src/lib/editor/rationale-validator.ts

/**
 * Rationale Purity Validator — lightweight Haiku check for garbage rationale input.
 * Catches: empty strings, random characters, "none", "n/a", "asdf", non-English, lazy input.
 */

import { getClaudeClient } from "../ai/claude-client";

export interface RationaleEntry {
  index: number;
  rationale: string;
}

export interface RationaleValidationResult {
  valid: boolean;
  rejected: Array<{ index: number; reason: string }>;
}

const SYSTEM_PROMPT = `You validate whether subtitle editing rationales are meaningful. Each rationale should briefly explain WHY a change was made to a subtitle.

REJECT if the rationale is:
- Empty, whitespace-only, or very short (1-2 characters)
- Random characters or keyboard smashing (e.g., "asdf", "jkl;", "xxx", "aaa")
- Placeholder text (e.g., "none", "n/a", "test", "todo", ".", "-", "...")
- Not in English
- Nonsensical or clearly not related to subtitle editing

ACCEPT if the rationale:
- Briefly explains a real editing decision (even short ones like "typo fix", "timing adjusted", "split for readability")
- References a specific change type (e.g., "fixed CPL", "corrected speaker", "adjusted overlap")

Respond with JSON:
\`\`\`json
{
  "results": [
    { "index": <number>, "valid": <boolean>, "reason": "<only if invalid>" }
  ]
}
\`\`\``;

// Quick local pre-filter for obviously bad rationales (no API call needed)
const GARBAGE_PATTERNS = [
  /^[\s\W]*$/, // Only whitespace/punctuation
  /^(.)\1{2,}$/, // Repeated single char: "aaa", "xxx"
  /^[a-z]{1,4}$/i, // 1-4 random letters
  /^(none|n\/a|na|test|todo|tbd|idk|ok|yes|no|null|undefined)$/i,
  /^\.{1,3}$/, // Just dots
  /^-{1,3}$/, // Just dashes
];

export function quickRejectRationale(rationale: string): string | null {
  const trimmed = rationale.trim();
  if (trimmed.length === 0) return "Empty rationale";
  if (trimmed.length < 3) return "Too short — must explain the change";
  for (const pattern of GARBAGE_PATTERNS) {
    if (pattern.test(trimmed)) return "Invalid rationale — must explain why the change was made";
  }
  return null; // Passes quick check
}

export async function validateRationales(
  entries: RationaleEntry[],
): Promise<RationaleValidationResult> {
  // Quick local rejection first
  const rejected: Array<{ index: number; reason: string }> = [];
  const needsAiCheck: RationaleEntry[] = [];

  for (const entry of entries) {
    const quickReason = quickRejectRationale(entry.rationale);
    if (quickReason) {
      rejected.push({ index: entry.index, reason: quickReason });
    } else {
      needsAiCheck.push(entry);
    }
  }

  // If all rejected locally, skip AI
  if (needsAiCheck.length === 0) {
    return { valid: rejected.length === 0, rejected };
  }

  // AI check with Haiku for the rest
  try {
    const { client } = getClaudeClient();
    const prompt = needsAiCheck
      .map((e) => `Index ${e.index}: "${e.rationale}"`)
      .join("\n");

    const response = await client.messages.create({
      model: "claude-haiku-4-5-20251001",
      max_tokens: 1024,
      system: SYSTEM_PROMPT,
      messages: [{ role: "user", content: `Validate these ${needsAiCheck.length} rationales:\n\n${prompt}` }],
    });

    const text = response.content
      .filter((b: any) => b.type === "text")
      .map((b: any) => b.text)
      .join("");

    const jsonMatch = text.match(/\{[\s\S]*\}/);
    if (jsonMatch) {
      const parsed = JSON.parse(jsonMatch[0]);
      if (parsed.results) {
        for (const r of parsed.results) {
          if (!r.valid) {
            rejected.push({ index: r.index, reason: r.reason || "Rejected by AI review" });
          }
        }
      }
    }
  } catch (err) {
    console.error("[RationaleValidator] AI check failed, accepting all:", err);
    // Non-fatal — if Haiku is unavailable, accept (only local checks apply)
  }

  return { valid: rejected.length === 0, rejected };
}