@steedos-labs/content-compliance 0.8.36 → 1.0.1

package/scripts/audit-legacy-tags.js

@@ -0,0 +1,246 @@
+/**
+ * 存量标签数据体检 —— issue #448 上线前必跑。
+ *
+ * 用法(直接对着历史库跑, 只读, 不写任何数据):
+ *   mongosh "mongodb://127.0.0.1:27017/pepsico-9-17" steedos-packages/content-review/scripts/audit-legacy-tags.js
+ *
+ * 查七件事:
+ *   1. C2  存量 tagValId 是否只含字母数字下划线、≤64 —— 违规的将来被 40003 挡掉, 永远同步不了
+ *   2. C5  树深是否 ≤3 —— 超了的话给其下加子节点会被拒
+ *   3. C1  同一类目内 value 是否唯一 —— 重复会让定位键失效
+ *   4. 父级引用是否悬空 —— 悬空的在下拉里会被当成一级节点, 层级静默错位
+ *   5. 水位线  多少条缺 ch_changed_at —— 缺的没有乱序保护, 会被旧事件覆盖
+ *   6. 历史单据  引用了不可见/不存在选项的单据数 —— 方案 §10 要求历史单据不得显示为空
+ *   7. 硬编码路径  三个元数据文件里写死的 7 条素材类型路径是否仍然成立(见 P1)
+ *
+ * 判"可见"的口径照抄真实数据源, 不是想当然:
+ *   四个字段的 source 都是 picklist_options(filters:[["enable","=",true]]),
+ *   adaptor 再按 parent.value 自底向上拼树。于是
+ *     - enable 不等于 true(含字段缺失)的选项, 下拉里根本没有;
+ *     - 父级被过滤掉时, 其下整棵子树都拼不进树, 即使子节点自己 enable=true。
+ *   只数"自己被停用"会显著少报受影响面。
+ */
+const WHITELIST = {
+    '223321001114213': '品牌/产品',
+    '223321001114222': '素材类型',
+    '223321001114944': '可使用渠道',
+    '223321022150011': '素材授权规则'
+};
+// 审批单上引用这四类标签的字段(值为 '-' 连接的完整路径, 可能是字符串或数组)
+const APPROVAL_FIELDS = {
+    '223321001114213': ['product'],
+    '223321001114222': ['material_type'],
+    '223321001114944': ['available_channels'],
+    '223321022150011': ['content_rule']
+};
+// 写死在元数据里、驱动"上线必填失效日期+授权规则+渠道"这条规则的 7 条素材类型路径。
+// 出处: fields/expiry_date.field.yml、fields/content_rule.field.yml、
+//       buttons/pepsico_confirm_online.button.yml(3 处)
+const HARDCODED_PATHS = [
+    '223321001114363-223321001114372',
+    '223321001114363-223321001114381',
+    '223321001114363-223321001114395',
+    '223321001114363-2233305175173759',
+    '223321001114468-22334121112868126',
+    '2233305104650610-2233305104650621',
+    '2233305104650610-2233305104650637'
+];
+const VALID = /^[A-Za-z0-9_]+$/;
+const MAX_LEN = 64, MAX_DEPTH = 3, SEP = '-';
+
+const line = (s) => print(s);
+const head = (s) => { print(''); print('═══ ' + s + ' ═══'); };
+
+let totalProblems = 0;
+const flag = (n) => { totalProblems += n; return n; };
+
+head('0. 托管类目');
+const picklists = {};
+Object.keys(WHITELIST).forEach((code) => {
+    const pl = db.picklists.findOne({ code: code });
+    if (!pl) { line('  ✗ ' + code + ' ' + WHITELIST[code] + ' —— 类目不存在！接口会返回 50000'); flag(1); return; }
+    picklists[code] = pl;
+    line('  ✓ ' + code + ' ' + WHITELIST[code] + '  (_id=' + pl._id + ')');
+});
+
+// 各类目的索引, 后面硬编码路径那节要跨类目找, 所以存下来
+const indexes = {};
+
+Object.keys(picklists).forEach((code) => {
+    const pl = picklists[code];
+    head(WHITELIST[code] + '  ' + code);
+
+    const opts = db.picklist_options.find({ picklist: pl._id }).toArray();
+    const byId = {}, byValue = {};
+    opts.forEach((o) => { byId[o._id] = o; (byValue[o.value] = byValue[o.value] || []).push(o); });
+
+    // parent 存的是 _id, 但历史数据里也见过直接存 value 的, 两种都认
