@sleepy-ai/cli 0.0.0-main-202607032053 → 0.1.4

package/src/provider/error.ts

@@ -0,0 +1,216 @@
+import { APICallError } from "ai"
+import { STATUS_CODES } from "http"
+import { iife } from "@/util/iife"
+import type { ProviderID } from "./schema"
+
+// Adapted from overflow detection patterns in:
+// https://github.com/badlogic/pi-mono/blob/main/packages/ai/src/utils/overflow.ts
+const OVERFLOW_PATTERNS = [
+  /prompt is too long/i, // Anthropic
+  /input is too long for requested model/i, // Amazon Bedrock
+  /exceeds the context window/i, // OpenAI (Completions + Responses API message text)
+  /input token count.*exceeds the maximum/i, // Google (Gemini)
+  /maximum prompt length is \d+/i, // xAI (Grok)
+  /reduce the length of the messages/i, // Groq
+  /maximum context length is \d+ tokens/i, // OpenRouter, DeepSeek, vLLM
+  /exceeds the limit of \d+/i, // GitHub Copilot
+  /exceeds the available context size/i, // llama.cpp server
+  /greater than the context length/i, // LM Studio
+  /context window exceeds limit/i, // MiniMax
+  /exceeded model token limit/i, // Kimi For Coding, Moonshot
+  /context[_ ]length[_ ]exceeded/i, // Generic fallback
+  /request entity too large/i, // HTTP 413
+  /context length is only \d+ tokens/i, // vLLM
+  /input length.*exceeds.*context length/i, // vLLM
+  /prompt too long; exceeded (?:max )?context length/i, // Ollama explicit overflow error
+  /too large for model with \d+ maximum context length/i, // Mistral
+  /model_context_window_exceeded/i, // z.ai non-standard finish_reason surfaced as error text
+]
+
+function isOpenAiErrorRetryable(e: APICallError) {
+  const status = e.statusCode
+  if (!status) return e.isRetryable
+  // openai sometimes returns 404 for models that are actually available
+  return status === 404 || e.isRetryable
+}
+
+// Providers not reliably handled in this function:
+// - z.ai: can accept overflow silently (needs token-count/context-window checks)
+function isOverflow(message: string) {
+  if (OVERFLOW_PATTERNS.some((p) => p.test(message))) return true
+
+  // Providers/status patterns handled outside of regex list:
+  // - Cerebras: often returns "400 (no body)" / "413 (no body)"
+  // - Mistral: often returns "400 (no body)" / "413 (no body)"
+  return /^4(00|13)\s*(status code)?\s*\(no body\)/i.test(message)
+}
+
+// Provider IDs served by the Sleepy model gateway. Its error bodies carry
+// non-standard semantics (e.g. moderation/risk-control blocks under HTTP 400),
+// so the gateway-specific handling below is scoped to these providers and leaves
+// every other provider's error flow untouched.
+const SLEEPY_GATEWAY_PROVIDERS = new Set(["sleepy", "sleepy"])
+
+// Sleepy gateway error.code values worth relabeling: moderation (421) and
+// risk-control (441) blocks arrive under a generic HTTP 400.
+const FRIENDLY_GATEWAY_CODES: Record<string, string> = {
+  "421": "Request blocked by content moderation",
+  "441": "Request blocked by risk control",
+}
+
+function message(providerID: ProviderID, e: APICallError) {
+  return iife(() => {
+    // Sleepy gateway: relabel known block codes and surface error.param (the real
+    // reason often lives there while error.message stays generic). json() returns
+    // undefined for non-JSON, so HTML/proxy error pages fall through to the
+    // original handling below.
+    const gw = SLEEPY_GATEWAY_PROVIDERS.has(providerID) ? json(e.responseBody)?.error : undefined
+    if (gw && typeof gw === "object") {
+      const base = FRIENDLY_GATEWAY_CODES[String(gw.code)] ?? (typeof gw.message === "string" ? gw.message : "")
+      if (base) return typeof gw.param === "string" && gw.param !== base ? `${base}: ${gw.param}` : base
+    }
+
+    const msg = e.message
+    if (msg === "") {
+      if (e.responseBody) return e.responseBody
+      if (e.statusCode) {
+        const err = STATUS_CODES[e.statusCode]
+        if (err) return err
+      }
+      return "Unknown error"
+    }
+
+    if (!e.responseBody || (e.statusCode && msg !== STATUS_CODES[e.statusCode])) {
+      return msg
+    }
+
+    try {
+      const body = JSON.parse(e.responseBody)
+      // try to extract common error message fields
+      const errMsg = body.message || body.error || body.error?.message
