From af2e4988e8a1bb9bbc1e5c07712fe2d41eda0dc0 Mon Sep 17 00:00:00 2001 From: Claude Date: Thu, 19 Nov 2026 19:36:30 +0800 Subject: [PATCH] fix: report and synastry reader text loses its last backstage wording (BUG-2038) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit A pronoun whose planet was in a removed sentence takes that planet; only a planet back in its natal sign is given to the reading; the 现在这几年 section leads with the stage of life and dates only the main dasha; the health reminder is one fixed program sentence; the synastry appendix no longer doubles 所以 and shows the technical summary with names, no machine tags, minutes as 分 and 秒. Co-Authored-By: Claude Fable 5.2 Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8 --- docs/BUG_HISTORY.md | 12 ++ .../synastry/synastry-report-view.tsx | 17 +- .../src/lib/personal-report-health-notice.ts | 35 ++++ .../src/lib/personal-report-reading-audit.ts | 2 +- .../src/lib/personal-report-reading-rebind.ts | 28 ++- .../lib/personal-report-reading-wording.ts | 18 +- frontend/src/lib/personal-report-reading.ts | 19 +- frontend/src/lib/synastry-analysis-reader.ts | 33 ++++ frontend/src/lib/synastry-report-export.ts | 10 +- frontend/tests/reader-leftovers-2038.test.tsx | 180 ++++++++++++++++++ frontend/tests/report-leftovers-2021.test.ts | 10 +- .../tests/report-reading-20261027.test.tsx | 6 +- ...ort-reading-claim-binding-20261112.test.ts | 8 +- .../report-reading-snapshot-20261113.test.ts | 2 +- .../tests/report-user-layer-2028.test.tsx | 3 +- references/product_runtime_contract.md | 28 ++- scripts/report_render/opening_guide.py | 7 +- scripts/report_render/reading_facts.py | 8 +- tests/test_reader_leftovers_2038.py | 48 +++++ tests/test_report_leftovers_2021.py | 10 +- .../test_report_narayana_shadbala_20261111.py | 7 +- tests/test_report_opening_supplement.py | 7 +- 22 files changed, 461 insertions(+), 37 deletions(-) create mode 100644 frontend/src/lib/personal-report-health-notice.ts create mode 100644 frontend/tests/reader-leftovers-2038.test.tsx create mode 100644 tests/test_reader_leftovers_2038.py diff --git a/docs/BUG_HISTORY.md b/docs/BUG_HISTORY.md index 73636fd94..2f2478f66 100644 --- a/docs/BUG_HISTORY.md +++ b/docs/BUG_HISTORY.md @@ -27380,3 +27380,15 @@ - 验证与防复发:该测试通过。上限仍在,事实包再明显变大时会再次报出来,届时先查冗余再议上限。 - 关联记录:BUG-2013、BUG-2019、BUG-2021、BUG-2024、BUG-2028。 - 修复版本:`claude/report-facts-budget-2036`。 + +## BUG-2038 | 个人报告与合盘附录里读者看到的零碎后台写法(复测第九轮) + +- 状态:fixed-pending-verify(待测试环境重新生成报告确认)。 +- 首次发现 / 最近更新:2026-11-20 / 2026-11-20。 +- 触发与现象:测试环境复测第九轮。个人报告:性格一节有一段以「它与落陷的木星互换了星座」开头,「它」没有所指;「现在这几年」大半是几套大运的日期清单,并写出「行运木星……不在同一个星座」;开篇时间线带算法版本出处;正文出现「这套结果还没有完成外部核对」「双行运只查了第7宫,没有激活」「六力值 6.0765 Rupa」「按规则筛选出来的候选里」;健康一节连着两句提醒。合盘附录:每条「所以:」都读成「所以:所以……」;「技术依据摘要」里仍有带下划线机器名的括注、甲方 / 乙方和「前 6.5 分钟」。 +- 触发条件:生成个人报告;打开合盘报告的附录。 +- 根因:(1) 程序按句去掉越界引用(BUG-2019)时,被删句的下一句以代词开头,主语跟着丢了。(2) 报告事实把不同座的行运也列成一行,规则没有说「现在这几年」先讲什么、哪些体系不在正文列日期。(3) 中文开篇照抄了算法标签。(4) 规则没有禁止这几种后台说明进各节正文。(5) 健康一节的提醒交给模型写,规则只限「每节最多一句」。(6) 合盘存下来的句子自己以「所以」开头,页面又加了一次标签;BUG-2034 的显示处理只用在「分析逻辑」,没用到「技术依据摘要」。 +- 修复:被删句之后紧跟的句子若以指代行星的代词开头,换成被删句最先点名的行星;事实里只列回到本命星座的行星;规则写明「现在这几年」先说处在什么阶段,正文只给 Vimshottari 的日期、Narayana 至多一句;中文开篇不带算法标签;三种后台说明在各节正文里去掉(「依据与边界」不动),规则同步;候选格局在正文用日常话;健康一节里模型写的提醒去掉,由程序在节末补一句固定的话(历史人物回放不补);合盘显示时去掉句首重复的「所以」,「技术依据摘要」同样换称呼、去机器括注、分钟写成几分几秒。计算、等级、保存的合盘文字都没有改。 +- 验证与防复发:`frontend/tests/reader-leftovers-2038.test.tsx`、`tests/test_reader_leftovers_2038.py`,用的是复测第九轮存下的句子;公开盘重新生成一次个人报告,九节一次通过,各节正文里这些写法为 0,健康一节只有程序补的一句。合盘两处只有显示层测试,没有重新生成合盘。 +- 关联记录:BUG-2019、BUG-2021、BUG-2024、BUG-2028、BUG-2034。 +- 修复版本:`claude/reader-leftovers-2038`。 diff --git a/frontend/src/components/synastry/synastry-report-view.tsx b/frontend/src/components/synastry/synastry-report-view.tsx index 95e88b534..c48958b3c 100644 --- a/frontend/src/components/synastry/synastry-report-view.tsx +++ b/frontend/src/components/synastry/synastry-report-view.tsx @@ -3,7 +3,7 @@ import { ReportBirthNotices } from "@/components/personal-report/report-birth-notices"; import { readReportBirthScenarios } from "@/lib/report-birth-marking"; import { buildSynastryExportBlocks, synastryParagraphs } from "@/lib/synastry-report-export"; -import { synastryAnalysisReaderText } from "@/lib/synastry-analysis-reader"; +import { synastryAnalysisReaderText, synastrySoReaderText, synastryTechnicalReaderText } from "@/lib/synastry-analysis-reader"; import { ReportExportButton } from "@/components/personal-report/report-actions"; import { ReportExportDrawer } from "@/components/personal-report/report-export-drawer"; import { Button } from "@/components/ui/button"; @@ -79,7 +79,8 @@ function chartLine(language: "zh" | "en", basis: string | undefined, names?: { a } function soLine(language: "zh" | "en", plain: string | undefined, names?: { a?: string; b?: string }) { - const text = synastryAnalysisReaderText(plain, names, language); + // BUG-2038: the stored sentence opens with 「所以」 itself; the label is not doubled. + const text = synastrySoReaderText(plain, names, language); return language === "zh" ? `所以:${text}` : `So: ${text}`; } @@ -162,6 +163,8 @@ export function SynastryReportView(props: { if (!body) return null; const names = snapshot.names ?? {}; const reportNames = { a: names.a || "甲", b: names.b || "乙" }; + // BUG-2038: 「技术依据摘要」 is shown with the two names, without machine tags, minutes as 分 / 秒. + const technical = (value: string) => synastryTechnicalReaderText(value, names, language); const birthScenarios = readReportBirthScenarios(snapshot.birthScenarios); const structure = body.structure ?? snapshot.structure; const order = synastrySectionOrder(structure); @@ -360,20 +363,20 @@ export function SynastryReportView(props: {

{block.title}

{block.precision?.map(({ title, disclosure, summary, rows }) => (
-

{title}

- {disclosure ?

{disclosure}

: null} +

{technical(title)}

+ {disclosure ?

{technical(disclosure)}

: null} {rows.length ? (
{rows.map((row) => - + )}
项目档案时刻候选时刻
{row.item}{row.reference}{row.branches.map((branch) =>
{branch}
)}
{row.branches.map((branch) =>
{technical(branch)}
)}
) : null} - {summary ?

{summary}

: null} + {summary ?