+    const parentOf = (o) => (o.parent ? (byId[o.parent] || (byValue[o.parent] && byValue[o.parent][0]) || null) : null);
+    const hasParentRef = (o) => o.parent !== null && o.parent !== undefined && o.parent !== '';
+
+    // value 允许含 '-'(恰恰是 C2 要查的违规), 所以解析末级不能 split('-').pop(),
+    // 那样 'LEGACY-OLD-01' 会被切成 '01'。改成在已知 value 里做最长后缀匹配。
+    const valuesByLen = Object.keys(byValue).sort((a, b) => b.length - a.length);
+    const leafCache = {};
+    const resolveLeaf = (path) => {
+        const p = String(path);
+        if (leafCache[p] !== undefined) return leafCache[p];
+        let hit = null;
+        for (let i = 0; i < valuesByLen.length; i++) {
+            const v = valuesByLen[i];
+            if (p === v || (p.length > v.length && p.slice(-(v.length + SEP.length)) === SEP + v)) { hit = byValue[v][0]; break; }
+        }
+        leafCache[p] = hit;
+        return hit;
+    };
+
+    // 可见性: 照抄数据源口径 —— 自己 enable=true, 且所有祖先都 enable=true
+    const visibilityOf = (o) => {
+        let c = o, guard = 0;
+        while (c && guard++ < 20) {
+            if (c.enable !== true) return c === o ? 'self_disabled' : 'ancestor_disabled';
+            if (!hasParentRef(c)) return 'visible';
+            const p = parentOf(c);
+            if (!p) return 'root_promoted';   // 父级悬空: 下拉里会被当一级节点, 层级错位但仍可见
+            c = p;
+        }
+        return 'cycle';
+    };
+
+    indexes[code] = { pl: pl, opts: opts, byId: byId, byValue: byValue, parentOf: parentOf, resolveLeaf: resolveLeaf, visibilityOf: visibilityOf };
+
+    const enabled = opts.filter((o) => o.enable === true);
+    line('  选项总数 ' + opts.length + ' (enable=true ' + enabled.length +
+         ' / 其余 ' + (opts.length - enabled.length) + ', 其余含 enable 字段缺失, 下拉里同样不出现)');
+    const invisible = opts.filter((o) => { const v = visibilityOf(o); return v === 'self_disabled' || v === 'ancestor_disabled'; });
+    const byAncestor = opts.filter((o) => visibilityOf(o) === 'ancestor_disabled');
+    line('  下拉里不可见 ' + invisible.length + ' 条, 其中 ' + byAncestor.length + ' 条自己是启用的, 只因父级被停用而整棵子树消失');
+
+    // --- C2 字符集与长度 ---
+    const badChar = opts.filter((o) => !VALID.test(String(o.value)));
+    const tooLong = opts.filter((o) => String(o.value).length > MAX_LEN);
+    if (badChar.length) {
+        line('  ✗ C2 字符集违规 ' + flag(badChar.length) + ' 条 —— 这些标签将来无法同步(40003):');
+        badChar.slice(0, 10).forEach((o) => line('      ' + o.value + '   ' + o.name));
+        if (badChar.length > 10) line('      … 共 ' + badChar.length + ' 条');
+    } else { line('  ✓ C2 字符集'); }
+    if (tooLong.length) { line('  ✗ C2 超长 ' + flag(tooLong.length) + ' 条'); } else { line('  ✓ C2 长度'); }
+
+    // --- C1 value 唯一 ---
+    const dups = Object.keys(byValue).filter((v) => byValue[v].length > 1);
+    if (dups.length) {
+        line('  ✗ C1 value 重复 ' + flag(dups.length) + ' 个 —— 定位键失效:');
+        dups.slice(0, 10).forEach((v) => line('      ' + v + '  ×' + byValue[v].length +
