[{"_1":2,"_214":-5,"_215":-5},"loaderData",{"_3":4,"_91":92,"_193":194},"root",{"_5":6,"_34":35,"_64":65},"ctx",{"_7":8,"_23":-5,"_24":25,"_26":27,"_28":22,"_29":22,"_30":22,"_31":32,"_33":22},"auth",{"_9":10,"_11":-5,"_12":-5,"_13":-5,"_14":-5,"_15":-5,"_16":-5,"_17":-5,"_18":-5,"_19":-5,"_20":-5,"_21":22},"tokenType","session_token","sessionClaims","sessionId","sessionStatus","userId","actor","orgId","orgRole","orgSlug","orgPermissions","factorVerificationAge","isAuthenticated",false,"user","preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","notificationCount",0,"flagDenols","ENV",{"_36":37,"_38":39,"_40":41,"_42":43,"_44":45,"_46":47,"_48":49,"_50":51,"_52":53,"_54":55,"_56":57,"_58":59,"_60":61,"_62":63},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","VAL_RUNTIME_NAME","prod","RENDER_GIT_COMMIT","1351d8f1f447bd5e330991ab31e39fcdf7d5b4b4","RENDER_GIT_BRANCH","main","CLERK_PUBLISHABLE_KEY","pk_live_Y2xlcmsudmFsLnRvd24k","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","API_URL","https://api.val.town","clerkState",{"_66":67},"__internal_clerk_state",{"_68":69,"_70":45,"_71":72,"_73":72,"_74":22,"_75":72,"_76":72,"_77":72,"_78":72,"_79":80,"_89":22,"_90":22},"__clerk_ssr_state",{"_9":10,"_11":-5,"_12":-5,"_13":-5,"_14":-5,"_15":-5,"_16":-5,"_17":-5,"_18":-5,"_19":-5,"_20":-5,"_21":22},"__publishableKey","__proxyUrl","","__domain","__isSatellite","__signInUrl","__signUpUrl","__afterSignInUrl","__afterSignUpUrl","__clerk_debug",{"_81":22,"_21":22,"_82":72,"_83":84,"_85":72,"_86":45,"_87":22,"_88":72},"isSignedIn","proxyUrl","reason","session-token-and-uat-missing","message","publishableKey","isSatellite","domain","__telemetryDisabled","__telemetryDebug","routes/_app.x.$handle.$name",{"_93":94,"_95":96,"_97":-7,"_98":72,"_99":-7,"_100":101,"_146":147,"_169":170,"_171":172,"_181":182,"_186":-5,"_187":-5,"_188":-5,"_189":22,"_190":22,"_191":192},"handle","zst","name","untitled-686","branchName","path","version","project",{"_102":103,"_95":96,"_104":105,"_106":-5,"_107":22,"_108":109,"_110":-5,"_111":-5,"_14":112,"_113":-5,"_114":22,"_115":-5,"_116":32,"_16":117,"_118":119,"_120":119,"_121":122,"_123":122,"_124":122,"_125":119,"_126":119,"_23":127,"_130":131,"_133":134,"_145":22},"id","d473c530-ae21-11f0-bb98-0224a6c84d84","createdAt",["D",1761012076994],"deletedAt","legacyEnvVars","privacy","public","description","imageUrl","019a027c-c8e2-73b1-ac0a-aff5ebdfa026","pinnedFileId","pinPreview","screenshotVersion","likeCount","019a027c-c8e2-73b1-ac0a-b22002b937c7","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm",{"_93":94,"_128":129},"profileImageUrl","https://img.clerk.com/eyJ0eXBlIjoiZGVmYXVsdCIsImlpZCI6Imluc18yTm10cGhQUklwU0tqZHNiOHNnNGd2ZjJURFUiLCJyaWQiOiJ1c2VyXzM0TDFxQ0tJaUVXeFV5VnZaYmZ6MTNpOHZkSCJ9","org",{"_93":94,"_132":129,"_14":112},"avatar","branch",{"_99":135,"_95":43,"_102":136,"_137":103,"_138":-5,"_139":-5,"_140":-5,"_141":-5,"_142":-5,"_143":96,"_144":94},4,"d4742dfe-ae21-11f0-bb98-0224a6c84d84","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","isOwned","tree",[148],{"_102":149,"_95":150,"_151":152,"_153":-5,"_154":136,"_155":135,"_156":157,"_104":158,"_99":159,"_160":161,"_162":163,"_164":22,"_165":-5,"_166":167,"_98":168},"e254fb74-ae21-11f0-945f-0224a6c84d84","main.ts","fileId","d476d9f0-ae21-11f0-bb98-0224a6c84d84","parentId","branchId","branchVersion","latest",true,["D",1761012100255],3,"fileType","http","content","// Akash Network OpenAI API Proxy - Val Town 真流式 SSE 版本\n// 本版仅修改“输出格式/解析”部分：只使用上游返回中以 \"0:\" 开头的行进行拼接（thinking 与正文均来自 0: 行）。