{technical(summary)}

: null}
))} {block.audit?.map(({ title, rows }) => ( @@ -390,7 +393,7 @@ export function SynastryReportView(props: { ))} {block.ramanComparison?.length ? ( <> diff --git a/frontend/src/lib/personal-report-health-notice.ts b/frontend/src/lib/personal-report-health-notice.ts new file mode 100644 index 000000000..e400a8853 --- /dev/null +++ b/frontend/src/lib/personal-report-health-notice.ts @@ -0,0 +1,35 @@ +/** + * BUG-2038 (staging 复测第九轮, 2026-11-20; 产品负责人 2026-11-16「不写推责的话」): the health section + * ended with two reminders of the model's own (「这些是结构上的条件,不是身体的结论。」「以上是传统结构上的 + * 读法,不构成医疗判断;有具体不适要找医生。」). One medical reminder is right for this section; it is now + * the program's: the model's are taken out, and one fixed sentence closes the section. The wording follows + * the report notice already in use (「……不构成医疗、法律或投资建议」, personal-report-generation.ts). + */ +export const HEALTH_NOTICE_ZH = "这一节不构成医疗建议,身体有具体不适请找医生。"; + +const SENTENCES = /[^。!?\n]+[。!?]*\n*|\n+/gu; +// A sentence whose job is the reminder: it disclaims medical meaning or sends the reader to a doctor. +const REMINDER = /不构成医疗|不是医疗(?:意见|判断|建议|诊断)|不属于医疗|不是身体的结论|不是(?:对)?身体(?:状况)?的(?:判断|诊断|结论)|不(?:能|可)?(?:代替|替代)(?:医生|医疗|诊断|就医)|(?:请|要|应|需要|记得|及时)(?:去)?(?:找|看|咨询)医生|(?:请|及时|尽快)就医/u; +/** Longer sentences carry content of their own and are left to the ordinary checks. */ +const REMINDER_MAX_CHARS = 60; + +/** The paragraph without reminder sentences; "" when it was nothing else. */ +export function stripHealthReminders(text: string): string { + const sentences = text.match(SENTENCES) ?? [text]; + return sentences + .filter((sentence) => !(sentence.trim().length <= REMINDER_MAX_CHARS && REMINDER.test(sentence))) + .join("") + .trim(); +} + +/** + * The fixed sentence at the end of the paragraph. A program line that must stay last (the other-scenario + * line, BUG-2013) keeps its place: `tail` is that line, already part of `text`. + */ +export function withHealthNotice(text: string, tail: string, maxChars: number): string | null { + const body = tail && text.endsWith(tail) ? text.slice(0, text.length - tail.length) : text; + const end = body === text ? "" : tail; + const joiner = /[。!?]$/.test(body.trimEnd()) ? "" : "。"; + const combined = `${body.trimEnd()}${joiner}${HEALTH_NOTICE_ZH}${end}`; + return combined.length <= maxChars ? combined : null; +} diff --git a/frontend/src/lib/personal-report-reading-audit.ts b/frontend/src/lib/personal-report-reading-audit.ts index a3b31edf8..88ad211a4 100644 --- a/frontend/src/lib/personal-report-reading-audit.ts +++ b/frontend/src/lib/personal-report-reading-audit.ts @@ -9,7 +9,7 @@ import type { ReadingSectionTransform, ReadingTransformLog } from "./personal-re // 原值: 6;新值: 7;原因: BUG-2021,正文里的内部状态词(参数敏感、双体系确认、命中候选……)在检查前换成读者说法;财富两类结论可引用 AL 起第 2、11 宫。 // 原值: 7;新值: 8;原因: BUG-2024,正文里抄出的四句规则话(同一句要并读、只作这一天观察不外推、被路由分给、用途稳定性没有评估)换成读者说法;落陷取消的标签括注不算尊贵状态。 // 原值: 8;新值: 9;原因: BUG-2028,段末的出生时间说明改用日常话(不写分盘和秒数);正文里的开发用词(具名规则、证据链、方法版本)换成读者说法。 -export const REPORT_READING_POSTCHECK_VERSION = 10; // 原值 9;BUG-2034 正文的写法收尾(建议句后的说明、日期写法、窗口级引子) +export const REPORT_READING_POSTCHECK_VERSION = 11; // 原值 10;BUG-2038 被删句之后的代词补主语、健康一节的提醒改由程序补、三处后台说明不进各节正文 function record(value: unknown): Record { return value !== null && typeof value === "object" && !Array.isArray(value) ? value as Record : {}; diff --git a/frontend/src/lib/personal-report-reading-rebind.ts b/frontend/src/lib/personal-report-reading-rebind.ts index fe82ec07e..fb36021aa 100644 --- a/frontend/src/lib/personal-report-reading-rebind.ts +++ b/frontend/src/lib/personal-report-reading-rebind.ts @@ -89,6 +89,25 @@ function textWithinAllowed(text: string, allowed: ReadonlySet, catalog: } const SENTENCES = /[^。;!?\n]+[。;!?]*\n*|\n+/g; +const PLANET_NAMES_ZH = ["太阳", "月亮", "火星", "水星", "木星", "金星", "土星", "罗睺", "计都"] as const; +const LEADING_PRONOUN = /^(\s*)(?:它(?!们)|这颗行星|这颗星|该星)/u; + +/** + * BUG-2038 (staging 复测第九轮, 2026-11-20): 「它与落陷的木星互换了星座……」 opened a paragraph with nothing + * for 「它」 to point at, because the sentence before it (which named the planet) had been removed for + * citing a technique outside the claim. When the sentence right after a removed one opens with a pronoun + * for a planet, the pronoun is replaced by the planet the removed sentence named first. Nothing else of + * the removed sentence comes back; with no planet in it the sentence is left as written. + */ +export function withSubjectOf(sentence: string, removed: string): string { + const lead = LEADING_PRONOUN.exec(sentence); + if (!lead) return sentence; + const first = PLANET_NAMES_ZH + .map((name) => ({ name, at: removed.indexOf(name) })) + .filter(({ at }) => at >= 0) + .sort((left, right) => left.at - right.at)[0]; + return first ? `${lead[1]}${first.name}${sentence.slice(lead[0].length)}` : sentence; +} /** * BUG-2019 (占星师 2026-11-19 第 8 条:「程序应拦截不受支持的具体断言,避免因一句引用问题扣掉整节」). @@ -121,11 +140,16 @@ export function dropOutsideReferences( const keptRefs = refs.filter((id) => allowed.has(id) && catalog.states.has(id)); const kept: string[] = []; let removed = false; + let justDropped: string | null = null; for (const sentence of text.match(SENTENCES) ?? []) { - if (textWithinAllowed(sentence, allowed, catalog)) kept.push(sentence); - else { + if (!sentence.trim()) kept.push(sentence); + else if (textWithinAllowed(sentence, allowed, catalog)) { + kept.push(justDropped === null ? sentence : withSubjectOf(sentence, justDropped)); + justDropped = null; + } else { dropped.push(sentence.trim()); removed = true; + justDropped = sentence; } } if (!removed && keptRefs.length === refs.length) break; diff --git a/frontend/src/lib/personal-report-reading-wording.ts b/frontend/src/lib/personal-report-reading-wording.ts index 3d4721b9c..631ab1019 100644 --- a/frontend/src/lib/personal-report-reading-wording.ts +++ b/frontend/src/lib/personal-report-reading-wording.ts @@ -75,6 +75,18 @@ const ADVICE_LABEL_LEAD = /(^|[。!?;]\s*)一般建议[::]\s*/gu; const ADVICE_LABEL_CLAUSE = /(?:——)?这(?:一点|一条|条)?(?:同样|也)?(?:只)?是一般(?:性的?)?建议(?:[,,]\s*不涉及[^;。,,]*)?[,,]?/gu; const SENTENCES = /[^。!?]+[。!?]?/gu; +/** + * BUG-2038 (staging 复测第九轮, 2026-11-20; 占星师分层答复): three backstage notes in the sections that + * give conclusions. Rules R-03 / R-05 now say not to write them; this catches what is still written. + * 「依据与边界」 is where execution status belongs, so that section is left alone. + */ +// 「……,但这套结果还没有完成外部核对,不作为事件先后的依据」: whether an outside check was done. +const EXTERNAL_CHECK = /[,,;;]?\s*(?:但|不过|只是)?(?:这套结果|这一套|这套|它)?(?:目前)?还?没有完成外部(?:核对|核验|验证)/gu; +// 「观察日这一天的双行运只查了第7宫,没有激活;」: a check that found nothing. +const DOUBLE_TRANSIT_IDLE = /(?:观察日)?(?:这一天)?(?:的)?(?:双行运|Double Transit)[^,。;]{0,16}只(?:查|检查|看)了?第\s*\d+\s*宫[,,]\s*(?:结论是)?没有激活[;;,,]?