+            '  (' + byValue[v].map((o) => o.name).join(' / ') + ')'));
+    } else { line('  ✓ C1 value 唯一'); }
+
+    // --- 父级悬空 ---
+    const dangling = opts.filter((o) => hasParentRef(o) && !parentOf(o));
+    if (dangling.length) {
+        line('  ✗ 父级悬空 ' + flag(dangling.length) + ' 条 —— parent 指向的选项在本类目里找不到,');
+        line('    adaptor 会把它们当一级节点渲染, 层级静默错位:');
+        dangling.slice(0, 10).forEach((o) => line('      ' + o.value + '  ' + o.name + '  parent=' + o.parent));
+    } else { line('  ✓ 无父级悬空'); }
+
+    // --- C5 层级 ---
+    const depthOf = (o) => { let d = 1, c = o, guard = 0; while (c && parentOf(c) && guard++ < 20) { c = parentOf(c); d++; } return d; };
+    const deep = opts.filter((o) => depthOf(o) > MAX_DEPTH);
+    const maxDepth = opts.reduce((m, o) => Math.max(m, depthOf(o)), 0);
+    line('  实际最大层级 ' + maxDepth);
+    if (deep.length) {
+        line('  ✗ C5 超过 ' + MAX_DEPTH + ' 层 ' + flag(deep.length) + ' 条:');
+        deep.slice(0, 10).forEach((o) => line('      ' + o.value + '  ' + o.name + '  第' + depthOf(o) + '层'));
+    } else { line('  ✓ C5 层级'); }
+
+    // --- 水位线 ---
+    const noWm = opts.filter((o) => !o.ch_changed_at).length;
+    if (noWm) {
+        line('  ⚠ 缺 ch_changed_at ' + noWm + ' 条 —— 没有乱序保护, 首个到达的事件无论多旧都会被应用。');
+        line('    (四个类目的补基线命令统一打印在最后)');
+    } else { line('  ✓ 水位线齐全'); }
+
+    // --- 历史单据引用 ---
+    const fields = APPROVAL_FIELDS[code] || [];
+    fields.forEach((f) => {
+        const q = {}; q[f] = { $exists: true, $ne: null };
+        const proj = { _id: 1, name: 1 }; proj[f] = 1;
+        const rows = db.pepsico_material_approval.find(q, proj).toArray();
+        const invisibleHit = [], orphanHit = [];
+        let refCount = 0;
+        rows.forEach((r) => {
+            const vals = Array.isArray(r[f]) ? r[f] : [r[f]];
+            vals.forEach((path) => {
+                if (!path) return;
+                refCount++;
+                const hit = resolveLeaf(path);
+                if (!hit) { orphanHit.push(r._id + ' → ' + path); return; }
+                const vis = visibilityOf(hit);
+                if (vis === 'self_disabled' || vis === 'ancestor_disabled') {
+                    invisibleHit.push(r._id + ' → ' + path + ' (' + hit.name + ', ' +
+                        (vis === 'self_disabled' ? '本身停用' : '父级停用') + ')');
+                }
+            });
+        });
+        line('  历史单据字段 ' + f + ': ' + rows.length + ' 张单据, ' + refCount + ' 处引用');
+        if (orphanHit.length) {
+            line('    ✗ 孤儿引用 ' + flag(orphanHit.length) + ' 处 —— 选项表里找不到, 表单已无法回显:');
+            orphanHit.slice(0, 5).forEach((s) => line('        ' + s));
+            if (orphanHit.length > 5) line('        … 共 ' + orphanHit.length + ' 处');
+        } else { line('    ✓ 无孤儿引用'); }
+        if (invisibleHit.length) {
+            line('    ⚠ 引用下拉里已不可见的选项 ' + invisibleHit.length + ' 处 —— 数据源过滤 enable=true,');
