#!/usr/bin/env python3
"""
EIP4 Schema 组件真实性审计基线扫描器
只读扫描,量化 6 类已确认缺陷,产出缺陷矩阵报告。
用法:
python3 schema_reality_audit.py [--schema-dir
] [--report-out ]
"""
import argparse
import json
import os
import re
import sys
from collections import defaultdict
from datetime import datetime
from pathlib import Path
# ── 默认路径 ──────────────────────────────────────────────────────────────────
DEFAULT_SCHEMA_DIR = (
"/Users/zhengwr/workspace/eip3/eip4-platform/platform-converter/转换产物/schema"
)
DEFAULT_REPORT_OUT = (
"/Users/zhengwr/workspace/eip3/docs/F2-17-组件真实性审计基线.md"
)
# ── 缺陷 1:人员列正则 ────────────────────────────────────────────────────────
PERSON_NAME_RE = re.compile(
r"(?i)(USER_?ID|_USER$|CREATEBY|UPDATEBY|.*PEOPLE.*|.*PERSON.*|HANDLER|DRAFT)"
)
PERSON_LABEL_KEYWORDS = ("人", "经办", "拟稿", "接收人", "负责人")
# ── 缺陷 3:长文本字段正则 ────────────────────────────────────────────────────
LONGTEXT_NAME_RE = re.compile(
r"(?i)(内容|描述|备注|说明|详情|正文|CONTENT|REMARK|MEMO|DESC|DETAIL)"
)
# ─────────────────────────────────────────────────────────────────────────────
# 通用递归工具
# ─────────────────────────────────────────────────────────────────────────────
def walk_nodes(node):
"""深度优先遍历 JSON 节点,yield 每个 dict。"""
if isinstance(node, dict):
yield node
for v in node.values():
yield from walk_nodes(v)
elif isinstance(node, list):
for item in node:
yield from walk_nodes(item)
def walk_strings(node):
"""递归 yield 所有字符串值。"""
if isinstance(node, str):
yield node
elif isinstance(node, dict):
for v in node.values():
yield from walk_strings(v)
elif isinstance(node, list):
for item in node:
yield from walk_strings(item)
def collect_form_fields(node):
"""从任意节点递归收集 form.body 里的 input-text 字段。"""
results = []
for n in walk_nodes(node):
if n.get("type") in ("form", "dialog") and "body" in n:
for field in (n["body"] if isinstance(n["body"], list) else [n["body"]]):
_collect_input_text(field, results)
return results
def _collect_input_text(node, out):
if isinstance(node, dict):
if node.get("type") == "input-text":
out.append(node)
for v in node.values():
_collect_input_text(v, out)
elif isinstance(node, list):
for item in node:
_collect_input_text(item, out)
# ─────────────────────────────────────────────────────────────────────────────
# 各缺陷检测函数(返回 list[dict] 命中条目)
# ─────────────────────────────────────────────────────────────────────────────
def detect_defect1_person_column(schema) -> list:
"""缺陷 1:crud.columns 人员列裸显示 id。"""
hits = []
for node in walk_nodes(schema):
if node.get("type") != "crud":
continue
for col in node.get("columns", []):
if not isinstance(col, dict):
continue
col_type = col.get("type", "text")
# 已有 mapping 或 tpl 则认为已处理,跳过
if col_type in ("mapping", "tpl"):
continue
name = col.get("name", "")
label = col.get("label", "")
matched = False
reason = ""
if PERSON_NAME_RE.search(name):
matched = True
reason = f"name 命中正则: {name}"