\n// 其他逻辑保持不变：仅发送最后一条 user 消息 + 保留最后一条 system；温度/TopP 动态传递并夹逼 temperature=0 为 0.01。\nimport { v4 as uuidv4 } from \"npm:uuid\";\nimport { fetch } from \"https://esm.town/v/std/fetch\";\n\nclass AkashOpenAIAPI {\n private baseUrl: string;\n private authUrl: string;\n private modelsUrl: string;\n private sessionId: string;\n private headers: Record;\n private availableModels: string[] = [];\n\n constructor(baseUrl: string = \"https://chat.akash.network/api/chat/\") {\n this.baseUrl = baseUrl;\n this.authUrl = \"https://chat.akash.network/api/auth/session/\";\n this.modelsUrl = \"https://chat.akash.network/api/models/\";\n this.sessionId = uuidv4();\n this.headers = {\n \"User-Agent\":\n \"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/139.0.0.0 Safari/537.36 Edg/139.0.0.0\",\n \"Content-Type\": \"application/json\",\n \"origin\": \"https://chat.akash.network\",\n \"referer\": \"https://chat.akash.network/\",\n \"accept-language\": \"zh-CN,zh;q=0.9,en;q=0.8,en-GB;q=0.7,en-US;q=0.6\",\n // IMPORTANT: 替换成你的有效 Cookie，或通过 /v1/auth/session POST 注入\n \"Cookie\": \"_ga=...; session_token=...;\",\n };\n }\n\n /**\n * 【已修复】从 Akash Network API 获取所有可用模型\n * @returns {Promise} 可用模型的 ID 列表\n */\n private async fetchAvailableModels(): Promise {\n console.log(\"🔄 正在从 Akash Network 获取可用模型列表...\");\n try {\n const resp = await fetch(this.modelsUrl, {\n method: \"GET\",\n headers: this.headers,\n });\n\n if (!resp.ok) {\n throw new Error(`获取模型列表失败: ${resp.status} ${resp.statusText}`);\n }\n\n // API 返回的是直接的模型数组 [ {...}, {...} ]\n const modelsArray: any[] = await resp.json();\n\n // 健壮性检查：确保返回的是一个数组\n if (!Array.isArray(modelsArray)) {\n console.error(\"❌ API 返回的不是预期的数组格式。实际收到的数据:\", modelsArray);\n throw new Error(\"API 返回的数据格式无效\");\n }\n\n // 直接从数组中筛选和映射\n const models = modelsArray\n .filter((model: any) => model.available === true)\n .map((model: any) => model.id);\n\n console.log(`✅ 成功获取 ${models.length} 个可用模型:`, models.join(\", \"));\n return models;\n } catch (error) {\n console.error(\"❌ 获取模型列表时发生错误，将使用回退模型列表:\", error);\n // 提供一个已知可用的回退列表，确保服务不中断\n return [\n \"Qwen3-Next-80B-A3B-Instruct\",\n \"DeepSeek-V3.1\",\n \"Qwen3-235B-A22B-Instruct-2507-FP8\",\n \"openai-gpt-oss-120b\",\n \"meta-llama-Llama-4-Maverick-17B-128E-Instruct-FP8\",\n ];\n }\n }\n\n /**\n * 获取并缓存可用模型列表\n * @returns {Promise}\n */\n public async getAvailableModels(): Promise {\n if (this.availableModels.length === 0) {\n this.availableModels = await this.fetchAvailableModels();\n }\n return this.availableModels;\n }\n\n /**\n * 兼容多种 content 结构，提取纯文本\n */\n private extractTextFromContent(content: any): string {\n if (typeof content === \"string\") return content;\n if (Array.isArray(content)) {\n return content\n .map((p) => {\n if (!p) return \"\";\n if (typeof p === \"string\") return p;\n if (typeof p.text === \"string\") return p.text;\n