\s*/gu; +// 「……入「极友」,六力值 6.0765 Rupa,同时受……」: the raw Shadbala total as an aside between commas. +const SHADBALA_ASIDE = /[,,]\s*(?:六力值|六力|Shadbala)(?:([^)]{0,30}))?\s*(?:为|是)?\s*\d+(?:\.\d+)?\s*Rupa(?=[,,。;;])/gu; + /** Drops a sentence that only labels the advice; a paragraph that is nothing else is left alone. */ function withoutAdviceMeta(text: string): string { const sentences = text.match(SENTENCES) ?? [text]; @@ -83,8 +95,10 @@ function withoutAdviceMeta(text: string): string { } /** Reader wording for one paragraph; text without these phrases comes back unchanged. */ -export function readerWording(text: string): string { - const replaced = withoutAdviceMeta(text) +export function readerWording(text: string, section?: string): string { + const conclusions = section === "basis" ? text + : text.replace(EXTERNAL_CHECK, "").replace(DOUBLE_TRANSIT_IDLE, "").replace(SHADBALA_ASIDE, ""); + const replaced = withoutAdviceMeta(conclusions) .replace(ADVICE_LABEL_LEAD, "$1") .replace(ADVICE_LABEL_CLAUSE, "") .replace(YOGA_CANDIDATE, YOGA_CANDIDATE_READER) diff --git a/frontend/src/lib/personal-report-reading.ts b/frontend/src/lib/personal-report-reading.ts index ba351b311..89bb5442d 100644 --- a/frontend/src/lib/personal-report-reading.ts +++ b/frontend/src/lib/personal-report-reading.ts @@ -42,6 +42,7 @@ import { type BoundaryDependence, } from "./personal-report-boundary-line"; import { readerWording } from "./personal-report-reading-wording"; +import { stripHealthReminders, withHealthNotice } from "./personal-report-health-notice"; export { REPORT_READING_BASIS_LABEL, @@ -421,6 +422,14 @@ function checkSection( allowExtraParagraphs = false, boundary: BoundaryDependence | null = null, ): Checked { + // BUG-2038: the health section's medical reminder is the program's (added below); the model's own are removed. + if (section.id === "health") { + const kept = section.paragraphs.flatMap((paragraph) => { + const stripped = stripHealthReminders(paragraph.text); + return stripped ? [stripped === paragraph.text.trim() ? paragraph : { ...paragraph, text: stripped }] : []; + }); + if (kept.length) section = { ...section, paragraphs: kept }; + } const inferredRefs: string[][] = []; let text = ""; const snippetFor = (matches?: (sentence: string) => boolean) => { @@ -449,7 +458,7 @@ function checkSection( // BUG-2013: the other-scenario line is the program's; it is taken off before the wording checks // and put back below on the first paragraph of the section that uses a boundary-dependent varga (BUG-2019). // BUG-2021: internal status words are put in reader wording before the checks, so the stored text has them too. - text = readerWording(stripBoundaryLine(raw.text.trim(), boundary).trim()); + text = readerWording(stripBoundaryLine(raw.text.trim(), boundary).trim(), section.id); if (!text) return withheld("too_short"); if (text.length > MAX_PARAGRAPH_CHARS) return withheld("too_long"); const fieldName = (sentence: string) => /[_={}<>#*|`]/.test(sentence); @@ -511,6 +520,14 @@ function checkSection( if (stored === null) return withheld("too_long"); paragraphs.push({ text: stored, claims: types, grade: paragraph.grade, refs: [...refs], refs_from_text: addedRefs }); } + // BUG-2038: one fixed medical reminder closes the health section (not in a replay of a historical figure, + // where current advice is not given). It goes in front of a trailing other-scenario line, which stays last. + const last = paragraphs[paragraphs.length - 1]; + if (section.id === "health" && facts?.current_advice_allowed !== false && last) { + const tail = boundary ? last.text.slice(stripBoundaryLine(last.text, boundary).length) : ""; + const noticed = withHealthNotice(last.text, tail, MAX_PARAGRAPH_CHARS); + if (noticed) paragraphs[paragraphs.length - 1] = { ...last, text: noticed }; + } if (disclaimers > 1) { text = section.paragraphs.flatMap(({ text }) => text.split(SENTENCE_END)) .filter((sentence) => (sentence.match(DISCLAIMER)?.length ?? 0) > 0).join("。"); diff --git a/frontend/src/lib/synastry-analysis-reader.ts b/frontend/src/lib/synastry-analysis-reader.ts index 3fda2eae0..8594442cf 100644 --- a/frontend/src/lib/synastry-analysis-reader.ts +++ b/frontend/src/lib/synastry-analysis-reader.ts @@ -24,3 +24,36 @@ export function synastryAnalysisReaderText(text: string | undefined, names: Name const escaped = own.map((name) => name.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")).join("|"); return cleaned.split(new RegExp(`(${escaped})`, "u")).map((part) => (own.includes(part) ? part : relabel(part))).join(""); } + +/** + * BUG-2038 (staging 复测第九轮, 2026-11-20): every 「所以:」 line read 「所以:所以……」, because the stored + * sentence already opens with 「所以」 and the page adds the label. The label stays; the sentence's own + * opening connective is taken off where it is shown. + */ +export function synastrySoReaderText(text: string | undefined, names: Names | undefined, language: "zh" | "en"): string { + const shown = synastryAnalysisReaderText(text, names, language); + return language === "zh" ? shown.replace(/^\s*所以[,,::]?\s*/u, "") : shown.replace(/^\s*so[,:]?\s+/iu, (lead) => (/^\s*so\s+that\b/iu.test(shown) ? lead : "")); +} + +// 「前 6.5 分钟」「偏晚 2.23 分钟」: minutes with a decimal part, written the way the personal report writes them (BUG-2034). +const DECIMAL_MINUTES = /(\d+)\.(\d+)\s*分钟/gu; + +/** 「6.5 分钟」 → 「6 分 30 秒」, 「0.5 分钟」 → 「30 秒」; whole minutes and other text are unchanged. */ +export function synastryMinutesReaderText(text: string): string { + return text.replace(DECIMAL_MINUTES, (_match, whole: string, fraction: string) => { + const total = Math.round(Number(`${whole}.${fraction}`) * 60); + const minutes = Math.floor(total / 60); + const seconds = total % 60; + if (minutes === 0) return `${seconds} 秒`; + return seconds === 0 ? `${minutes} 分钟` : `${minutes} 分 ${seconds} 秒`; + }); +} + +/** + * A line of 「技术依据摘要」 as a reader sees it: the same naming and tag removal as 「分析逻辑」, plus the + * minute wording. Stored blocks are unchanged. + */ +export function synastryTechnicalReaderText(text: string | undefined, names: Names | undefined, language: "zh" | "en"): string { + const shown = synastryAnalysisReaderText(text, names, language); + return language === "zh" ? synastryMinutesReaderText(shown) : shown; +} diff --git a/frontend/src/lib/synastry-report-export.ts b/frontend/src/lib/synastry-report-export.ts index eac8d033a..ceedf82d6 100644 --- a/frontend/src/lib/synastry-report-export.ts +++ b/frontend/src/lib/synastry-report-export.ts @@ -1,6 +1,6 @@ // 2026-11-16 owner: the synastry report gets the same plain-text .md export as the annual report. import type { BirthInputState } from "./birth-input-revision"; -import { synastryAnalysisReaderText } from "./synastry-analysis-reader"; +import { synastryAnalysisReaderText, synastrySoReaderText, synastryTechnicalReaderText } from "./synastry-analysis-reader"; import { reportBirthExportPreamble, type ReportBirthScenarios } from "./report-birth-marking.ts"; import { withExportBirthInputState, type ReportExportBlock } from "./report-export-blocks"; @@ -83,8 +83,10 @@ export function buildSynastryExportBlocks(input: SynastryExportInput): ReportExp if (!section) continue; // BUG-2034: the analysis lines name the two people and drop machine tags, as on the page. const read = (value: unknown) => synastryAnalysisReaderText(text(value), input.names, input.language); + const so = (value: unknown) => synastrySoReaderText(text(value), input.names, input.language); const reasons = rows(section.reasons).map((reason) => ( - zh ? `- 盘面:${read(reason.basis)}\n 所以:${read(reason.plain)}` : `- Chart: ${read(reason.basis)}\n So: ${read(reason.plain)}` + // BUG-2038: the stored sentence opens with 「所以」 itself; the label is not doubled. + zh ? `- 盘面:${read(reason.basis)}\n 所以:${so(reason.plain)}` : `- Chart: ${read(reason.basis)}\n So: ${so(reason.plain)}` )); blocks.push(block(`body:${id}`, "body", title, [ ...synastryParagraphs(section.text), @@ -93,7 +95,9 @@ export function buildSynastryExportBlocks(input: SynastryExportInput): ReportExp } for (const item of rows(input.technical)) { const source = zh ? item : { ...item, ...record(item.english) }; - const lines = Array.isArray(source.lines) ? source.lines.map(text).filter(Boolean).map((line) => `- ${line}`) : []; + // BUG-2038: technical lines read as on the page (two names, no machine tags, minutes as 分 / 秒). + const lines = Array.isArray(source.lines) + ? source.lines.map(text).filter(Boolean).map((line) => `- ${synastryTechnicalReaderText(line, input.names, input.language)}`) : []; if (!text(source.title) || lines.length === 0) continue; blocks.push(block(`table:${text(item.id)}`, "table", text(source.title), [lines.join("\n")])); } diff --git a/frontend/tests/reader-leftovers-2038.test.tsx b/frontend/tests/reader-leftovers-2038.test.tsx new file mode 100644 index 000000000..96121e405 --- /dev/null +++ b/frontend/tests/reader-leftovers-2038.test.tsx @@ -0,0 +1,180 @@ +// BUG-2038 (staging 复测第九轮, 2026-11-20): small reader-text leftovers in the personal report and +// the synastry appendix. Texts are the ones saved from that run. +import assert from "node:assert/strict"; +import { readFileSync } from "node:fs"; +import test from "node:test"; + +import { HEALTH_NOTICE_ZH, stripHealthReminders, withHealthNotice } from "../src/lib/personal-report-health-notice"; +import { acceptReportReadingDraft } from "../src/lib/personal-report-reading"; +import { REPORT_READING_POSTCHECK_VERSION } from "../src/lib/personal-report-reading-audit"; +import { + EVIDENCE_GRADE_CONTRACT, REPORT_READING_SECTIONS, type ReportReadingSectionId, +} from "../src/lib/personal-report-reading-contract"; +import { withSubjectOf } from "../src/lib/personal-report-reading-rebind"; +import { readerWording } from "../src/lib/personal-report-reading-wording"; +import { productContractBlock } from "../src/lib/product-runtime-contract"; +import { + synastryMinutesReaderText, synastrySoReaderText, synastryTechnicalReaderText, +} from "../src/lib/synastry-analysis-reader"; + +const SKILL = { name: "reading", version: "test", loaded: [{ path: "SKILL.md", sha256: "0".repeat(64) }] }; +const NEUTRAL = "本段复述虚构记录中的顺序、参与者位置以及需要核对的几条线索。读者可对照各行说明追溯原记录。".repeat(3); +const SUFFIX = "另一种情况下的结论还没有评估。"; + +function claim(cap: string, sections: string[], required: string[], checked: string[] = []) { + return { cap, sections, required: required.map((id) => ({ id })), checked: checked.map((id) => ({ id })), auxiliary: [] }; +} + +function facts(extra: Record = {}, boundary?: unknown) { + return { + ...extra, + evidence_grades: { + contract: EVIDENCE_GRADE_CONTRACT, + claims: { + character_structure: claim("C", ["character"], ["v:D1"], ["s:shadbala"]), + health_pressure: claim("C", ["health"], ["v:D1", "v:D6"], ["s:shadbala"]), + }, + dependency_catalog: { + "v:D1": { state: "clear_at_tolerance", aliases: ["本命盘"] }, + "v:D6": { state: boundary ? "boundary_sensitive" : "clear_at_tolerance", aliases: ["D6"] }, + "p:marriage:UL": { state: "executed", aliases: ["UL"] }, + "s:shadbala": { state: "executed", aliases: ["六力"] }, + }, + ...(boundary ? { boundary } : {}), + }, + }; +} + +function draft(overrides: Partial[]>>) { + return { sections: REPORT_READING_SECTIONS.map(({ id }) => ({ + id, paragraphs: [{ text: NEUTRAL, kind: "fact" }, { text: NEUTRAL, kind: "fact" }, ...(overrides[id] ?? [])], + })) }; +} + +function check(id: ReportReadingSectionId, paragraphs: Record[], evidence = facts()) { + const result = acceptReportReadingDraft(draft({ [id]: paragraphs }), evidence, SKILL); + assert.ok(result.ok, JSON.stringify(result.diagnostics)); + const section = result.reading.sections.find((entry) => entry.id === id)!; + return { section, stored: section.status === "ok" ? section.paragraphs.map(({ text }) => text) : [] }; +} + +test("a pronoun left without its planet takes the planet the removed sentence named", () => { + // Round nine, 性格 paragraph 3: 「它与落陷的木星互换了星座……」 after the sentence naming Saturn was removed. + assert.equal( + withSubjectOf("它与落陷的木星互换了星座——木星在土星掌管的摩羯座。", "土星落在第九宫双鱼座,UL 的位置由它照看。"), + "土星与落陷的木星互换了星座——木星在土星掌管的摩羯座。", + ); + assert.equal(withSubjectOf("这颗星的相位落在第三宫。", "月亮受土星相位,UL 在旁。"), "月亮的相位落在第三宫。"); + // No planet in the removed sentence, a plural, or no pronoun: left as written. + for (const sentence of ["它与落陷的木星互换了星座。", "它们一起落在第三宫。", "木星在第七宫。"]) { + assert.equal(withSubjectOf(sentence, sentence.startsWith("它与") ? "这一层没有取值。" : "土星落在第九宫。"), sentence); + } + const { stored } = check("character", [{ + text: "土星落在第九宫,UL 的位置由它照看。它与落陷的木星互换了星座,耐力够、成就感来得慢。", + claims: ["character_structure"], grade: "C", refs: ["v:D1"], + }]); + assert.equal(stored.at(-1), "土星与落陷的木星互换了星座,耐力够、成就感来得慢。"); +}); + +test("backstage notes stay out of the sections that give conclusions; 依据与边界 keeps them", () => { + assert.equal( + readerWording("另一套大运(Narayana)只作参考,它同样把金星放在主导位置,但这套结果还没有完成外部核对,不作为事件先后的依据。"), + "另一套大运(Narayana)只作参考,它同样把金星放在主导位置,不作为事件先后的依据。", + ); + assert.equal( + readerWording("观察日这一天的双行运只查了第7宫,没有激活;另一套大运(Narayana)只作参考,但不作为婚期依据。"), + "另一套大运(Narayana)只作参考,但不作为婚期依据。", + ); + assert.equal( + readerWording("月亮这颗上升主星落在处女座第3宫,入「极友」,六力值 6.0765 Rupa,同时受木星与土星相位。"), + "月亮这颗上升主星落在处女座第3宫,入「极友」,同时受木星与土星相位。", + ); + const status = "Double Transit 只查了第 7 宫,没有激活;月亮六力值 6.0765 Rupa,达到本方法的最低线。"; + assert.equal(readerWording(status, "basis"), status); + // A sentence that uses the value as its subject is not an aside and is left to the rule. + const subject = "月亮的六力值 6.0765 Rupa 高于本方法的最低线。"