HealthLog/src/lib/ai/openai-wire.ts

import type {
  AiContentPart,
  AiMessage,
  AiToolCall,
  AiToolDef,
  CompletionResult,
} from "./types";

/**
 * Shared OpenAI Chat-Completions wire mapping. Both the hosted OpenAI client
 * and the local OpenAI-compatible client (Ollama / LM Studio / vLLM) speak this
 * shape, so the message → wire translation and the response parsing live here
 * once. The two clients differ only in how they treat `responseFormat` and the
 * endpoint/auth, not in the message structure.
 */

type OpenAIChatContent =
  | string
  | Array<
      | { type: "text"; text: string }
      | { type: "image_url"; image_url: { url: string } }
    >;

interface OpenAIChatMessage {
  role: "system" | "user" | "assistant" | "tool";
  content: OpenAIChatContent | null;
  tool_calls?: Array<{
    id: string;
    type: "function";
    function: { name: string; arguments: string };
  }>;
  tool_call_id?: string;
}

/**
 * Map a single `AiMessage` content into the Chat-Completions content shape.
 * A plain string stays a string (byte-identical to the text-only wire we shipped
 * before the refactor); parts become the multimodal array. PDF document parts
 * are dropped — the Chat-Completions wire has no document block (the OCR route
 * gates PDFs to Anthropic).
 */
function mapContent(content: string | AiContentPart[]): OpenAIChatContent {
  if (typeof content === "string") return content;
  const out: Array<
    | { type: "text"; text: string }
    | { type: "image_url"; image_url: { url: string } }
  > = [];
  for (const part of content) {
    if (part.type === "text") {
      out.push({ type: "text", text: part.text });
    } else if (part.type === "image") {
      out.push({
        type: "image_url",
        image_url: { url: `data:${part.mediaType};base64,${part.dataBase64}` },
      });
    }
    // `document` parts are not representable on this wire; drop them.
  }
  return out;
}

/**
 * Build the full Chat-Completions `messages[]` array: the system turn first,
 * then every conversation turn mapped 1:1. Assistant `toolCalls` and the
 * `role:"tool"` result turns map to the OpenAI tool wire so a multi-round
 * tool loop (F1) round-trips.
 */
export function buildOpenAIMessages(
  system: string,
  messages: AiMessage[],
): OpenAIChatMessage[] {
  const out: OpenAIChatMessage[] = [{ role: "system", content: system }];
  for (const m of messages) {
    if (m.role === "tool") {
      out.push({
        role: "tool",
        tool_call_id: m.toolCallId,
        content:
          typeof m.content === "string" ? m.content : mapContent(m.content),
      });
      continue;
    }
    const mapped: OpenAIChatMessage = {
      role: m.role,
      content: mapContent(m.content),
    };
    if (m.role === "assistant" && m.toolCalls && m.toolCalls.length > 0) {
      mapped.tool_calls = m.toolCalls.map((tc) => ({
        id: tc.id,
        type: "function",
        function: { name: tc.name, arguments: tc.arguments },
      }));
    }
    out.push(mapped);
  }
  return out;
}

/** Map tool defs into the OpenAI `tools` array. */
export function buildOpenAITools(tools: AiToolDef[]) {
  return tools.map((t) => ({
    type: "function" as const,
    function: {
      name: t.name,
      description: t.description,
      parameters: t.parameters,
    },
  }));
}

interface OpenAIChoice {
  message?: {
    content?: string | null;
    tool_calls?: Array<{
      id?: string;
      function?: { name?: string; arguments?: string };
    }>;
  };
  finish_reason?: string;
}

interface OpenAIResponseJson {
  choices?: OpenAIChoice[];
  usage?: {
    total_tokens?: number;
    prompt_tokens_details?: { cached_tokens?: number };
  };
}

/** Map the OpenAI `finish_reason` onto the provider-neutral finishReason. */
export function mapFinishReason(
  reason: string | undefined,
): CompletionResult["finishReason"] {
  if (reason === "tool_calls") return "tool_calls";
  if (reason === "length") return "length";
  if (reason === "stop") return "stop";
  return undefined;
}

