From 38272c48d00b1e024cf48ed51331341c7a42f247 Mon Sep 17 00:00:00 2001
From: arkohut <39525455+arkohut@users.noreply.github.com>
Date: Thu, 11 Jul 2024 12:12:54 +0800
Subject: [PATCH] feat(indexing): add api fetch document in typesense

---
 memos/indexing.py             | 32 +++++++++++++++++++++++++++++++-
 memos/initialize_typesense.py |  6 +++---
 memos/server.py               | 17 +++++++++++++++++
 3 files changed, 51 insertions(+), 4 deletions(-)

diff --git a/memos/indexing.py b/memos/indexing.py
index b45ce67..18e985f 100644
--- a/memos/indexing.py
+++ b/memos/indexing.py
@@ -137,7 +137,8 @@ def search_entities(
 
         search_parameters = {
             "q": q,
-            "query_by": "tags,metadata_entries,filepath,filename,embedding",
+            "query_by": "filename,filepath,tags,metadata_entries,embedding",
+            "infix": "always,always,off,off,off",
             "filter_by": f"{filter_by_str} && file_type_group:=image" if filter_by_str else "file_type_group:=image",
             "per_page": limit,
             "page": offset // limit + 1,
@@ -174,3 +175,32 @@ def search_entities(
         raise Exception(
             f"Failed to search entities: {str(e)}",
         )
+
+
+def fetch_entity_by_id(client, id: str) -> EntityIndexItem:
+    try:
+        document = client.collections["entities"].documents[id].retrieve()
+        return EntitySearchResult(
+            id=document["id"],
+            filepath=document["filepath"],
+            filename=document["filename"],
+            size=document["size"],
+            file_created_at=document["file_created_at"],
+            file_last_modified_at=document["file_last_modified_at"],
+            file_type=document["file_type"],
+            file_type_group=document["file_type_group"],
+            last_scan_at=document.get("last_scan_at"),
+            library_id=document["library_id"],
+            folder_id=document["folder_id"],
+            tags=document["tags"],
+            metadata_entries=[
+                MetadataIndexItem(
+                    key=entry["key"], value=entry["value"], source=entry["source"]
+                )
+                for entry in document["metadata_entries"]
+            ],
+        )
+    except Exception as e:
+        raise Exception(
+            f"Failed to fetch document by id: {str(e)}",
+        )
diff --git a/memos/initialize_typesense.py b/memos/initialize_typesense.py
index 8726a4c..35fa317 100644
--- a/memos/initialize_typesense.py
+++ b/memos/initialize_typesense.py
@@ -21,8 +21,8 @@ schema = {
     "name": "entities",
     "enable_nested_fields": True,
     "fields": [
-        {"name": "filepath", "type": "string"},
-        {"name": "filename", "type": "string"},
+        {"name": "filepath", "type": "string", "infix": True},
+        {"name": "filename", "type": "string", "infix": True},
         {"name": "size", "type": "int32"},
         {"name": "file_created_at", "type": "int64", "facet": False},
         {"name": "file_last_modified_at", "type": "int64", "facet": False},
@@ -55,7 +55,7 @@ schema = {
             "optional": True,
         },
     ],
-    "token_separators": [":", "/", ".", " ", "-", "\\"],
+    "token_separators": [":", "/", " ", "\\"],
 }
 
 
diff --git a/memos/server.py b/memos/server.py
index d8c71ec..16e9a30 100644
--- a/memos/server.py
+++ b/memos/server.py
@@ -317,6 +317,23 @@ async def sync_entity_to_typesense(entity_id: int, db: Session = Depends(get_db)
     return None
 
 
+@app.get(
+    "/entities/{entity_id}/index",
+    response_model=EntitySearchResult,
+    tags=["entity"],
+)
+async def get_entity_index(entity_id: int) -> EntityIndexItem:
+    try:
+        entity_index_item = indexing.fetch_entity_by_id(client, entity_id)
+    except Exception as e:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=str(e),
+        )
+
+    return entity_index_item
+
+
 @app.delete(
     "/entities/{entity_id}/index",
     status_code=status.HTTP_204_NO_CONTENT,