; + assert.equal(readerWording(subject), subject); +}); + +test("the health section closes with one fixed reminder; the model's own are taken out", () => { + // Round nine: two reminders in a row at the end of 健康. + assert.equal(stripHealthReminders("第六宫主落在第七宫,要多留意的是长期消耗。这些是结构上的条件,不是身体的结论。"), "第六宫主落在第七宫,要多留意的是长期消耗。"); + assert.equal(stripHealthReminders("以上是传统结构上的读法,不构成医疗判断;有具体不适要找医生。"), ""); + const content = "金星小运到 2027 年 5 月 15 日,这段时间作息容易乱,要多留意恢复。"; + assert.equal(stripHealthReminders(content), content); + assert.equal(withHealthNotice("要多留意恢复。", "", 650), `要多留意恢复。${HEALTH_NOTICE_ZH}`); + assert.equal(withHealthNotice(`要多留意恢复。出生时间前后几分钟,会影响这部分判断;${SUFFIX}`, `出生时间前后几分钟,会影响这部分判断;${SUFFIX}`, 650), + `要多留意恢复。${HEALTH_NOTICE_ZH}出生时间前后几分钟,会影响这部分判断;${SUFFIX}`); + assert.equal(withHealthNotice("长".repeat(640), "", 650), null); + + const { stored } = check("health", [ + { text: "第六宫主落在第七宫,长期消耗是要留意的地方。这些是结构上的条件,不是身体的结论。", claims: ["health_pressure"], grade: "C", refs: ["v:D1"] }, + { text: "以上是传统结构上的读法,不构成医疗判断;有具体不适要找医生。", kind: "fact" }, + ]); + assert.equal(stored.length, 3); + assert.equal(stored.at(-1), `第六宫主落在第七宫,长期消耗是要留意的地方。${HEALTH_NOTICE_ZH}`); + assert.equal(stored.join("").split("医生").length - 1, 1); + // Checked again (a stored reading, or a repair round), the sentence is not doubled. + const again = check("health", [{ text: stored.at(-1)!, claims: ["health_pressure"], grade: "C", refs: ["v:D1"] }]); + assert.equal(again.stored.at(-1), stored.at(-1)); + // A replay of a historical figure gives no current advice, so no reminder to see a doctor either. + const replay = check("health", [{ text: "第六宫主落在第七宫。", kind: "fact" }], facts({ report_mode: "historical_replay", current_advice_allowed: false })); + assert.ok(!replay.stored.join("").includes(HEALTH_NOTICE_ZH)); + // Other sections are untouched. + assert.ok(!check("character", [{ text: "本命盘的上升在巨蟹座。", kind: "fact" }]).stored.join("").includes(HEALTH_NOTICE_ZH)); +}); + +test("the other-scenario line stays last when the health reminder is added", () => { + const boundary = { + contract: "report_boundary_dependence_v1", other_side_line_zh: SUFFIX, + vargas: { D6: { kind: "sensitivity_range", label_zh: "D6", direction: "earlier", birth_time_seconds: 15 } }, + refs: { "v:D6": ["D6"] }, + }; + const { stored } = check("health", [ + { text: "D6 里第六宫的压力偏重,长期消耗是要留意的地方。", claims: ["health_pressure"], grade: "C", refs: ["v:D6"] }, + ], facts({}, boundary)); + assert.equal(stored.at(-1), `D6 里第六宫的压力偏重,长期消耗是要留意的地方。${HEALTH_NOTICE_ZH}出生时间前后几分钟,会影响这部分判断;${SUFFIX}`); +}); + +test("the rules say what the 「现在这几年」 section leads with and what stays out of the body", () => { + const contract = readFileSync(new URL("../../references/product_runtime_contract.md", import.meta.url), "utf8"); + const rules = productContractBlock("report-reading", { CLAIM_TYPES: "x" }, contract); + assert.match(rules, /现在这几年:先用两三句说当事人这几年处在什么阶段/); + assert.match(rules, /其他大运体系(Yogini、Kalachakra、Tribhagi、Shattimshatsama、Shastihayani、年盘的 Mudda 等)不在正文逐个列日期/); + assert.match(rules, /transit_natal_sign 只列回到出生时所在星座的行星/); + assert.match(rules, /没有列出的行星不写它「不在同一个星座」/); + assert.match(rules, /Shadbala 的 Rupa 数值不写进各节正文/); + assert.match(rules, /这一节的医疗提醒由系统在节末补一句固定的话/); + assert.match(rules, /它没有激活时,各节正文不提它/); + assert.match(rules, /不写它「还没有完成外部核对」/); + assert.match(rules, /这一点目前还不能明确判断」,不写「候选」「筛出」「检出」/); + assert.equal(REPORT_READING_POSTCHECK_VERSION, 11); +}); + +test("synastry appendix: the 「所以」 label is not doubled", () => { + const names = { a: "岳辰", b: "林瑶" }; + // Round nine: 「所以:所以岳辰在关系里要的是可靠……」 ×20. + assert.equal(synastrySoReaderText("所以甲在关系里要的是可靠、能一起规划生活的伴侣。", names, "zh"), "岳辰在关系里要的是可靠、能一起规划生活的伴侣。"); + assert.equal(synastrySoReaderText("所以,乙把喜欢放在明面上。", names, "zh"), "林瑶把喜欢放在明面上。"); + assert.equal(synastrySoReaderText("甲更看重稳定,所以节奏慢。", names, "zh"), "岳辰更看重稳定,所以节奏慢。"); + assert.equal(synastrySoReaderText("So they settle slowly.", names, "en"), "they settle slowly."); + assert.equal(synastrySoReaderText("So that both can plan, they talk first.", names, "en"), "So that both can plan, they talk first."); +}); + +test("synastry technical summary: names, no machine tags, minutes as 分 and 秒", () => { + const names = { a: "岳辰", b: "林瑶" }; + assert.equal( + synastryTechnicalReaderText("乙方的月亮落甲方的第4宫(person_b_to_person_a,recipient_house 4);第4宫关联熟悉感。", names, "zh"), + "林瑶的月亮落岳辰的第4宫;第4宫关联熟悉感。", + ); + assert.equal(synastryTechnicalReaderText("甲的七宫主 土星 落第 9 宫 双鱼座", names, "zh"), "岳辰的七宫主 土星 落第 9 宫 双鱼座"); + assert.equal( + synastryTechnicalReaderText("甲方按生时纠正选定的区间(前 6.5 分钟、后 1.7 分钟)", names, "zh"), + "岳辰按生时纠正选定的区间(前 6 分 30 秒、后 1 分 42 秒)", + ); + assert.equal(synastryMinutesReaderText("偏晚 2.23 分钟:天蝎座"), "偏晚 2 分 14 秒:天蝎座"); + assert.equal(synastryMinutesReaderText("偏早 0.5 分钟"), "偏早 30 秒"); + assert.equal(synastryMinutesReaderText("偏早 3.0 分钟;按前后各 30 分钟试算"), "偏早 3 分钟;按前后各 30 分钟试算"); + // English stays as stored apart from the machine tag. + assert.equal(synastryTechnicalReaderText("Selected range (6.5 minutes before)", names, "en"), "Selected range (6.5 minutes before)"); +}); diff --git a/frontend/tests/report-leftovers-2021.test.ts b/frontend/tests/report-leftovers-2021.test.ts index ebeab2e42..a47aed90e 100644 --- a/frontend/tests/report-leftovers-2021.test.ts +++ b/frontend/tests/report-leftovers-2021.test.ts @@ -186,13 +186,19 @@ test("the page's fact tables name a known planet from the product's table, not t test("the report rules ask for the natal-sign return and keep status words out of the body", () => { const contract = readFileSync(new URL("../../references/product_runtime_contract.md", import.meta.url), "utf8"); const block = contract.slice(contract.indexOf(""), contract.indexOf("")); - assert.match(block, /transit_natal_sign 里 same_sign 为 true 的行,要写明这颗星回到了出生时它所在的星座/); + // 原值: /transit_natal_sign 里 same_sign 为 true 的行,要写明这颗星回到了出生时它所在的星座/ + // 新值: 事实里只列回到本命星座的行星,规则照此写 + // 原因: BUG-2038,复测第九轮正文把不同座的一行写成「不在同一个星座」 + assert.match(block, /transit_natal_sign 只列回到出生时所在星座的行星:列出的要写明这颗星回到了出生时它所在的星座/); // 原值: /saturn_from_moon 写出行运土星在本命月亮起第几宫/(规则让模型「同一句里照 saturn_from_moon 写出」) // 新值: 规则指向事实包里成句的 summary_zh // 原因: BUG-2024,复测第六轮正文把「同一句要并读」抄了出来;怎么并列由程序写成事实句 assert.match(block, /用 transits\.sade_sati\.summary_zh 给出的事实/); assert.match(block, /行运土星在本命月亮起第几宫、是否属于 Kantaka 或 Ashtama Shani/); - assert.match(block, /state 为 candidate 的只说它是按规则筛出的候选/); + // 原值: /state 为 candidate 的只说它是按规则筛出的候选/ + // 新值: 候选格局各节正文不拿它下判断,提到时用日常话 + // 原因: BUG-2038(占星师 2026-11-19 分层答复:读者看到的是「这一点目前还不能明确判断」);复测第九轮正文写「按规则筛选出来的候选里……被检出」 + assert.match(block, /state 为 candidate 的不说成已经形成,各节正文不拿它下判断/); assert.doesNotMatch(block, /现实里是否应验另说;/); assert.doesNotMatch(block, /参数敏感的层/); }); diff --git a/frontend/tests/report-reading-20261027.test.tsx b/frontend/tests/report-reading-20261027.test.tsx index c002fa925..606bb7957 100644 --- a/frontend/tests/report-reading-20261027.test.tsx +++ b/frontend/tests/report-reading-20261027.test.tsx @@ -138,7 +138,11 @@ test("a section with field names, fate, disease, investment or stacked disclaime ["love:banned_phrase", draft({ love: [`${PARAGRAPH}明年一定会结婚。