if (typeof p.content === \"string\") return p.content;\n return \"\";\n })\n .join(\"\");\n }\n if (content && typeof content === \"object\") {\n if (typeof content.text === \"string\") return content.text;\n if (typeof content.content === \"string\") return content.content;\n }\n return \"\";\n }\n\n /**\n * Akash API 只需要用户输入，这里仅取“最后一条 user 消息”\n */\n private buildAkashMessages(messages: any[]): any[] {\n if (!Array.isArray(messages)) return [];\n const lastUser = [...messages].reverse().find((m) => m && m.role === \"user\");\n const userText = this.extractTextFromContent(lastUser?.content || \"\");\n return [\n {\n role: \"user\",\n content: userText,\n parts: [{ type: \"text\", text: userText }],\n },\n ];\n }\n\n /**\n * 取“最后一条 system 文本”，若无则默认\n */\n private getSystemMessage(messages: any[]): string {\n const lastSystem = [...(messages || [])].reverse().find((m) => m.role === \"system\");\n if (lastSystem) {\n return this.extractTextFromContent(lastSystem.content);\n }\n\n return `You are a helpful assistant with thinking and LaTeX support. \n\nWhen you need to think through a problem, use ... tags to show your reasoning process.\n\nFor mathematical content, use LaTeX formatting:\n1. Use single dollar signs ($...$) for inline math: e.g., $x^2 + y^2 = r^2$\n2. Use double dollar signs ($$...$$) for block/display math\n\nAlways show your thinking process when solving complex problems.`;\n }\n\n /**\n * 夹逼与规范化 temperature（避免上游将 0 判为非法）\n */\n private normalizeTemperature(temp: any): number {\n const EPS = 0.01;\n const n = Number(temp);\n let num = Number.isFinite(n) ? n : 0.7;\n if (num <= 0) num = EPS;\n if (num > 2) num = 2;\n return Math.round(num * 100) / 100;\n }\n\n /**\n * 规范化 topP\n */\n private normalizeTopP(topP: any): number {\n const n = Number(topP);\n let num = Number.isFinite(n) ? n : 0.95;\n if (num < 0) num = 0;\n if (num > 1) num = 1;\n return Math.round(num * 100) / 100;\n }\n\n /**\n * 将 OpenAI 模型名映射到 Akash 可用模型\n * 动态映射，确保只映射到实际可用的模型\n * @param openaiModel - OpenAI 模型名\n * @returns {string} Akash 模型名\n */\n public async modelNameMapping(openaiModel: string): Promise {\n const availableModels = await this.getAvailableModels();\n\n // 优先使用精确匹配\n if (availableModels.includes(openaiModel)) {\n return openaiModel;\n }\n\n // 定义通用模型名到 Akash 模型的映射\n const mapping: Record = {\n \"gpt-4\": \"openai-gpt-oss-120b\",\n \"gpt-4o\": \"DeepSeek-V3.1\",\n \"gpt-4-turbo\": \"openai-gpt-oss-120b\",\n \"gpt-3.5-turbo\": \"openai-gpt-oss-120b\",\n \"gpt-4o-mini\": \"DeepSeek-V3.1\",\n };\n\n const mappedModel = mapping[openaiModel];\n\n // 如果找到了映射模型，且它在可用列表中，则返回\n if (mappedModel && availableModels.includes(mappedModel)) {\n console.log(`🔗 模型映射: ${openaiModel} -> ${mappedModel}`);\n return mappedModel;\n }\n\n // 如果找不到映射或映射的模型不可用，则返回第一个可用模型\n if (availableModels.length > 0) {\n const fallbackModel = availableModels[0];\n console.log(\n `⚠️ 模型 \"${openaiModel}\" 未找到直接映射，正在使用第一个可用模型: ${fallbackModel}`,\n );\n return fallbackModel;\n }\n\n // 极端情况下，如果没有可用模型，返回一个硬编码的默认值\n console.error(\n \"❌ 没有任何可用模型，将使用硬编码默认值。这表明获取模型列表失败且无回退列表。\",\n );\n return \"openai-gpt-oss-120b\";\n }\n\n // ---------- 仅拼接 0: 行的解析 ----------\n private parseZeroLines(responseText: string): string {\n const lines = responseText.trim().split(/\\r?