Almaatla commited on
Commit
0bb9e0a
·
verified ·
1 Parent(s): 22ca6a8

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +78 -145
app.py CHANGED
@@ -104,184 +104,117 @@ async def get_proxy():
104
  return HTMLResponse("""
105
  <html>
106
  <body>
107
- <h1>Proxy Client (LLM Gateway)</h1>
108
- <div style="margin-bottom: 20px;">
109
- <input type="password" id="apiKey" placeholder="Enter API Key" style="width: 300px;">
110
- <button onclick="initializeClient()">Fetch Models</button>
111
- </div>
112
- <div style="margin-bottom: 20px;">
113
- <select id="modelSelect" style="width: 300px;">
114
- <option value="" disabled selected>-- Select Model --</option>
115
- </select>
116
- </div>
117
- <div id="status"></div>
118
 
119
- <script>
120
- let agentClient = null;
121
- let currentModel = null;
122
- const systemPrompt = "You are a helpful AI assistant. Respond concisely and accurately.";
123
- const conversationHistory = [];
 
 
 
124
 
125
- function showStatus(message, type = 'info') {
126
- const statusDiv = document.getElementById('status');
127
- statusDiv.innerHTML = `<div style="color: ${type === 'error' ? 'red' : 'orange'}">${message}</div>`;
 
 
 
 
 
128
  }
 
 
 
 
 
 
 
 
 
 
 
 
129
 
130
- function initializeClient() {
131
- const apiKey = document.getElementById('apiKey').value;
132
- if (!apiKey) {
133
- showStatus("Please enter an API key", 'error');
134
- return;
135
- }
136
-
137
- agentClient = new ConversationalAgentClient(apiKey);
138
- agentClient.populateLLMModels()
139
- .then(models => {
140
- agentClient.updateModelSelect('modelSelect', models.find(m => m.includes("gemini-2.5")));
141
- currentModel = document.getElementById('modelSelect').value;
142
- showStatus(`Loaded ${models.length} models. Default: ${currentModel}`);
143
- })
144
- .catch(error => {
145
- showStatus(`Error fetching models: ${error.message}`, 'error');
146
- });
147
  }
148
 
149
- // WebSocket setup
150
  const ws = new WebSocket('wss://' + window.location.host + '/ws');
151
- ws.onopen = () => {
152
- ws.send(JSON.stringify({ source: 'proxy' }));
153
- };
154
-
155
  ws.onmessage = async e => {
156
  const msg = JSON.parse(e.data);
 
 
157
  if (msg.destination === 'proxy') {
 
 
 
158
  try {
159
- showStatus("Processing user query...");
160
- const llmResponse = await agentClient.call(
161
- currentModel,
162
- msg.content,
163
- systemPrompt,
164
- conversationHistory
165
- );
166
 
167
  const responseMsg = {
168
  content: llmResponse.response,
169
  source: 'proxy',
170
- destination: 'user'
171
  };
172
  ws.send(JSON.stringify(responseMsg));
173
- showStatus("Response sent successfully");
 
174
  } catch (error) {
175
- console.error("LLM Error:", error);
176
  const errorResponse = {
177
- content: `Error processing request: ${error.message}`,
178
  source: 'proxy',
179
- destination: 'user'
180
  };
181
  ws.send(JSON.stringify(errorResponse));
182
- showStatus(`Error: ${error.message}`, 'error');
183
  }
184
  }
185
  };
186
 
