fix: resolve 422 from Replicate + add local Cog Docker mode

- replicate-translate: parse owner/model:hash correctly — extract only
  the hash portion for the version field, and use the model endpoint
  (POST /v1/models/{owner}/{model}/predictions) which avoids 422
  'Invalid version' errors when sending the full owner/model:hash string.

- Add local Cog mode: when replicateMode="local", calls the local Docker
  container directly (no Replicate API key needed), default endpoint
  http://localhost:5030/predictions (host port 5030 → container port 5000).

- settings-store: add replicateMode ("cloud"|"local") and localEndpoint
  fields with env var fallbacks REPLICATE_MODE and LOCAL_MODEL_ENDPOINT.

- admin panel: Radio selector for Cloud vs Local mode; shows docker run
  command snippet and local endpoint URL field when local is selected;
  hides Replicate API token field in local mode (not needed).

Local model startup:
  docker run -d -p 5030:5000 \
    r8.im/jigsawstack/text-translate@sha256:454df4c...

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
2026-03-10 08:50:40 +01:00
parent e034771087
commit 0be8b0b0f0
4 changed files with 243 additions and 138 deletions

View File

@@ -1,30 +1,75 @@
import { readSettings } from "./settings-store";
type CogOutput = {
status?: string;
output?: unknown;
error?: string;
};
type ReplicateOutput = string | string[] | { translation?: string; translated_text?: string; output?: string };
export async function replicateTranslate(
function extractText(output: unknown): string {
if (typeof output === "string") return output;
if (Array.isArray(output)) return output.join("");
if (output && typeof output === "object") {
const o = output as { translation?: string; translated_text?: string; output?: string };
return o.translation ?? o.translated_text ?? o.output ?? JSON.stringify(output);
}
throw new Error("Unexpected output format from model");
}
/**
* Parse modelVersion string into components.
* Accepts:
* "jigsawstack/text-translate:454df4c..." → { owner: "jigsawstack", model: "text-translate", hash: "454df4c..." }
* "jigsawstack/text-translate" → { owner: "jigsawstack", model: "text-translate", hash: undefined }
* "454df4c..." → { owner: undefined, model: undefined, hash: "454df4c..." }
*/
function parseModelVersion(mv: string) {
const [ownerModel, hash] = mv.split(":");
if (ownerModel.includes("/")) {
const [owner, model] = ownerModel.split("/");
return { owner, model, hash };
}
// Bare hash
return { owner: undefined, model: undefined, hash: ownerModel };
}
async function translateViaCloud(
text: string,
targetLanguage: string
): Promise<string> {
const settings = readSettings();
if (!settings.replicateApiToken) {
throw new Error("Replicate API token not configured");
}
if (!settings.jigsawApiKey) {
throw new Error("JigsawStack API key not configured");
if (!settings.replicateApiToken) throw new Error("Replicate API token not configured");
if (!settings.jigsawApiKey) throw new Error("JigsawStack API key not configured");
const { owner, model, hash } = parseModelVersion(settings.modelVersion);
let url: string;
let body: Record<string, unknown>;
if (owner && model && !hash) {
// No version pinned — use the model's deployment endpoint (latest)
url = `https://api.replicate.com/v1/models/${owner}/${model}/predictions`;
body = { input: { text, api_key: settings.jigsawApiKey, target_language: targetLanguage } };
} else if (owner && model && hash) {
// Pinned version — use model endpoint with version (avoids 422 from bare /predictions)
url = `https://api.replicate.com/v1/models/${owner}/${model}/predictions`;
body = {
version: hash,
input: { text, api_key: settings.jigsawApiKey, target_language: targetLanguage }
};
} else {
// Bare hash only — use the generic predictions endpoint
url = "https://api.replicate.com/v1/predictions";
body = {
version: hash,
input: { text, api_key: settings.jigsawApiKey, target_language: targetLanguage }
};
}
const body = {
version: settings.modelVersion,
input: {
text,
api_key: settings.jigsawApiKey,
target_language: targetLanguage
}
};
const response = await fetch("https://api.replicate.com/v1/predictions", {
const response = await fetch(url, {
method: "POST",
headers: {
"Authorization": `Bearer ${settings.replicateApiToken}`,
@@ -40,21 +85,55 @@ export async function replicateTranslate(
}
const data = await response.json();
if (data.error) throw new Error(`Replicate model error: ${data.error}`);
if (data.error) {
throw new Error(`Replicate model error: ${data.error}`);
return extractText(data.output as ReplicateOutput);
}
async function translateViaLocal(
text: string,
targetLanguage: string
): Promise<string> {
const settings = readSettings();
if (!settings.jigsawApiKey) throw new Error("JigsawStack API key not configured");
const endpoint = (settings.localEndpoint || "http://localhost:5030/predictions").replace(/\/$/, "");
const response = await fetch(endpoint, {
method: "POST",
headers: { "Content-Type": "application/json" },
body: JSON.stringify({
input: {
text,
api_key: settings.jigsawApiKey,
target_language: targetLanguage
}
})
});
if (!response.ok) {
const err = await response.text();
throw new Error(`Local model error: ${response.status} ${err}`);
}
// Extract translated text from various output formats
const output: ReplicateOutput = data.output;
const data: CogOutput = await response.json();
if (typeof output === "string") return output;
if (Array.isArray(output)) return output.join("");
if (output && typeof output === "object") {
return output.translation ?? output.translated_text ?? output.output ?? String(output);
}
if (data.error) throw new Error(`Local model error: ${data.error}`);
throw new Error("Unexpected output format from Replicate");
// Cog wraps output in { status, output } or returns output directly
const output = data.output !== undefined ? data.output : data;
return extractText(output);
}
export async function replicateTranslate(
text: string,
targetLanguage: string
): Promise<string> {
const { replicateMode } = readSettings();
return replicateMode === "local"
? translateViaLocal(text, targetLanguage)
: translateViaCloud(text, targetLanguage);
}
// Batch translate using separator trick to minimize API calls
@@ -72,12 +151,9 @@ export async function replicateTranslateBatch(
const joined = texts.join(SEPARATOR);
const translated = await replicateTranslate(joined, targetLanguage);
// Try to split on the separator; fall back to individual calls if it got translated
const parts = translated.split(SEPARATOR);
if (parts.length === texts.length) {
return parts;
}
if (parts.length === texts.length) return parts;
// Fallback: translate individually
// Fallback: translate individually if separator got mangled
return Promise.all(texts.map(t => replicateTranslate(t, targetLanguage)));
}

View File

@@ -1,11 +1,15 @@
import fs from "fs";
import path from "path";
export type ReplicateMode = "cloud" | "local";
export type Settings = {
replicateApiToken: string;
jigsawApiKey: string;
modelVersion: string;
replicateEnabled: boolean;
replicateMode: ReplicateMode;
localEndpoint: string;
adminPasswordHash: string;
};
@@ -14,6 +18,8 @@ const DEFAULT_SETTINGS: Settings = {
jigsawApiKey: process.env["JIGSAWSTACK_API_KEY"] ?? "",
modelVersion: "jigsawstack/text-translate:454df4c49941c05dea05175bd37686d0872c73c1f9366d1c2505db32ade52a89",
replicateEnabled: false,
replicateMode: (process.env["REPLICATE_MODE"] as ReplicateMode) ?? "cloud",
localEndpoint: process.env["LOCAL_MODEL_ENDPOINT"] ?? "http://localhost:5030/predictions",
adminPasswordHash: process.env["ADMIN_PASSWORD"] ?? "admin"
};