\\n/);\n const parts: string[] = [];\n\n for (const line of lines) {\n if (!line.startsWith(\"0:\")) continue;\n const payload = line.slice(line.indexOf(\":\") + 1).trim();\n let content = payload;\n if (payload.startsWith('\"') && payload.endsWith('\"')) {\n try {\n content = JSON.parse(payload);\n } catch {\n content = payload.slice(1, -1);\n }\n }\n parts.push(content);\n }\n\n return parts.join(\"\");\n }\n\n private splitThinking(full: string): [string, string] {\n const thinkRe = /Thinking([\\s\\S]*?)<\\/thinking>/;\n const m = full.match(thinkRe);\n if (!m) return [\"\", full];\n const thinking = (m[1] || \"\").trim();\n const regular = full.replace(thinkRe, \"\").trim();\n return [thinking, regular];\n }\n\n // 非流式：一次性返回\n async chatCompletionsCreate(\n model: string = \"gpt-4\",\n messages: any[] = [],\n temperature: number = 0.7,\n topP: number = 0.95,\n stream: boolean = false,\n ): Promise {\n if (stream) {\n throw new Error(\"Use streamChatCompletionsSSE for streaming.\");\n }\n\n await this.refreshSessionToken();\n\n const akashModel = await this.modelNameMapping(model);\n const akashMessages = this.buildAkashMessages(messages);\n const akashPayload = {\n id: this.sessionId,\n messages: akashMessages,\n model: akashModel,\n system: this.getSystemMessage(messages),\n temperature: this.normalizeTemperature(temperature),\n topP: this.normalizeTopP(topP),\n context: [],\n };\n\n const resp = await fetch(this.baseUrl, {\n method: \"POST\",\n headers: this.headers,\n body: JSON.stringify(akashPayload),\n });\n\n if (!resp.ok) {\n const text = await resp.text().catch(() => \"\");\n throw new Error(\n `Request failed: ${resp.status} ${resp.statusText} ${text}`,\n );\n }\n\n const responseText = await resp.text();\n\n // 仅拼接上游 0: 行，再做 Thinking 分离\n const fullFromZero = this.parseZeroLines(responseText);\n const [thinkingContent, regularContent] = this.splitThinking(fullFromZero);\n\n let full = \"\";\n if (thinkingContent) {\n full += `Thinking\\n${thinkingContent}\\n\\n\\n`;\n }\n full += regularContent;\n\n return {\n id: `chatcmpl-${uuidv4().replace(/-/g, \"\").slice(0, 10)}`,\n object: \"chat.completion\",\n created: Math.floor(Date.now() / 1000),\n model: akashModel,\n choices: [\n {\n index: 0,\n message: { role: \"assistant\", content: full },\n finish_reason: \"stop\",\n },\n ],\n usage: { prompt_tokens: 100, completion_tokens: 50, total_tokens: 150 },\n };\n }\n\n // 真·流式：返回一个 ReadableStream，边读 Akash 边转 SSE（仅处理 0: 行）\n async streamChatCompletionsSSE(\n model: string = \"gpt-4\",\n messages: any[] = [],\n temperature: number = 0.7,\n topP: number = 0.95,\n ): Promise> {\n await this.refreshSessionToken();\n\n const akashModel = await this.modelNameMapping(model);\n const akashMessages = this.buildAkashMessages(messages);\n const akashPayload = {\n id: this.sessionId,\n messages: akashMessages,\n model: akashModel,\n system: this.getSystemMessage(messages),\n temperature: this.normalizeTemperature(temperature),\n topP: this.normalizeTopP(topP),\n context: [],\n };\n\n const created = Math.floor(Date.now() / 1000);\n const encoder = new TextEncoder();\n\n