`] })], // #192 eval 2026-11-12: the draft wrote that web and case calibration were not done (BUG-1794: backstage only). ["basis:banned_phrase", draft({ basis: [`${PARAGRAPH}全球网络资料采集与真实案例校准在本次都未完成。`] })], - ["health:disclaimer", draft({ health: [`${PARAGRAPH}仅供参考。`, `${PARAGRAPH}不构成医疗意见。`] })], + // 原值: ["health:disclaimer", draft({ health: [`${PARAGRAPH}仅供参考。`, `${PARAGRAPH}不构成医疗意见。`] })] + // 新值: 同一条检查放在事业一节 + // 原因: BUG-2038,健康一节里模型自己写的医疗提醒由程序去掉、节末补一句固定的话,不再因此扣下整节; + // 其他各节叠两句免责仍然扣下 + ["career:disclaimer", draft({ career: [`${PARAGRAPH}仅供参考。`, `${PARAGRAPH}不能替代专业判断。`] })], // 原值:PARAGRAPH.repeat(5)(690 字,超过旧上限 420);新值:repeat(6)(828 字,超过新上限 700) // 原因:产品 2026-11-07 占星师方案:段落上限随专业写法放宽 ["career:too_long", draft({ career: [PARAGRAPH.repeat(6)] })], diff --git a/frontend/tests/report-reading-claim-binding-20261112.test.ts b/frontend/tests/report-reading-claim-binding-20261112.test.ts index 1399339b4..ec5e9fed1 100644 --- a/frontend/tests/report-reading-claim-binding-20261112.test.ts +++ b/frontend/tests/report-reading-claim-binding-20261112.test.ts @@ -10,6 +10,7 @@ import { type ReportReadingSectionId, } from "../src/lib/personal-report-reading-contract"; import { buildReportExportBlocks } from "../src/lib/report-export-blocks"; +import { HEALTH_NOTICE_ZH } from "../src/lib/personal-report-health-notice"; const SKILL = { name: "reading", version: "test", loaded: [{ path: "SKILL.md", sha256: "0".repeat(64) }] }; const NEUTRAL = "本段复述虚构记录中的顺序、参与者位置以及需要核对的几条线索。读者可对照各行说明追溯原记录。".repeat(3); @@ -199,7 +200,12 @@ test("rebind adds a permitted dependency and timing claim, converts a fact, and assert.ok(sections.every(({ status }) => status === "ok")); for (const section of sections) { assert.ok(section.status === "ok"); - assert.deepEqual(section.paragraphs.map(({ text }) => text), body.sections.find(({ id }) => id === section.id)!.paragraphs.map(({ text }) => text)); + // 原值: 各节保存的正文与草稿逐段相同 + // 新值: 健康一节最后一段末尾多一句程序补的医疗提醒 + // 原因: BUG-2038,健康一节的提醒由程序补一句固定的话,模型不写 + const written = body.sections.find(({ id }) => id === section.id)!.paragraphs.map(({ text }) => text); + assert.deepEqual(section.paragraphs.map(({ text }) => text), + section.id === "health" ? [...written.slice(0, -1), `${written.at(-1)}${HEALTH_NOTICE_ZH}`] : written); } assert.deepEqual(log.auto_rebound?.map(({ section, rule, added_claims, grade_from, grade_to }) => [section, rule, added_claims, grade_from, grade_to]), [ diff --git a/frontend/tests/report-reading-snapshot-20261113.test.ts b/frontend/tests/report-reading-snapshot-20261113.test.ts index 855fef491..29aa7f2e9 100644 --- a/frontend/tests/report-reading-snapshot-20261113.test.ts +++ b/frontend/tests/report-reading-snapshot-20261113.test.ts @@ -104,7 +104,7 @@ test("the stored snapshot binds the selected raw response and records a failed r system_prompt_sha256: reportReadingSha256("原始提示"), raw_response_sha256: canonicalReportReadingSha256(rawResponse), stored_sha256: canonicalReportReadingSha256(stored), - rules: { evidence_grade_contract: EVIDENCE_GRADE_CONTRACT, postcheck_version: 10, git_commit: COMMIT }, // 原值 9;新值 10;原因: BUG-2034 正文写法收尾(建议句后的说明、日期写法、窗口级引子);此前各次见 personal-report-reading-audit.ts + rules: { evidence_grade_contract: EVIDENCE_GRADE_CONTRACT, postcheck_version: 11, git_commit: COMMIT }, // 原值 9;新值 10;原因: BUG-2034 正文写法收尾(建议句后的说明、日期写法、窗口级引子);此前各次见 personal-report-reading-audit.ts;BUG-2038 再由 10 改为 11 }); assert.equal(stored.transform_log.selected_attempt, 1); assert.equal(stored.transform_log.retries, 1); diff --git a/frontend/tests/report-user-layer-2028.test.tsx b/frontend/tests/report-user-layer-2028.test.tsx index 24b28ffd8..89a90ba7d 100644 --- a/frontend/tests/report-user-layer-2028.test.tsx +++ b/frontend/tests/report-user-layer-2028.test.tsx @@ -53,5 +53,6 @@ test("developer words the reading picked up are put in reader words before the c assert.equal(readerWording("这条具名规则成立,但证据链还不完整,两种方法版本结论不同。"), "这条规则成立,但依据还不完整,两种算法结论不同。"); assert.equal(readerWording("木星落陷,不过不利影响有缓解。"), "木星落陷,不过不利影响有缓解。"); // 原值 8;新值 9;原因: BUG-2028 - assert.equal(REPORT_READING_POSTCHECK_VERSION, 10); // 原值 9;新值 10;原因: BUG-2034 正文写法收尾(建议句后的说明、日期写法、窗口级引子) + // 原值 9;新值 10;原因: BUG-2034 正文写法收尾(建议句后的说明、日期写法、窗口级引子) + assert.equal(REPORT_READING_POSTCHECK_VERSION, 11); // 原值 10;新值 11;原因: BUG-2038 代词补主语、健康提醒由程序补、后台说明不进各节正文 }); diff --git a/references/product_runtime_contract.md b/references/product_runtime_contract.md index ce8db04ee..563f97c2b 100644 --- a/references/product_runtime_contract.md +++ b/references/product_runtime_contract.md @@ -293,6 +293,8 @@ evidence_card.backstage records, for the backend only, that global web evidence + + 你在给一份个人星盘报告写开头的「解读」。用户消息里的盘面事实(report-reading-facts-v3)是引擎已经算好的全部材料,只用这些,不编数据,不另起算法,不凭印象重算尊贵度、宫位或日期。事实里没有的层就当没有。 @@ -301,10 +303,10 @@ evidence_card.backstage records, for the backend only, that global web evidence 九节固定,按顺序:总评、性格、事业、财富、感情、健康、现在这几年、几条建议、依据与边界。 - 总评:先用三到六句回答这张盘最突出的地方,再简短给依据。原始数值和逐项技法放在后面的计算依据里,正文不必每段先列一长串依据,但用到的数值必须照事实原值写。 -- 性格:只写静态结构(上升、月亮、上升主等),不要求每项性格都配大运。 +- 性格:只写静态结构(上升、月亮、上升主等),不要求每项性格都配大运。Shadbala 的 Rupa 数值不写进各节正文(数值在后面的计算依据里)。 - 事业、财富、感情、健康:先结论,再简短依据。讲时间时按「本命承诺 → 激活 → 落地 → 应期」推理;落地的依据不足时,只写条件窗口,并把条件直接写进这句判断里(例如「这段时间更有利于争取岗位调整,是否形成实际变动还取决于……」),不在句前句后另加「应期只到窗口级」「只到窗口级」「不写成已经落地」这类说明。财富分开写「能不能挣到」和「能不能留下」。感情先分清是心动、成对还是法律婚,不把所有关系需求都当成婚姻。 -- 健康:只写有依据的传统条件和要多留意的时段,不从 6、8、12 宫直接推出免疫、睡眠这类具体身体事实,不写病名。D30 能不能用看 birth_precision 和 evidence_grades,不只看 minute_confirmed。 -- 现在这几年:写明实际覆盖哪几年。行运写成「observation_date 这一天的行运位置」,只当作那一天的位置来写,不由它推出几个月或几年的结论,也不在正文里解释这条限制;时间粒度由证据决定。transits.transit_natal_sign 里 same_sign 为 true 的行,要写明这颗星回到了出生时它所在的星座,并照写 end 给出的离开日期(已有的换座记录;end 为 not_available 就不写日期),只陈述位置和日期,不由此另加推断。写到 Sade Sati 时,用 transits.sade_sati.summary_zh 给出的事实(Sade Sati 是否启动,以及行运土星在本命月亮起第几宫、是否属于 Kantaka 或 Ashtama Shani、哪一天离开),不只写「未启动」;把它写进自己的句子里,不整句照抄,也不用「Sade Sati 的说法是」这类引子,同一件事全篇说一次就够。正文里的日期一律写成「2027 年 5 月 20 日」这种形式。 +- 健康:只写有依据的传统条件和要多留意的时段,不从 6、8、12 宫直接推出免疫、睡眠这类具体身体事实,不写病名。这一节的医疗提醒由系统在节末补一句固定的话,正文不要自己写「不构成医疗判断」「不是身体的结论」「有不适要找医生」这类提醒。D30 能不能用看 birth_precision 和 evidence_grades,不只看 minute_confirmed。 +- 现在这几年:先用两三句说当事人这几年处在什么阶段、主题落在生活的哪一块(主运 Vimshottari 当前大运、小运的主星在这张盘上管什么),再写明实际覆盖哪几年。正文里的日期只给 Vimshottari 的大运、小运和接下来一两个小运;Narayana 至多一句并说明只作参考;其他大运体系(Yogini、Kalachakra、Tribhagi、Shattimshatsama、Shastihayani、年盘的 Mudda 等)不在正文逐个列日期,它们在后面的计算依据里。行运写成「observation_date 这一天的行运位置」,只写正文用到的那几颗星,不逐颗罗列度数;只当作那一天的位置来写,不由它推出几个月或几年的结论,也不在正文里解释这条限制;时间粒度由证据决定。transits.transit_natal_sign 只列回到出生时所在星座的行星:列出的要写明这颗星回到了出生时它所在的星座,并照写 end 给出的离开日期(已有的换座记录;end 为 not_available 就不写日期),只陈述位置和日期,不由此另加推断;没有列出的行星不写它「不在同一个星座」。