/**
 * Phase 1 — context-window alignment.
 *
 * The budgeter must clamp to the model's ACTUALLY-loaded window when it's
 * smaller than the user's `contextLength` setting, so a model loaded with a
 * smaller window than the setting never silently overflows the server.
 */

import { computeBudgetedRequest } from '../src/agent/handlePrompt/computeBudgetedRequest';
import type { ChatMessage } from '../src/agent';

const baseConfig = {
    contextLength: 32768,
    maxOutputTokens: 4096,
    contextSafetyMargin: 512,
    smallModelContextCap: 0, // disabled
    autoCompactHistory: false,
};

function run(overrides: { actualContextLength?: number; config?: Partial<typeof baseConfig> } = {}) {
    const reqMessages: ChatMessage[] = [{ role: 'user', content: 'hello' }];
    return computeBudgetedRequest({
        fullSystemPrompt: 'You are a helpful assistant.',
        reqMessages,
        actualModel: 'some-13b-model',
        config: { ...baseConfig, ...overrides.config },
        imageCount: 0,
        actualContextLength: overrides.actualContextLength,
    });
}

describe('computeBudgetedRequest — real-window alignment', () => {
    test('clamps to the actual loaded window when it is smaller than the setting', () => {
        const r = run({ actualContextLength: 8192 });
        expect(r.windowMismatch).toBe(true);
        expect(r.effectiveContextLength).toBe(8192);
        expect(r.ctxLimits.contextLength).toBe(8192);
    });

    test('keeps the configured window when the actual window is unknown', () => {
        const r = run({ actualContextLength: undefined });
        expect(r.windowMismatch).toBe(false);
        expect(r.effectiveContextLength).toBe(32768);
        expect(r.ctxLimits.contextLength).toBe(32768);
    });

    test('does not raise the window when the actual window is larger than the setting', () => {
        const r = run({ actualContextLength: 131072 });
        expect(r.windowMismatch).toBe(false);
        expect(r.effectiveContextLength).toBe(32768); // setting is the lower bound here
    });

    test('ignores a non-positive / non-finite actual window (falls back to setting)', () => {
        expect(run({ actualContextLength: 0 }).effectiveContextLength).toBe(32768);
        expect(run({ actualContextLength: -5 }).effectiveContextLength).toBe(32768);
        expect(run({ actualContextLength: NaN }).effectiveContextLength).toBe(32768);
    });
});

// ────────────────────────────────────────────────────────────────────────────
// [v2.2.311] KV 캐시 분리 — dynamicContextTail 삽입 계약.
// message[0] 은 불변 정적 프롬프트로 남고, 동적 컨텍스트는 마지막 user 메시지
// 직전의 internal system 메시지로 들어가야 프롬프트 프리픽스 캐시가 산다.
// ────────────────────────────────────────────────────────────────────────────
describe('computeBudgetedRequest — dynamicContextTail (KV cache split)', () => {
    const HEAD = 'STATIC HEAD PROMPT';
    const TAIL = '[CURRENT DATE/TIME]\nToday: 2026-07-20\n\n[CONTEXT]\nRAG chunk A\n[/CONTEXT]';

    // 주의: default parameter 는 명시적 undefined 에도 적용되므로 쓰지 않는다 —
    // "tail 없음" 케이스를 정확히 표현하기 위해 항상 명시적으로 넘긴다.
    function runSplit(reqMessages: ChatMessage[], tail: string | undefined) {
        return computeBudgetedRequest({
            fullSystemPrompt: HEAD,
            dynamicContextTail: tail,
            reqMessages,
            actualModel: 'some-13b-model',
            config: baseConfig,
            imageCount: 0,
        });
    }

    test('tail 은 마지막 user 메시지 직전에 internal system 으로 삽입된다', () => {
        const msgs: ChatMessage[] = [
            { role: 'user', content: '첫 질문' },
            { role: 'assistant', content: '첫 답' },
            { role: 'user', content: '두번째 질문' },
        ];
        const r = runSplit(msgs, TAIL);
        expect(r.messagesForRequest[0]).toMatchObject({ role: 'system', content: HEAD });
        const idx = r.messagesForRequest.findIndex((m) => m.content === TAIL);
        expect(idx).toBeGreaterThan(0);
        expect(r.messagesForRequest[idx].role).toBe('system');
        expect(r.messagesForRequest[idx + 1]).toMatchObject({ role: 'user', content: '두번째 질문' });
    });

    test('continuation(마지막 user 뒤 액션 결과 system)이 있어도 tail 은 user 앞', () => {
        const msgs: ChatMessage[] = [
            { role: 'user', content: '조사해줘' },
            { role: 'system', content: '[Result of read_file] ...', internal: true },
        ];
        const r = runSplit(msgs, TAIL);
        const tailIdx = r.messagesForRequest.findIndex((m) => m.content === TAIL);
        const userIdx = r.messagesForRequest.findIndex((m) => m.role === 'user');
        const actionIdx = r.messagesForRequest.findIndex((m) => String(m.content).startsWith('[Result of'));
        expect(tailIdx).toBeLessThan(userIdx);
        expect(actionIdx).toBeGreaterThan(userIdx);
    });

    test('user 메시지가 없으면 tail 을 히스토리 끝에 붙인다', () => {
        const msgs: ChatMessage[] = [{ role: 'system', content: 'internal only', internal: true }];
        const r = runSplit(msgs, TAIL);
        const last = r.messagesForRequest[r.messagesForRequest.length - 1];
        expect(last.content).toBe(TAIL);
    });

    test('tail 미지정이면 종전 단일-시스템 동작 그대로', () => {
        const msgs: ChatMessage[] = [{ role: 'user', content: 'hello' }];
        const r = runSplit(msgs, undefined);
        expect(r.messagesForRequest).toHaveLength(2);
        expect(r.messagesForRequest[0].content).toBe(HEAD);
    });

    test('예산 초과 시 [CONTEXT] truncation 은 tail 에 적용되고 head 는 보존', () => {
        const hugeTail = `[CONTEXT]\n${'긴 배경 지식 청크. '.repeat(20000)}\n[/CONTEXT]`;
        const msgs: ChatMessage[] = [{ role: 'user', content: 'hello' }];
        const r = computeBudgetedRequest({
            fullSystemPrompt: HEAD,
            dynamicContextTail: hugeTail,
            reqMessages: msgs,
            actualModel: 'some-13b-model',
            config: { ...baseConfig, contextLength: 8192 },
            imageCount: 0,
        });
        expect(r.systemTruncated).toBe(true);
        expect(r.messagesForRequest[0].content).toBe(HEAD);
        const tailMsg = r.messagesForRequest.find((m) => String(m.content).includes('[CONTEXT]'));
        expect(tailMsg).toBeDefined();
        expect(String(tailMsg!.content).length).toBeLessThan(hugeTail.length);
    });

    test('systemTokens 는 head+tail 합산으로 보고된다', () => {
        const msgs: ChatMessage[] = [{ role: 'user', content: 'hello' }];
        const withTail = runSplit(msgs, TAIL);
        const withoutTail = runSplit(msgs, undefined);
        expect(withTail.systemTokens).toBeGreaterThan(withoutTail.systemTokens);
    });
});