return new ReadableStream({\n start: async (controller) => {\n const sendJSON = (obj: any) => {\n const sseData = `data: ${JSON.stringify(obj)}\\n\\n`;\n console.log(\"📤 [SSE Sent]:\", sseData.trim());\n controller.enqueue(encoder.encode(sseData));\n };\n\n const sendComment = (comment = \"🔄 heartbeat\") => {\n const sseComment = `: ${comment}\\n\\n`;\n console.log(\"💓 [SSE Heartbeat]:\", sseComment.trim());\n controller.enqueue(encoder.encode(sseComment));\n };\n\n const sendDone = () => {\n const sseDone = \"data: [DONE]\\n\\n\";\n console.log(\"🏁 [SSE Finished]:\", sseDone.trim());\n controller.enqueue(encoder.encode(sseDone));\n };\n\n const sendInitialRole = () => {\n sendJSON({\n id: `chatcmpl-${uuidv4().replace(/-/g, \"\").slice(0, 10)}`,\n object: \"chat.completion.chunk\",\n created,\n model: akashModel,\n choices: [{\n index: 0,\n delta: { role: \"assistant\" },\n finish_reason: null,\n }],\n });\n };\n\n const sendContentDelta = (text: string) => {\n if (!text) return;\n sendJSON({\n id: `chatcmpl-${uuidv4().replace(/-/g, \"\").slice(0, 10)}`,\n object: \"chat.completion.chunk\",\n created,\n model: akashModel,\n choices: [{\n index: 0,\n delta: { content: text },\n finish_reason: null,\n }],\n });\n };\n\n const sendFinalStop = () => {\n sendJSON({\n id: `chatcmpl-${uuidv4().replace(/-/g, \"\").slice(0, 10)}`,\n object: \"chat.completion.chunk\",\n created,\n model: akashModel,\n choices: [{ index: 0, delta: {}, finish_reason: \"stop\" }],\n });\n };\n\n let heartbeatTimer: number | undefined;\n const beginHeartbeat = () => {\n heartbeatTimer = setInterval(\n () => sendComment(),\n 15000,\n ) as unknown as number;\n };\n const endHeartbeat = () => {\n if (heartbeatTimer) clearInterval(heartbeatTimer);\n };\n\n try {\n const resp = await fetch(this.baseUrl, {\n method: \"POST\",\n headers: this.headers,\n body: JSON.stringify(akashPayload),\n });\n\n if (!resp.ok || !resp.body) {\n const errText = await resp.text().catch(() => \"\");\n throw new Error(\n `Upstream failed: ${resp.status} ${resp.statusText} ${errText}`,\n );\n }\n\n sendInitialRole();\n beginHeartbeat();\n\n const reader = resp.body.getReader();\n const decoder = new TextDecoder();\n let buf = \"\";\n\n while (true) {\n const { value, done } = await reader.read();\n if (done) break;\n\n buf += decoder.decode(value, { stream: true });\n\n let idx: number;\n while ((idx = buf.indexOf(\"\\n\")) >= 0) {\n const line = buf.slice(0, idx);\n buf = buf.slice(idx + 1);\n\n // 仅处理以 0: 开头的行\n if (!line.startsWith(\"0:\")) continue;\n\n let rawPayload = line.slice(2).trim();\n if (rawPayload.startsWith('\"') && rawPayload.endsWith('\"')) {\n try {\n rawPayload = JSON.parse(rawPayload);\n } catch {\n rawPayload = rawPayload.slice(1, -1);\n }\n }\n\n sendContentDelta(rawPayload);\n }\n }\n\n // 收尾：处理缓冲区残留的最后一行\n if (buf.trim().length > 0) {\n const line = buf.trim();\n if (line.startsWith(\"0:\")) {\n let rawPayload = line.slice(2).trim();\n if (rawPayload.startsWith('\"') && rawPayload.endsWith('\"')) {\n try {\n rawPayload = JSON.parse(rawPayload);\n } catch {\n rawPayload = rawPayload.slice(1, -1);\n }\n }\n sendContentDelta(rawPayload);\n }\n }\n\n sendFinalStop();\n sendDone();\n