187
- // Model selection change handler
188
- document.getElementById('modelSelect').addEventListener('change', function() {
189
- currentModel = this.value;
190
- showStatus(`Model changed to: ${currentModel}`);
191
- });
192
-
193
- // --- Include provided client classes here ---
194
- // --- API Client Classes --- (Keep existing classes BaseAgentClient, ConversationalAgentClient)
195
- class BaseAgentClient {
196
- constructor(apiKey, apiUrl = 'https://llm.synapse.thalescloud.io/v1/') { this.apiKey = apiKey; this.apiUrl = apiUrl; this.models = []; this.maxCallsPerMinute = 4; this.callTimestamps = []; }
197
- async fetchLLMModels() { if (!this.apiKey) throw new Error("API Key is not set."); console.log("Fetching models from:", this.apiUrl + 'models'); try { const response = await fetch(this.apiUrl + 'models', { method: 'GET', headers: { 'Authorization': `Bearer ${this.apiKey}` } }); if (!response.ok) { const errorText = await response.text(); console.error("Fetch models error response:", errorText); throw new Error(`HTTP error! Status: ${response.status} - ${errorText}`); } const data = await response.json(); console.log("Models fetched:", data.data); const filteredModels = data.data.map(model => model.id).filter(id => !id.toLowerCase().includes('embed') && !id.toLowerCase().includes('image')); return filteredModels; } catch (error) { console.error('Error fetching LLM models:', error); throw new Error(`Failed to fetch models: ${error.message}`); } }
198
- async populateLLMModels(defaultModel = "gemini-2.5-pro-exp-03-25") { try { const modelList = await this.fetchLLMModels(); const sortedModels = modelList.sort((a, b) => { if (a === defaultModel) return -1; if (b === defaultModel) return 1; return a.localeCompare(b); }); const finalModels = []; if (sortedModels.includes(defaultModel)) { finalModels.push(defaultModel); sortedModels.forEach(model => { if (model !== defaultModel) finalModels.push(model); }); } else { finalModels.push(defaultModel); finalModels.push(...sortedModels); } this.models = finalModels; console.log("Populated models:", this.models); return this.models; } catch (error) { console.error("Error populating models:", error); this.models = [defaultModel]; throw error; } }
199
- updateModelSelect(elementId = 'modelSelect', selectedModel = null) { const select = document.getElementById(elementId); if (!select) { console.warn(`Element ID ${elementId} not found.`); return; } const currentSelection = selectedModel || select.value || this.models[0]; select.innerHTML = ''; if (this.models.length === 0 || (this.models.length === 1 && this.models[0] === "gemini-2.5-pro-exp-03-25" && !this.apiKey)) { const option = document.createElement('option'); option.value = ""; option.textContent = "-- Fetch models first --"; option.disabled = true; select.appendChild(option); return; } this.models.forEach(model => { const option = document.createElement('option'); option.value = model; option.textContent = model; if (model === currentSelection) option.selected = true; select.appendChild(option); }); if (!select.value && this.models.length > 0) select.value = this.models[0]; }
200
- async rateLimitWait() { const currentTime = Date.now(); this.callTimestamps = this.callTimestamps.filter(ts => currentTime - ts <= 60000); if (this.callTimestamps.length >= this.maxCallsPerMinute) { const waitTime = 60000 - (currentTime - this.callTimestamps[0]); const waitSeconds = Math.ceil(waitTime / 1000); const waitMessage = `Rate limit (${this.maxCallsPerMinute}/min) reached. Waiting ${waitSeconds}s...`; console.log(waitMessage); showGenerationStatus(waitMessage, 'warn'); await new Promise(resolve => setTimeout(resolve, waitTime + 100)); showGenerationStatus('Resuming after rate limit wait...', 'info'); this.callTimestamps = this.callTimestamps.filter(ts => Date.now() - ts <= 60000); } }
