From f27444fe48e579f4f45e94423359dd8f825b641f Mon Sep 17 00:00:00 2001
From: Alex <a@tushynski.me>
Date: Thu, 18 Jun 2026 13:35:37 +0100
Subject: [PATCH] feat: mini llm fixes

---
 application/llm/google_ai.py | 23 +++++++++++++++++++----
 application/llm/openai.py    | 16 ++++++++--------
 2 files changed, 27 insertions(+), 12 deletions(-)

diff --git a/application/llm/google_ai.py b/application/llm/google_ai.py
index e1fcf7bc..21bd1ed3 100644
--- a/application/llm/google_ai.py
+++ b/application/llm/google_ai.py
@@ -545,7 +545,6 @@ class GoogleLLM(BaseLLM):
         **kwargs,
     ):
         """Generate content using Google AI API without streaming."""
-        client = genai.Client(api_key=self.api_key)
         system_instruction = None
         if formatting == "openai":
             messages, system_instruction = self._clean_messages_google(messages)
@@ -562,7 +561,7 @@ class GoogleLLM(BaseLLM):
         if response_schema:
             config.response_schema = response_schema
             config.response_mime_type = "application/json"
-        response = client.models.generate_content(
+        response = self.client.models.generate_content(
             model=model,
             contents=messages,
             config=config,
@@ -585,7 +584,6 @@ class GoogleLLM(BaseLLM):
         **kwargs,
     ):
         """Generate content using Google AI API with streaming."""
-        client = genai.Client(api_key=self.api_key)
         system_instruction = None
         if formatting == "openai":
             messages, system_instruction = self._clean_messages_google(messages)
@@ -607,7 +605,24 @@ class GoogleLLM(BaseLLM):
             config.response_mime_type = "application/json"
         # Check if we have both tools and file attachments
 
-        response = client.models.generate_content_stream(
+        has_attachments = False
+        for message in messages:
+            parts = getattr(message, "parts", None) or []
+            for part in parts:
+                if hasattr(part, "file_data") and part.file_data is not None:
+                    has_attachments = True
+                    break
+            if has_attachments:
+                break
+        messages_summary = self._summarize_messages_for_log(messages)
+        logging.info(
+            "GoogleLLM: Starting stream generation. Model: %s, Messages: %s, Has attachments: %s",
+            model,
+            messages_summary,
+            has_attachments,
+        )
+
+        response = self.client.models.generate_content_stream(
             model=model,
             contents=messages,
             config=config,
diff --git a/application/llm/openai.py b/application/llm/openai.py
index 1ae59632..1ae7c070 100644
--- a/application/llm/openai.py
+++ b/application/llm/openai.py
@@ -1076,8 +1076,6 @@ class OpenAILLM(BaseLLM):
         Returns:
             str: OpenAI file_id for the uploaded file.
         """
-        import logging
-
         if "openai_file_id" in attachment:
             return attachment["openai_file_id"]
         file_path = attachment.get("path")
@@ -1085,12 +1083,14 @@ class OpenAILLM(BaseLLM):
         if not self.storage.file_exists(file_path):
             raise FileNotFoundError(f"File not found: {file_path}")
         try:
-            file_id = self.storage.process_file(
-                file_path,
-                lambda local_path, **kwargs: self.client.files.create(
-                    file=open(local_path, "rb"), purpose="assistants"
-                ).id,
-            )
+            def _upload(local_path, **_kwargs):
+                with open(local_path, "rb") as uploaded_file:
+                    return self.client.files.create(
+                        file=uploaded_file,
+                        purpose="assistants",
+                    ).id
+
+            file_id = self.storage.process_file(file_path, _upload)
 
             # Cache the OpenAI file id on the attachment row so we don't
             # re-upload the same blob on the next LLM call. Prefer the PG