endHeartbeat();\n controller.close();\n } catch (e) {\n console.error(\"Stream processing error:\", e);\n sendJSON({\n id: `chatcmpl-${uuidv4().replace(/-/g, \"\").slice(0, 10)}`,\n object: \"chat.completion.chunk\",\n created,\n model: akashModel,\n choices: [\n {\n index: 0,\n delta: {\n content: `\\n[Proxy Error] ${\n e instanceof Error ? e.message : String(e)\n }\\n`,\n },\n finish_reason: null,\n },\n ],\n });\n sendFinalStop();\n sendDone();\n controller.close();\n } finally {\n endHeartbeat();\n }\n },\n cancel: () => {\n console.log(\"Stream canceled by client.\");\n },\n });\n }\n\n // ---- 认证 & 刷新 session_token ----\n async getAuthSession(customCookies?: string): Promise<{\n status: \"success\" | \"error\";\n statusCode?: number | null;\n data?: string;\n headers?: Record;\n message?: string;\n }> {\n try {\n const authHeaders: Record = {\n \"User-Agent\":\n \"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/139.0.0.0 Safari/537.36 Edg/139.0.0.0\",\n \"origin\": \"https://chat.akash.network\",\n \"referer\": \"https://chat.akash.network/\",\n \"accept-language\": \"zh-CN,zh;q=0.9,en;q=0.8,en-GB;q=0.7,en-US;q=0.6\",\n \"Cookie\": customCookies || this.headers[\"Cookie\"] || \"\",\n };\n\n const resp = await fetch(this.authUrl, {\n method: \"GET\",\n headers: authHeaders,\n });\n const text = await resp.text().catch(() => \"\");\n\n let setCookies: string[] = [];\n const headersAny = resp.headers as any;\n if (typeof headersAny.getSetCookie === \"function\") {\n setCookies = headersAny.getSetCookie();\n } else {\n const sc = resp.headers.get(\"set-cookie\");\n if (sc) setCookies = [sc];\n }\n\n const headersObj: Record = {};\n resp.headers.forEach((v, k) => (headersObj[k] = v));\n if (setCookies.length) headersObj[\"set-cookie\"] = setCookies;\n\n return {\n status: \"success\",\n statusCode: resp.status,\n data: text,\n headers: headersObj,\n };\n } catch (e) {\n return {\n status: \"error\",\n message: `认证请求失败: ${e instanceof Error ? e.message : String(e)}`,\n statusCode: null,\n };\n }\n }\n\n private extractSessionTokenFromHeaders(\n headers?: Record,\n ): string | null {\n if (!headers) return null;\n const sc = headers[\"set-cookie\"] || headers[\"Set-Cookie\"];\n const candidates = Array.isArray(sc) ? sc : (sc ? [sc] : []);\n for (const cookieStr of candidates) {\n const m = cookieStr.match(/session_token=([^;,]+)/);\n if (m && m[1]) return m[1];\n }\n return null;\n }\n\n private updateCookieSessionToken(newToken: string) {\n const current = this.headers[\"Cookie\"] || \"\";\n if (current.includes(\"session_token=\")) {\n this.headers[\"Cookie\"] = current.replace(\n /session_token=[^;,]+/,\n `session_token=${newToken}`,\n );\n } else {\n this.headers[\"Cookie\"] = current\n ? `${current}; session_token=${newToken}`\n : `session_token=${newToken}`;\n }\n }\n\n async refreshSessionToken(): Promise {\n try {\n const authResult = await this.getAuthSession();\n if (authResult.status !== \"success\") return false;\n const newToken = this.extractSessionTokenFromHeaders(authResult.headers);\n if (!newToken) return false;\n this.updateCookieSessionToken(newToken);\n console.log(\"🔑 Session token refreshed successfully.