#!/usr/bin/env python3 # -*- coding: utf-8 -*- """ 02_build_evidence.py — 153 定制视图「带证据的终判草案」批量生成器(可复跑) 输入: - 153视图全量清单.txt(id|名称|engineclass,权威全量来源 = 盘点报告) - framework 源码树(只读 grep 复杂度) - DB 侧证据 CSV(由 01_db_evidence.sql / 本脚本内联导出生成) 输出: - 终判建议表.csv(153 行) - stats.json(四分类/工时/MISSING/Top10 汇总,供评审材料引用) 判定逻辑(基于证据,非类名):见 decide() 注释,落地 v1 文档第一节决策树。 """ import csv, json, os, re, subprocess, sys from collections import defaultdict # ---- 路径 ---- ROOT = "/Users/zhengwr/workspace/eip3" TXT = f"{ROOT}/docs/M0交付/153视图全量清单.txt" # 【2026-06-11 全见源码升级】源码基准从 eip3/framework(仅核心平台)扩展到苏州五建完整 OSGi 源码全集 EIP-3.0/。 # EIP-3.0 = 完整源码(带 git,最近提交「苏州五建包更新版本号」):hr/pm/oa/dm 业务 addon + framework 核心。 # 原 106 个 MISSING 的真相 = 业务 addon「在另一仓库」(EIP-3.0),并非废弃。现 152/153 全见源码,1 个疑废弃。 # 可用环境变量 EIP3_SRC 覆盖源码根(默认 /Users/zhengwr/workspace/EIP-3.0),脚本对路径参数化。 EIP3_SRC = os.environ.get("EIP3_SRC", "/Users/zhengwr/workspace/EIP-3.0") SRC_ROOTS = [f"{EIP3_SRC}/hr", f"{EIP3_SRC}/pm", f"{EIP3_SRC}/oa", f"{EIP3_SRC}/dm", f"{EIP3_SRC}/framework"] SRC_ROOTS = [p for p in SRC_ROOTS if os.path.isdir(p)] # 兜底:若 EIP-3.0 不可用,退回旧的 framework-only 口径(会复现 106 MISSING,仅作降级保护) if not SRC_ROOTS: SRC_ROOTS = [f"{ROOT}/framework"] OUTDIR = f"{ROOT}/docs/M0交付/153终判证据包" DB_AGG = "/tmp/db_ec_agg.csv" # engineclass,template_cnt,xml_bytes_max,xml_bytes_sum,any_active DB_MENU = "/tmp/db_ec_menu.csv" # engineclass,menu_ref_cnt # ⚠️ 已知坑:上方两个 CSV 须由 00_run_all.sh(或手工 \copy)从 dbname=postgres 导出。 # 5435 上存在两库:postgres(153 条全量,正确口径)/ ceshi(103 条旧副本,数据不全)。 # 若 CSV 来自 ceshi 库,将出现大量"菜单引用=无数据/XML=无数据"假象。曾犯此错,已订正。 # 疑废弃类:客户库 view_template 有此 engineclass,但 EIP-3.0 无对应 .java;同模块有近似类(疑改名/废弃)。 SUSPECT_DEPRECATED = { "com.dsasp.pm.materialsuppliermanagement.client.view.SupplierResearchCaseView": "同模块 com.dsasp.pm.materialsuppliermanagement 存在近似类 SupplierResearchView/SupplierResearchWindowView/SupplierResearchCaseWindow,疑改名或废弃", } # ---- 组件谱:Vaadin / dsasp 自定义 / 外部库 关键词(按 import 与 new 检测)---- # 重组件(推高复杂度,倾向重写) HEAVY_COMPONENTS = { "Tree": [r"\bTree\b", r"\bTreeTable\b", r"ExTree", r"TreeGrid"], "Upload": [r"\bUpload\b", r"MultiFileUpload", r"easyuploads", r"swfupload"], "Chart": [r"\bChart\b", r"invient", r"highcharts", r"DCharts"], "Gantt": [r"Gantt", r"SchedulerGantt"], "Canvas": [r"dscanvas", r"\bCanvas\b"], "BrowserFrame/iframe": [r"BrowserFrame", r"\bEmbedded\b", r"ExBrowserFrame"], "Window/Dialog": [r"\bWindow\b", r"\bDialog\b"], "Wizard/MultiStep": [r"Wizard", r"Stepper"], # TabSheet 过于常见,不计入重组件 "TabSheet": [r"TabSheet"], "Calendar": [r"\bCalendar\b", r"DateField"], } # CRUD 包装信号(只用这些 => 低复杂、可降级) CRUD_COMPONENTS = [r"PagedTable", r"\bTable\b", r"BeanFieldGroup", r"FieldGroup", r"FormLayout", r"\bForm\b", r"SearchComponent"] # 外部依赖(打印/设备/第三方) EXTERNAL_DEPS = { "PDF(itext/lowagie)": [r"import com\.itextpdf", r"import com\.lowagie"], "Excel(poi/jxl)": [r"import org\.apache\.poi", r"import