Copy the full-data packet once per report, not once per domain (issue #122, BUG-1471)

read_full_data_domain deep-copied the shared worksheets on every call; a long
report reads five domains twice, so ten copies for read-only use. The shared
part is now copied once (full_data_domain_base) and reused for every domain's
facts and their leaf check; each domain's facts are still copied out so domains
share no objects. Long reports byte-identical on four fictional charts; with
#121, -24–29% end to end.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoz1Qk3y9buDv1mkmTu

CHANGELOG.md

@@ -1,5 +1,11 @@
 # 印度占星 Skill 更新日志
 
+## 2026-10-28 — 个人长报告生成快约四分之一
+
+- 第一次生成一份完整的个人长报告，本机实测从约 26 秒降到约 20 秒；报告内容一字不变。
+- 两处改动：报告出稿的提速开关默认打开（issue #121）；报告和咨询数据卡不再为每个领域把整份数据各复制一遍（issue #122）。
+- Skill 版本不变（BUG-1335、BUG-1471）。
+
 ## 2026-10-28 — 星盘页页头整齐了，深色模式下几处灰底和白边去掉
 
 - 首页输入框在深色模式下不再有一块灰底。

docs/BUG_HISTORY.md

@@ -20065,3 +20065,18 @@
 - 复发自：BUG-1466（深色变体未覆盖）
 - 修复版本：分支 `codex/ui-polish-140-20261028`。
 
+## BUG-1471 | 长报告与咨询数据卡每个领域都把整份数据包深拷贝一遍
+
+- 状态：resolved（代码 + 回归测试；待部署）
+- 首次发现 / 最近更新：2026-10-28 / 2026-10-28（issue #118 代码体检 → #122）
+- 影响面：个人长报告（模型写的「读盘」前置事实）、咨询数据卡；`scripts/pl9_full_data_export.py` `read_full_data_domain`、`scripts/consult_card_domain_facts.py`、`scripts/report_render/reading_facts.py`。
+- 现象：长报告冷生成慢；profile 口径下整包 deepcopy 合计约 9 s，其中 `read_full_data_domain` 约 5 s。
+- 触发条件：任何长报告生成、任何带数据卡的咨询轮。
+- 根因：`read_full_data_domain` 每调一次就把 `worksheets` 等共享部分整份深拷贝。长报告读 5 个领域，生成事实一遍、逐条核对又一遍，共 10 份；读的人全是只读。
+- 修复：新增 `full_data_domain_base`，同一个数据包只深拷一次；`read_full_data_domain(..., base=)` 复用它。`build_consult_card_facts` 每次调用只建一次，长报告的事实与核对共用同一份。各领域产出的事实再各自深拷贝（体量小），保持「不同领域的结果互不共享对象」与改前一致。
+- 验证：`scripts/research/report_reference_perf_ab.py --mask-pyjhora`（四张虚构盘各三次）改前 / 改后、开关开 / 关全部组合下完整报告逐字节相同；与改前相比，叠加 #121 后端到端快 24%～29%（中位约 26%），单论本单在开关打开时再快 1%～7%。JSON 与参考版两项附加检查在改前代码上连跑两次即不同（输出自身跨进程不确定），无法用跨代码 sha 比对，已注明。新测试 `tests/test_consult_card_full_source.py::test_all_domains_share_one_packet_copy_and_still_match_the_per_domain_path`：只建一次共享拷贝、结果与逐领域路径一致、领域间不共享对象。Python 快速门全部 Python 步骤通过（前端 npm 步因本工作树无依赖未跑，本单无前端改动）。
+- 防复发：上面的测试；只读用途的「视图」不要每次整份深拷贝，多处读同一包时共享一份。
+- 关联：BUG-1335（同一份报告的出稿与哈希重复劳动，#121 已开快路径）。溯源哈希仍在出稿后由质量门重算一次以核对，这是有意的完整性检查，未合并。
+- 复发自：无
+- 修复版本：分支 `codex/report-fast-default-20261028`。
+

scripts/consult_card_domain_facts.py

@@ -8,10 +8,12 @@ judgment table; the card builder does that from these leaves.
 
 from __future__ import annotations
 
+import copy
 import json
+from collections.abc import Mapping
 from typing import Any
 
-from pl9_full_data_export import read_full_data_domain
+from pl9_full_data_export import full_data_domain_base, read_full_data_domain
 