+      if (errMsg && typeof errMsg === "string") {
+        return `${msg}: ${errMsg}`
+      }
+    } catch {}
+
+    // If responseBody is HTML (e.g. from a gateway or proxy error page),
+    // provide a human-readable message instead of dumping raw markup
+    if (/^\s*<!doctype|^\s*<html/i.test(e.responseBody)) {
+      if (e.statusCode === 401) {
+        return "Unauthorized: request was blocked by a gateway or proxy. Your authentication token may be missing or expired — try running `opencode auth login <your provider URL>` to re-authenticate."
+      }
+      if (e.statusCode === 403) {
+        return "Forbidden: request was blocked by a gateway or proxy. You may not have permission to access this resource — check your account and provider settings."
+      }
+      return msg
+    }
+
+    return `${msg}: ${e.responseBody}`
+  }).trim()
+}
+
+function json(input: unknown) {
+  if (typeof input === "string") {
+    try {
+      const result = JSON.parse(input)
+      if (result && typeof result === "object") return result
+      return undefined
+    } catch {
+      return undefined
+    }
+  }
+  if (typeof input === "object" && input !== null) {
+    return input
+  }
+  return undefined
+}
+
+export type ParsedStreamError =
+  | {
+      type: "context_overflow"
+      message: string
+      responseBody: string
+    }
+  | {
+      type: "api_error"
+      message: string
+      isRetryable: false
+      responseBody: string
+    }
+
+export function parseStreamError(input: unknown): ParsedStreamError | undefined {
+  const body = json(input)
+  if (!body) return
+
+  const responseBody = JSON.stringify(body)
+  if (body.type !== "error") return
+
+  switch (body?.error?.code) {
+    case "context_length_exceeded":
+      return {
+        type: "context_overflow",
+        message: "Input exceeds context window of this model",
+        responseBody,
+      }
+    case "insufficient_quota":
+      return {
+        type: "api_error",
+        message: "Quota exceeded. Check your plan and billing details.",
+        isRetryable: false,
+        responseBody,
+      }
+    case "usage_not_included":
+      return {
+        type: "api_error",
+        message: "To use Codex with your ChatGPT plan, upgrade to Plus: https://chatgpt.com/explore/plus.",
+        isRetryable: false,
+        responseBody,
+      }
+    case "invalid_prompt":
+      return {
+        type: "api_error",
+        message: typeof body?.error?.message === "string" ? body?.error?.message : "Invalid prompt.",
+        isRetryable: false,
+        responseBody,
+      }
+  }
+}
+
+export type ParsedAPICallError =
+  | {
+      type: "context_overflow"
+      message: string
+      responseBody?: string
+    }
+  | {
+      type: "api_error"
+      message: string
+      statusCode?: number
+      isRetryable: boolean
+      responseHeaders?: Record<string, string>
+      responseBody?: string
+      metadata?: Record<string, string>
+    }
+
+export function parseAPICallError(input: { providerID: ProviderID; error: APICallError }): ParsedAPICallError {
+  const m = message(input.providerID, input.error)
+  const body = json(input.error.responseBody)
+  if (isOverflow(m) || input.error.statusCode === 413 || body?.error?.code === "context_length_exceeded") {
+    return {
+      type: "context_overflow",
+      message: m,
+      responseBody: input.error.responseBody,
+    }
+  }
+
+  const metadata = input.error.url ? { url: input.error.url } : undefined
+  return {
+    type: "api_error",
+    message: m,
+    statusCode: input.error.statusCode,
+    isRetryable: input.providerID.startsWith("openai") ? isOpenAiErrorRetryable(input.error) : input.error.isRetryable,
+    responseHeaders: input.error.responseHeaders,
+    responseBody: input.error.responseBody,
+    metadata,
+  }
+}

package/src/provider/index.ts

@@ -0,0 +1,5 @@
+export * as Provider from "./provider"
+export * as ProviderAuth from "./auth"
+export * as ProviderError from "./error"
+export * as ModelsDev from "./models"
+export * as ProviderTransform from "./transform"

package/src/provider/models-snapshot.d.ts

@@ -0,0 +1,2 @@
+// Auto-generated by build.ts - do not edit
+export declare const snapshot: Record<string, unknown>