autosnippet/dist/lib/external/mcp/handlers/bootstrap/pipeline/dimension-configs.js

Extract code patterns into a knowledge base for AI coding assistants

/**
 * dimension-configs.js — v3.0 维度配置 + Tier Reflection
 *
 * **v2: 从统一维度注册表 (DimensionRegistry) 派生**
 *
 * 从 orchestrator.js 拆分，包含:
 * - DIMENSION_CONFIGS_V3: 维度的 outputType + allowedKnowledgeTypes
 *   现在从 DimensionRegistry 自动生成，无需手动维护
 * - getFullDimensionConfig(): 合并 baseDimensions + V3 专属配置
 * - buildTierReflection: Tier 级反思聚合 (规则化, 不需要 AI)
 *
 * @module pipeline/dimension-configs
 */
import { getDimensionFocusKeywords, getDimensionSOP } from '#domain/dimension/DimensionSop.js';
import { DIMENSION_REGISTRY, getDimension } from '#domain/dimension/index.js';
import { baseDimensions } from '../base-dimensions.js';
// ──────────────────────────────────────────────────────────────────
// v3.0 维度配置 — 从统一注册表自动生成
// ──────────────────────────────────────────────────────────────────
/**
 * 从统一注册表生成 V3 配置映射
 * 所有维度统一为 candidate-only outputType
 */
export const DIMENSION_CONFIGS_V3 = Object.fromEntries(DIMENSION_REGISTRY.map((dim) => [
    dim.id,
    {
        outputType: 'candidate',
        allowedKnowledgeTypes: [...dim.allowedKnowledgeTypes],
    },
]));
// ──────────────────────────────────────────────────────────────────
// 完整维度配置获取（合并 baseDimensions + V3 专属 + SOP）
// ──────────────────────────────────────────────────────────────────
/**
 * 获取完整维度配置（合并 baseDimensions + V3 专属配置 + SOP）
 *
 * @param dimId 维度 ID
 * @returns 完整维度配置，或 null（未知维度）
 */
export function getFullDimensionConfig(dimId) {
    // 优先从统一注册表获取
    const unified = getDimension(dimId);
    // 回退到旧 baseDimensions（兼容 Enhancement Pack 动态维度）
    const base = unified
        ? {
            id: unified.id,
            label: unified.label,
            guide: unified.extractionGuide,
            knowledgeTypes: [...unified.allowedKnowledgeTypes],
        }
        : baseDimensions.find((d) => d.id === dimId);
    const v3 = DIMENSION_CONFIGS_V3[dimId];
    if (!base) {
        return null;
    }
    const sop = getDimensionSOP(dimId);
    return {
        id: dimId,
        label: base.label,
        guide: base.guide,
        outputType: v3?.outputType || 'candidate',
        allowedKnowledgeTypes: v3?.allowedKnowledgeTypes || base.knowledgeTypes || [],
        skillWorthy: false,
        dualOutput: false,
        knowledgeTypes: base.knowledgeTypes || [],
        // SOP 结构化分析步骤
        sopSteps: sop?.steps || null,
        commonMistakes: sop?.commonMistakes || [],
        timeEstimate: sop?.timeEstimate || null,
        // 关键关注域词汇（用于 EpisodicMemory 跨维度 findings 相关性匹配）
        focusKeywords: getDimensionFocusKeywords(dimId, base.guide),
    };
}
/**
 * 构建 Tier 级 Reflection — 在每个 Tier 完成后调用
 *
 * 无需 AI 调用，通过规则化聚合维度发现:
 * - 收集所有维度的关键发现并按重要性排序
 * - 检测跨维度重复模式
 * - 为下一 Tier 生成建议
 *
 * @param tierIndex Tier 索引 (0-based)
 * @param tierResults 本 Tier 的维度结果
 * @returns TierReflection
 */
export function buildTierReflection(tierIndex, tierResults, sessionStore) {
    const completedDimensions = [...tierResults.keys()];
    // 收集本 Tier 所有维度的 findings
    const allFindings = [];
    for (const dimId of completedDimensions) {
        const report = sessionStore.getDimensionReport(dimId);
        if (report?.findings) {
            for (const f of report.findings) {
                allFindings.push({ dimId, ...f });
            }
        }
    }
    // Top findings by importance
    const topFindings = allFindings
        .sort((a, b) => (b.importance || 5) - (a.importance || 5))
        .slice(0, 10);
    // 检测跨维度模式 (多个维度提到同一文件/关键词)
    const fileMentions = {};
    const keywordMentions = {};
    for (const f of allFindings) {
        // 统计文件引用频率
        if (f.evidence) {
            const ev = typeof f.evidence === 'string' ? f.evidence : String(f.evidence);
            const file = ev.split(':')[0];
            if (file) {
                fileMentions[file] = (fileMentions[file] || 0) + 1;
            }
        }
        // 统计关键词
        const words = (f.finding || '').split(/[\s,，。.]+/).filter((w) => w.length > 3);
        for (const w of words) {
            keywordMentions[w] = (keywordMentions[w] || 0) + 1;
        }
    }
    const crossDimensionPatterns = [];
    // 多维度引用的文件 = 跨维度热点
    for (const [file, count] of Object.entries(fileMentions)) {
        if (count >= 2) {
            crossDimensionPatterns.push(`文件 "${file}" 被 ${count} 个维度引用 — 可能是系统核心组件`);
        }
    }
    // 多维度提及的关键词
    for (const [word, count] of Object.entries(keywordMentions)) {
        if (count >= 3) {
            crossDimensionPatterns.push(`关键词 "${word}" 出现 ${count} 次 — 跨维度关联主题`);
        }
    }
    // 为下一 Tier 生成建议
    const suggestionsForNextTier = [];
    // 找出 gaps (各维度报告的未覆盖方面)
    for (const dimId of completedDimensions) {
        const report = sessionStore.getDimensionReport(dimId);
        const gaps = report?.digest?.gaps || [];
        for (const gap of gaps) {
            if (gap && typeof gap === 'string' && gap.length > 5) {
                suggestionsForNextTier.push(`[${dimId}] 未覆盖: ${gap}`);
            }
        }
    }
    // remainingTasks
    for (const dimId of completedDimensions) {
        const report = sessionStore.getDimensionReport(dimId);
        const remaining = report?.digest?.remainingTasks || [];
        for (const task of remaining) {
            if (task?.signal) {
                suggestionsForNextTier.push(`[${dimId}] 遗留信号: ${task.signal} (${task.reason || '未处理'})`);
            }
        }
    }
    return {
        tierIndex,
        completedDimensions,
        topFindings,
        crossDimensionPatterns: crossDimensionPatterns.slice(0, 5),
        suggestionsForNextTier: suggestionsForNextTier.slice(0, 8),
    };
}