more configs for chat

shinkai-bin/shinkai-node/src/llm_provider/providers/ollama.rs

@@ -1,7 +1,7 @@
 use crate::llm_provider::execution::chains::inference_chain_trait::{FunctionCall, LLMInferenceResponse};
 use crate::llm_provider::llm_stopper::LLMStopper;
 use crate::llm_provider::providers::shared::ollama_api::{
-    ollama_prepare_messages, ollama_conversation_prepare_messages_with_tooling, OllamaAPIStreamingResponse,
+    ollama_conversation_prepare_messages_with_tooling, ollama_prepare_messages, OllamaAPIStreamingResponse,
 };
 use crate::managers::model_capabilities_manager::PromptResultEnum;
 
@@ -105,8 +105,11 @@ impl LLMService for Ollama {
                 }
             }
 
-            // Conditionally add functions to the payload if tools_json is not empty
-            if !tools_json.is_empty() {
+            // Add this line to extract the use_tools flag from the config
+            let use_tools = config.as_ref().map_or(true, |c| c.use_tools.unwrap_or(true));
+
+            // Conditionally add tools to the payload if use_tools is true
+            if use_tools && !tools_json.is_empty() {
                 payload["tools"] = serde_json::Value::Array(tools_json);
             }
 

shinkai-bin/shinkai-node/src/llm_provider/providers/openai.rs

@@ -4,9 +4,7 @@ use std::sync::Arc;
 use super::super::error::LLMProviderError;
 use super::shared::openai_api::{openai_prepare_messages, MessageContent, OpenAIResponse};
 use super::LLMService;
-use crate::llm_provider::execution::chains::inference_chain_trait::{
-    LLMInferenceResponse, FunctionCall,
-};
+use crate::llm_provider::execution::chains::inference_chain_trait::{FunctionCall, LLMInferenceResponse};
 use crate::llm_provider::llm_stopper::LLMStopper;
 use crate::managers::model_capabilities_manager::PromptResultEnum;
 use async_trait::async_trait;
@@ -100,8 +98,11 @@ impl LLMService for OpenAI {
                     "max_tokens": result.remaining_tokens,
                 });
 
-                // Conditionally add functions to the payload if tools_json is not empty
-                if !tools_json.is_empty() {
+                // Add this line to extract the use_tools flag from the config
+                let use_tools = config.as_ref().map_or(true, |c| c.use_tools.unwrap_or(true));
+
+                // Conditionally add tools to the payload if use_tools is true
+                if use_tools && !tools_json.is_empty() {
                     payload["functions"] = serde_json::Value::Array(tools_json);
                 }
 

shinkai-bin/shinkai-node/src/network/v2_api/api_v2_commands_jobs.rs

@@ -704,6 +704,8 @@ impl Node {
                     top_p: None,
                     stream: None,
                     other_model_params: None,
+                    use_tools: None,
+                    web_search: None,
                 });
                 let _ = res.send(Ok(config)).await;
                 Ok(())

shinkai-bin/shinkai-node/src/network/v2_api/api_v2_commands_prompts.rs

@@ -179,7 +179,7 @@ impl Node {
         let start_time = Instant::now();
 
         // Perform the internal search using LanceShinkaiDb
-        match lance_db.read().await.prompt_vector_search(&query, 5).await {
+        match lance_db.read().await.prompt_vector_search(&query, 10).await {
             Ok(prompts) => {
                 // Set embeddings to None before returning
                 let prompts_without_embeddings: Vec<CustomPrompt> = prompts

shinkai-libs/shinkai-lancedb/Cargo.toml

@@ -11,7 +11,7 @@ uuid = { version = "1.6.1", features = ["v4"] }
 shinkai_tools_primitives = { workspace = true }
 shinkai_vector_resources = { workspace = true }
 shinkai_message_primitives = { workspace = true }
-shinkai_tools_runner = { version = "0.7.14", features = ["built-in-tools"] }
+shinkai_tools_runner = { version = "0.7.15", features = ["built-in-tools"] }
 regex = "1"
 base64 = "0.22.0"
 lancedb = "0.10.0"

shinkai-libs/shinkai-lancedb/src/lance_db/shinkai_prompt_db.rs

@@ -90,7 +90,7 @@ impl LanceShinkaiDb {
         Ok(())
     }
 
-    fn convert_batch_to_prompt(batch: &RecordBatch) -> Option<CustomPrompt> {
+    fn convert_batch_to_prompt(batch: &RecordBatch) -> Vec<CustomPrompt> {
         let name_array = batch
             .column_by_name(ShinkaiPromptSchema::name_field())
             .unwrap()
@@ -134,13 +134,15 @@ impl LanceShinkaiDb {
             .downcast_ref::<FixedSizeListArray>()
             .unwrap();
 
-        if name_array.len() > 0 {
-            let embedding = if vector_array.is_null(0) {
+        let mut prompts = Vec::new();
+
+        for i in 0..name_array.len() {
+            let embedding = if vector_array.is_null(i) {
                 None
             } else {
                 Some(
                     vector_array
-                        .value(0)
+                        .value(i)
                         .as_any()
                         .downcast_ref::<Float32Array>()
                         .unwrap()
@@ -149,18 +151,18 @@ impl LanceShinkaiDb {
                 )
             };
 