if not matched:
for kw in PERSON_LABEL_KEYWORDS:
if kw in str(label):
matched = True
reason = f"label 含关键字 '{kw}': {label}"
break
if matched:
hits.append({"name": name, "label": label, "reason": reason})
return hits
def detect_defect2_dead_button(schema) -> list:
"""缺陷 2:按钮死桩(未配编辑表单 / 该单据未配)。"""
hits = []
for s in walk_strings(schema):
if "未配编辑表单" in s or "该单据未配" in s:
hits.append({"text": s[:120]})
return hits
def detect_defect3_longtext_mismatch(schema) -> list:
"""缺陷 3:长文本/富文本错配单行 input-text。"""
hits = []
for node in walk_nodes(schema):
if node.get("type") != "input-text":
continue
name = node.get("name", "")
label = node.get("label", "")
combined = name + "|" + str(label)
if LONGTEXT_NAME_RE.search(combined):
hits.append({"name": name, "label": label})
return hits
def detect_defect4_dialog_embed_page(schema) -> list:
"""缺陷 4:弹窗嵌整页(service + schemaApi 含 /api/page/schema/)。"""
hits = []
for node in walk_nodes(schema):
if node.get("type") != "service":
continue
schema_api = node.get("schemaApi", "") or node.get("schemaAPI", "")
if isinstance(schema_api, dict):
schema_api = schema_api.get("url", "")
if isinstance(schema_api, str) and "/api/page/schema/" in schema_api:
hits.append({"schemaApi": schema_api[:200]})
return hits
def detect_defect5_picker(schema) -> list:
"""缺陷 5:依赖 picker 后端字段。同时收集 dialogid。"""
hits = []
picker_guid_re = re.compile(r"/api/picker/([0-9a-f\-]{36})", re.IGNORECASE)
for node in walk_nodes(schema):
source = node.get("source", "") or ""
if isinstance(source, dict):
source = source.get("url", "") or ""
if isinstance(source, str) and "/api/picker/" in source:
m = picker_guid_re.search(source)
dialog_id = m.group(1) if m else "unknown"
hits.append({
"name": node.get("name", ""),
"label": node.get("label", ""),
"dialogId": dialog_id,
"source": source[:200],
})
return hits
def detect_defect6_crud_no_pagination(schema) -> bool:
"""缺陷 6:body.type=='crud' 且有 api(全局后端不分页缺陷)。"""
body = schema.get("body", {})
if isinstance(body, dict):
if body.get("type") == "crud" and body.get("api"):
return True
elif isinstance(body, list):
for item in body:
if isinstance(item, dict) and item.get("type") == "crud" and item.get("api"):
return True
return False
# ─────────────────────────────────────────────────────────────────────────────
# 运行时已修项(审计盲区):静态产物里仍呈未修形态,但运行时已解决,不计真实缺口。
# 升级要点(EIP4-架构与缺口现状-2026-06-13 §4.3 / §7-2):D1/D4/D5/D6 改用 e2e 验证,
# 静态计数仅供回归观察,不再当作覆盖缺口。D2/D3 仍是静态真缺口。
# ─────────────────────────────────────────────────────────────────────────────
RUNTIME_RESOLVED = {
"D1": "人员/引用列裸 id → 运行时 UserColumnResolver 解析任意引用实体的 _DISPLAY__(662828c+2a6c1d0)",
"D4": "弹窗嵌整页 → 运行时 format=amis 剥 page 壳,弹窗只渲表单本体(a021f33)",
"D5": "依赖 picker 后端 → SelectDialog 已弹窗化为 amis picker 多列对照,运行时 /api/picker 供列(9d92df2)",
"D6": "CRUD 不分页 → 运行时 FlowController 分页翻译(_SEARCH 合成流)",
