fix: parsing file extension

backend/api/quivr_api/modules/knowledge/dto/inputs.py

@@ -26,14 +26,13 @@ class CreateKnowledgeProperties(BaseModel):
 class AddKnowledge(BaseModel):
     file_name: Optional[str] = None
     url: Optional[str] = None
-    extension: str = ".txt"
+    is_folder: bool = False
     source: str = "local"
     source_link: Optional[str] = None
-    metadata: Optional[Dict[str, str]] = None
-    is_folder: bool = False
     parent_id: UUID | None = None
     sync_id: int | None = None
     sync_file_id: str | None = None
+    metadata: Optional[Dict[str, str]] = None
 
 
 class KnowledgeUpdate(BaseModel):

backend/api/quivr_api/modules/knowledge/service/knowledge_service.py

@@ -35,6 +35,7 @@
 )
 from quivr_api.modules.sync.entity.sync_models import SyncFile, SyncType
 from quivr_api.modules.upload.service.upload_file import check_file_exists
+from quivr_api.utils.knowledge_utils import parse_file_extension
 
 logger = get_logger(__name__)
 
@@ -95,7 +96,6 @@ async def create_or_link_sync_knowledge(
                 knowledge_to_add=AddKnowledge(
                     file_name=sync_file.name,
                     is_folder=sync_file.is_folder,
-                    extension=sync_file.extension,
                     source=parent_knowledge.source,  # same as parent
                     source_link=sync_file.web_view_link,
                     parent_id=parent_knowledge.id,
@@ -183,13 +183,19 @@ async def create_knowledge(
                     if b.brain_id not in {b.brain_id for b in brains}
                 ]
             )
+        # TODO: slugify url names here !!
+        extension = (
+            parse_file_extension(knowledge_to_add.file_name)
+            if knowledge_to_add.file_name
+            else ""
+        )
 
         knowledgedb = KnowledgeDB(
             user_id=user_id,
             file_name=knowledge_to_add.file_name,
             is_folder=knowledge_to_add.is_folder,
             url=knowledge_to_add.url,
-            extension=knowledge_to_add.extension,
+            extension=extension,
             source=knowledge_to_add.source,
             source_link=knowledge_to_add.source_link,
             file_size=upload_file.size if upload_file else 0,

backend/api/quivr_api/modules/knowledge/tests/test_knowledge_service.py

@@ -549,7 +549,7 @@ async def test_create_knowledge_folder(session: AsyncSession, user: User):
     service = KnowledgeService(repository, storage)
 
     km_to_add = AddKnowledge(
-        file_name="test",
+        file_name="test.txt",
         source="local",
         is_folder=True,
         parent_id=None,
@@ -566,9 +566,9 @@ async def test_create_knowledge_folder(session: AsyncSession, user: User):
     assert km.id
     # Knowledge properties
     assert km.file_name == km_to_add.file_name
+    assert km.extension == ".txt"
     assert km.is_folder == km_to_add.is_folder
     assert km.url == km_to_add.url
-    assert km.extension == km_to_add.extension
     assert km.source == km_to_add.source
     assert km.file_size == 128
     assert km.metadata_ == km_to_add.metadata
@@ -613,7 +613,6 @@ def _send_task(*args, **kwargs):
     assert km.file_name == km_to_add.file_name
     assert km.is_folder == km_to_add.is_folder
     assert km.url == km_to_add.url
-    assert km.extension == km_to_add.extension
     assert km.source == km_to_add.source
     assert km.file_size == 128
     assert km.metadata_ == km_to_add.metadata

backend/api/quivr_api/utils/knowledge_utils.py

@@ -0,0 +1,10 @@
+from quivr_core.files.file import FileExtension
+
+
+def parse_file_extension(file_name: str) -> FileExtension | str:
+    if file_name.startswith(".") and file_name.count(".") == 1:
+        return ""
+    if "." not in file_name or file_name.endswith("."):
+        return ""
+
+    return FileExtension(f".{file_name.split('.')[-1]}")