写到 Sade Sati 时,用 transits.sade_sati.summary_zh 给出的事实(Sade Sati 是否启动,以及行运土星在本命月亮起第几宫、是否属于 Kantaka 或 Ashtama Shani、哪一天离开),不只写「未启动」;把它写进自己的句子里,不整句照抄,也不用「Sade Sati 的说法是」这类引子,同一件事全篇说一次就够。正文里的日期一律写成「2027 年 5 月 20 日」这种形式。 - 几条建议:两到四条,每条只能落在前面已经有依据的主题上,直接写要做什么、为什么值得做;前面没有依据的一般建议不写。不在句后说明这条建议的来历或性质(是不是从星盘来的、算不算一般建议、具体还要看现实条件、不宜下确定结论这类话都不写;依据在文末的「逐条依据」里);不从数值门槛推出具体作息时刻、贷款或资产安排。 - 依据与边界:功能吉星、功能凶星及与自然吉凶的冲突;实际用到的技法;没有使用的层及原因(按 participation 的 state 和 reason_code 写:not_routed 是这一层放在别的章节里使用,这里不重复,正文写成「已放在某某部分使用」,不写「路由」;not_in_domain_route 是占星师这一领域的必查模块没有列它,带 computed 的已经算出数值,要写「已算出,本报告不用它下结论」,不能说参考资料里没有、没有来源或没有计算;not_computed 才是没有计算);只作参考的层(事实里 state 为 parameter_sensitive 的层,正文写成「只作参考」,不写「参数敏感」);出生时间精度与 birth_precision 里接近边界的分盘;岁差、交点、宫制。 各层的执行状态(已算出但不用、没有取值、内容为空、没有路由、没有计算)只写在「依据与边界」一节;其他各节只写用上的依据和结论,不叙述某一层有没有算、有没有取值。 @@ -313,11 +315,11 @@ evidence_card.backstage records, for the backend only, that global web evidence 九节逐段区分判断和纯事实。判断段填写 claims 和 grade,claims 至少一类。evidence_grades.claims 列出各类的上限 cap(A 最高,D 最低)及 required、checked。混合多类判断可拆段,也可合写并取最严格的上限。refs 可选,列出主要依据即可,只能从所列各类的 required 和 checked 并集中取。正文点名的层由系统按 dependency_catalog 自动绑定到这些结论允许的依赖,并计入等级判断。引用的依赖缺失或未评估,等级只能是 D。evidence_grades.boundary.vargas 列出在另一种出生情景下上升星座会不同的分盘(本命盘 D1 不在其中,D1 跨边界时相关结论仍是 D):这不让整节停写。不用这些分盘的判断照常写、照常分级;用到其中某个分盘的上升、宫位或宫主的判断,按当前这一侧的盘面写出结论,等级不因此降为 D,段末由系统补一句出生时间的说明(「出生时间前后几分钟,会影响这部分判断;另一种情况下的结论还没有评估」),正文不要自己再写这句,也不要把整节写成暂不能判断。说到离换座点多近时,照 birth_precision 里该分盘 summary_zh 的写法(出生时间提前或推后几分几秒),不写带小数的分钟,也不写成角度的分。分盘代号不加书名号。判断段提到四位年份、数字连写的月份、大运、小运、次小运或窗口时,claims 必须含 timing_window。总评和建议只能引用本稿性格、事业、财富、感情、健康、现在这几年六节已经分级的判断,沿用其中最严格的等级。只复述盘面事实、交代方法或给一般操作说明时,填 kind 为 fact,正文不夹带「意味着」「适合」等推断,省略 claims、grade 和 refs。 等级含义:C 是有星盘来源和可计算的依据、还没有经过案例校准;D 是这条结论需要的关键依据缺失或无法判别。D 的段落用日常话写「这一点目前还不能明确判断」,再说缺了什么,不写结果。不要在正文里写「置信度:」或「证据等级:」,也不写 A、B、C、D 这些字母;等级只在后台保存,不给读者看。must_disclose 里的缺口和只作参考的层,在「依据与边界」一节各说一次;其中某一项会改变某条结论的方向或时间时,在那条结论的句子里把这个条件说出来,其他段落不重复。只有某类结论的必需依据全部可用时,才能说它「没有关键阻塞」。 -两套大运(Vimshottari、Narayana)的主星或星座不同,本身不是冲突;同名主星也不自动算一致。只有对应同一件事、同一段时间、同一个判断方向时,才比较是否一致。KP 与整宫制结果不同,写成方法差异。Double Transit 只对 checked_houses 里的宫位有结论,不推到别的宫或别的领域。 +两套大运(Vimshottari、Narayana)的主星或星座不同,本身不是冲突;同名主星也不自动算一致。只有对应同一件事、同一段时间、同一个判断方向时,才比较是否一致。KP 与整宫制结果不同,写成方法差异。Double Transit 只对 checked_houses 里的宫位有结论,不推到别的宫或别的领域;它没有激活时,各节正文不提它,也不写「只查了第几宫」(这属于执行状态,只在「依据与边界」说)。另一套大运只作参考时,写「只作参考」就够,不写它「还没有完成外部核对」。 -格局按 yogas 每一行的形成状态写:满足了来源的形成条件就可以说形成,不把未经现实验证写成未形成;state 为 candidate 的只说它是按规则筛出的候选,不说成已经形成,也不再加「本次只按候选记录,不作结论」这类说明。状态原词(「命中候选」「尚未完整确认」)不照抄进正文,句后也不另加「现实是否应验另说」这类说明。尊贵度照 dignity_zh 原词写(入旺、入庙、落陷……),不自己翻译。dignity_zh 以「落陷;这一配置的不利影响有缓解因素」开头的行星:用自己的话讲清它落陷、不利影响有缓解、但困难不会因此全部消失,结合它所在的宫位说这对当事人意味着什么;不写「落陷取消」「取消条件已满足」,也不整句照抄这条状态。 +格局按 yogas 每一行的形成状态写:满足了来源的形成条件就可以说形成,不把未经现实验证写成未形成;state 为 candidate 的不说成已经形成,各节正文不拿它下判断;要提到时用日常话,例如「盘里有某某格局的迹象,这一点目前还不能明确判断」,不写「候选」「筛出」「检出」,也不再加「本次只按候选记录,不作结论」这类说明。状态原词(「命中候选」「尚未完整确认」)不照抄进正文,句后也不另加「现实是否应验另说」这类说明。尊贵度照 dignity_zh 原词写(入旺、入庙、落陷……),不自己翻译。dignity_zh 以「落陷;这一配置的不利影响有缓解因素」开头的行星:用自己的话讲清它落陷、不利影响有缓解、但困难不会因此全部消失,结合它所在的宫位说这对当事人意味着什么;不写「落陷取消」「取消条件已满足」,也不整句照抄这条状态。 -可以写技法名和分盘代号(Vimshottari、Narayana、D10、A10、UL、AK、AmK、DK、KP、Shadbala、Ashtakavarga、Sade Sati),第一次出现时在括号里用半句中文说明它看什么。行星、星座用中文;交点只叫罗睺、计都。不写字段名、下划线、花括号、井号、代码,不写「本仓」「检测器」「字段」「口径」「参数敏感」「双体系确认」「具名规则」「证据链」「方法版本」「候选分支」这类内部说法。审计行里网上检索和案例校准这两行只在后台记录(产品 2026-11-08 决定,BUG-1794),正文不写它们有没有做;等级不给读者看。不写病名、诊断和身体部位的疾病;不给投资指令,不点名股票、基金、币;不写「注定」「先苦后甜」「宿命」「一定会」;不预测死亡、必然离婚、怀孕与孩子性别、某天必然发财;婚姻、生死不下确定结论。每节最多一句提醒,不逐句免责。 +可以写技法名和分盘代号(Vimshottari、Narayana、D10、A10、UL、AK、AmK、DK、KP、Shadbala、Ashtakavarga、Sade Sati),第一次出现时在括号里用半句中文说明它看什么。行星、星座用中文;交点只叫罗睺、计都。不写字段名、下划线、花括号、井号、代码,不写「本仓」「检测器」「字段」「口径」「参数敏感」「双体系确认」「具名规则」「证据链」「方法版本」「候选分支」这类内部说法。审计行里网上检索和案例校准这两行只在后台记录(产品 2026-11-08 决定,BUG-1794),正文不写它们有没有做;等级不给读者看。不写病名、诊断和身体部位的疾病;不给投资指令,不点名股票、基金、币;不写「注定」「先苦后甜」「宿命」「一定会」;不预测死亡、必然离婚、怀孕与孩子性别、某天必然发财;婚姻、生死不下确定结论。每节最多一句提醒,不逐句免责;健康一节的提醒由系统补,正文不写。 每节一到六段,每段不超过六百五十字,全篇三千到一万五千字。只输出 JSON,顶层为 {"sections":[…]}。九节的 id 依次是 overview、character、career、wealth、love、health、now、advice、basis,各节的 paragraphs 按段选择一种形态,例如 {"text":"水星落第五宫。","kind":"fact"} 或 {"text":"本命盘显示……","claims":["character_structure"],"grade":"C","refs":["v:D1"]}。claims 的每项只能取 {{CLAIM_TYPES}} 之一,grade 只能取 A、B、C、D。 收到章节修改请求时,只按退回原因与命中片段重写指定章节。沿用同一份事实包、段落标注和证据等级规则。仍用 `sections` 数组,按请求顺序返回指定章节,各返回一次。 @@ -756,3 +758,17 @@ evidence_card.backstage records, for the backend only, that global web evidence - 「逐条依据」:每行不列 `v:` 开头的上升稳定性检查(每类结论都要过,尤其 `v:D1`);只引用了它的段落写「依据:本命盘」。受影响的段落仍标「受出生时间前后几分钟影响」。等级计算和存下来的引用不变。 - 年运报告:评级项下面不再有「证据等级 X:……」一行,只列原因(「这一点目前还不能明确判断:……」),没有原因就不显示;附录「证据等级上限」一块在页面和导出里叫「各类结论的依据情况」,每行写「所需依据已算出」或「这一点目前还不能明确判断」。快照里的等级、原因和附录原行不变。 - 合盘报告:正文和导出里没有等级行,本条没有改动;「分析逻辑」显示时把「甲」「乙」换成两个人的称呼并去掉带下划线机器名的括注(存下来的文字不变,写作核对仍按原文比对);技术附录里「研究扫描 v1」写成「探索性扫描(版本 v1)」,「随参数变化」写成「只作参考」。 + +## AR. 读者正文的几处收尾(产品 2026-11-20,BUG-2038) + +依据:占星师 2026-11-19 分层答复;产品负责人 2026-11-16「不写推责的话」;测试环境复测第九轮。 + +- 个人报告:程序按句去掉越界引用(§AM)之后,紧跟的句子若以「它」「这颗星」「该星」开头,换成被删句最先点名的行星;被删句没有点名行星时不动。 +- 报告事实 `transits.transit_natal_sign` 只列回到本命星座的行星;不同座的不列,正文也不写「不在同一个星座」(规则 R-03,公告 G-7)。 +- 「现在这几年」先说处在什么阶段;正文日期只给 Vimshottari,Narayana 至多一句,其他大运体系与逐颗行运度数留在后面的计算依据(规则 R-03)。 +- 中文开篇时间线的「Narayana 当前大运」不带算法标签;标签在 Narayana 章节和附录。英文版不变。 +- 各节正文不写:外部核对有没有做、没有激活的 Double Transit、Shadbala 的 Rupa 数值(规则 R-03、R-05;残留由程序去掉)。「依据与边界」一节不受这条影响。 +- 候选格局:各节正文不拿它下判断,提到时用日常话「这一点目前还不能明确判断」(规则 R-06)。数据章节里的状态文字不变。 +- 健康一节:模型不写医疗提醒,程序在节末补一句「这一节不构成医疗建议,身体有具体不适请找医生。」;历史人物回放不补。 +- 合盘显示:「所以:」后面不重复「所以」;「技术依据摘要」与「分析逻辑」同样处理(两个人的称呼、去掉带下划线机器名的括注),带小数的分钟写成几分几秒。保存的文字不变。 + diff --git a/scripts/report_render/opening_guide.py b/scripts/report_render/opening_guide.py index c17c0e19e..17043854d 100644 --- a/scripts/report_render/opening_guide.py +++ b/scripts/report_render/opening_guide.py @@ -183,8 +183,11 @@ def timeline(packet: dict, language: str) -> str: if md.get("sign"): label = narayana.get("algorithm_label" if zh else "algorithm_label_en") heading = "Narayana 当前大运" if zh else "Current Narayana period" - if label: - heading += f"({label})" if zh else f" ({label})" + # BUG-2038 原值:中文开篇也带模块给出的算法标签(版本与出处,写在括号里)。 + # 新值:中文开篇不带;标签留在后面的 Narayana 章节和附录。英文版不变。 + # 原因:复测第九轮;开篇是读者正文,版本出处是后台信息(占星师 2026-11-19 分层答复)。 + if label and not zh: + heading += f" ({label})" current_periods = [(heading, md)] if ad.get("sign") and current.get("ad_status") != "blocked": current_periods.append(("小运" if zh else "Antardasha", ad)) diff --git a/scripts/report_render/reading_facts.py b/scripts/report_render/reading_facts.py index 9dfb95e09..b4d1ab37c 100644 --- a/scripts/report_render/reading_facts.py +++ b/scripts/report_render/reading_facts.py @@ -737,8 +737,12 @@ def _transits(packet: dict, full_reading: dict | None = None) -> dict[str, Any]: returns = transit_natal_sign_facts( dig(packet, "core_chart.planets"), out.get("planets"), key_dates, out.get("target_date"), ) - if returns: - out["transit_natal_sign"] = returns + # BUG-2038 原值:两颗星各一行,same_sign 为 false 的也给。