cont : cleanup (#16)

Author: ggerganov

package.json

@@ -114,26 +114,11 @@
           "default": true,
           "description": "If code completion should be trggered automatically (true) or only by pressing Ctrl+l."
         },
-       "llama-vscode.api_key": {
+        "llama-vscode.api_key": {
           "type": "string",
           "default": "",
           "description": "llama.cpp server api key or OpenAI endpoint API key (optional)"
         },
-        "llama-vscode.is_openai_compatible": {
-          "type": "boolean",
-          "default": false,
-          "description": "If the server exposes an OpenAI API compatible endpoint."
-        },
-        "llama-vscode.openai_client_model": {
-          "type": "string",
-          "default": "",
-          "description": "The FIM friendly model supported by your OpenAI compatible endpoint to be used (e.g., Qwen2.5-Coder-14B-4-bit)"
-        },
-        "llama-vscode.openai_prompt_template": {
-          "type": "string",
-          "default": "<|fim_prefix|>{inputPrefix}{prompt}<|fim_suffix|>{inputSuffix}<|fim_middle|>",
-          "description": "The prompt template to be used for the OpenAI compatible endpoint."
-        },
         "llama-vscode.n_prefix": {
           "type": "number",
           "default": 256,
@@ -204,7 +189,6 @@
           "default": true,
           "description": "Enable/disable completions"
         },
-      
         "llama-vscode.languageSettings": {
             "type": "object",
             "default": {
@@ -214,8 +198,23 @@
               "type": "boolean"
             },
             "description": "Enable/disable suggestions for specific languages"
-          }
+        },
+        "llama-vscode.use_openai_endpoint": {
+          "type": "boolean",
+          "default": false,
+          "description": "[EXPERIMENTAL] Use OAI endpoint. Slow and poor quality - avoid using"
+        },
+        "llama-vscode.openai_client_model": {
+          "type": "string",
+          "default": "",
+          "description": "The FIM friendly model supported by your OpenAI compatible endpoint to be used (e.g., Qwen2.5-Coder-14B-4-bit)"
+        },
+        "llama-vscode.openai_prompt_template": {
+          "type": "string",
+          "default": "<|fim_prefix|>{inputPrefix}{prompt}<|fim_suffix|>{inputSuffix}<|fim_middle|>",
+          "description": "The prompt template to be used for the OpenAI compatible endpoint."
         }
+      }
     }
   },
   "scripts": {

src/architect.ts

@@ -677,10 +677,10 @@ export class Architect {
 
     // cut part of the suggestion in some special cases
     updateSuggestion = (suggestionLines: string[], lineSuffix: string) => {
-        if (lineSuffix.trim() != ""){
+        if (lineSuffix.trim() != "") {
             if (suggestionLines[0].endsWith(lineSuffix)) return suggestionLines[0].slice(0, -lineSuffix.length);
             if (suggestionLines.length > 1) return suggestionLines[0];
-        } 
+        }
 
         return suggestionLines.join("\n");
     }

src/configuration.ts

@@ -5,10 +5,6 @@ export class Configuration {
     // extension configs
     enabled = true;
     endpoint = "http=//127.0.0.1:8012";
-    is_openai_compatible = false;
-    openai_client: OpenAI | null = null;
-    openai_client_model: string = "";
-    openai_prompt_template: string = "<|fim_prefix|>{inputPrefix}{prompt}<|fim_suffix|>{inputSuffix}<|fim_middle|>";
     auto = true;
     api_key = "";
     n_prefix = 256;
@@ -24,9 +20,19 @@ export class Configuration {
     ring_scope = 1024;
     ring_update_ms = 1000;
     language = "en";
+
+    // experimental - avoid using
+    use_openai_endpoint = false;
+    openai_client: OpenAI | null = null;
+    openai_client_model: string = "";
+    openai_prompt_template: string = "<|fim_prefix|>{inputPrefix}{prompt}<|fim_suffix|>{inputSuffix}<|fim_middle|>";
+
     // additional configs
+    // TODO: change to snake_case for consistency
     axiosRequestConfig = {};
     disabledLanguages: string[] = [];
+
+    // TODO: change to snake_case for consistency
     RING_UPDATE_MIN_TIME_LAST_COMPL = 3000;
     MIN_TIME_BETWEEN_COMPL = 600;
     MAX_LAST_PICK_LINE_DISTANCE = 32;
@@ -82,7 +88,7 @@ export class Configuration {
     private updateConfigs = (config: vscode.WorkspaceConfiguration) => {
         // TODO Handle the case of wrong types for the configuration values
         this.endpoint = this.trimTrailingSlash(String(config.get<string>("endpoint")));
-        this.is_openai_compatible = Boolean(config.get<boolean>("is_openai_compatible"));
+        this.use_openai_endpoint = Boolean(config.get<boolean>("use_openai_endpoint"));
         this.openai_client_model = String(config.get<string>("openai_client_model"));
         this.openai_prompt_template = String(config.get<string>("openai_prompt_template"));
         this.auto = Boolean(config.get<boolean>("auto"));
@@ -139,7 +145,7 @@ export class Configuration {
 
     setOpenAiClient = () => {
         this.openai_client = null;
-        if (this.is_openai_compatible) {
+        if (this.use_openai_endpoint) {
             const openai = new OpenAI({
                 apiKey: this.api_key || "empty",
                 baseURL: this.endpoint,

src/llama-server.ts

@@ -103,7 +103,7 @@ export class LlamaServer {
         nindent: number
     ): Promise<LlamaResponse | undefined> => {
         // If the server is OpenAI compatible, use the OpenAI API to get the completion
-        if (this.extConfig.is_openai_compatible) {
+        if (this.extConfig.use_openai_endpoint) {
             const response = await this.handleOpenAICompletion(chunks, inputPrefix, inputSuffix, prompt);
             return response || undefined;
         }
@@ -120,7 +120,7 @@ export class LlamaServer {
 
     prepareLlamaForNextCompletion = (chunks: any[]): void => {
         // If the server is OpenAI compatible, use the OpenAI API to prepare for the next FIM
-        if (this.extConfig.is_openai_compatible) {
+        if (this.extConfig.use_openai_endpoint) {
             // wtg 20250207 - per @igardev ... "This makes sense only if there is a server cache"
             // this.handleOpenAICompletion(chunks, "", "", "", true);
             return;