}
STATIC_REAL = {
"D2": "按钮死桩(未配编辑表单/该单据未配)—— 仍是静态真缺口(诚实降级,待配 EDITFORM 或低代码补)",
"D3": "长文本/富文本错配单行 input-text —— 富文本已大面修(37db794),残余按 e2e 复核",
}
# 覆盖缺口候选 uitype(§4.2「真丢失控件」)—— 频次统计重点关注,决定 §7-4 优先级。
LOSTLIKE_UITYPES = ("TwinColSelect", "PasswordField")
# 高级模板结构标记(§4.2「高级模板类型缺失」)—— 子串探测(与 DB 实测口径一致:
# 这些标记在 XML 里以元素名/属性出现,不一定带尖括号根形态,故用大小写不敏感子串匹配,
# 避免漏报。实测存量:editgrid≈55 / calendar≈2 / workflow≈1 / mastersubgrid·treegrid·subgrid=0。
ADVANCED_TEMPLATE_MARKERS = {
"EditGrid(子表)": ("editgrid",),
"MasterSubGrid": ("mastersubgrid",),
"SubGrid": ("subgrid",),
"TreeGrid": ("treegrid",),
"Calendar": ("calendar",),
"Wf/Workflow": (" str:
"""取 XML 第一个元素标签名(root),忽略 声明与注释。"""
s = re.sub(r"<\?.*?\?>", "", xml, flags=re.DOTALL)
s = re.sub(r"", "", s, flags=re.DOTALL)
m = _ROOT_TAG_RE.search(s)
return m.group(1).lower() if m else "(none)"
def uitype_template_frequency(dsn: str) -> dict:
"""
DB 视角的 uitype × 模板根类型 × 频次统计(§7-2)。
直接扫 view_template.xml(authoritative 存量真实用量),让 §4.2 的覆盖缺口能按真实频次排序。
返回:
by_uitype: {uitype: {"occ": 出现次数, "tpls": 模板数}}
by_root: {root_tag: 模板数}
by_uitype_root:{(uitype, root): 模板数}
advanced: {marker: 模板数} 高级模板结构存量
lostlike: {uitype: {"occ","tpls"}} §4.2 真丢失控件存量
total_templates, parsed
"""
import psycopg2 # 仅在 --db 模式按需导入
by_uitype = defaultdict(lambda: {"occ": 0, "tpls": 0})
by_root = defaultdict(int)
by_uitype_root = defaultdict(int)
advanced = defaultdict(int)
total = 0
parsed = 0
conn = psycopg2.connect(dsn)
try:
conn.set_session(readonly=True, autocommit=True)
cur = conn.cursor()
cur.execute(
"SELECT view_template_id, xml::text FROM eip3.view_template "
"WHERE xml IS NOT NULL AND xml::text <> ''"
)
for _tid, xml in cur.fetchall():
total += 1
if not xml:
continue
parsed += 1
root = _root_tag(xml)
by_root[root] += 1
uitypes_in_tpl = set(_UITYPE_RE.findall(xml))
for ut in _UITYPE_RE.findall(xml):
by_uitype[ut]["occ"] += 1
for ut in uitypes_in_tpl:
by_uitype[ut]["tpls"] += 1
by_uitype_root[(ut, root)] += 1
for marker, pats in ADVANCED_TEMPLATE_MARKERS.items():
if any(re.search(p, xml, re.IGNORECASE | re.MULTILINE) for p in pats):
advanced[marker] += 1
finally:
conn.close()
lostlike = {ut: by_uitype[ut] for ut in LOSTLIKE_UITYPES if ut in by_uitype}
return {
"by_uitype": dict(by_uitype),
"by_root": dict(by_root),
"by_uitype_root": dict(by_uitype_root),
"advanced": dict(advanced),
"lostlike": lostlike,
"total_templates": total,
"parsed": parsed,
}
def format_frequency_section(freq: dict) -> str:
"""把 uitype/模板频次统计渲染为 Markdown(接在主报告之后)。"""
by_uitype = freq["by_uitype"]
lines = [
"",
"## 十、uitype × 模板类型 × 使用频次(DB 存量真实用量,§7-2)",
"",
f"> 源:`view_template.xml` 全量扫描({freq['parsed']}/{freq['total_templates']} 有 xml)。",
f"> 用途:让 §4.2 的「高级模板类型」「真丢失控件」覆盖缺口按**存量真实频次**排序,而非按清单平铺。",
"",
"### 10.1 uitype 频次 Top(出现次数 / 涉及模板数)",
"",
"| uitype | 出现次数 | 模板数 |",
"|--------|---------|--------|",