\");\n return true;\n } catch {\n console.warn(\"⚠️ Failed to refresh session token.\");\n return false;\n }\n }\n}\n\n// ---- 全局实例 ----\nconst akashApi = new AkashOpenAIAPI();\n\n// ---- Val Town 入口 ----\nexport default async function (req: Request): Promise {\n const url = new URL(req.url);\n const method = req.method;\n const path = url.pathname;\n\n const corsHeaders = {\n \"Access-Control-Allow-Origin\": \"*\",\n \"Access-Control-Allow-Methods\": \"GET, POST, OPTIONS\",\n \"Access-Control-Allow-Headers\": \"Content-Type, Authorization, Cookie\",\n };\n\n if (method === \"OPTIONS\") {\n return new Response(null, { headers: corsHeaders });\n }\n\n try {\n if (path === \"/v1/models\" && method === \"GET\") {\n return await listModels(corsHeaders);\n }\n if (path === \"/v1/chat/completions\" && method === \"POST\") {\n return await chatCompletions(req, corsHeaders);\n }\n if (\n path === \"/v1/auth/session\" && (method === \"GET\" || method === \"POST\")\n ) {\n return await authSession(req, corsHeaders);\n }\n if (path === \"/\" && method === \"GET\") {\n return index(corsHeaders);\n }\n\n return new Response(\"Not Found\", { status: 404, headers: corsHeaders });\n } catch (error) {\n console.error(\"Unhandled Error in main handler:\", error);\n return new Response(\n JSON.stringify({\n error: {\n message: error instanceof Error ? error.message : String(error),\n type: \"invalid_request_error\",\n param: null,\n code: null,\n },\n }),\n {\n status: 400,\n headers: { \"Content-Type\": \"application/json\", ...corsHeaders },\n },\n );\n }\n}\n\n// ---- 路由实现 ----\nasync function listModels(cors: Record): Promise {\n try {\n const models = await akashApi.getAvailableModels();\n const modelData = models.map((id) => ({\n id: id,\n object: \"model\",\n created: 1699000000,\n owned_by: \"akash-network\",\n permission: [],\n root: id,\n parent: null,\n }));\n\n const data = {\n object: \"list\",\n data: modelData,\n };\n\n return new Response(JSON.stringify(data), {\n headers: { \"Content-Type\": \"application/json\", ...cors },\n });\n } catch (error) {\n console.error(\"Failed to list models:\", error);\n return new Response(JSON.stringify({ error: \"Failed to fetch models\" }), {\n status: 500,\n headers: { \"Content-Type\": \"application/json\", ...cors },\n });\n }\n}\n\nasync function chatCompletions(\n req: Request,\n cors: Record,\n): Promise {\n const body = await req.json().catch(() => ({}));\n console.log(\"--- [Chat Completions Request] ---\");\n console.log(JSON.stringify(body, null, 2));\n console.log(\"------------------------------------\\n\");\n\n const model = body.model || \"gpt-4\";\n const messages = body.messages || [];\n const temperature = body.temperature ?? 0.7;\n const topP = body.topP ?? body.top_p ?? 