Upload 36 files

kimi-js/kimi-llm-manager.js

@@ -933,6 +933,8 @@ class KimiLLMManager {
             ? window.getUnifiedDefaults()
             : { temperature: 0.9, maxTokens: 400, top_p: 0.9, frequency_penalty: 0.9, presence_penalty: 0.8 };
 
+        const enableStreaming = await this.db.getPreference("enableStreaming", true);
+
         const llmSettings = {
             temperature: await this.db.getPreference("llmTemperature", unifiedDefaults.temperature),
             maxTokens: await this.db.getPreference("llmMaxTokens", unifiedDefaults.maxTokens),
@@ -944,7 +946,7 @@ class KimiLLMManager {
         const payload = {
             model: this.currentModel,
             messages: messages,
-            stream: true, // Enable streaming
+            stream: enableStreaming, // Use user preference for streaming
             temperature: typeof options.temperature === "number" ? options.temperature : llmSettings.temperature,
             max_tokens: typeof options.maxTokens === "number" ? options.maxTokens : llmSettings.maxTokens,
             top_p: typeof options.topP === "number" ? options.topP : llmSettings.top_p,
@@ -1061,6 +1063,8 @@ class KimiLLMManager {
             ? window.getUnifiedDefaults()
             : { temperature: 0.9, maxTokens: 400, top_p: 0.9, frequency_penalty: 0.9, presence_penalty: 0.8 };
 
+        const enableStreaming = await this.db.getPreference("enableStreaming", true);
+
         const llmSettings = {
             temperature: await this.db.getPreference("llmTemperature", unifiedDefaults.temperature),
             maxTokens: await this.db.getPreference("llmMaxTokens", unifiedDefaults.maxTokens),
@@ -1072,7 +1076,7 @@ class KimiLLMManager {
         const payload = {
             model: this.currentModel,
             messages: messages,
-            stream: true,
+            stream: enableStreaming,
             temperature: typeof options.temperature === "number" ? options.temperature : llmSettings.temperature,
             max_tokens: typeof options.maxTokens === "number" ? options.maxTokens : llmSettings.maxTokens,
             top_p: typeof options.topP === "number" ? options.topP : llmSettings.top_p,
@@ -1171,6 +1175,7 @@ class KimiLLMManager {
 
     async chatWithLocalStreaming(userMessage, onToken, options = {}) {
         const systemPromptContent = await this.assemblePrompt(userMessage);
+        const enableStreaming = await this.db.getPreference("enableStreaming", true);
 
         const payload = {
             model: this.currentModel || "llama2",
@@ -1179,7 +1184,7 @@ class KimiLLMManager {
                 ...this.conversationContext.slice(-this.maxContextLength),
                 { role: "user", content: userMessage }
             ],
-            stream: true
+            stream: enableStreaming
         };
 
         try {
@@ -1195,36 +1200,47 @@ class KimiLLMManager {
                 throw new Error("Ollama not available");
             }
 
-            const reader = response.body.getReader();
-            const decoder = new TextDecoder();
             let fullResponse = "";
 
-            try {
-                while (true) {
-                    const { done, value } = await reader.read();
-                    if (done) break;
+            if (enableStreaming) {
+                // Streaming mode
+                const reader = response.body.getReader();
+                const decoder = new TextDecoder();
 
-                    const chunk = decoder.decode(value, { stream: true });
-                    const lines = chunk.split("\n").filter(line => line.trim());
+                try {
+                    while (true) {
+                        const { done, value } = await reader.read();
+                        if (done) break;
 
