Files
AgentLab/backend/tests/fixtures/xlsx/flow.py
T
ThearchyHelios 2d5e50fa88 Excel 导入改造期 2:配方引擎与确认向导——表从哪一行哪一列开始、哪些是合计、日期怎么补年份,都写进一份可校验的配方;首次导入经人确认,之后每期按配方重放并核对
配方语言(agentlab-recipe/2,新 recipe_types、recipe)
- 封闭、不含正则、保存时静态校验:常量只能从表里的候选文字里选,占位符只能映射成空值,单位从固定词表里取
- 交叉表和列表两种版式;分段各自定位、不依赖先后顺序;合计按标签语义区间核对,关系恒等式、口径不同的声明都写在配方里
- 固定的解析器集合(parsers/2):全角、各类横线、千分位、月日写法宽容归一,规范写法存库、原文另记

执行器(新 xlsx_cells、recipe_engine)
- 两遍读取逐行对账,每个非空格都要有去向(格子账);交叉表逆透视成长表,按统计期补年份并断言覆盖整期
- 合计行改作核对并另存一张表,不和明细混在一列;区域外文字、隐藏行列、合并单元格、空行之后的文字都单独报出
- STRICT 加主键写临时库;按锚点段记溯源,供后续证据下钻用;大表流式读取,扫描在 sheetData 处停止

校验与说明(新 recipe_checks、recipe_notes)
- SQL 核对分通过、不一致、无法核对三种结果,失败分结构、数据质量、需确认三类;结构类直接拒收
- 数据质量类可以写理由接受,接受不改库、只改状态和说明
- 表的说明按核对结果生成、不含数字,经过数字检查;没核对过的不写「已核对」

起草(新 recipe_suggest、recipe_ai)
- 先用规则起草:建议卡片加待确认问题(关系要不要登记、占位符表示什么)
- 规则起草不完整时才提示 AI 兜底:界面先展示要发出的全部文字,用户同意才发;只发结构,数字、隐藏行列、遮罩列不发,公式先脱敏;同意绑定模型和服务方;AI 只写配方,有限次修订,记 token 和费用;截断单独识别
- 无论规则还是 AI 起草,都必须人工确认才能启用

导入流程(新 recipe_imports、recipe_confirm、recipe_diff、api/table_imports)
- 暂存 → 起草 → 回答问题或改配方 → 试运行 → 勾确认项、写接受理由 → 确认并启用;接入期 1 的构建、导入记录、快照和运行固定版本
- 上传新一期按已确认的配方重放,不调用模型,出回执和差异卡,手动启用
- 改配方报破坏性变更(breaking:*)、单位变化、工作表改名的兜底,都要逐项勾选确认

前端导入向导(新 pages/import)
- 原始网格(行列号、去向着色)、建议卡片、配方面板(表单,不露正则)、试运行回执、确认清单、差异卡、AI 起草的同意对话框
2026-10-01 22:00:07 +08:00

273 lines
13 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
"""「结构仿照客流表」的合成交叉表(P2-SPEC 9.1)。名字全是假名(分区甲 / 分区乙 / 分区丙),数字随机。
坐标与真实报表一致(B2 统计期、第 4 行日期、第 5–7 行指标、第 8 行空、第 9 / 21 行分段标题、第 28–30 行
公式合计),这样执行器、核对和说明对着它跑出来的格子账、行数、核对结果才有可比的期望(FLOW_EXPECT)。
variant 取 9.2 的用例代号,在同一套生成逻辑上改出各种漂移:行的增删挪动都在「行清单」上做,行号和公式
引用跟着重新算,不在 XML 里硬改坐标。
"""
from __future__ import annotations
import datetime as dt
import random
from dataclasses import dataclass
from typing import Any
from openpyxl import Workbook
from openpyxl.utils import get_column_letter
from . import excel_saved, save
SHEET = "客流汇总"
TITLE = "客流汇总表"
PLACEHOLDER = "·"
TOTAL_LABEL, ZONE_A_LABEL, ZONE_B_LABEL = "全日客流(人次)", "分区甲(人次)", "分区乙(人次)"
ZONE_C_LABEL = "分区丙(人次)"
DAY_TITLE = "日间时段客流(人次)"
NIGHT_TITLE = "夜间时段客流(人次)"
DAY_HOURS = tuple(range(7, 18))
NIGHT_HOURS = tuple(range(18, 24))
#: 表内合计:(标签, 起始小时, 结束小时)
TOTALS = (("18-22 时合计", 18, 22), ("22-24 时合计", 22, 24), ("18-24 时合计", 18, 24))
LABEL_COL, FIRST_COL = 2, 3 # B 列是标签,C 列起是日期
EXTRA_SHEET = "说明"
#: 只换起止日期、版式与 D00 相同的代号
_PLAIN = {"D00", "D01", "D02", "D03", "D03b"}
#: 9.2 的全部代号,外加 noperiod(把 B2 删掉,9.7 第 6 步的人工录入统计期用)
VARIANTS = frozenset(_PLAIN | {
"D04", "D05", "D06", "D07", "D08", "D09", "D10", "D11", "D12", "D13", "D14", "D15", "D16", "D17", "D18",
"D19", "D20", "D21", "D22", "D23", "D24", "D25", "D26", "D27", "D28", "D28b", "noperiod",
})
#: D00(2026-08-01 起 31 天、参考配方)的期望,照 P2-SPEC 9.1 抄录,WP-7 验收时逐项比
FLOW_EXPECT: dict[str, Any] = {
