connectai/tests/contextManager.test.ts

import {
    estimateTokens,
    estimateMessagesTokens,
    computeOutputBudget,
    trimHistoryToBudget,
    truncateSystemPromptContext,
    classifyStopReason,
    shouldShowTruncationNotice,
    estimateModelParamsB,
    estimateActiveParamsB,
    CONTEXT_OPEN_MARKER,
    CONTEXT_CLOSE_MARKER,
    type BudgetMessage,
} from '../src/lib/contextManager';

describe('contextManager.estimateModelParamsB', () => {
    it('reads common naming schemes', () => {
        expect(estimateModelParamsB('qwen2.5-7b-instruct')).toBe(7);
        expect(estimateModelParamsB('llama-3.1-8b')).toBe(8);
        expect(estimateModelParamsB('google/gemma-3n-e2b-it')).toBe(2);
        expect(estimateModelParamsB('gemma4:e2b')).toBe(2);
        expect(estimateModelParamsB('Qwen3-30B-A3B')).toBe(30);
    });
    it('returns null when there is no clear parameter hint', () => {
        expect(estimateModelParamsB('phi-3-mini')).toBeNull();
        expect(estimateModelParamsB('gpt-4o')).toBeNull();
        expect(estimateModelParamsB('')).toBeNull();
        expect(estimateModelParamsB('llama-q4bit')).toBeNull(); // quantization, not params
        expect(estimateModelParamsB('mixtral-8x7b')).toBeNull(); // MoE size is ambiguous — don't guess
    });
});

describe('contextManager.estimateActiveParamsB', () => {
    it('prefers active params for MoE naming (a/e prefix)', () => {
        expect(estimateActiveParamsB('gemma-4-26b-a4b-it')).toBe(4);   // 활성 4B (총 26B 아님)
        expect(estimateActiveParamsB('Qwen3-30B-A3B')).toBe(3);        // 활성 3B
        expect(estimateActiveParamsB('google/gemma-3n-e2b-it')).toBe(2);
    });
    it('falls back to total params when no active hint', () => {
        expect(estimateActiveParamsB('llama-3.1-8b')).toBe(8);
        expect(estimateActiveParamsB('qwen2.5-7b-instruct')).toBe(7);
    });
    it('returns null when there is no parameter hint', () => {
        expect(estimateActiveParamsB('phi-3-mini')).toBeNull();
        expect(estimateActiveParamsB('')).toBeNull();
    });
});

describe('contextManager.computeOutputBudget', () => {
    const limits = { contextLength: 32768, maxOutputTokens: 4096, safetyMargin: 2048, minOutputTokens: 512 };
    it('caps at maxOutputTokens when there is plenty of room', () => {
        const r = computeOutputBudget(1000, limits);
        expect(r.maxOutputTokens).toBe(4096);
        expect(r.tight).toBe(false);
    });
    it('shrinks output as input grows', () => {
        const r = computeOutputBudget(30000, limits); // 32768 - 30000 - 2048 = 720
        expect(r.maxOutputTokens).toBe(720);
        expect(r.tight).toBe(false);
    });
    it('flags tight and floors at minOutputTokens when input nearly fills the window', () => {
        const r = computeOutputBudget(31000, limits); // available 32768-31000-2048 = -280 ≤ 512
        expect(r.maxOutputTokens).toBe(512);
        expect(r.tight).toBe(true);
    });
});