-            Some(CustomPrompt {
-                name: name_array.value(0).to_string(),
-                prompt: prompt_array.value(0).to_string(),
-                is_system: is_system_array.value(0),
-                is_enabled: is_enabled_array.value(0),
-                version: version_array.value(0).to_string(),
-                is_favorite: is_favorite_array.value(0),
+            prompts.push(CustomPrompt {
+                name: name_array.value(i).to_string(),
+                prompt: prompt_array.value(i).to_string(),
+                is_system: is_system_array.value(i),
+                is_enabled: is_enabled_array.value(i),
+                version: version_array.value(i).to_string(),
+                is_favorite: is_favorite_array.value(i),
                 embedding,
-            })
-        } else {
-            None
+            });
         }
+
+        prompts
     }
 
     pub async fn get_prompt(&self, name: &str) -> Result<Option<CustomPrompt>, ShinkaiLanceDBError> {
@@ -188,7 +190,8 @@ impl LanceShinkaiDb {
             .map_err(|e| ShinkaiLanceDBError::DatabaseError(e.to_string()))?;
 
         for batch in results {
-            if let Some(prompt) = Self::convert_batch_to_prompt(&batch) {
+            let prompts = Self::convert_batch_to_prompt(&batch);
+            for prompt in prompts {
                 return Ok(Some(prompt));
             }
         }
@@ -281,11 +284,8 @@ impl LanceShinkaiDb {
         let mut prompts = Vec::new();
         let mut res = query;
         while let Some(Ok(batch)) = res.next().await {
-            for _i in 0..batch.num_rows() {
-                if let Some(prompt) = Self::convert_batch_to_prompt(&batch) {
-                    prompts.push(prompt);
-                }
-            }
+            let prompts_batch = Self::convert_batch_to_prompt(&batch);
+            prompts.extend(prompts_batch);
         }
 
         Ok(prompts)
@@ -311,7 +311,12 @@ impl LanceShinkaiDb {
         let fts_query_builder = self
             .prompt_table
             .query()
-            .full_text_search(FullTextSearchQuery::new(query.to_owned()))
+            .full_text_search(FullTextSearchQuery {
+                columns: vec!["prompt".to_string()],
+                query: query.to_owned(),
+                limit: Some(num_results as i64),
+                wand_factor: Some(1.0),
+            })
             .select(Select::columns(&[
                 ShinkaiPromptSchema::name_field(),
                 ShinkaiPromptSchema::prompt_field(),
@@ -357,16 +362,14 @@ impl LanceShinkaiDb {
 
         let mut fts_res = fts_query;
         while let Some(Ok(batch)) = fts_res.next().await {
-            if let Some(prompt) = Self::convert_batch_to_prompt(&batch) {
-                fts_results.push(prompt);
-            }
+            let prompts = Self::convert_batch_to_prompt(&batch);
+            fts_results.extend(prompts);
         }
 
         let mut vector_res = vector_query;
         while let Some(Ok(batch)) = vector_res.next().await {
-            if let Some(prompt) = Self::convert_batch_to_prompt(&batch) {
-                vector_results.push(prompt);
-            }
+            let prompts = Self::convert_batch_to_prompt(&batch);
+            vector_results.extend(prompts);
         }
 
         // Merge results using interleave and remove duplicates
@@ -399,6 +402,21 @@ impl LanceShinkaiDb {
             }
         }
 
+        // Continue to add results from the remaining iterator if needed
+        while combined_results.len() < num_results as usize {
+            if let Some(fts_item) = fts_iter.next() {
+                if seen.insert(fts_item.name.clone()) {
+                    combined_results.push(fts_item);
+                }
+            } else if let Some(vector_item) = vector_iter.next() {
+                if seen.insert(vector_item.name.clone()) {
+                    combined_results.push(vector_item);
+                }
+            } else {
+                break;
+            }
+        }
+
         Ok(combined_results)
     }
 
@@ -422,11 +440,8 @@ impl LanceShinkaiDb {
         let mut prompts = Vec::new();
         let mut res = query;
         while let Some(Ok(batch)) = res.next().await {
-            for _i in 0..batch.num_rows() {
-                if let Some(prompt) = Self::convert_batch_to_prompt(&batch) {
-                    prompts.push(prompt);
-                }
-            }
+            let prompts_batch = Self::convert_batch_to_prompt(&batch);
+            prompts.extend(prompts_batch);
         }
 
         Ok(prompts)

shinkai-libs/shinkai-message-primitives/src/schemas/job_config.rs

@@ -13,4 +13,6 @@ pub struct JobConfig {
     pub top_p: Option<f64>,
     pub stream: Option<bool>,
     pub other_model_params: Option<Value>,
+    pub use_tools: Option<bool>,
+    pub web_search: Option<bool>,
 }