# -*- coding: utf-8 -*- """ 从 ent schema 源文件抽取 表注释(类型 doc) + 列注释(.Comment 或字段上方 inline //), 与真实数据库目录(information_schema 导出 TSV) join,生成可重复执行的 COMMENT ON SQL。 仅对库中真实存在的 表/列 生成语句;不猜测业务含义。 """ import os, re, sys ROOT = "D:/hardman/bj_power" SCHEMA_DIRS = { "mes": os.path.join(ROOT, "bj_power_mes", "schema"), "wms": os.path.join(ROOT, "bj_power_wms", "schema"), } CATALOG = { "mes": os.path.join(ROOT, "sql", "_mes_catalog.tsv"), "wms": os.path.join(ROOT, "sql", "_wms_catalog.tsv"), } OUT = { "mes": os.path.join(ROOT, "sql", "db_comments_mes.sql"), "wms": os.path.join(ROOT, "sql", "db_comments_wms.sql"), } # ---------- helpers ---------- def snake(name): out = [] for i, ch in enumerate(name): if ch.isupper() and i > 0: out.append('_') out.append(ch.lower()) return "".join(out) def pluralize(s): if s.endswith('y') and s[-2:-1] not in 'aeiou': return s[:-1] + 'ies' if s.endswith(('s', 'x', 'ch', 'sh')): return s + 'es' return s + 's' SYS_MAP = { 'id': '主键ID', 'created_at': '创建时间', 'updated_at': '更新时间', 'deleted_at': '逻辑删除时间', 'created_by': '创建人', 'updated_by': '更新人', } # 兜底:声明了但既无 .Comment 也无 inline // 的列,按列名给出安全中文(字面翻译,低误判) NAME_MAP = { 'remark': '备注', 'operator': '操作人', 'order_no': '关联工单号', 'material_name': '物料名称', 'spec': '规格', 'unit': '单位', 'icon': '图标', 'sort': '排序', 'target_dock': '目标接驳台', 'process_name': '工序名称', 'step_name': '工序名称', 'criterion_name': '判据名称', 'criterion_logic': '判据逻辑', 'is_ok': '是否合格', 'started_at': '开始时间', 'ended_at': '结束时间', 'resolved_at': '处理时间', 'parent_code': '父级编码', 'target_type': '目标类型', 'target_id': '目标ID', 'material_code': '物料编码', 'zone_code': '区域编码', 'shelf_no': '货架号', 'layer_no': '层号', 'position_no': '位置号', 'book_qty': '账面数量', 'scanned_qty': '扫描数量', 'diff_qty': '差异数量', 'counted': '是否已盘点', 'counted_by': '盘点人', 'counted_at': '盘点时间', 'product_name': '产品名称', 'completed_at': '完成时间', } def find_block(text, brace_idx): depth = 0 for i in range(brace_idx, len(text)): if text[i] == '{': depth += 1 elif text[i] == '}': depth -= 1 if depth == 0: return text[brace_idx+1:i] return "" FIELD_RE = re.compile(r'^\s*field\.\w+\(\s*"(\w+)"') COMMENT_RE = re.compile(r'\.Comment\(\s*"((?:[^"\\]|\\.)*)"') TABLE_RE = re.compile(r'entsql\.Annotation\{\s*Table:\s*"(\w+)"') TYPE_RE = re.compile(r'^type\s+(\w+)\s+struct', re.MULTILINE) def parse_file(path): with open(path, encoding='utf-8') as f: text = f.read() results = [] # list of (type_name, doc, table_name, [(col, comment), ...]) for tm in TYPE_RE.finditer(text): type_name = tm.group(1) type_line = text[:tm.start()].count('\n') # doc comment: walk up from type line over consecutive // lines lines = text.split('\n') doc = [] j = type_line - 1 while j >= 0 and lines[j].strip().startswith('//'): doc.insert(0, lines[j].strip()[2:].strip()) j -= 1 doc_text = doc[0] if doc else type_name # table name ann_m = re.search(r'func\s*\(\w+\)\s*Annotations\(\)\s*\[\]schema\.Annotation\s*\{', text) table = None if ann_m: ab = find_block(text, ann_m.end()-1) tm2 = TABLE_RE.search(ab) if tm2: table = tm2.group(1) if not table: table = pluralize(snake(type_name)) # fields block fm = re.search(r'func\s*\(\w+\)\s*Fields\(\)\s*\[\]ent\.Field\s*\{', text) fields = [] if fm: fb = find_block(text, fm.end()-1) flines = fb.split('\n') starts = [(i, FIELD_RE.match(flines[i]).group(1)) for i in range(len(flines)) if FIELD_RE.match(flines[i])] for k, (i, fname) in enumerate(starts): end = starts[k+1][0] if k+1 < len(starts) else len(flines) seg = '\n'.join(flines[i:end]) cm = COMMENT_RE.search(seg) if cm: comment = cm.group(1) else: # inline // immediately above the field start line (within fb) buf = [] jj = i - 1 while jj >= 0 and flines[jj].strip().startswith('//'): buf.insert(0, flines[jj].strip()[2:].strip()) jj -= 1 comment = " ".join(buf) if buf else "" col = snake(fname) if not comment: comment = SYS_MAP.get(col, "") fields.append((col, comment)) results.append((type_name, doc_text, table, fields)) return results def clean(s): s = s.replace("\\", "\\\\") s = s.replace("'", "''") s = re.sub(r'\s+', ' ', s).strip() return s def main(): for side in ("mes", "wms"): # parse all schema files schema_map = {} # table_name -> (doc, {col: comment}) unparsed_files = [] for fn in sorted(os.listdir(SCHEMA_DIRS[side])): if not fn.endswith('.go'): continue if fn in ('gen.go', 'mixin.go'): continue parsed = parse_file(os.path.join(SCHEMA_DIRS[side], fn)) for type_name, doc, table, fields in parsed: colmap = {} for col, comment in fields: colmap[col] = comment schema_map[table] = (doc, colmap, type_name) # read catalog cat = {} # table -> [(col, dtype, existing_comment)] with open(CATALOG[side], encoding='utf-8') as f: header = f.readline() for line in f: line = line.rstrip('\n') if not line: continue parts = line.split('\t') if len(parts) < 4: continue t, c, dt, ec = parts[0], parts[1], parts[2], parts[3] cat.setdefault(t, []).append((c, dt, ec)) # generate SQL out_lines = [] out_lines.append("-- ============================================================") out_lines.append(f"-- {side.upper()} 数据库 表/列 注释(COMMENT ON),可重复执行") out_lines.append("-- 由 ent schema 源文件抽取生成;表名/列名均以真实库目录为准") out_lines.append("-- ============================================================\n") tables_done = set() table_comments_emitted = 0 col_comments_emitted = 0 cols_without = [] # (table, col) no comment available for t in sorted(cat.keys()): if t in schema_map: doc, colmap, type_name = schema_map[t] # table comment if doc: out_lines.append(f'COMMENT ON TABLE "{t}" IS \'{clean(doc)}\';') table_comments_emitted += 1 # column comments for (c, dt, ec) in cat[t]: comment = colmap.get(c, "") or NAME_MAP.get(c, "") or SYS_MAP.get(c, "") if comment: out_lines.append(f'COMMENT ON COLUMN "{t}"."{c}" IS \'{clean(comment)}\';') col_comments_emitted += 1 else: cols_without.append((t, c)) tables_done.add(t) out_lines.append("") with open(OUT[side], 'w', encoding='utf-8') as f: f.write('\n'.join(out_lines)) # report unmatched = [t for t in cat.keys() if t not in schema_map] print(f"[{side.upper()}] schema tables matched: {len(tables_done)}") print(f"[{side.upper()}] table comments emitted: {table_comments_emitted}") print(f"[{side.upper()}] column comments emitted: {col_comments_emitted}") print(f"[{side.upper()}] columns WITHOUT comment (skipped): {len(cols_without)}") for t, c in cols_without: print(f" NO-COMMENT {t}.{c}") print(f"[{side.upper()}] DB tables WITHOUT schema file: {unmatched}") print(f"[{side.upper()}] -> wrote {OUT[side]}\n") if __name__ == '__main__': main()