From d01e1b720f13f4119f9f7b5c34f278765bdb9222 Mon Sep 17 00:00:00 2001 From: lhl Date: Tue, 25 Aug 2026 23:30:24 +0800 Subject: [PATCH] =?UTF-8?q?feat(writer):=20=E8=BE=93=E5=87=BA=E8=AF=AD?= =?UTF-8?q?=E8=A8=80=E4=B8=80=E8=87=B4=E6=80=A7=E4=BF=9D=E9=9A=9C=20+=20?= =?UTF-8?q?=E7=94=A8=E6=88=B7=E5=8F=AF=E9=80=89=E8=BE=93=E5=87=BA=E8=AF=AD?= =?UTF-8?q?=E8=A8=80?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - 步骤0: 新增中文镜像模板 scripts/make_zh_template.py 与 samples/概要设计书模板_中文.docx（7章锚点原样保留） - 步骤1: config.WriterConfig.output_language→Settings.writer；GenerationContext.output_language + to_vars.language_instruction（zh/ja/auto）；writer_agent【语言约束】改引变量；context_builder/orchestrator/run_trial 透传 --output-language - 步骤A: 新建 src/genesis/writer/language.py（detect_script/resolve_expected_language/find_language_violations）；WriterAgent.generate_chapter 按期望语言强制、违规重试、耗尽硬失败；max_retries 默认 1→2 - 步骤B: _format_impact 影响调查标签按 output_language 本地化（zh 新建/变更/删除/警告） - 步骤C: eval scorer 第 11 维度 language_consistency（不可验证=满分，不拉低总分）；ChapterArtifact.expected_language；QAValidator.validate_doc 透传；QALoop.run 透传 output_language - 测试: test_zh_template/test_language_plumbing/test_writer_language/test_scorer_language/test_language_coverage，并更新 test_phase5_e2e - 全量 pytest 424 passed / 99.15%（覆盖率门槛 99% 达标） --- _AI_USAGE_LOG.md | 1 + .../plans/2026-08-24-language-consistency.md | 155 ++++++++++++++++ samples/概要设计书模板_中文.docx | Bin 0 -> 35681 bytes scripts/make_zh_template.py | 82 +++++++++ scripts/run_trial.py | 3 + src/genesis/config.py | 16 +- src/genesis/eval/scorer.py | 45 +++++ src/genesis/qa/qa_loop.py | 15 +- src/genesis/qa/validator.py | 28 ++- src/genesis/writer/context_builder.py | 7 +- src/genesis/writer/language.py | 104 +++++++++++ src/genesis/writer/models.py | 45 ++++- src/genesis/writer/orchestrator.py | 3 +- src/genesis/writer/writer_agent.py | 31 +++- tests/test_language_coverage.py | 102 +++++++++++ tests/test_language_plumbing.py | 93 ++++++++++ tests/test_phase5_e2e.py | 9 +- tests/test_scorer_language.py | 92 ++++++++++ tests/test_writer_language.py | 172 ++++++++++++++++++ tests/test_zh_template.py | 77 ++++++++ 20 files changed, 1049 insertions(+), 31 deletions(-) create mode 100644 docs/superpowers/plans/2026-08-24-language-consistency.md create mode 100644 samples/概要设计书模板_中文.docx create mode 100644 scripts/make_zh_template.py create mode 100644 src/genesis/writer/language.py create mode 100644 tests/test_language_coverage.py create mode 100644 tests/test_language_plumbing.py create mode 100644 tests/test_scorer_language.py create mode 100644 tests/test_writer_language.py create mode 100644 tests/test_zh_template.py diff --git a/_AI_USAGE_LOG.md b/_AI_USAGE_LOG.md index cca3292..19fb769 100644 --- a/_AI_USAGE_LOG.md +++ b/_AI_USAGE_LOG.md @@ -115,3 +115,4 @@ | 2026-08-24 13:18 | Agent 实现 | 修复无子节章 LLM 自造标题：prompt【小节约束】禁止空结构时输出 heading 块（尽力约束）+ WriterAgent.generate_chapter 程序化强制（sub_headings 为空的章剔除生成结果中的 heading 块，模板结构为准）。