"nonempty": 771,
"bounds": "B2:AG30",
"merged": 5,
"formulas": 93,
#: 格子账去向(契约 Role):数据值、合计格、合计标签、列表头(日期)、行标签、分段标题、统计期、区域外文字
"ledger": {"value": 620, "derived_value": 93, "derived_label": 3, "col_header": 31, "row_label": 20,
"section_title": 2, "context": 1, "outside_text": 1},
"rows": {"日客流": 31, "时段客流": 527, "时段客流_表内合计": 93},
"placeholders": {PLACEHOLDER: 31},
"outside_text": [("客流汇总!B3", TITLE, "text")],
"period_cells": ["客流汇总!B2"],
"checks": {"C1": "passed", "C2": "passed", "K1": "passed", "G1": "passed", "R1": "passed", "R2": "info",
"N1": "passed", "N2": "passed", "N3": "passed", "P1": "passed", "P2": "passed", "P3": "passed"},
#: R2:31 天中 0 天相等
"r2_equal_days": 0,
"canonicalized": 0,
"confirms": frozenset({
"placeholder:·", "derived:夜间合计", "year_from:交叉表", "relation:R1", "relation:R2",
"unit:日客流.全日客流", "unit:日客流.分区甲", "unit:日客流.分区乙", "unit:时段客流.客流",
"unit:时段客流_表内合计.客流", "mode",
}),
"note_has": ("已逐日核对",),
"note_lacks": ("本期原表附有说明文字",),
}
@dataclass
class _Row:
#: empty / period / text / axis / measure / blank / section / hour / total / subtotal / note / raw
kind: str
label: Any = None
values: list[Any] | None = None
merged: bool = False
#: hour:小时;total / subtotal:(起始小时, 结束小时)
key: Any = None
r: int = 0
def flow_filename(start: dt.date, days: int, variant: str = "D00") -> str:
end = start + dt.timedelta(days=days - 1)
if variant == "D23":
return f"{start.month}月客流.xlsx"
return f"月报导出_{start.isoformat()}_{end.isoformat()}.xlsx"
def _hour_label(h: int, variant: str) -> str:
if variant == "D11":
return f"{h:02d}:00-{h + 1:02d}:00"
if variant == "D24":
return f"{h}–{h + 1}" # en dash
return f"{h}-{h + 1}"
def _numbers(days: int, seed: int, variant: str):
rnd = random.Random(seed)
zone_a = [rnd.randint(3000, 9000) for _ in range(days)]
zone_b = [rnd.randint(2000, 8000) for _ in range(days)]
hours: dict[int, list[Any]] = {}
for h in DAY_HOURS + NIGHT_HOURS:
# 第 10 行(7-8)整行是占位符:真实报表早上这一档没有数
hours[h] = [PLACEHOLDER] * days if h == 7 else [rnd.randint(50, 900) for _ in range(days)]
# 变体专用的数另起一个随机源,基线的数不跟着变
extra = random.Random(seed * 7919 + 13)
zone_c = [extra.randint(1000, 5000) for _ in range(days)] if variant == "D13" else None
total = [zone_a[i] + zone_b[i] + (zone_c[i] if zone_c else 0) for i in range(days)]
if variant == "D26" and days >= 10:
total[9] += 7 # 第 10 个日期:全日 = 甲 + 乙 + 7
# 保证每个日期「各时段之和 ≠ 全日客流」(R2 的口径不同在每一天都看得出来)
for i in range(days):
s = sum(v[i] for h, v in hours.items() if h != 7)
if s == total[i]:
hours[8][i] += 1 if hours[8][i] < 900 else -1
return zone_a, zone_b, zone_c, total, hours
def _period_text(start: dt.date, end: dt.date, variant: str) -> str:
if variant == "D12": # 第二个日期省略年份
return f"统计时间范围:{start.year}年{start.month}月{start.day}日至{end.month}月{end.day}日"