201
- async callAgent(model, messages, temperature = 0.7) { await this.rateLimitWait(); const startTime = Date.now(); console.log("Calling Agent:", model); try { const response = await fetch(this.apiUrl + 'chat/completions', { method: 'POST', headers: { 'Content-Type': 'application/json', 'Authorization': `Bearer ${this.apiKey}` }, body: JSON.stringify({ model: model, messages: messages, temperature: temperature }) }); const endTime = Date.now(); this.callTimestamps.push(endTime); console.log(`API call took ${endTime - startTime} ms`); if (!response.ok) { const errorData = await response.json().catch(() => ({ error: { message: response.statusText } })); console.error("API Error:", errorData); throw new Error(errorData.error?.message || `API failed: ${response.status}`); } const data = await response.json(); if (!data.choices || !data.choices[0]?.message) throw new Error("Invalid API response structure"); console.log("API Response received."); return data.choices[0].message.content; } catch (error) { this.callTimestamps.push(Date.now()); console.error('Error calling agent:', error); throw error; } }
202
- setMaxCallsPerMinute(value) { const parsedValue = parseInt(value, 10); if (!isNaN(parsedValue) && parsedValue > 0) { console.log(`Max calls/min set to: ${parsedValue}`); this.maxCallsPerMinute = parsedValue; return true; } console.warn(`Invalid max calls/min: ${value}`); return false; }
203
- }
204
- class ConversationalAgentClient extends BaseAgentClient {
205
- constructor(apiKey, apiUrl = 'https://llm.synapse.thalescloud.io/v1/') {
206
- super(apiKey, apiUrl);
207
- }
208
-
209
- async call(model, userPrompt, systemPrompt, conversationHistory = [], temperature = 0.7) {
210
- const messages = [
211
- { role: 'system', content: systemPrompt },
212
- ...conversationHistory,
213
- { role: 'user', content: userPrompt }
214
- ];
215
-
216
- const assistantResponse = await super.callAgent(model, messages, temperature);
217
-
218
- const updatedHistory = [
219
- ...conversationHistory,
220
- { role: 'user', content: userPrompt },
221
- { role: 'assistant', content: assistantResponse }
222
- ];
223
-
224
- return {
225
- response: assistantResponse,
226
- history: updatedHistory
227
- };
228
- }
229
-
230
- async callWithCodeContext(
231
- model,
232
- userPrompt,
233
- systemPrompt,
234
- selectedCodeVersionsData = [],
235
- conversationHistory = [],
236
- temperature = 0.7
237
- ) {
238
- let codeContext = "";
239
- let fullSystemPrompt = systemPrompt || "";
240
-
241
- if (selectedCodeVersionsData && selectedCodeVersionsData.length > 0) {
242
- codeContext = `Code context (chronological):\n\n`;
243
-
244
- selectedCodeVersionsData.forEach((versionData, index) => {
245
- if (versionData && typeof versionData.code === 'string') {
246
- codeContext += `--- Part ${index + 1} (${versionData.version || '?'}) ---\n`;
247
- codeContext += `${versionData.code}\n\n`;
248
- } else {
249
- console.warn(`Invalid context version data at index ${index}`);
250
- }
251
- });
252
-
253
- codeContext += `-------- end context ---\n\nUser request based on context:\n\n`;
254
- }
255
-
256
- const fullPrompt = codeContext + userPrompt;
257
-
258
- const messages = [
259
- { role: 'system', content: fullSystemPrompt },
260
- ...conversationHistory,
261
- { role: 'user', content: fullPrompt }
262
- ];
263
-
264
- const assistantResponse = await super.callAgent(model, messages, temperature);
265
-
266
- const updatedHistory = [
267
- ...conversationHistory,
268
- { role: 'user', content: fullPrompt },
269
- { role: 'assistant', content: assistantResponse }
270
- ];
271
-
272
- return {
273
- response: assistantResponse,
274
- history: updatedHistory
275
- };
276
  }
277
- }
278
-
279
  </script>
280
  </body>
281
  </html>
282
  """)
283
 