-                    for (const line of lines) {
-                        try {
-                            const parsed = JSON.parse(line);
-                            const content = parsed.message?.content;
-                            if (content) {
-                                fullResponse += content;
-                                onToken(content);
-                            }
-                            if (parsed.done) {
-                                break;
+                        const chunk = decoder.decode(value, { stream: true });
+                        const lines = chunk.split("\n").filter(line => line.trim());
+
+                        for (const line of lines) {
+                            try {
+                                const parsed = JSON.parse(line);
+                                const content = parsed.message?.content;
+                                if (content) {
+                                    fullResponse += content;
+                                    onToken(content);
+                                }
+                                if (parsed.done) {
+                                    break;
+                                }
+                            } catch (parseError) {
+                                console.warn("Failed to parse Ollama streaming chunk:", parseError);
                             }
-                        } catch (parseError) {
-                            console.warn("Failed to parse Ollama streaming chunk:", parseError);
                         }
                     }
+                } finally {
+                    reader.releaseLock();
+                }
+            } else {
+                // Non-streaming mode
+                const data = await response.json();
+                fullResponse = data.message?.content || "";
+                if (fullResponse && onToken) {
+                    onToken(fullResponse);
                 }
-            } finally {
-                reader.releaseLock();
             }
 
             // Add to context

kimi-js/kimi-module.js

@@ -777,7 +777,8 @@ async function loadSettingsData() {
             "llmMaxTokens",
             "llmTopP",
             "llmFrequencyPenalty",
-            "llmPresencePenalty"
+            "llmPresencePenalty",
+            "enableStreaming"
         ];
         const preferences = await kimiDB.getPreferencesBatch(preferenceKeys);
 
@@ -796,6 +797,7 @@ async function loadSettingsData() {
         const llmTopP = preferences.llmTopP !== undefined ? preferences.llmTopP : 0.9;
         const llmFrequencyPenalty = preferences.llmFrequencyPenalty !== undefined ? preferences.llmFrequencyPenalty : 0.9;
         const llmPresencePenalty = preferences.llmPresencePenalty !== undefined ? preferences.llmPresencePenalty : 0.8;
+        const enableStreaming = preferences.enableStreaming !== undefined ? preferences.enableStreaming : true;
 
         // Update UI with voice settings
         const languageSelect = document.getElementById("language-selection");
@@ -811,6 +813,17 @@ async function loadSettingsData() {
         updateSlider("llm-frequency-penalty", llmFrequencyPenalty);
         updateSlider("llm-presence-penalty", llmPresencePenalty);
 
+        // Update streaming toggle
+        const streamingToggle = document.getElementById("enable-streaming");
+        if (streamingToggle) {
+            if (enableStreaming) {
+                streamingToggle.classList.add("active");
+            } else {
+                streamingToggle.classList.remove("active");
+            }
+            streamingToggle.setAttribute("aria-checked", String(enableStreaming));
+        }
+
         // Batch load personality traits
         const traitNames = ["affection", "playfulness", "intelligence", "empathy", "humor", "romance"];
         const personality = await kimiDB.getPersonalityTraitsBatch(traitNames, selectedCharacter);
@@ -832,10 +845,10 @@ async function loadSettingsData() {
         await updateStats();
 
         // Update API key input
-        const apiKeyInput = document.getElementById("openrouter-api-key");
+        const apiKeyInput = document.getElementById("provider-api-key");
         if (apiKeyInput) {
             const keyPref = window.KimiProviderUtils ? window.KimiProviderUtils.getKeyPrefForProvider(provider) : null;
-            const providerKey = keyPref && preferences[keyPref] ? preferences[keyPref] : genericKey;
+            const providerKey = keyPref && preferences[keyPref] ? preferences[keyPref] : apiKey;
             apiKeyInput.value = providerKey || "";
         }
         const providerSelect = document.getElementById("llm-provider");

kimi-js/kimi-script.js

@@ -140,7 +140,14 @@ document.addEventListener("DOMContentLoaded", async function () {
             const baseUrlInput = ApiUi.baseUrlInput();
             const modelIdInput = ApiUi.modelIdInput();
             const apiKeyInput = ApiUi.apiKeyInput();
-            if (baseUrlInput) baseUrlInput.value = baseUrl || "";
+
+            // Set base URL based on modifiability
+            if (baseUrlInput) {
+                const isModifiable = isUrlModifiable(provider);
+                baseUrlInput.value = baseUrl || "";
+                baseUrlInput.disabled = !isModifiable;
+                baseUrlInput.style.opacity = isModifiable ? "1" : "0.6";
+            }
             // Only prefill model for OpenRouter, others should show placeholder only
             if (modelIdInput) {
                 if (provider === "openrouter") {
@@ -193,6 +200,11 @@ document.addEventListener("DOMContentLoaded", async function () {
         }
     });
 