新值:只给回到本命星座的行。 + # 原因:复测第九轮正文写出「行运木星从出生时的摩羯座走到狮子座,不在同一个星座」这种没有内容的话; + # 问答的关键事实行同样只列 same_sign 为 true 的(BUG-2006)。`returns` 仍完整,用于下面取土星离开日期。 + back = [row for row in returns if row.get("same_sign") is True] + if back: + out["transit_natal_sign"] = back sade = dig(packet, SADE_SATI) if isinstance(sade, dict): out["sade_sati"] = _pick(sade, ("is_active", "phase", "moon_sign", "phases")) diff --git a/tests/test_reader_leftovers_2038.py b/tests/test_reader_leftovers_2038.py new file mode 100644 index 000000000..94bab8a01 --- /dev/null +++ b/tests/test_reader_leftovers_2038.py @@ -0,0 +1,48 @@ +"""BUG-2038 (staging 复测第九轮, 2026-11-20): reader-text leftovers on the Python side of the personal report.""" + +from __future__ import annotations + +from scripts.report_render import reading_facts +from scripts.report_render.opening_guide import timeline + + +def _packet(saturn_sign: str) -> dict: + timing = { + "transit_positions": {"target_date": "2026-11-19", "planets": { + "Saturn": {"sign": saturn_sign, "degree_in_sign": 15.6}, "Jupiter": {"sign": "Leo", "degree_in_sign": 3.3}}}, + "transit_multi_reference": {"transit_analysis": {"Saturn": {"house_from_ref": {"Chandra_Lagna": {"house": 7}}}}}, + } + return { + "core_chart": {"planets": {"Saturn": {"sign": "Pisces"}, "Jupiter": {"sign": "Capricorn"}}}, + "worksheets": { + "timing_and_predictive_systems": timing, + "advanced_systems": {"yogas_doshas": {"sade_sati": {"is_active": False, "phase": "none", "moon_sign": "Virgo"}}}, + }, + } + + +def test_only_a_planet_back_in_its_natal_sign_is_given_to_the_reading() -> None: + # 复测第九轮正文:「行运木星从出生时的摩羯座走到狮子座,不在同一个星座。」 + rows = reading_facts._transits(_packet("Pisces"), {})["transit_natal_sign"] + assert [row["planet"] for row in rows] == ["Saturn"] and rows[0]["same_sign"] is True + away = reading_facts._transits(_packet("Aries"), {}) + assert "transit_natal_sign" not in away + # Saturn's house from the Moon is still stated when it is not in its natal sign. + assert away["sade_sati"]["saturn_from_moon"]["house"] == 7 + + +def _narayana_packet() -> dict: + return {"worksheets": {"timing_and_predictive_systems": {"narayana_dasha": { + "algorithm_label": "Sanjay Rath 版(默认;按原书 p.43 正文(第七轮裁定))", + "algorithm_label_en": "Sanjay Rath version (default)", + "current_dasha": {"md": {"sign": "Libra", "start": "2018-08-08", "end": "2028-08-07", "start_age": 21.0, "end_age": 31.0}}, + }}}} + + +def test_the_chinese_opening_names_the_narayana_period_without_the_profile_label() -> None: + # 复测第九轮开篇:「Narayana 当前大运(Sanjay Rath 版(默认;按原书 p.43 正文(第七轮裁定))):天秤座 …」 + zh = timeline(_narayana_packet(), "zh") + assert "Narayana 当前大运:天秤座" in zh + assert "Sanjay Rath" not in zh and "第七轮裁定" not in zh + # The English opening is unchanged. + assert "Current Narayana period (Sanjay Rath version (default)): " in timeline(_narayana_packet(), "en") diff --git a/tests/test_report_leftovers_2021.py b/tests/test_report_leftovers_2021.py index 72d06c8d5..1576a7bc3 100644 --- a/tests/test_report_leftovers_2021.py +++ b/tests/test_report_leftovers_2021.py @@ -101,8 +101,11 @@ def _packet(saturn_sign: str, house_from_moon: int) -> dict: def test_report_facts_name_the_return_and_saturn_from_the_moon_next_to_sade_sati(monkeypatch) -> None: monkeypatch.setattr(reading_facts, "slow_transit_key_dates", lambda packet, full, day: INGRESS) transits = reading_facts._transits(_packet("Pisces", 7), {}) - saturn, jupiter = transits["transit_natal_sign"] - assert saturn["same_sign"] is True and saturn["end"] == "2027-05-20" and jupiter["same_sign"] is False + # 原值: 土星、木星各一行,木星 same_sign 为 False 也在事实里 + # 新值: 只列回到本命星座的行(这里只有土星) + # 原因: BUG-2038,复测第九轮正文把不同座的那一行写成「不在同一个星座」,没有内容 + (saturn,) = transits["transit_natal_sign"] + assert saturn["same_sign"] is True and saturn["end"] == "2027-05-20" sade = transits["sade_sati"] # 原值:只有 is_active / phase / moon_sign / phases,正文写成「Sade Sati 未启动」一句 # 新值:同一条里带土星在月亮起第几宫、是否 Kantaka、到哪一天 @@ -122,7 +125,8 @@ def test_saturn_outside_its_natal_sign_still_gets_its_house_from_the_moon_and_le rows = [{"kind": "ingress", "date": "2029-08-08", "body": "saturn", "from_sign": "Aries", "to_sign": "Taurus"}] monkeypatch.setattr(reading_facts, "slow_transit_key_dates", lambda packet, full, day: rows) transits = reading_facts._transits(_packet("Aries", 8), {}) - assert transits["transit_natal_sign"][0]["same_sign"] is False + # 原值: transits["transit_natal_sign"][0]["same_sign"] is False;新值: 不同座时没有这个键;原因: BUG-2038(同上) + assert "transit_natal_sign" not in transits from_moon = transits["sade_sati"]["saturn_from_moon"] assert (from_moon["house"], from_moon["kantaka"], from_moon["ashtama_shani"], from_moon["until"]) == (8, False, True, "2029-08-08") diff --git a/tests/test_report_narayana_shadbala_20261111.py b/tests/test_report_narayana_shadbala_20261111.py index 6fc28ea90..f5364edaf 100644 --- a/tests/test_report_narayana_shadbala_20261111.py +++ b/tests/test_report_narayana_shadbala_20261111.py @@ -76,7 +76,12 @@ def test_current_periods_keep_module_label_and_exact_engine_dates(case_id): for language, report, label in (("zh", zh, "algorithm_label"), ("en", en, "algorithm_label_en")): opening = timeline(packet, language) assert opening in report - assert module[label] in opening + # 原值: 中英文开篇都带模块的算法标签(assert module[label] in opening) + # 新值: 中文开篇不带,标签仍在报告后面的 Narayana 章节;英文开篇照旧 + # 原因: BUG-2038,复测第九轮;开篇是读者正文,算法版本与出处是后台信息(占星师 2026-11-19 分层答复) + assert module[label] in (report if language == "zh" else opening) + if language == "zh": + assert module[label] not in opening audit = _dual_dasha_note(packet["worksheets"][TIMING], language) difference = next(row[1] for row in _difference_rows(packet, language) if row[0] == "Narayana Dasha") for key in ("md", "ad"): diff --git a/tests/test_report_opening_supplement.py b/tests/test_report_opening_supplement.py index cebc52447..1899b56b8 100644 --- a/tests/test_report_opening_supplement.py +++ b/tests/test_report_opening_supplement.py @@ -77,7 +77,12 @@ def test_supplement_has_guide_timeline_and_domain_tables() -> None: assert "MD" not in text.split("### 时间线")[0] assert "- 土星:2027-03-28 → 2027-09-28" in text assert "- 水星:2027-09-28 → 2028-03-09" in text - assert "Narayana 当前大运(任务包算法):巨蟹座 2022-08-01 → 2031-08-01" in text + # 原值: "Narayana 当前大运(任务包算法):巨蟹座 2022-08-01 → 2031-08-01" + # 新值: 中文开篇不带算法标签 + # 原因: BUG-2038,复测第九轮开篇出现「Sanjay Rath 版(默认;按原书 p.43 正文(第七轮裁定))」; + # 版本出处是后台信息,留在 Narayana 章节。英文版仍带标签。 + assert "Narayana 当前大运:巨蟹座 2022-08-01 → 2031-08-01" in text + assert "任务包算法" not in text.split("### 时间线")[1].split("###")[0] assert "起运星座 水瓶座" in text # Career: 10th lord Saturn + AmK Mars in D10 and D9; A10 Taurus is natal house 2. assert "| D10 | 狮子座 | 土星 | 天秤座 | 3 |" in text