284
 
 
285
  @app.post("/v1/chat/completions", response_model=ChatCompletionResponse)
286
  async def chat_completions(request: ChatCompletionRequest):
287
  request_id = str(uuid.uuid4())
 
104
  return HTMLResponse("""
105
  <html>
106
  <body>
107
+ <h1>Proxy Client (Message Gateway)</h1>
108
+ <div style="display: grid; grid-template-columns: 1fr 1fr; gap: 20px; height: 80vh;">
109
+ <!-- Connection Panel -->
110
+ <div style="border-right: 1px solid #ccc; padding-right: 20px;">
111
+ <div style="margin-bottom: 20px;">
112
+ <input type="password" id="apiKey" placeholder="Enter API Key" style="width: 100%;">
113
+ <button onclick="initializeClient()" style="margin-top: 10px;">Fetch Models</button>
114
+ </div>
115
+ <select id="modelSelect" style="width: 100%; margin-bottom: 20px;"></select>
116
+ <div id="systemStatus" style="color: #666; font-size: 0.9em;"></div>
117
+ </div>
118
 
119
+ <!-- Message Flow Visualization -->
120
+ <div style="display: flex; flex-direction: column; height: 100%;">
121
+ <div id="messageFlow" style="flex: 1; border: 1px solid #eee; padding: 10px; overflow-y: auto; background: #f9f9f9;">
122
+ <div style="text-align: center; color: #999; margin-bottom: 10px;">Message Flow</div>
123
+ </div>
124
+ <div id="detailedStatus" style="color: #666; font-size: 0.9em; margin-top: 10px;"></div>
125
+ </div>
126
+ </div>
127
 
128
+ <style>
129
+ .message-entry {
130
+ margin: 5px 0;
131
+ padding: 8px;
132
+ border-radius: 8px;
133
+ background: white;
134
+ box-shadow: 0 2px 4px rgba(0,0,0,0.1);
135
+ font-family: monospace;
136
  }
137
+ .incoming { border-left: 4px solid #4CAF50; }
138
+ .outgoing { border-left: 4px solid #2196F3; }
139
+ .system { border-left: 4px solid #9C27B0; }
140
+ .error { border-left: 4px solid #F44336; }
141
+ .message-header {
142
+ display: flex;
143
+ justify-content: space-between;
144
+ font-size: 0.8em;
145
+ color: #666;
146
+ margin-bottom: 4px;
147
+ }
148
+ </style>
149
 
150
+ <script>
151
+ function addMessageEntry(direction, source, destination, content) {
152
+ const flowDiv = document.getElementById('messageFlow');
153
+ const timestamp = new Date().toLocaleTimeString();
154
+
155
+ const entry = document.createElement('div');
156
+ entry.className = `message-entry ${direction}`;
157
+ entry.innerHTML = `
158
+ <div class="message-header">
159
+ <span>${source} ${destination}</span>
160
+ <span>${timestamp}</span>
161
+ </div>
162
+ <div style="white-space: pre-wrap;">${content}</div>
163
+ `;
164
+
165
+ flowDiv.appendChild(entry);
166
+ flowDiv.scrollTop = flowDiv.scrollHeight;
167
  }
168
 
169
+ // Modified WebSocket handler
170
  const ws = new WebSocket('wss://' + window.location.host + '/ws');
 
 
 
 
171
  ws.onmessage = async e => {
172
  const msg = JSON.parse(e.data);
173
+
174
+ // Display incoming messages
175
  if (msg.destination === 'proxy') {
176
+ addMessageEntry('incoming', msg.source, 'proxy', msg.content);
177
+ document.getElementById('detailedStatus').textContent = `Processing ${msg.source} request...`;
178
+
179
  try {
180
+ const llmResponse = await agentClient.call(currentModel, msg.content, systemPrompt, conversationHistory);
181
+
182
+ // Display outgoing response
183
+ addMessageEntry('outgoing', 'proxy', msg.source, llmResponse.response);
 
 
 
184
 
185
  const responseMsg = {
186
  content: llmResponse.response,
187
  source: 'proxy',
188
+ destination: msg.source
189
  };
190
  ws.send(JSON.stringify(responseMsg));
191
+
192
+ document.getElementById('detailedStatus').textContent = `Response sent to ${msg.source}`;
193
  } catch (error) {
194
+ addMessageEntry('error', 'system', 'proxy', `Error: ${error.message}`);
195
  const errorResponse = {
196
+ content: `Error: ${error.message}`,
197
  source: 'proxy',
198
+ destination: msg.source
199
  };
200
  ws.send(JSON.stringify(errorResponse));
 
201
  }
202
  }
203
  };
204
 
205
+ // Modified model initialization
206
+ function showStatus(message, type = 'info') {
207
+ const statusDiv = document.getElementById('systemStatus');
208
+ statusDiv.innerHTML = `<div style="color: ${type === 'error' ? '#F44336' : '#4CAF50'}">${message}</div>`;
209
+ addMessageEntry('system', 'system', 'proxy', message);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
210
  }
 
 
211
  </script>
212
  </body>
213
  </html>
214
  """)
215
 
216
 
217
+
218
  @app.post("/v1/chat/completions", response_model=ChatCompletionResponse)
219
  async def chat_completions(request: ChatCompletionRequest):
220
  request_id = str(uuid.uuid4())