]
ranked = sorted(by_uitype.items(), key=lambda kv: kv[1]["occ"], reverse=True)
for ut, st in ranked[:25]:
lines.append(f"| {ut} | {st['occ']} | {st['tpls']} |")
lines += [
"",
"### 10.2 模板根类型分布",
"",
"| 根元素 | 模板数 |",
"|--------|--------|",
]
for root, n in sorted(freq["by_root"].items(), key=lambda kv: kv[1], reverse=True):
lines.append(f"| `{root}` | {n} |")
lines += [
"",
"### 10.3 高级模板结构存量(§4.2「高级模板类型缺失」按频次排序)",
"",
"| 结构标记 | 涉及模板数 |",
"|---------|-----------|",
]
for marker, n in sorted(freq["advanced"].items(), key=lambda kv: kv[1], reverse=True):
lines.append(f"| {marker} | {n} |")
lines += [
"",
"### 10.4 §4.2「真丢失控件」存量(决定 §7-4 是否值得做)",
"",
"| 控件 uitype | 出现次数 | 模板数 | 结论 |",
"|------------|---------|--------|------|",
]
for ut in LOSTLIKE_UITYPES:
st = freq["lostlike"].get(ut)
if st:
verdict = "有存量,需评估优先级" if st["tpls"] > 0 else "零存量,可不做"
lines.append(f"| {ut} | {st['occ']} | {st['tpls']} | {verdict} |")
else:
lines.append(f"| {ut} | 0 | 0 | 零存量,可不做 |")
lines.append("")
return "\n".join(lines)
# ─────────────────────────────────────────────────────────────────────────────
# 主扫描逻辑
# ─────────────────────────────────────────────────────────────────────────────
def get_page_type(schema) -> str:
body = schema.get("body", {})
if isinstance(body, dict):
return body.get("type", "unknown")
if isinstance(body, list):
types = [b.get("type", "") for b in body if isinstance(b, dict)]
if "crud" in types:
return "crud"
if "form" in types:
return "form"
return "mixed"
return "unknown"
def scan(schema_dir: str) -> dict:
schema_dir = Path(schema_dir)
files = sorted(schema_dir.glob("*.json"))
total = len(files)
results = {
"total": total,
"crud_pages": 0,
"form_pages": 0,
"parse_errors": [],
# 每类:{template_id: [hits...]}
"d1": defaultdict(list),
"d2": defaultdict(list),
"d3": defaultdict(list),
"d4": defaultdict(list),
"d5": defaultdict(list),
"d6": [], # list of template_ids
"all_dialog_ids": set(),
}
for fp in files:
tid = fp.stem
try:
with open(fp, "r", encoding="utf-8") as f:
schema = json.load(f)
except Exception as e:
results["parse_errors"].append({"file": fp.name, "error": str(e)})
continue
ptype = get_page_type(schema)
if ptype == "crud":
results["crud_pages"] += 1
elif ptype == "form":
results["form_pages"] += 1
h1 = detect_defect1_person_column(schema)
if h1:
results["d1"][tid].extend(h1)
h2 = detect_defect2_dead_button(schema)
if h2:
results["d2"][tid].extend(h2)
h3 = detect_defect3_longtext_mismatch(schema)
if h3:
results["d3"][tid].extend(h3)
h4 = detect_defect4_dialog_embed_page(schema)
if h4:
results["d4"][tid].extend(h4)
h5 = detect_defect5_picker(schema)
if h5:
results["d5"][tid].extend(h5)
for item in h5:
if item["dialogId"] != "unknown":
results["all_dialog_ids"].add(item["dialogId"])
if detect_defect6_crud_no_pagination(schema):
results["d6"].append(tid)