0.95;\n const stream = !!body.stream;\n\n if (stream) {\n console.log(\"🚀 Initiating SSE stream for model:\", model);\n const streamRS = await akashApi.streamChatCompletionsSSE(\n model,\n messages,\n temperature,\n topP,\n );\n\n return new Response(streamRS, {\n headers: {\n \"Content-Type\": \"text/event-stream; charset=utf-8\",\n \"Cache-Control\": \"no-cache, no-transform\",\n \"Connection\": \"keep-alive\",\n \"X-Accel-Buffering\": \"no\",\n ...cors,\n },\n });\n }\n\n console.log(\"📄 Initiating non-streaming request for model:\", model);\n const result = await akashApi.chatCompletionsCreate(\n model,\n messages,\n temperature,\n topP,\n false,\n );\n\n console.log(\"\\n--- [Chat Completions Response] ---\");\n console.log(JSON.stringify(result, null, 2));\n console.log(\"----------------------------------\\n\");\n\n return new Response(JSON.stringify(result), {\n headers: { \"Content-Type\": \"application/json\", ...cors },\n });\n}\n\nasync function authSession(\n req: Request,\n cors: Record,\n): Promise {\n let customCookies: string | undefined = undefined;\n if (req.method === \"POST\") {\n const data = await req.json().catch(() => ({}));\n if (data && typeof data.cookies === \"string\") {\n customCookies = data.cookies;\n }\n }\n\n const result = await akashApi.getAuthSession(customCookies);\n\n if (customCookies) {\n (akashApi as any)[\"headers\"][\"Cookie\"] = customCookies;\n }\n\n return new Response(JSON.stringify(result), {\n headers: { \"Content-Type\": \"application/json\", ...cors },\n });\n}\n\nfunction index(cors: Record): Response {\n const data = {\n message: \"Akash Network OpenAI API Proxy\",\n version: \"1.7.2-ZeroOnly\",\n endpoints: {\n models: \"/v1/models\",\n chat_completions: \"/v1/chat/completions\",\n auth_session: \"/v1/auth/session\",\n },\n documentation:\n \"This proxy converts OpenAI API requests to Akash Network format (Val Town version, with real SSE streaming). Output parser now strictly concatenates only lines starting with '0:'.\",\n };\n\n return new Response(JSON.stringify(data), {\n headers: { \"Content-Type\": \"application/json\", ...cors },\n });\n}","deleted","label","children",[],[150],"files",[148],"counts",{"_173":32,"_174":175,"_176":32,"_177":178},"pulls","branches",1,"forks","envVars",{"_179":32,"_180":32},"count","emptyCount","contents",[183],{"_102":149,"_95":150,"_151":152,"_153":-5,"_154":136,"_155":135,"_156":157,"_104":184,"_99":159,"_160":161,"_162":163,"_164":22,"_165":-5,"_98":185},["D",1761012100255],[150],"file","upstreamProject","upstreamBranch","emptyEnvVars","isNew","flags",[],"routes/_app.x.$handle.$name._index",{"_93":94,"_95":96,"_97":-7,"_99":-7,"_100":195,"_181":200,"_203":-7,"_204":-5,"_205":-5,"_169":206,"_210":211,"_212":-5,"_213":-5},{"_102":103,"_95":96,"_104":196,"_106":-5,"_107":22,"_108":109,"_110":-5,"_111":-5,"_14":112,"_113":-5,"_114":22,"_115":-5,"_116":32,"_16":117,"_118":119,"_120":119,"_121":122,"_123":122,"_124":122,"_125":119,"_126":119,"_23":197,"_130":198,"_133":199,"_145":22},["D",1761012076994],{"_93":94,"_128":129},{"_93":94,"_132":129,"_14":112},{"_99":135,"_95":43,"_102":136,"_137":103,"_138":-5,"_139":-5,"_140":-5,"_141":-5,"_142":-5,"_143":96,"_144":94},[201],{"_102":149,"_95":150,"_151":152,"_153":-5,"_154":136,"_155":135,"_156":157,"_104":184,"_99":159,"_160":161,"_162":163,"_164":22,"_165":-5,"_98":202},[150],"readme","upstream","pull",[207],{"_102":149,"_95":150,"_151":152,"_153":-5,"_154":136,"_155":135,"_156":157,"_104":158,"_99":159,"_160":161,"_162":163,"_164":22,"_165":-5,"_166":208,"_98":209},[],[150],"previews",[207],"ahead","behind","actionData","errors"]