import { eq, sql } from 'drizzle-orm';
import { crawlState } from '@rareindex/database';
import type { BudgetStore } from '@rareindex/connectors';
import { db } from './db.ts';

/**
 * Crawl budget engine (§170): per-URL content hash + change-interval estimate.
 * nextFetchAt = lastFetched + clamp(estimatedChangeInterval / 2, minH, maxH).
 * Unchanged fetches lengthen the interval (×1.5), changes shorten it (×0.5).
 */
export function createBudgetStore(opts: { connectorId: string; minHours?: number; maxHours?: number }): BudgetStore {
  const minH = opts.minHours ?? 1;
  const maxH = opts.maxHours ?? 24 * 14;
  return {
    async get(urlHash) {
      const [row] = await db().select({ contentHash: crawlState.contentHash, nextFetchAt: crawlState.nextFetchAt }).from(crawlState).where(eq(crawlState.urlHash, urlHash)).limit(1);
      return row ? { contentHash: row.contentHash, nextFetchAt: row.nextFetchAt } : null;
    },
    async record(input) {
      const now = new Date();
      const [prev] = await db().select().from(crawlState).where(eq(crawlState.urlHash, input.urlHash)).limit(1);
      let interval = prev?.changeIntervalHours ?? 24;
      if (prev) interval = input.changed ? Math.max(minH, interval * 0.5) : Math.min(maxH, interval * 1.5);
      const failures = input.status && input.status >= 400 ? (prev?.failures ?? 0) + 1 : 0;
      const backoff = failures > 0 ? Math.min(maxH, 2 ** failures) : interval / 2;
      const nextFetchAt = new Date(now.getTime() + Math.max(minH, backoff) * 3600_000);
      await db()
        .insert(crawlState)
        .values({
          urlHash: input.urlHash,
          connectorId: input.connectorId,
          url: input.url,
          contentHash: input.contentHash,
          lastFetchedAt: now,
          lastChangedAt: input.changed ? now : null,
          fetchCount: 1,
          changeCount: input.changed ? 1 : 0,
          changeIntervalHours: interval,
          nextFetchAt,
          lastStatus: input.status,
          failures,
        })
        .onConflictDoUpdate({
          target: crawlState.urlHash,
          set: {
            contentHash: input.contentHash,
            lastFetchedAt: now,
            lastChangedAt: input.changed ? now : sql`${crawlState.lastChangedAt}`,
            fetchCount: sql`${crawlState.fetchCount} + 1`,
            changeCount: sql`${crawlState.changeCount} + ${input.changed ? 1 : 0}`,
            changeIntervalHours: interval,
            nextFetchAt,
            lastStatus: input.status,
            failures,
          },
        });
    },
  };
}