Skip to content

🪢 Open source agent evals & observability: Trace, evaluate, and improve LLM applications with one open platform.

2.3 KB
Raw Blame
import { z } from "zod";
import {
  ChatMessageType,
  PlaceholderMessageSchema,
  PromptChatMessageListSchema,
  PromptNameSchema,
  TextPromptContentSchema,
  COMMIT_MESSAGE_MAX_LENGTH,
  PromptType,
} from "@langfuse/shared";

const NewPromptBaseSchema = z.object({
  name: PromptNameSchema,
  isActive: z.boolean({
    error: "Enter whether the prompt should go live",
  }),
  config: z.string().refine(validateJson, "Config needs to be valid JSON"),
  commitMessage: z
    .string()
    .trim()
    .max(COMMIT_MESSAGE_MAX_LENGTH)
    .transform((val) => (val === "" ? undefined : val))
    .optional(),
});

const NewChatPromptSchema = NewPromptBaseSchema.extend({
  type: z.literal(PromptType.Chat),
  chatPrompt: z
    .array(z.any())
    .refine(
      (messages: Array<{ type?: ChatMessageType; content?: string }>) =>
        messages.every((message) => {
          const isPlaceholder = message?.type === ChatMessageType.Placeholder;
          return (
            !isPlaceholder ||
            PlaceholderMessageSchema.safeParse(message).success
          );
        }),
      "Placeholder name must start with a letter and contain only alphanumeric characters and underscores",
    )
    .refine(
      (messages: Array<{ type?: ChatMessageType; content?: string }>) =>
        messages.every((message) => {
          const isPlaceholder = message?.type === ChatMessageType.Placeholder;
          return isPlaceholder || Boolean(message?.content?.trim()?.length);
        }),
      "Enter a chat message or remove the empty message",
    ),
  textPrompt: z.string(),
});

const NewTextPromptSchema = NewPromptBaseSchema.extend({
  type: z.literal(PromptType.Text),
  chatPrompt: z.array(z.any()),
  textPrompt: TextPromptContentSchema,
});

export const NewPromptFormSchema = z.discriminatedUnion("type", [
  NewChatPromptSchema,
  NewTextPromptSchema,
]);
export type NewPromptFormSchemaType = z.infer<typeof NewPromptFormSchema>;

export const PromptVariantSchema = z.union([
  z.object({
    type: z.literal(PromptType.Chat),
    prompt: PromptChatMessageListSchema,
  }),
  z.object({
    type: z.literal(PromptType.Text),
    prompt: z.string(),
  }),
]);
export type PromptVariant = z.infer<typeof PromptVariantSchema>;

function validateJson(content: string): boolean {
  try {
    JSON.parse(content);

    return true;
  } catch (_e) {
    return false;
  }
}