 PRECISION_SENTENCE = "出生时间精度不够，不参与"
 RAHU_NOTE = "罗计不单独定功能吉凶"
@@ -783,10 +785,16 @@ def _factors(domain: str, timing: dict[str, Any], karakas: dict[str, dict[str, A
     return factors
 
 
-def build_domain_facts(packet: dict, domain: str, *, minute_confirmed: bool) -> dict[str, Any]:
+def build_domain_facts(
+    packet: dict,
+    domain: str,
+    *,
+    minute_confirmed: bool,
+    base: Mapping | None = None,
+) -> dict[str, Any]:
     if domain not in PACKET_DOMAIN:
         raise KeyError(domain)
-    view = read_full_data_domain(packet, PACKET_DOMAIN[domain])
+    view = read_full_data_domain(packet, PACKET_DOMAIN[domain], base=base)
     varga = {
         "career": "D10", "wealth": "D2", "marriage": "D9", "health": "D6",
         "children": "D7", "parents": "D12", "education": "D24", "migration": "D4",
@@ -824,7 +832,9 @@ def build_domain_facts(packet: dict, domain: str, *, minute_confirmed: bool) ->
     encoded = json.dumps(facts, ensure_ascii=False, default=str)
     if "tajika_named_yogas" in encoded:
         raise RuntimeError("consult card facts must not read tajika_named_yogas")
-    return facts
+    # Leaves may be objects of the view. With a shared base, two domains' facts
+    # would share them; each domain's facts stay their own, as before.
+    return facts if base is None else copy.deepcopy(facts)
 
 
 def build_consult_card_facts(
@@ -833,13 +843,16 @@ def build_consult_card_facts(
     *,
     minute_confirmed: bool | None,
     body: dict | None = None,
+    base: Mapping | None = None,
 ) -> dict[str, Any]:
     confirmed = minute_confirmed_from_body(body) if minute_confirmed is None else minute_confirmed
     rows = []
+    if base is None and any(domain in PACKET_DOMAIN for domain in domains):
+        base = full_data_domain_base(packet)
     for domain in domains:
         if domain not in PACKET_DOMAIN:
             continue
-        rows.append(build_domain_facts(packet, domain, minute_confirmed=confirmed))
+        rows.append(build_domain_facts(packet, domain, minute_confirmed=confirmed, base=base))
     return {
         "schema": "consult-card-facts-v1",
         "minute_confirmed": confirmed,
@@ -848,9 +861,9 @@ def build_consult_card_facts(
     }
 
 
-def assert_fact_leaves(packet: dict, facts: dict[str, Any]) -> None:
+def assert_fact_leaves(packet: dict, facts: dict[str, Any], *, base: Mapping | None = None) -> None:
     """Every copied leaf equals the packet. Derived selections are not leaves."""
-    view = read_full_data_domain(packet, facts["packet_domain"])
+    view = read_full_data_domain(packet, facts["packet_domain"], base=base)
 
     def check(item: Any) -> None:
         if isinstance(item, list):

scripts/pl9_full_data_export.py

@@ -2472,19 +2472,18 @@ def _append_annual_method_reference(markdown: str, packet: dict, language: str)
     return markdown.rstrip() + "\n\n" + extra + "\n"
 
 
-def read_full_data_domain(packet: Mapping, domain: str) -> dict:
-    """Deep-copied packet view for one consult domain.
+def full_data_domain_base(packet: Mapping) -> dict:
+    """The detached, domain-independent part of every domain view.
 