return f"统计时间范围:{start.year}年{start.month}月{start.day}日至{end.year}年{end.month}月{end.day}日"
def _title_text(start: dt.date, end: dt.date, variant: str) -> str:
if variant == "D17":
return f"{TITLE}({start.year}年{start.month}月)"
if variant == "D27":
return f"{TITLE}({start.year}年{start.month}月{start.day}日至{end.year}年{end.month}月{end.day}日)"
return TITLE
def _layout(start: dt.date, end: dt.date, days: int, seed: int, variant: str) -> list[_Row]:
zone_a, zone_b, zone_c, total, hours = _numbers(days, seed, variant)
rows: list[_Row] = []
if variant == "noperiod":
rows.append(_Row("empty"))
else:
rows.append(_Row("period", _period_text(start, end, variant), merged=True))
rows.append(_Row("text", _title_text(start, end, variant), merged=True))
rows.append(_Row("axis"))
measures = [_Row("measure", TOTAL_LABEL, total), _Row("measure", ZONE_A_LABEL, zone_a),
_Row("measure", ZONE_B_LABEL, zone_b)]
if zone_c is not None: # D13:多一个分区,全日 = 甲 + 乙 + 丙
measures.append(_Row("measure", ZONE_C_LABEL, zone_c))
if variant == "D19": # 顺序改成 甲、乙、全日
measures = [measures[1], measures[2], measures[0]]
blank = [] if variant == "D14" else [_Row("blank", merged=True)]
titled = variant != "D15" # D15:分段标题不合并
day = [_Row("section", "日间分时段客流(人次)" if variant == "D09" else DAY_TITLE, merged=titled)]
for h in DAY_HOURS:
if variant == "D04" and h == 7: # D04:日间少了 7-8
continue
day.append(_Row("hour", _hour_label(h, variant), hours[h], key=h))
if variant == "D18": # 日间末尾多一行公式小计
day.append(_Row("subtotal", "7-18 时合计", key=(7, 18)))
night = [_Row("section", NIGHT_TITLE, merged=titled)]
night += [_Row("hour", _hour_label(h, variant), hours[h], key=h) for h in NIGHT_HOURS]
for label, s, e in TOTALS:
night.append(_Row("total", f"{s}:00-{e}:00合计" if variant == "D22" else label, key=(s, e)))
if variant == "D05": # 夜间块(标题、时段、合计)移到日间块之前
rows += measures + blank + night + day
elif variant == "D06": # 日客流三行移到最后,前面加标题
rows += day + night + [_Row("empty"), _Row("text", "日客流(人次)")] + measures
else:
rows += measures + blank + day + night
if variant == "D21": # 表下空一行补录一行数据
rows += [_Row("empty"), _Row("raw", values=["补录(人次)", "1,234", "2,345"])]
elif variant == "D28":
rows += [_Row("empty"), _Row("note", f"注:{start.month}月15日闸机故障,当日客流为估算值")]
elif variant == "D28b":
rows += [_Row("empty"), _Row("note", f"注:{start.year}年{start.month}月数据为初步统计,待修订")]
for i, row in enumerate(rows):
row.r = 2 + i # 第 1 行空着,和真实报表一样从第 2 行开始
return rows
def flow_workbook(start: dt.date, days: int, *, seed: int = 0, variant: str = "D00") -> tuple[bytes, str]:
"""返回 (xlsx 字节, 文件名)。variant 见 VARIANTS(9.2 的用例代号)。"""
if variant not in VARIANTS:
raise ValueError(f"不认识的 variant:{variant}")
if days < 1:
raise ValueError("days 至少为 1")