+            line('      方案 §10 要求历史单据不得因停用而显示为空, 这些就是受影响面:');
+            invisibleHit.slice(0, 5).forEach((s) => line('        ' + s));
+            if (invisibleHit.length > 5) line('        … 共 ' + invisibleHit.length + ' 处');
+        } else { line('    ✓ 无引用不可见选项'); }
+    });
+});
+
+// --- 硬编码路径(P1) ---
+head('硬编码标签路径  7 条');
+line('  出处: fields/expiry_date.field.yml、fields/content_rule.field.yml、buttons/pepsico_confirm_online.button.yml');
+line('  这 7 条驱动"上线必填失效日期 + 授权规则 + 渠道", 形如 material_type.includes(\'…\')。');
+line('  匹配不上时必填校验静默失效, 没有任何报错 —— 所以这节要盯住"路径是否仍然成立"。');
+const codes = Object.keys(indexes);
+HARDCODED_PATHS.forEach((path) => {
+    let found = null, foundCode = null;
+    for (let i = 0; i < codes.length; i++) {
+        const hit = indexes[codes[i]].resolveLeaf(path);
+        if (hit) { found = hit; foundCode = codes[i]; break; }
+    }
+    if (!found) { line('  ✗ ' + path + '  末级在四个托管类目里都找不到 —— 规则已失效'); flag(1); return; }
+    const ix = indexes[foundCode];
+    // 从末级往上拼回完整路径, 与写死的那条逐字比对 —— 这同时验了父子关系(C4)
+    const chain = []; let c = found, guard = 0;
+    while (c && guard++ < 20) { chain.unshift(c); c = ix.parentOf(c); }
+    const actual = chain.map((o) => o.value).join(SEP);
+    const names = chain.map((o) => o.name).join(' > ');
+    const vis = ix.visibilityOf(found);
+    const ok = actual === path;
+    line('  ' + (ok && vis === 'visible' ? '✓' : '✗') + ' ' + path);
+    line('      名称  ' + names + '   (类目: ' + WHITELIST[foundCode] + ')');
+    if (!ok) { line('      ✗ 实际路径已变成 ' + actual + ' —— 父级被移动过, 公式匹配不上了'); flag(1); }
+    if (vis !== 'visible') { line('      ✗ 当前不可见(' + vis + ') —— 新单据选不到, 但公式仍在引用'); flag(1); }
+    const f = (APPROVAL_FIELDS[foundCode] || [])[0];
+    if (f) {
+        const q = {}; q[f] = path;
+        line('      历史单据精确引用此路径 ' + db.pepsico_material_approval.countDocuments(q) + ' 张');
+    }
+});
+
+// --- P0-b 补基线 ---
+head('上线补基线(P0-b, 存量记录缺水位线则乱序保护整体失效)');
+const ids = Object.keys(picklists).map((c) => '"' + picklists[c]._id + '"').join(', ');
+line('  初始导入完成、接口上线那一刻执行(把时间换成真实上线时刻):');
+line('');
+line('  db.picklist_options.updateMany(');
+line('    { picklist: { $in: [' + ids + '] }, ch_changed_at: { $exists: false } },');
+line('    { $set: { ch_changed_at: new Date("<接口上线时刻>") } })');
+line('');
+line('  不补的话: 代码判据是 if (existing && existing.ch_changed_at), 存量记录整个跳过水位线比较,');
+line('  半年前的改名事件推进来也会被直接应用, 正确名称被覆盖。');
+
+head('结论');
+line(totalProblems === 0
+    ? '  未发现阻塞性问题(⚠ 项仍需按上面的提示处理)。'
+    : '  发现 ' + totalProblems + ' 处需要处理的问题, 见上。');

package/scripts/probe-content-hub-tags.js

@@ -0,0 +1,134 @@
+#!/usr/bin/env node
+/**
+ * 拉取 Content Hub 真实标签树, 校验 issue #448 的 C1 / C2 / C5 约束。
+ *
+ * 用途: 回答 issue #448 待确认 #1「tagValId 是否满足 C1/C2」——
+ * 该项标注「不满足则方案重做」, 联调前必须先验。
+ *
+ * 凭据只从环境变量读, 不写进仓库:
+ *   CH_BASE_URL=https://cback.getmogic.com \