+    // Helper function to check if URL is modifiable for current provider
+    function isUrlModifiable(provider) {
+        return provider === "openai-compatible" || provider === "ollama";
+    }
+
     const providerSelectEl = document.getElementById("llm-provider");
     if (providerSelectEl) {
         providerSelectEl.addEventListener("change", async function (e) {
@@ -241,7 +253,21 @@ document.addEventListener("DOMContentLoaded", async function () {
             const p = placeholders[provider] || placeholders.openai;
             if (baseUrlInput) {
                 baseUrlInput.placeholder = p.url;
-                baseUrlInput.value = provider === "openrouter" ? placeholders.openrouter.url : p.url;
+                // Only allow URL modification for custom and ollama providers
+                const isModifiable = isUrlModifiable(provider);
+
+                if (isModifiable) {
+                    // For custom and ollama: load saved URL or use default
+                    const savedUrl = await window.kimiDB.getPreference("llmBaseUrl", p.url);
+                    baseUrlInput.value = savedUrl;
+                    baseUrlInput.disabled = false;
+                    baseUrlInput.style.opacity = "1";
+                } else {
+                    // For other providers: fixed URL, not modifiable
+                    baseUrlInput.value = p.url;
+                    baseUrlInput.disabled = true;
+                    baseUrlInput.style.opacity = "0.6";
+                }
             }
             if (apiKeyInput) {
                 apiKeyInput.placeholder = p.keyPh;
@@ -263,7 +289,7 @@ document.addEventListener("DOMContentLoaded", async function () {
             }
             if (window.kimiDB) {
                 await window.kimiDB.setPreference("llmProvider", provider);
-                await window.kimiDB.setPreference("llmBaseUrl", provider === "openrouter" ? placeholders.openrouter.url : p.url);
+
                 const apiKeyLabel = document.getElementById("api-key-label");
                 // Load provider-specific key into the input for clarity
                 const keyPref = window.KimiProviderUtils
@@ -296,6 +322,15 @@ document.addEventListener("DOMContentLoaded", async function () {
                     }
                 }
                 ApiUi.clearStatus();
+
+                // Save URL after all UI updates are complete
+                const isModifiable = isUrlModifiable(provider);
+                if (isModifiable && baseUrlInput) {
+                    await window.kimiDB.setPreference("llmBaseUrl", baseUrlInput.value);
+                } else {
+                    // For fixed providers, save the standard URL
+                    await window.kimiDB.setPreference("llmBaseUrl", p.url);
+                }
             }
         });
 
@@ -315,6 +350,27 @@ document.addEventListener("DOMContentLoaded", async function () {
                 }
             });
         }
+
+        // Listen for Base URL changes and save for modifiable providers
+        const baseUrlInput = ApiUi.baseUrlInput();
+        if (baseUrlInput) {
+            baseUrlInput.addEventListener("blur", async function (e) {
+                const providerSelect = ApiUi.providerSelect();
+                const provider = providerSelect ? providerSelect.value : "openrouter";
+                const isModifiable = isUrlModifiable(provider);
+
+                if (isModifiable && window.kimiDB) {
+                    const newUrl = e.target.value.trim();
+                    if (newUrl) {
+                        try {
+                            await window.kimiDB.setPreference("llmBaseUrl", newUrl);
+                        } catch (error) {
+                            console.warn("Failed to save base URL:", error.message);
+                        }
+                    }
+                }
+            });
+        }
     }
 
     // Loading screen management
@@ -773,7 +829,6 @@ document.addEventListener("DOMContentLoaded", async function () {
         [],
         500
     );
-
     kimiInit.register(
         "dataManager",
         async () => {