describe('contextManager.trimHistoryToBudget', () => {
    // v2.2.69: makeMarker now also receives the dropped messages array so callers can build a real summary.
    // Tests don't need the dropped payload — just keep the signature compatible.
    const marker = (n: number, _dropped?: BudgetMessage[]): BudgetMessage => ({ role: 'system', content: `[dropped ${n}]`, internal: true });
    it('keeps everything when under budget', () => {
        const msgs: BudgetMessage[] = [{ role: 'user', content: 'hi' }, { role: 'assistant', content: 'hello' }];
        const r = trimHistoryToBudget(msgs, 10_000, marker);
        expect(r.droppedCount).toBe(0);
        expect(r.messages).toEqual(msgs);
    });
    it('drops oldest messages and prepends a marker when over budget', () => {
        const msgs: BudgetMessage[] = Array.from({ length: 10 }, (_, i) => ({ role: i % 2 ? 'assistant' : 'user', content: 'x'.repeat(400) }));
        const r = trimHistoryToBudget(msgs, 250, marker); // each msg ≈ 400*0.3+4 = 124 tokens
        expect(r.droppedCount).toBeGreaterThan(0);
        expect(r.messages[0].content).toMatch(/^\[dropped \d+\]$/);
        // most recent message survives
        expect(r.messages[r.messages.length - 1]).toEqual(msgs[msgs.length - 1]);
        expect(r.tokensAfter).toBeLessThanOrEqual(250 + estimateMessagesTokens([marker(1, [])]));
    });
    it('passes the dropped messages array to the marker factory (v2.2.69)', () => {
        const msgs: BudgetMessage[] = Array.from({ length: 6 }, (_, i) => ({
            role: i % 2 ? 'assistant' : 'user',
            content: 'x'.repeat(400),
        }));
        let observedDropped: BudgetMessage[] | undefined;
        const factory = (n: number, dropped: BudgetMessage[]): BudgetMessage => {
            observedDropped = dropped;
            return { role: 'system', content: `[summary of ${n}: first=${dropped[0]?.role}]`, internal: true };
        };
        const r = trimHistoryToBudget(msgs, 250, factory);
        expect(r.droppedCount).toBeGreaterThan(0);
        expect(observedDropped).toBeDefined();
        expect(observedDropped!.length).toBe(r.droppedCount);
        // Dropped messages are the OLDEST ones, in order.
        expect(observedDropped![0]).toEqual(msgs[0]);
        expect(r.messages[0].content).toMatch(/^\[summary of \d+: first=user\]$/);
    });
    it('always keeps at least the last message even if it alone exceeds the budget', () => {
        const msgs: BudgetMessage[] = [{ role: 'user', content: 'short' }, { role: 'user', content: 'y'.repeat(5000) }];
        const r = trimHistoryToBudget(msgs, 10, marker);
        expect(r.messages.some(m => m.content === 'y'.repeat(5000))).toBe(true);
    });
});

describe('contextManager.truncateSystemPromptContext', () => {
    it('leaves a small prompt untouched', () => {
        const p = 'You are helpful.';
        expect(truncateSystemPromptContext(p, 1000)).toEqual({ prompt: p, truncated: false });
    });
    it('trims only the [CONTEXT]…[/CONTEXT] body, preserving head and tail', () => {
        const head = 'CORE INSTRUCTIONS that must never be dropped. ' + 'a'.repeat(200);
        const body = 'BIG BRAIN CONTEXT ' + 'b'.repeat(20_000);
        const tail = 'CRITICAL NEGATIVE CONSTRAINTS — also never dropped. ' + 'c'.repeat(200);
        const prompt = head + CONTEXT_OPEN_MARKER + body + CONTEXT_CLOSE_MARKER + tail;
        const out = truncateSystemPromptContext(prompt, 400);
        expect(out.truncated).toBe(true);
        expect(out.prompt).toContain('CORE INSTRUCTIONS');
        expect(out.prompt).toContain('CRITICAL NEGATIVE CONSTRAINTS');
        expect(out.prompt).toContain(CONTEXT_CLOSE_MARKER.trim());
        // The bulk of the body is gone
        expect(out.prompt.length).toBeLessThan(prompt.length / 2);
        expect(estimateTokens(out.prompt)).toBeLessThanOrEqual(400 + estimateTokens(tail) + 64);
    });
    it('falls back to a hard tail-cut when there is no [CONTEXT] marker', () => {
        const prompt = 'instructions ' + 'z'.repeat(50_000);
        const out = truncateSystemPromptContext(prompt, 200);
        expect(out.truncated).toBe(true);
        expect(out.prompt.length).toBeLessThan(prompt.length);
        expect(out.prompt.startsWith('instructions')).toBe(true);
    });
});

describe('contextManager.classifyStopReason', () => {
    it('maps engine-specific reasons to common kinds', () => {
        expect(classifyStopReason('eosFound')).toBe('complete');
        expect(classifyStopReason('stop')).toBe('complete');
        expect(classifyStopReason('length')).toBe('output-limit');
        expect(classifyStopReason('maxPredictedTokensReached')).toBe('output-limit');
        expect(classifyStopReason('contextLengthReached')).toBe('context-overflow');
        expect(classifyStopReason('userStopped')).toBe('user-stopped');
        expect(classifyStopReason('failed')).toBe('error');
        expect(classifyStopReason(undefined)).toBe('unknown');
    });
});

describe('contextManager.shouldShowTruncationNotice', () => {
    it('suppresses output-limit notices for visibly short answers', () => {
        expect(shouldShowTruncationNotice('output-limit', 80, 4096)).toBe(false);
    });

    it('shows output-limit notices when output consumed most of the budget', () => {
        expect(shouldShowTruncationNotice('output-limit', 3900, 4096)).toBe(true);
    });

    it('always surfaces context overflow and error stops', () => {
        expect(shouldShowTruncationNotice('context-overflow', 10, 4096)).toBe(true);
        expect(shouldShowTruncationNotice('error', 10, 4096)).toBe(true);
    });
});