end = start + dt.timedelta(days=days - 1)
dates = [start + dt.timedelta(days=i) for i in range(days)]
rows = _layout(start, end, days, seed, variant)
last = FIRST_COL + days - 1
last_letter = get_column_letter(last)
hour_row = {row.key: row.r for row in rows if row.kind == "hour"}
hour_vals = {row.key: row.values for row in rows if row.kind == "hour"}
book = Workbook()
ws = book.active
ws.title = SHEET
cached: dict[str, object] = {}
computed: dict[int, list[int]] = {} # 合计行 → 各日期的值(D16 的右侧合计列要用)
def merge(r: int) -> None:
ws.merge_cells(start_row=r, start_column=LABEL_COL, end_row=r, end_column=last)
for row in rows:
r = row.r
if row.kind == "empty":
continue
if row.kind in ("period", "text", "section", "note"):
ws.cell(r, LABEL_COL, row.label)
if row.merged:
merge(r)
elif row.kind == "blank":
if row.merged:
merge(r)
elif row.kind == "axis":
for i, d in enumerate(dates):
if variant == "D10": # 表头是日期格,显示成「9月1日」
cell = ws.cell(r, FIRST_COL + i, dt.datetime(d.year, d.month, d.day))
cell.number_format = 'm"月"d"日"'
else:
ws.cell(r, FIRST_COL + i, f"{d.month}月{d.day}日")
if variant == "D16":
ws.cell(r, last + 1, "合计")
elif row.kind in ("measure", "hour"):
ws.cell(r, LABEL_COL, row.label)
for i, v in enumerate(row.values or []):
ws.cell(r, FIRST_COL + i, v)
elif row.kind in ("total", "subtotal"):
ws.cell(r, LABEL_COL, row.label)
s, e = row.key
span = [h for h in range(s, e) if h in hour_row]
r1, r2 = hour_row[span[0]], hour_row[span[-1]]
assert r2 - r1 == len(span) - 1, "合计覆盖的时段行必须连续"
vals = []
for i in range(days):
v = sum(x for h in span if isinstance(x := hour_vals[h][i], int))
col = get_column_letter(FIRST_COL + i)
literal = variant in ("D07", "D08") and row.kind == "total"
if variant == "D08" and row.key == (18, 22) and i == 4:
v += 100 # 写死的「18-22 时合计」第 5 个日期多 100
vals.append(v)
if literal:
ws.cell(r, FIRST_COL + i, v)
else:
ws.cell(r, FIRST_COL + i, f"=SUM({col}{r1}:{col}{r2})")
cached[f"{col}{r}"] = v
computed[r] = vals
elif row.kind == "raw":
for j, v in enumerate(row.values or []):
ws.cell(r, LABEL_COL + j, v)
else: # pragma: no cover - 行清单只会有上面这些
raise AssertionError(row.kind)
if variant == "D16": # 右侧多一列「合计」:每个数据行一个 SUM 公式,有缓存
col = get_column_letter(last + 1)
for row in rows:
if row.kind in ("measure", "hour"):
vals = [v for v in row.values or [] if isinstance(v, int)]
elif row.kind in ("total", "subtotal"):
vals = computed[row.r]
else:
continue
ws.cell(row.r, last + 1, f"=SUM(C{row.r}:{last_letter}{row.r})")
cached[f"{col}{row.r}"] = sum(vals)
if variant == "D20": # 多一个有内容的可见工作表
extra = book.create_sheet(EXTRA_SHEET)
extra["A1"] = "本工作簿为合成测试数据,仅供测试使用"
raw = excel_saved(save(book), SHEET, cached, full_calc=variant == "D25")
return raw, flow_filename(start, days, variant)