-    Field selection belongs to the consult card. This entry only freezes the
-    domain name and detaches the copy from the live packet.
+    One deep copy of the packet's shared sections. ``read_full_data_domain``
+    used to make this copy once per domain, and a long report reads five
+    domains twice (facts, then the leaf check): ten copies of the whole
+    ``worksheets`` tree for read-only use (issue #122). Callers that read
+    several domains of one packet build this once and pass it as ``base``.
     """
-    if domain not in FULL_DATA_DOMAINS:
-        raise KeyError(domain)
     if not isinstance(packet, Mapping):
         raise TypeError("packet must be a mapping")
     copied = copy.deepcopy({
-        "domain": domain,
-        "label": FULL_DATA_DOMAINS[domain],
         "report_version": packet.get("report_version"),
         "birth_info": packet.get("birth_info"),
         "core_chart": packet.get("core_chart"),
@@ -2496,3 +2495,19 @@ def read_full_data_domain(packet: Mapping, domain: str) -> dict:
     if isinstance(timing, dict):
         timing.pop("pyjhora_annual_method_reference", None)
     return copied
+
+
+def read_full_data_domain(packet: Mapping, domain: str, *, base: Mapping | None = None) -> dict:
+    """Detached packet view for one consult domain.
+
+    Field selection belongs to the consult card. This entry only freezes the
+    domain name and detaches the view from the live packet. With ``base`` (from
+    ``full_data_domain_base`` of the same packet) the shared sections are that
+    one copy; views of different domains then share it and must be read only.
+    """
+    if domain not in FULL_DATA_DOMAINS:
+        raise KeyError(domain)
+    if not isinstance(packet, Mapping):
+        raise TypeError("packet must be a mapping")
+    shared = full_data_domain_base(packet) if base is None else base
+    return {"domain": domain, "label": FULL_DATA_DOMAINS[domain], **shared}

scripts/report_render/reading_facts.py

@@ -14,9 +14,11 @@ from typing import Any
 try:
     from scripts.consult_card_domain_facts import assert_fact_leaves, build_consult_card_facts, dig
     from scripts.consult_full_data_cache import _attach_graha_drishti
+    from scripts.pl9_full_data_export import full_data_domain_base
 except ModuleNotFoundError:  # pragma: no cover - direct scripts/ execution path
     from consult_card_domain_facts import assert_fact_leaves, build_consult_card_facts, dig
     from consult_full_data_cache import _attach_graha_drishti
+    from pl9_full_data_export import full_data_domain_base
 
 READING_FACTS_SCHEMA = "report-reading-facts-v1"
 READING_DOMAINS = ("general", "career", "wealth", "marriage", "health")
@@ -87,10 +89,12 @@ def build_report_reading_facts(
 ) -> dict[str, Any]:
     """Compact, packet-sourced facts for the reading. Raises on any non-packet leaf."""
     working = _attach_graha_drishti(copy.deepcopy(packet), full_reading)
-    raw = build_consult_card_facts(working, list(READING_DOMAINS), minute_confirmed=minute_confirmed)
+    # One detached copy of the packet serves every domain's facts and their leaf check (issue #122).
+    base = full_data_domain_base(working)
+    raw = build_consult_card_facts(working, list(READING_DOMAINS), minute_confirmed=minute_confirmed, base=base)
     rows = {row["domain"]: row for row in raw["domains"]}
     for row in rows.values():
-        assert_fact_leaves(working, row)
+        assert_fact_leaves(working, row, base=base)
     general = rows["general"]
     planets = {}
     for name, record in general["planets"].items():

tests/test_consult_card_full_source.py

@@ -425,3 +425,32 @@ def test_one_warm_at_a_time_and_same_key_user_waits_for_that_build(monkeypatch):
         warm_thread.join(timeout=5)
         user_thread.join(timeout=5)
         _drop_cache(same_identity)
+
+
+def test_all_domains_share_one_packet_copy_and_still_match_the_per_domain_path(monkeypatch):
+    """Issue #122: a report reads five domains twice; the packet is copied once, not ten times."""
+    import consult_card_domain_facts as facts_module
+    from consult_card_domain_facts import build_consult_card_facts
+
+    packet = _packet()
+    expected = [build_domain_facts(packet, domain, minute_confirmed=True) for domain in CONSULTATION_DOMAINS]
+    calls = []
+    real_base = facts_module.full_data_domain_base
+
+    def counting_base(value):
+        calls.append(1)
+        return real_base(value)
+
+    monkeypatch.setattr(facts_module, "full_data_domain_base", counting_base)
+    shared = build_consult_card_facts(packet, list(CONSULTATION_DOMAINS), minute_confirmed=True)
+    assert len(calls) == 1
+    assert json.dumps(shared["domains"], ensure_ascii=False, sort_keys=True, default=str) == json.dumps(
+        expected, ensure_ascii=False, sort_keys=True, default=str,
+    )
+    # Each domain's facts stay their own objects, as when every domain copied the packet.
+    first, second = shared["domains"][0], shared["domains"][1]
+    assert first["planets"]["Sun"] is not second["planets"]["Sun"]
+    assert first["timing"] is not second["timing"]
+    base = real_base(packet)
+    for row in shared["domains"]:
+        assert_fact_leaves(packet, row, base=base)