jxl"], "Report(jasper/jor)": [r"jasperreports", r"\bJasper", r"\.jrxml"], "Device(实名制/考勤机/串口)": [r"\bSerialPort\b", r"javax\.comm", r"gnu\.io", r"\bRXTX", r"考勤机", r"读卡器", r"刷卡"], "Print(打印)": [r"java\.awt\.print", r"PrinterJob", r"javax\.print", r"打印"], } def sh(args): try: return subprocess.run(args, capture_output=True, text=True, timeout=60).stdout except Exception: return "" # ---- 全见源码定位:预建「简单类名 -> [路径]」索引(一次 find 扫全树,比逐类 find 快得多)---- _SRC_INDEX = None def _build_src_index(): global _SRC_INDEX if _SRC_INDEX is not None: return _SRC_INDEX idx = defaultdict(list) for root in SRC_ROOTS: out = sh(["find", root, "-name", "*.java", "-type", "f"]) for p in out.splitlines(): if p.strip(): idx[os.path.basename(p)[:-5]].append(p) # 去 .java _SRC_INDEX = idx return idx def find_class_file(engineclass): """class fqn -> source file path or None。 全见源码版:在 EIP-3.0 hr/pm/oa/dm/framework 全树按简单类名定位; 同名多文件时用 engineclass 完整包路径做「最长后缀」消歧(已验证 17 个同名类 0 存疑)。 注意:OSGi addon 目录名可能与包名不同(如目录 com.dsasp.pm.projectprocess 内含包 com.dsasp.pm.process), 故以包路径片段匹配而非目录名匹配。""" idx = _build_src_index() simple = engineclass.split(".")[-1] paths = idx.get(simple, []) if not paths: return None if len(paths) == 1: return paths[0] segs = engineclass.split(".") best, best_len = None, 0 for path in paths: norm = path.replace("\\", "/") for k in range(len(segs), 1, -1): tail = "/".join(segs[len(segs) - k:]) # 末 k 段(含类名) if tail in norm: if k > best_len: best, best_len = path, k break return best or paths[0] def analyze_source(path): """返回 (loc, comp_spectrum_list, external_deps_list, is_crud_wrapper)""" try: with open(path, encoding="utf-8", errors="ignore") as f: text = f.read() except Exception: return (0, [], [], False) loc = text.count("\n") + 1 comps = [] for label, pats in HEAVY_COMPONENTS.items(): if any(re.search(p, text) for p in pats): comps.append(label) deps = [] for label, pats in EXTERNAL_DEPS.items(): if any(re.search(p, text) for p in pats): deps.append(label) # CRUD 包装判定:命中 CRUD 组件,且未命中任何 “真正的重组件”(Tree/Upload/Chart/Gantt/Canvas/BrowserFrame/Wizard) real_heavy = {"Tree", "Upload", "Chart", "Gantt", "Canvas", "BrowserFrame/iframe", "Wizard/MultiStep"} has_crud = any(re.search(p, text) for p in CRUD_COMPONENTS) has_real_heavy = bool(set(comps) & real_heavy) is_crud_wrapper = has_crud and not has_real_heavy and loc < 600 return (loc, comps, deps, is_crud_wrapper) def load_db(): agg, menu = {}, {} if os.path.exists(DB_AGG): for r in csv.DictReader(open(DB_AGG, encoding="utf-8")): agg[r["engineclass"]] = r if os.path.exists(DB_MENU): for r in csv.DictReader(open(DB_MENU, encoding="utf-8")): menu[r["engineclass"]] = int(r["menu_ref_cnt"]) return agg, menu NAME_DISCARD = ["未用", "暂时不用", "测试", "(测试)", "(测试)", "废"] def v1_bucket(ec): """v1 文档 §6.1 名称启发式初判归桶(A/B/C/D/E/F),用于回填「初判桶」列做初判→终判对照。""" low = ec.lower() if any(k in low for k in ["report", "pdf", "print", "account", "summary", "journal", "chart"]): return "A.