document",
+            operation = "upsert",
+            "[KWSEARCH] invariant_first_parent_self"
+        );
+    }
+
     match state
         .store
         .load_data_source(&project, &data_source_id)
@@ -2068,6 +2090,17 @@ async fn tables_upsert(
 ) -> (StatusCode, Json<APIResponse>) {
     let project = project::Project::new_from_id(project_id);
 
+    if payload.parents.get(0) != Some(&payload.table_id) {
+        info!(
+            data_source_id = data_source_id,
+            node_id = payload.table_id,
+            parents = ?payload.parents,
+            node_type = "table",
+            operation = "upsert",
+            "[KWSEARCH] invariant_first_parent_self"
+        );
+    }
+
     match state
         .store
         .upsert_data_source_table(
@@ -2305,6 +2338,17 @@ async fn tables_update_parents(
 ) -> (StatusCode, Json<APIResponse>) {
     let project = project::Project::new_from_id(project_id);
 
+    if payload.parents.get(0) != Some(&table_id) {
+        info!(
+            data_source_id = data_source_id,
+            node_id = table_id,
+            parents = ?payload.parents,
+            node_type = "table",
+            operation = "update_parents",
+            "[KWSEARCH] invariant_first_parent_self"
+        );
+    }
+
     match state
         .store
         .load_data_source_table(&project, &data_source_id, &table_id)
@@ -2690,6 +2734,17 @@ async fn folders_upsert(
 ) -> (StatusCode, Json<APIResponse>) {
     let project = project::Project::new_from_id(project_id);
 
+    if payload.parents.get(0) != Some(&payload.folder_id) {
+        info!(
+            data_source_id = data_source_id,
+            node_id = payload.folder_id,
+            parents = ?payload.parents,
+            node_type = "folder",
+            operation = "upsert",
+            "[KWSEARCH] invariant_first_parent_self"
+        );
+    }
+
     match state
         .store
         .upsert_data_source_folder(
diff --git a/core/src/data_sources/data_source.rs b/core/src/data_sources/data_source.rs
index 9ed0beb832fa..747c45f74fc0 100644
--- a/core/src/data_sources/data_source.rs
+++ b/core/src/data_sources/data_source.rs
@@ -26,7 +26,7 @@ use std::collections::HashMap;
 use std::fmt;
 use std::str::FromStr;
 use tokio_stream::{self as stream};
-use tracing::{error, info, warn};
+use tracing::{error, info};
 use uuid::Uuid;
 
 /// Section is used to represent the structure of document to be taken into account during chunking.
@@ -627,22 +627,6 @@ impl DataSource {
             ))?;
         }
 
-        if parents.is_empty() {
-            warn!(
-                document_id = document_id,
-                timestamp = ?timestamp,
-                parents = ?parents,
-                "Upserting a document without any parent"
-            );
-        } else if parents[0] != document_id {
-            warn!(
-                document_id = document_id,
-                timestamp = ?timestamp,
-                parents = ?parents,
-                "Upserting a document that is not self-referenced as its parent"
-            );
-        }
-
         let store = store.clone();
 
         let current_system_tags = if preserve_system_tags {