feat: completed image ingestion with system settings

styling

common/migrations/20250627_120000_add_image_processing_settings.surql

@@ -0,0 +1,7 @@
+DEFINE FIELD IF NOT EXISTS image_processing_model ON system_settings TYPE string;
+DEFINE FIELD IF NOT EXISTS image_processing_prompt ON system_settings TYPE string;
+
+UPDATE system_settings:current SET
+	image_processing_model =  "gpt-4o-mini",
+    image_processing_prompt = "Analyze this image and respond based on its primary content:\n - If the image is mainly text (document, screenshot, sign), transcribe the text verbatim.\n - If the image is mainly visual (photograph, art, landscape), provide a concise description of the scene.\n - For hybrid images (diagrams, ads), briefly describe the visual, then transcribe the text under a Text: heading.\n\n Respond directly with the analysis."
+WHERE image_processing_model == NONE && image_processing_prompt == NONE;

common/migrations/definitions/20250627_120000_add_image_processing_settings.json

@@ -0,0 +1 @@
+{"schemas":"--- original\n+++ modified\n@@ -157,10 +157,12 @@\n DEFINE FIELD IF NOT EXISTS require_email_verification ON system_settings TYPE bool;\n DEFINE FIELD IF NOT EXISTS query_model ON system_settings TYPE string;\n DEFINE FIELD IF NOT EXISTS processing_model ON system_settings TYPE string;\n+DEFINE FIELD IF NOT EXISTS image_processing_model ON system_settings TYPE string;\n DEFINE FIELD IF NOT EXISTS embedding_model ON system_settings TYPE string;\n DEFINE FIELD IF NOT EXISTS embedding_dimensions ON system_settings TYPE int;\n DEFINE FIELD IF NOT EXISTS query_system_prompt ON system_settings TYPE string;\n DEFINE FIELD IF NOT EXISTS ingestion_system_prompt ON system_settings TYPE string;\n+DEFINE FIELD IF NOT EXISTS image_processing_prompt ON system_settings TYPE string;\n\n # Defines the schema for the 'text_chunk' table.\n\n","events":null}

common/schemas/system_settings.surql

@@ -7,7 +7,9 @@ DEFINE FIELD IF NOT EXISTS registrations_enabled ON system_settings TYPE bool;
 DEFINE FIELD IF NOT EXISTS require_email_verification ON system_settings TYPE bool;
 DEFINE FIELD IF NOT EXISTS query_model ON system_settings TYPE string;
 DEFINE FIELD IF NOT EXISTS processing_model ON system_settings TYPE string;
+DEFINE FIELD IF NOT EXISTS image_processing_model ON system_settings TYPE string;
 DEFINE FIELD IF NOT EXISTS embedding_model ON system_settings TYPE string;
 DEFINE FIELD IF NOT EXISTS embedding_dimensions ON system_settings TYPE int;
 DEFINE FIELD IF NOT EXISTS query_system_prompt ON system_settings TYPE string;
 DEFINE FIELD IF NOT EXISTS ingestion_system_prompt ON system_settings TYPE string;
+DEFINE FIELD IF NOT EXISTS image_processing_prompt ON system_settings TYPE string;

common/src/storage/types/system_prompts.rs

@@ -54,3 +54,10 @@ Guidelines:
 7. Only create relationships between existing KnowledgeEntities.
 8. Entities that exist already in the database should NOT be created again. If there is only a minor overlap, skip creating a new entity.
 9. A new relationship MUST include a newly created KnowledgeEntity."#;
+
+pub static DEFAULT_IMAGE_PROCESSING_PROMPT: &str = r#"Analyze this image and respond based on its primary content:
+- If the image is mainly text (document, screenshot, sign), transcribe the text verbatim.
+- If the image is mainly visual (photograph, art, landscape), provide a concise description of the scene.
+- For hybrid images (diagrams, ads), briefly describe the visual, then transcribe the text under a "Text:" heading.
+
+Respond directly with the analysis."#;

common/src/storage/types/system_settings.rs

@@ -15,6 +15,8 @@ pub struct SystemSettings {
     pub embedding_dimensions: u32,
     pub query_system_prompt: String,
     pub ingestion_system_prompt: String,
+    pub image_processing_model: String,
+    pub image_processing_prompt: String,
 }
 
 impl StoredObject for SystemSettings {
@@ -78,6 +80,7 @@ mod tests {
         assert_eq!(settings.require_email_verification, false);
         assert_eq!(settings.query_model, "gpt-4o-mini");
         assert_eq!(settings.processing_model, "gpt-4o-mini");
+        assert_eq!(settings.image_processing_model, "gpt-4o-mini");
         // Dont test these for now, having a hard time getting the formatting exactly the same
         // assert_eq!(
         //     settings.query_system_prompt,