报表/PDF/台账(留旧)" if any(k in low for k in ["research", "select", "window", "picker", "tree", "dialog"]): return "B.选择窗/树/查询窗(降级)" if any(k in low for k in ["manage", "config", "main", "info", "list"]): return "C.管理CRUD(降级)" if any(k in low for k in ["task", "process", ".wf.", "monitor", "workflow"]): return "D.流程交互(重写)" if any(k in low for k in ["import", "export", "send", "message", "input"]): return "E.导入导出/消息(逐个)" return "F.其他(逐个)" def decide(row, src, db): """ 基于证据的终判建议。决策树落地 v1 文档第一节。 优先级:废弃标记 > 报表/PDF/台账(留旧) > 设备依赖(重写) > 流程交互(重写) > 富交互组件(重写) > 选择/查询窗(降级) > CRUD/管理(降级) > 导入导出(逐个) > 兜底 源码缺失(MISSING)时:仍按 名称+engineclass+DB 信号 走同一决策树,但工时记为「名义估算」, 并在证据摘要标注源码缺失(重大信号:留旧都需补丁包/可能已废弃)。 row: dict(id,name,engineclass);src: (loc,comps,deps,is_crud) 或 None;db: dict(menu_ref,xml_bytes,in_db) 返回 (bucket, mandays, reason);bucket ∈ {重写,降级,留旧,废弃} """ name = row["name"] ec = row["engineclass"] low = ec.lower() # 疑废弃:客户库有模板但 EIP-3.0 全树无 .java,且同模块有近似类 → 优先判废弃候选 if ec in SUSPECT_DEPRECATED: return ("废弃", 0.0, f"源码缺失-疑废弃:EIP-3.0 全树无对应 .java;{SUSPECT_DEPRECATED[ec]};建议客户确认是否已删/改名后剔除") missing = src is None if missing: # 全见源码升级后,此分支仅剩「真正定位不到源码」的兜底(正常情况不应触发,仅作保护) loc, comps, deps, is_crud = "MISSING", [], [], False src_tag = "源码MISSING(EIP-3.0全树未定位,需核查)" md_tag = "名义估算(源码缺失需核查)" else: loc, comps, deps, is_crud = src src_tag = f"源码{loc}行" md_tag = "" # 真·amis 无对应物的重组件(触发重写):甘特/画布/图表/内嵌iframe/向导。 # Tree/Upload 由 amis tree-select/input-file 覆盖,不单独触发重写(只作为复杂度旁证)。 real_heavy = {"Chart", "Gantt", "Canvas", "BrowserFrame/iframe", "Wizard/MultiStep"} heavy_hit = sorted(set(comps) & real_heavy) has_pdf_dep = any(("PDF" in d) or ("Report" in d) for d in deps) has_device = any("Device" in d for d in deps) def loc_lt(n): # 行数阈值:MISSING 时取保守上档 return (not missing) and loc < n # F0. 废弃候选:名称含废弃标记 if any(k in name for k in NAME_DISCARD): return ("废弃", 0.0, "名称含废弃标记,建议客户确认后剔除(连留旧成本都省)") # 工作流(审批) 与 项目进度计划(Gantt) 的区分: # - workflow 审批流 = .client.wf. 包 或 类名含 workflow/任务(task)审批语义 # - .pm.process.ProjectProcess* = 项目「进度计划」(甘特/计划表),非审批流,不强制重写 is_wf = (".client.wf." in low) or any(k in low for k in ["workflow", "flowservice", "processmonitor", "processdefinition", "processselect", "processformremove", "sendtaskcomponent", ".task.component"]) is_schedule = ".pm.process." in low # 进度计划族(ProjectProcess*) # 台账/日记账/journal/account 归报表族 is_journal = any(k in low for k in ["journal", "account", "buyaccount", "台账"]) or "台账" in name or "日记账" in name or "日志统计" in name # A. 报表/PDF/打印/台账/图表/日记账 → 留旧(待 P4 报表方案 C 类),审批流例外 is_report = (any(k in low for k in ["pdf", "report", "chart", "print", "summary"]) or has_pdf_dep or is_journal) if is_report and not is_wf: return ("留旧", 0.25, f"报表/PDF/台账/图表/日记账类({src_tag},依赖={deps or '无'})→ 留旧,归 P4 报表方案 C 类") # B. 