return results
# ─────────────────────────────────────────────────────────────────────────────
# 报告生成
# ─────────────────────────────────────────────────────────────────────────────
def format_report(results: dict, schema_dir: str) -> str:
now = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
d1, d2, d3, d4, d5, d6 = (
results["d1"], results["d2"], results["d3"],
results["d4"], results["d5"], results["d6"],
)
all_dids = sorted(results["all_dialog_ids"])
lines = [
f"# F2-17 EIP4 前端组件真实性审计基线",
f"",
f"> 扫描时间: {now} ",
f"> 扫描目录: `{schema_dir}` ",
f"> 本报告由 `schema_reality_audit.py` 自动生成,只读扫描,可重复运行。",
f"",
f"## 一、汇总表",
f"",
f"| 指标 | 数值 |",
f"|------|------|",
f"| 总页数 | {results['total']} |",
f"| CRUD 列表页 | {results['crud_pages']} |",
f"| Form 表单页 | {results['form_pages']} |",
f"| JSON 解析失败 | {len(results['parse_errors'])} |",
f"",
f"| 缺陷编号 | 缺陷描述 | 静态命中页数 | 类别 |",
f"|---------|---------|---------|------|",
f"| D1 | 人员/引用列裸显示 id | {len(d1)} | 🟢 运行时已修(审计盲区) |",
f"| D2 | 按钮死桩(未配编辑表单/该单据未配) | {len(d2)} | 🔴 静态真缺口 |",
f"| D3 | 长文本/富文本错配单行 input-text | {len(d3)} | 🟡 大面已修,残余 e2e 复核 |",
f"| D4 | 弹窗嵌整页(双确认取消风险) | {len(d4)} | 🟢 运行时已修(审计盲区) |",
f"| D5 | 依赖 picker 后端 | {len(d5)} | 🟢 已弹窗化(picker,审计盲区) |",
f"| D6 | CRUD 页受后端不分页缺陷影响 | {len(d6)} | 🟢 运行时已修(审计盲区) |",
f"",
f"> **审计盲区说明(§4.3/§7-2)**:🟢 标记项的静态计数是产物快照,运行时已解决,**不计真实缺口**,改用 e2e 验。",
]
for code, desc in RUNTIME_RESOLVED.items():
lines.append(f"> - {code}:{desc}")
lines.append(f"")
def section(title, data_dict, key_fn, top=10):
lines.append(f"## {title}")
lines.append(f"")
lines.append(f"命中页数: **{len(data_dict)}**,以下展示前 {top} 个 templateId 及首条命中:")
lines.append(f"")
lines.append(f"| templateId | 命中详情 |")
lines.append(f"|-----------|---------|")
for tid, hits in list(data_dict.items())[:top]:
detail = key_fn(hits[0]) if hits else ""
# 截断避免表格过宽
detail = detail[:120].replace("|", "|").replace("\n", " ")
lines.append(f"| `{tid}` | {detail} |")
lines.append(f"")
section(
"二、D1 人员列裸显示 id",
d1,
lambda h: f"name={h.get('name','')} label={h.get('label','')} | {h.get('reason','')}",
)
section(
"三、D2 按钮死桩",
d2,
lambda h: h.get("text", ""),
)
section(
"四、D3 长文本字段错配",
d3,
lambda h: f"name={h.get('name','')} label={h.get('label','')}",
)
section(
"五、D4 弹窗嵌整页",
d4,
lambda h: f"schemaApi={h.get('schemaApi','')}",
)
section(
"六、D5 依赖 picker 后端",
d5,
lambda h: f"name={h.get('name','')} dialogId={h.get('dialogId','')}",
)
lines.append(f"## 七、D6 受后端不分页缺陷影响的 CRUD 页(前 10)")
lines.append(f"")
lines.append(f"命中页数: **{len(d6)}**")
lines.append(f"")
lines.append(f"| templateId |")
lines.append(f"|-----------|")
for tid in d6[:10]:
lines.append(f"| `{tid}` |")
lines.append(f"")
lines.append(f"## 八、Picker dialogId 去重清单")
lines.append(f"")
lines.append(f"共 **{len(all_dids)}** 个唯一 dialogId(后端修 picker 时需逐一验证),以下展示前 20 个:")
lines.append(f"")
lines.append(f"| # | dialogId |")