/** Pull tool calls off an OpenAI choice into the provider-neutral shape. */
export function parseOpenAIToolCalls(
  choice: OpenAIChoice | undefined,
): AiToolCall[] | undefined {
  const calls = choice?.message?.tool_calls;
  if (!calls || calls.length === 0) return undefined;
  return calls.map((c) => ({
    id: c.id ?? "",
    name: c.function?.name ?? "",
    arguments: c.function?.arguments ?? "",
  }));
}

export function parseCachedTokens(json: OpenAIResponseJson): number | null {
  return json.usage?.prompt_tokens_details?.cached_tokens ?? null;
}

/**
 * Bound and key-redact an upstream body before it is attached to a thrown
 * error. Both clients inlined this next to their non-2xx handling; it lives
 * here once so the embedded-error excerpt below is sanitised by the very same
 * code rather than by a copy that can drift.
 */
export function sanitiseBodyExcerpt(raw: string): string {
  return raw
    .slice(0, 500)
    .replace(/sk-[A-Za-z0-9_-]{8,}/g, "sk-***redacted***")
    .replace(/Bearer\s+[A-Za-z0-9_.-]+/gi, "Bearer ***redacted***");
}

interface EmbeddedError {
  message: string;
  /** The gateway's own `error.code` when it is an HTTP error status, else 0. */
  httpStatus: number;
}

/** Longest slice of a remote error message that reaches a thrown message. */
const EMBEDDED_MESSAGE_MAX = 200;

function coerceEmbeddedError(candidate: unknown): EmbeddedError | null {
  if (typeof candidate === "string") {
    const trimmed = candidate.trim();
    return trimmed.length > 0 ? { message: trimmed, httpStatus: 0 } : null;
  }
  if (candidate === null || typeof candidate !== "object") return null;
  const raw = candidate as { code?: unknown; message?: unknown };
  const message =
    typeof raw.message === "string" && raw.message.trim().length > 0
      ? raw.message.trim()
      : "gateway reported a failure without a message";
  // Only an error-range status is adopted. A gateway that puts a vendor
  // string ("insufficient_quota") or a non-status number in `code` keeps the
  // 0 sentinel, which the chain classifier already treats as a hard failure.
  const code = raw.code;
  const httpStatus =
    typeof code === "number" &&
    Number.isInteger(code) &&
    code >= 400 &&
    code <= 599
      ? code
      : 0;
  return { message, httpStatus };
}

/**
 * Read a failure that a 2xx body carries INSIDE itself.
 *
 * OpenAI-compatible gateways (OpenRouter and friends) answer a provider-side
 * refusal with HTTP 200 and either a top-level `error` object and no `choices`
 * at all, or — when a generation dies mid-flight — `choices[0].error` beside
 * `finish_reason: "error"`. Returns the embedded failure, or null for a normal
 * reply. `api.openai.com` never answers this way; every failure there is a
 * non-2xx status.
 */
export function readEmbeddedError(json: unknown): EmbeddedError | null {
  if (json === null || typeof json !== "object") return null;
  const body = json as {
    error?: unknown;
    choices?: Array<{ error?: unknown } | null> | null;
  };
  return (
    coerceEmbeddedError(body.error) ??
    coerceEmbeddedError(body.choices?.[0]?.error)
  );
}

/**
 * Throw the same shaped error the non-2xx path throws when a 2xx body carries
 * an embedded failure; return quietly when it does not. The gateway's own
 * message leads so the hop reason says WHY the call failed instead of
 * "returned empty content"; it is bounded and key-redacted on the way in,
 * because remote text is never trusted to be short or clean.
 */
export function raiseEmbeddedError(
  json: unknown,
  ctx: { upstream: string; label: string; model: string },
): void {
  const embedded = readEmbeddedError(json);
  if (!embedded) return;
  const message = sanitiseBodyExcerpt(embedded.message).slice(
    0,
    EMBEDDED_MESSAGE_MAX,
  );
  const err = new Error(`${message} (${ctx.label})`);
  Object.assign(err, {
    httpStatus: embedded.httpStatus,
    kind: "embedded_error",
    upstream: ctx.upstream,
    model: ctx.model,
    bodyExcerpt: sanitiseBodyExcerpt(JSON.stringify(json) ?? ""),
  });
  throw err;
}

export type { OpenAIChoice, OpenAIResponseJson };