设备/终端依赖(实名制考勤机/显示屏/串口)→ 重写 if has_device or any(k in low for k in ["shiming.view", "screen", "prjdevice"]): md = 3.0 if loc_lt(400) else 4.0 return ("重写", md, f"设备/终端依赖({src_tag},组件={heavy_hit or comps or '名义'})→ 重写,需确认 PC/移动终端适配{(';'+md_tag) if md_tag else ''}") # C. 工作流审批交互类(待办/已办/审批/监控/流程定义)→ 重写 if is_wf: md = 3.0 if loc_lt(300) else 4.0 return ("重写", md, f"工作流审批交互类({src_tag},组件={heavy_hit or comps or '名义'})→ 重写,amis 无现成对应,待 P3 Flowable REST 稳定{(';'+md_tag) if md_tag else ''}") # C2. 项目进度计划族(甘特/计划表)→ 重写(有甘特组件) 或 降级(纯表格计划) if is_schedule: if (not missing and heavy_hit): md = 3.0 if loc_lt(500) else 4.0 return ("重写", md, f"项目进度计划+富组件 {heavy_hit}({src_tag})→ 重写(甘特/计划交互)") # MISSING 或无甘特证据:保守降级(计划表本质为结构化表格),止损预案优先压低成本 return ("降级", 1.0, f"项目进度计划族({src_tag})→ 降级为 amis 表格/计划视图;若实为甘特需升级重写(待过堂){(';'+md_tag) if md_tag else ''}") # D. 真·重组件命中(甘特/画布/图表/iframe/向导,amis 无对应物)→ 重写 # 注意:放在选择窗/CRUD 之前,因这类组件即便出现在管理视图里也无法降级。 if heavy_hit: md = 2.0 if loc_lt(250) else (3.0 if loc_lt(500) else 4.0) return ("重写", md, f"amis 无对应物的重组件命中 {heavy_hit}({src_tag})→ 重写为 React/amis 自定义渲染器") # E. 选择窗/树/查询窗 → 降级(amis tree-select/picker 标准对应物,Tree 也由其覆盖) if any(k in low for k in ["select", "window", "picker", "dialog", "research"]) or "选择" in name: tree_note = "(含组织树,amis tree-select 覆盖)" if (comps and "Tree" in comps) else "" return ("降级", 0.75, f"选择/查询窗类{tree_note}({src_tag})→ 降级为 amis tree-select/picker,需确认多级过滤{(';'+md_tag) if md_tag else ''}") # F. CRUD 包装 / 标准管理列表(含树形 CRUD)→ 降级 if is_crud or any(k in low for k in ["manage", "config", "main", "info", "list", "edit", "setting"]): tree_note = "(树形列表,amis tree/crud 覆盖)" if (comps and "Tree" in comps) else "" return ("降级", 0.75, f"标准 CRUD/管理列表{tree_note}({src_tag},CRUD包装={'是' if is_crud else '否/名义'})→ 降级走批量转换器+手工配置{(';'+md_tag) if md_tag else ''}") # G. 导入导出/消息 → 逐个:仅当行数很高(复杂字段映射UI)才重写,否则降级 # Upload/Excel 由 amis input-file/input-excel 覆盖,不单独触发重写。 if any(k in low for k in ["import", "export", "upload", "input", "send", "message"]): if (not missing and loc > 500): return ("重写", 2.5, f"导入导出类但交互重({src_tag},组件={comps},疑含复杂字段映射UI)→ 重写") return ("降级", 1.0, f"导入导出/消息类({src_tag})→ 降级用 amis input-excel/input-file{(';'+md_tag) if md_tag else ''}") # H. 兜底:低复杂降级、否则留旧 if not missing and loc < 200: return ("降级", 0.75, f"低复杂未命中启发式({src_tag})→ 降级候选,终判过堂确认") return ("留旧", 0.25, f"未命中启发式({src_tag})→ 留旧候选,终判过堂确认{(';'+md_tag) if md_tag else ''}") def main(): rows = [] with open(TXT, encoding="utf-8") as f: for line in f: line = line.strip() if "|" not in line: continue # 跳过 SET 头 parts = line.split("|") if len(parts) < 3: continue rows.append({"id": parts[0].strip(), "name": parts[1].strip(), "engineclass": parts[2].strip()}) db_agg, db_menu = load_db() out_rows, stats = [], { "total": 0, "buckets": defaultdict(int), "mandays": defaultdict(float), "missing": [], "deprecated": [], "rewrite_detail": [], "in_db": 0, "not_in_db": 0, "src_located": 