真实试运行验证：自造同名 H2 清零、H2 总数 14→6（仅模板定义的子节）、无子节章正文全为段落 | src/genesis/writer/writer_agent.py, tests/test_phase5_writer_agent.py | x-preview-f-free (opencode) | | 2026-08-24 14:02 | 架构设计 | 输出语言一致性保障方案落盘：用户可选输出语言（auto/zh/ja）+ 中文模板镜像 + 确定性脚本检测与重试强制（A）+ impact 标签本地化（B）+ QA 第11维度 language_consistency（C）；关键决策：表格照抄原文、双模板锚点不变、违规硬失败 | docs/superpowers/plans/2026-08-24-language-consistency.md | x-preview-f-free (opencode) | | 2026-08-24 14:35 | 反馈迭代 | 修复 _AI_USAGE_LOG.md 表格渲染断裂（用户报告 2026-08-09 03:30 起不以表格显示）：删除表内 4 处空行与 1 处误写入的工具输出残留行（End of file）；L55 摘要内嵌错误码表的 6 个管道符转义为反斜杠管；L107 被拼接的两条 2026-08-23 记录拆分为独立两行；零日志行删除，修复后全表 116 行均为有效 5 列 | _AI_USAGE_LOG.md | x-preview-f-free (opencode) | +| 2026-08-25 10:20 | Agent 实现 | 输出语言一致性 + 可选输出语言（实施中，暂停于步骤 A 修正点）：步骤0 新增 scripts/make_zh_template.py 与 samples/概要设计书模板_中文.docx（7章锚点镜像）+ tests/test_zh_template.py；步骤1 config.WriterConfig.output_language→Settings.writer、GenerationContext.output_language+to_vars.language_instruction、writer_agent【语言约束】引变量、context_builder/orchestrator/run_trial 透传 --output-language + tests/test_language_plumbing.py；步骤A 新建 src/genesis/writer/language.py（detect_script/resolve_expected_language/find_language_violations）并接入 WriterAgent.generate_chapter 违规重试、max_retries 默认1→2 + tests/test_writer_language.py（待修正 resolve_expected_language 标题仅假名可信） | src/genesis/config.py; src/genesis/writer/models.py; src/genesis/writer/writer_agent.py; src/genesis/writer/context_builder.py; src/genesis/writer/orchestrator.py; src/genesis/writer/language.py; scripts/make_zh_template.py; scripts/run_trial.py; tests/test_zh_template.py; tests/test_language_plumbing.py; tests/test_writer_language.py; samples/概要设计书模板_中文.docx | x-preview-f-free (opencode) | diff --git a/docs/superpowers/plans/2026-08-24-language-consistency.md b/docs/superpowers/plans/2026-08-24-language-consistency.md new file mode 100644 index 0000000..f48abae --- /dev/null +++ b/docs/superpowers/plans/2026-08-24-language-consistency.md @@ -0,0 +1,155 @@ +# 输出语言一致性保障 + 用户可选输出语言（2026-08-24） + +## 背景 + +真实 LLM 试运行暴露：生成正文曾出现中日混杂（早期第 1-2 章中文、第 3-7 章日文；prompt 渲染缺陷 +修复后全日文，但属概率性改善）。盘点确认： + +1. **唯一保障是 prompt 层【语言约束】**，无任何程序化检测（grep 全代码库证实） +2. **输入侧存在中文诱导源**：`_format_impact()` 固定标签（`受影响:`、`[警告]`）进入每章 prompt +3. design.md §7.2 QA 十项清单**没有语言一致性校验项** +4. 日文章标题多为纯汉字（「5. DB設計」无假名），仅凭标题无法判定期望语言 + +用户需求升级：**让用户可选择输出中文还是日文**（参数化），并配套中文模板。 + +## 关键用户决策（已确认） + +| # | 决策点 | 结论 | +|---|--------|------| +| 1 | Excel 表格数据是否翻译 | **照抄原文**——数据是源数据忠实呈现，翻译违反 §7.2 内容准确性/可追溯性 | +| 2 | 子节标题语言 | **双模板方案**——提供中文/日文两套模板，按用户选择使用；锚点 id 语言中立不变 | +| 3 | 违规处理 | 重试耗尽后**硬失败**（WriterGenerationError），与既有失败语义一致 | +| 4 | 实施范围 | A（强制校验）+ B（impact 标签本地化）+ C（QA 第 11 维度）+ 参数贯通，全做 | + +## 核心设计：模板 × 语言参数 双轨制 + +``` +中文输出 → samples/概要设计书模板_中文.docx（新建镜像）+ --output-language zh +日文输出 → 既有日文模板 + --output-language ja（或不传，走 auto） +``` + +| 层 | 日文模板（现状） | 中文模板（新增镜像） | +|---|---|---| +| H1/H2 标题 | 1. はじめに / 2.1 機能一覧表 | 1. 前言 / 2.1 功能一览表 | +| 锚点 id | `section:function_list` … | **完全相同（不变）** → 映射/注入器零改动 | +| 封面 | 概要設計書 / バージョン | 概要设计书 / 版本 | + +## 实施清单（TDD，每项先 RED 后 GREEN） + +### 0. 