lines.append(f"|---|---------|")
for i, did in enumerate(all_dids[:20], 1):
lines.append(f"| {i} | `{did}` |")
lines.append(f"")
if results["parse_errors"]:
lines.append(f"## 九、JSON 解析失败文件")
lines.append(f"")
lines.append(f"| 文件名 | 错误 |")
lines.append(f"|--------|------|")
for e in results["parse_errors"]:
lines.append(f"| {e['file']} | {str(e['error'])[:100]} |")
lines.append(f"")
return "\n".join(lines)
def print_summary(results: dict):
d1, d2, d3, d4, d5, d6 = (
results["d1"], results["d2"], results["d3"],
results["d4"], results["d5"], results["d6"],
)
print("=" * 56)
print("EIP4 Schema 组件真实性审计 — 汇总")
print("=" * 56)
print(f" 总页数 : {results['total']}")
print(f" CRUD 列表页 : {results['crud_pages']}")
print(f" Form 表单页 : {results['form_pages']}")
print(f" JSON 解析失败 : {len(results['parse_errors'])}")
print("-" * 56)
print(f" D1 人员列裸显示 id : {len(d1)} 页")
print(f" D2 按钮死桩 : {len(d2)} 页")
print(f" D3 长文本字段错配 : {len(d3)} 页")
print(f" D4 弹窗嵌整页 : {len(d4)} 页")
print(f" D5 依赖 picker : {len(d5)} 页")
print(f" D6 CRUD 不分页受影响: {len(d6)} 页")
print(f" Picker dialogId 数 : {len(results['all_dialog_ids'])}")
print("=" * 56)
# ─────────────────────────────────────────────────────────────────────────────
# 入口
# ─────────────────────────────────────────────────────────────────────────────
def main():
parser = argparse.ArgumentParser(
description="EIP4 Schema 组件真实性审计基线扫描器(只读)"
)
parser.add_argument(
"--schema-dir",
default=DEFAULT_SCHEMA_DIR,
help="amis schema JSON 目录(默认: 转换产物/schema)",
)
parser.add_argument(
"--report-out",
default=DEFAULT_REPORT_OUT,
help="报告输出路径(Markdown,默认: docs/F2-17-...)",
)
parser.add_argument(
"--db",
default=None,
help="(§7-2)DB DSN,提供则附加 uitype×模板类型×频次统计(存量真实用量)。"
"例: 'host=localhost port=5433 dbname=eip4dev user=zhengwr'",
)
args = parser.parse_args()
print(f"[扫描目录] {args.schema_dir}")
results = scan(args.schema_dir)
print_summary(results)
report_text = format_report(results, args.schema_dir)
if args.db:
print(f"[频次统计] 连接 DB 扫 view_template.xml …")
try:
freq = uitype_template_frequency(args.db)
report_text += "\n" + format_frequency_section(freq)
print(f" uitype 种类: {len(freq['by_uitype'])} | 模板: {freq['parsed']}/{freq['total_templates']} "
f"| 高级结构标记: {dict(sorted(freq['advanced'].items(), key=lambda kv: -kv[1]))}")
print(f" 丢失控件存量: " + ", ".join(
f"{ut}={freq['lostlike'].get(ut, {}).get('tpls', 0)}页" for ut in LOSTLIKE_UITYPES))
except Exception as e:
print(f" [警告] 频次统计失败(跳过): {e}")
report_path = Path(args.report_out)
report_path.parent.mkdir(parents=True, exist_ok=True)
with open(report_path, "w", encoding="utf-8") as f:
f.write(report_text)
print(f"\n[报告] {report_path}")
if results["parse_errors"]:
print(f"\n[警告] {len(results['parse_errors'])} 个文件解析失败:")
for e in results["parse_errors"][:10]:
print(f" {e['file']}: {e['error']}")
if __name__ == "__main__":
main()