0, "src_missing": 0, } detail = [] # 供 S1 止损场景与去重分析 for r in rows: ec = r["engineclass"] is_deprecated = ec in SUSPECT_DEPRECATED path = None if is_deprecated else find_class_file(ec) src = analyze_source(path) if path else None in_db = ec in db_agg menu_ref = db_menu.get(ec, "") # xml_bytes_max 为空字符串表示模板存在但 xml 字段为 NULL(23 条已知情况) # 用 "null(xml为空)" 区分"模板不存在"的情况,避免与旧版 ceshi 库 "无数据" 混淆 if in_db: raw_xml = db_agg[ec]["xml_bytes_max"] xml_bytes = raw_xml if raw_xml != "" else "null(模板存在,xml为空)" else: xml_bytes = "无数据(模板不存在)" bucket, md, reason = decide(r, src, {"menu_ref": menu_ref, "xml_bytes": xml_bytes, "in_db": in_db}) # 源码绝对路径(全见源码新列);疑废弃标注,真缺失标注 if is_deprecated: src_path = "源码缺失-疑废弃" elif path: src_path = path else: src_path = "未定位" loc = src[0] if src else ("废弃" if is_deprecated else "MISSING") comps = "/".join(src[1]) if src else "" deps = "/".join(src[2]) if src else "" crud = ("是" if src and src[3] else "否") if src else "" # 待人工确认点 confirm = [] if is_deprecated: confirm.append("疑废弃:确认是否已删/改名(同模块有近似类)后剔除") elif src is None: confirm.append("源码未定位,需核查 engineclass 拼写/包路径") if bucket == "重写": confirm.append("交互细节/终端适配/REST 依赖") if bucket == "降级": confirm.append("是否有特殊批量/权限/多级过滤") if menu_ref == 0: confirm.append("菜单引用=0,是否仍在用") if menu_ref == "": confirm.append("本快照无该模板,使用频率无数据") out_rows.append({ "id": r["id"], "name": r["name"], "engineclass": ec, "源码路径": src_path, # 全见源码新列:.java 绝对路径 / 源码缺失-疑废弃 / 未定位 "源码行数": loc, "组件谱": comps, "外部依赖": deps, "CRUD包装": crud, "菜单引用数": menu_ref if menu_ref != "" else "无数据", "XML大小(bytes)": xml_bytes, # 已在上方赋值时区分:数字/null(模板存在,xml为空)/无数据(模板不存在) "初判桶": v1_bucket(ec), # v1 §6.1 名称启发式归桶,供初判→终判对照 "终判建议": bucket, "工时(人日)": md, "证据摘要": reason, "待人工确认点": ";".join(confirm) or "无", }) stats["total"] += 1 stats["buckets"][bucket] += 1 stats["mandays"][bucket] += md if in_db: stats["in_db"] += 1 else: stats["not_in_db"] += 1 if is_deprecated: stats["deprecated"].append({"id": r["id"], "name": r["name"], "engineclass": ec, "note": SUSPECT_DEPRECATED[ec]}) elif path: stats["src_located"] += 1 else: stats["src_missing"] += 1 stats["missing"].append({"id": r["id"], "name": r["name"], "engineclass": ec}) if bucket == "重写": stats["rewrite_detail"].append({"name": r["name"], "engineclass": ec, "loc": loc, "comps": comps, "md": md}) detail.append({"name": r["name"], "ec": ec, "bucket": bucket, "md": md, "menu_ref": (menu_ref if menu_ref != "" else None), "in_db": in_db, "missing_src": src is None}) # 写 CSV os.makedirs(OUTDIR, exist_ok=True) csv_path = f"{OUTDIR}/终判建议表.csv" cols = ["id","name","engineclass","源码路径","源码行数","组件谱","外部依赖","CRUD包装", "菜单引用数","XML大小(bytes)","初判桶","终判建议","工时(人日)","证据摘要","待人工确认点"] with open(csv_path, "w", encoding="utf-8-sig", newline="") as f: w = csv.DictWriter(f, fieldnames=cols) w.writeheader() w.writerows(out_rows) # 汇总 stats["buckets"] = dict(stats["buckets"]) stats["mandays"] = {k: round(v, 2) for k, v in stats["mandays"].items()} stats["total_mandays"] = round(sum(stats["mandays"].values()), 2) stats["total_manmonths"] = round(stats["total_mandays"] / 21.0, 2) # 21 人日/月 stats["rewrite_detail"].sort(key=lambda x: (-x["md"], str(x["loc"])), reverse=False) # Top10 最贵重写:按工时降序,再按行数降序 def loc_num(v): try: return int(v) except: return -1 stats["top10_rewrite"] = sorted( stats["rewrite_detail"], key=lambda x: (x["md"], loc_num(x["loc"])), reverse=True )[:10] # ---- 去重红利:同一 engineclass 被多个视图复用,重写/降级只做一次,复用项按登记成本 ---- by_ec = defaultdict(list) for d in detail: by_ec[d["ec"]].append(d) dup_classes = {ec: ds for ec, ds in by_ec.items() if len(ds) > 1} dedup_saving = 0.0 dup_report = [] for ec, ds in dup_classes.items(): if ds[0]["bucket"] in ("重写", "降级"): # 首个全价,其余按 0.25 登记成本 extra = len(ds) - 1 saved = sum(d["md"] for d in ds[1:]) - extra * 0.25 dedup_saving += max(saved, 0) dup_report.append({"ec": ec, "count": len(ds), "bucket": ds[0]["bucket"], "saved": round(max(saved,0),2)}) stats["dedup_classes"] = sorted(dup_report, key=lambda x: -x["saved"]) stats["dedup_saving_mandays"] = round(dedup_saving, 2) # ---- S1 止损场景:base 超 3 人月时,按止损预案把「低价值降级」降为留旧 ---- # 低价值判据:降级 且 (菜单引用=0 或 本快照无数据[无法转换]) → 留旧(0.25) base_md = stats["total_mandays"] s1_rows = [] s1_md = base_md s1_demoted = 0 for d in detail: if d["bucket"] == "降级": # 低价值判据(postgres 库全量 153 条都在库,所以 not in_db 不再触发): # 仅剩 menu_ref=0(菜单无引用,疑似未用或纯内嵌组件) low_value = (d["menu_ref"] == 0) if low_value: s1_md += (0.25 - d["md"]) # 降级md -> 留旧0.25 s1_demoted += 1 s1_rows.append({"name": d["name"], "ec": d["ec"], "why": "菜单引用=0,疑似未用或内嵌选择框/弹窗(需老员工区分)"}) s1_md = round(s1_md, 2) stats["s1_scenario"] = { "trigger": base_md / 21.0 > 3.0, "demoted_count": s1_demoted, "demoted_to_legacy": s1_rows, "mandays_after": s1_md, "manmonths_after": round(s1_md / 21.0, 2), "mandays_after_dedup": round(s1_md - dedup_saving, 2), "manmonths_after_dedup": round((s1_md - dedup_saving) / 21.0, 2), } stats["stop_loss_line_manmonths"] = 3.0 with open(f"{OUTDIR}/stats.json", "w", encoding="utf-8") as f: json.dump(stats, f, ensure_ascii=False, indent=2) # 控制台摘要 print("=== 四分类 ===") for k in ["重写","降级","留旧","废弃"]: print(f" {k}: {stats['buckets'].get(k,0)} 个 / {stats['mandays'].get(k,0)} 人日") print(f"基线总工时: {stats['total_mandays']} 人日 = {stats['total_manmonths']} 人月 (止损线 3 人月)") s1 = stats["s1_scenario"] print(f"S1止损场景: 降级->留旧 {s1['demoted_count']} 个 => {s1['manmonths_after']} 人月; 叠加去重红利 {stats['dedup_saving_mandays']} 人日 => {s1['manmonths_after_dedup']} 人月") print(f"源码定位(全见): {stats['src_located']} 个 / 疑废弃: {len(stats['deprecated'])} 个 / 真缺失: {stats['src_missing']} 个") print(f"本快照有DB数据: {stats['in_db']} / 无: {stats['not_in_db']}") print(f"重复 engineclass(可批量): {len(stats['dedup_classes'])} 组") print(f"CSV: {csv_path}") if __name__ == "__main__": main()