+ *   CH_APP_ID=... CH_APP_SECRET=... CH_TENANT_ID=... CH_WORKSPACE_ID=... \
+ *   node scripts/probe-content-hub-tags.js
+ */
+const https = require('https');
+const { URL } = require('url');
+
+const BASE = (process.env.CH_BASE_URL || '').replace(/\/+$/, '');
+const APP_ID = process.env.CH_APP_ID;
+const APP_SECRET = process.env.CH_APP_SECRET;
+const TENANT_ID = process.env.CH_TENANT_ID;
+const WORKSPACE_ID = process.env.CH_WORKSPACE_ID;
+
+if (!BASE || !APP_ID || !APP_SECRET || !TENANT_ID || !WORKSPACE_ID) {
+    console.error('缺少环境变量: CH_BASE_URL / CH_APP_ID / CH_APP_SECRET / CH_TENANT_ID / CH_WORKSPACE_ID');
+    process.exit(2);
+}
+
+// 与服务端实现保持一致(见 src/methods/content_hub_tag.ts)
+const TAG_VAL_ID_PATTERN = /^[A-Za-z0-9_]+$/;
+const MAX_TAG_VAL_ID_LENGTH = 64;
+const MAX_TAG_DEPTH = 3;
+
+const post = (path, body, headers) => new Promise((resolve, reject) => {
+    const payload = JSON.stringify(body);
+    const url = new URL(BASE + path);
+    const req = https.request({
+        hostname: url.hostname,
+        port: url.port || 443,
+        path: url.pathname + url.search,
+        method: 'POST',
+        headers: Object.assign({
+            'Content-Type': 'application/json;charset=UTF-8',
+            'Content-Length': Buffer.byteLength(payload)
+        }, headers || {})
+    }, (res) => {
+        let raw = '';
+        res.on('data', (c) => { raw += c; });
+        res.on('end', () => {
+            try {
+                resolve({ status: res.statusCode, body: JSON.parse(raw) });
+            } catch (e) {
+                resolve({ status: res.statusCode, body: raw.slice(0, 400) });
+            }
+        });
+    });
+    req.on('error', reject);
+    req.write(payload);
+    req.end();
+});
+
+const TAGS = [
+    { key: 'brand-product',  label: '品牌/产品',     tagId: '223321001114213' },
+    { key: 'channel',        label: '可使用渠道',    tagId: '223321001114944' },
+    { key: 'material-type',  label: '素材类型',      tagId: '223321001114222' }
+    // 注意: 对方 API v2 文档里没有「素材授权规则」(223321022150011) 的查询接口,
+    // 但它在 issue #448 的托管白名单内 —— 需要向 Content Hub 确认。
+];
+
+(async () => {
+    console.log('=== T0 换取 tenant-access-token ===');
+    const auth = await post('/creativeSaas/auth/v1/tenantAccessToken/internal', { appId: APP_ID, appSecret: APP_SECRET });
+    const token = auth.body && auth.body.data && auth.body.data.tenantAccessToken;
+    if (!token) {
+        console.error('取 token 失败:', JSON.stringify(auth.body).slice(0, 300));
+        process.exit(1);
+    }
+    console.log('  ok, expire =', auth.body.data.expire);
+
+    const problems = [];
+    for (const tag of TAGS) {
+        console.log(`\n=== ${tag.label} (${tag.tagId}) ===`);
+        const res = await post(`/creativeSaas/pepsico/compliance/v1/tags/${tag.key}`,
+            { tenantId: TENANT_ID, workspaceId: WORKSPACE_ID, keyword: null },
+            { 'tenant-access-token': token });
+        const list = res.body && res.body.data && (res.body.data.list || res.body.data);
+        if (!Array.isArray(list)) {
+            console.error('  返回结构不是数组:', JSON.stringify(res.body).slice(0, 300));
+            problems.push(`${tag.label}: 返回结构异常`);
+            continue;
+        }
+
+        const seen = new Map();
+        let total = 0, maxDepth = 0, disabled = 0;
+        const badCharset = [], tooLong = [], tooDeep = [], duplicated = [];
+
+        const walk = (nodes, depth) => {
+            for (const node of nodes || []) {
+                total += 1;
+                maxDepth = Math.max(maxDepth, depth);
+                const id = String(node.id);
+                if (!TAG_VAL_ID_PATTERN.test(id)) badCharset.push(`${id} (${node.value})`);
+                if (id.length > MAX_TAG_VAL_ID_LENGTH) tooLong.push(id);
+                if (depth > MAX_TAG_DEPTH) tooDeep.push(`${id} 第 ${depth} 层 (${node.value})`);
+                if (seen.has(id)) duplicated.push(`${id}: ${seen.get(id)} / ${node.value}`);
+                else seen.set(id, node.value);
+                if (node.disable === true || node.status === 0) disabled += 1;
+                walk(node.childTagList, depth + 1);
+            }
+        };
+        walk(list, 1);
+
+        console.log(`  节点总数 ${total}, 最大层级 ${maxDepth}, 停用节点 ${disabled}`);
+        const report = (name, arr, constraint) => {
+            if (arr.length) {
+                console.log(`  ✗ ${constraint} 违反 ${arr.length} 处: ${arr.slice(0, 5).join('; ')}${arr.length > 5 ? ' …' : ''}`);
+                problems.push(`${tag.label} ${constraint}: ${arr.length} 处`);
+            } else {
+                console.log(`  ✓ ${constraint}`);
+            }
+        };
+        report('charset', badCharset, 'C2 只含字母数字下划线');
+        report('length', tooLong, `C2 长度 ≤ ${MAX_TAG_VAL_ID_LENGTH}`);
+        report('depth', tooDeep, `C5 层级 ≤ ${MAX_TAG_DEPTH}`);
+        report('dup', duplicated, 'C1 ID 唯一');
+    }
+
+    console.log('\n================ 结论 ================');
+    if (problems.length) {
+        console.log('发现约束冲突, 联调前必须先解决:');
+        problems.forEach((p) => console.log('  - ' + p));
+        process.exit(1);
+    }
+    console.log('三个标签类目均满足 C1 / C2 / C5, 可以按现方案联调。');
+    console.log('另需向 Content Hub 确认: 素材授权规则 223321022150011 的查询接口(v2 文档未提供)。');
+})().catch((e) => { console.error('探测失败:', e.message); process.exit(1); });

package/main/default/objects/pepsico_material_approval/fields/has_detail_page.field.yml

@@ -1,18 +0,0 @@
-name: has_detail_page
-label: 素材是否包含商品详情页
-type: select
-data_type: text
-group: 基本信息
-inlineHelpText: 涉及到商品详情页的素材，请关联pls文件
-sort_no: 805
-is_wide: true
-required: true
-amis:
-  type: button-group-select
-  multiple: false
-options:
-  - label: 是
-    value: 'yes'
-  - label: 否
-    value: 'no'
-visible_on: (classification == 'MD-PO1-PO15' || classification == 'SLS-PO2-PO25') && form_version == 'v1'