中文样本模板 +- 新建 `scripts/make_zh_template.py`：python-docx 镜像日文模板结构 + （Title/H1×7/H2×6/锚点原样/封面字段），产出 `samples/概要设计书模板_中文.docx` 入库 +- 测试：解析 zh 模板断言 7 章 anchor id 与 ja 模板逐一对应 + +### 1. 语言参数贯通 +- `config.py`：新增 `WriterConfig(output_language: "auto"|"zh"|"ja" = "auto")` → `Settings.writer` +- `models.py`：`GenerationContext.output_language` 字段；`to_vars()` 输出动态 + `{{language_instruction}}`（zh=必须使用中文撰写 / ja=必須日本語で記述 / auto=与标题一致） +- `writer_agent.py`：【语言约束】段落改引该变量 +- `context_builder.py` / `orchestrator.py` / `run_trial.py`：`--output-language {auto,zh,ja}` 逐层透传 + +### A. 确定性脚本检测 + 失败重试【核心】 +- 新建 `src/genesis/writer/language.py`： + - `detect_script(text)`：含假名（U+3040–30FF）→ "ja"；CJK 汉字零假名 → "zh"；否则 None + - `resolve_expected_language(explicit, title, fallback_texts)`：显式 > 标题假名 > 规则文档主导脚本； + 单一事实来源（A 的重试校验与 C 的 QA 维度共用） + - `violations(...)`：仅检 paragraph/note/list 正文块；**表格 rows 不检（照抄原文）、heading 不检（跟随模板）**； + 违规判定 = 含 CJK 汉字零假名且长度 ≥12（防误杀短术语） +- `writer_agent.py`：`generate_chapter` 重试循环内后置校验，违规按生成失败重试； + `max_retries` 默认 1→2（现默认无重试机会，已确认既有测试均显式传参） + +### B. `_format_impact()` 标签本地化 +- 按 `output_language` 选标签集（zh=受影响 / ja=影響；auto 默认 ja）；只影响 prompt 变量， + 不影响 impact-report.json 独立产物 + +### C. QA 第 11 维度 `language_consistency` +- `ChapterArtifact.expected_language: str = ""`（上层解析后传入，scorer 不自行推导） +- `qa/validator.py` 从 orchestrator 透传期望语言；歧义记 unverifiable 计入 detail +- `DEFAULT_THRESHOLDS["language_consistency"] = 1.0` +- ⚠️ 已识别风险：新维度改变 score() 总分均值，`test_eval_scorer.py` / `test_phase5_scorer.py` + 具体分数断言需同步修正（诚实集成，不做旁路方法） + +## 验收标准 + +```bash +# 中文概要设计书 +python scripts/run_trial.py --template samples/概要设计书模板_中文.docx --output-language zh +# → 正文全中文、表格照抄日文原文、零中日混杂段落（程序化保证） + +# 日文（现状默认，行为不变） +python scripts/run_trial.py +``` + +1. 全量 pytest ≥99% 覆盖率（当前基线 389 passed / 99.04%） +2. 真实 LLM 双跑（zh/ja 各一次）+ 扫描脚本确认 output.docx 无违规件 +3. 提交 commit + 追加 `_AI_USAGE_LOG.md` + +## 涉及文件 + +- 新增：`scripts/make_zh_template.py`、`samples/概要设计书模板_中文.docx`、 + `src/genesis/writer/language.py`、`tests/test_writer_language.py` +- 修改：`config.py`、`writer/models.py`、`writer/writer_agent.py`、`writer/context_builder.py`、 + `writer/orchestrator.py`、`eval/scorer.py`、`qa/validator.py`、`scripts/run_trial.py` +- 测试更新：`tests/test_phase5_writer_agent.py`、`tests/test_phase5_models.py`、 + `tests/test_eval_scorer.py`、`tests/test_phase5_scorer.py`、`tests/test_run_trial.py` + +## 状态 + +- [x] 计划已获用户批准（等待二次确认开工） +- [x] 计划评审（2026-08-24，对照真实代码逐条核验，见下节） +- [x] 步骤 0-3 实施 +- [ ] 验收 + +## 验收记录（2026-08-25） + +- pytest 全量：**424 passed / 99.15%**（覆盖率门槛 99% 达标） +- Fake 模式双模板试运行：zh 模板 + `--output-language zh` → 7 章 docx；ja 模板默认 → 7 章 docx，均无崩溃 +- 程序化扫描 `output/zh_output.docx`：25 段落，**0 日文假名混入**（语言一致性成立） +- 约束链路单测覆盖：writer 重试/硬失败、QA 第 11 维度、validator 透传、qa_loop 透传 +- ⚠️ **未执行项**：真实 LLM 双语双跑需 `GENESIS_INFERENCE__API_KEY`（本环境无 key），确定性强制逻辑已由 e2e + 单测 + 覆盖率间接验证；真实双跑待 key 就绪后补。 + +## 实施进度（2026-08-25 暂停于步骤 A 修正点） + +### 已完成 +- **步骤 0（中文模板）**：`scripts/make_zh_template.py` + `samples/概要设计书模板_中文.docx`（已生成入库）+ `tests/test_zh_template.py`（4 passed）。锚点 id 与 ja 模板逐一对应、译文正确。 +- **步骤 1（参数贯通）**：`config.py` 新增 `WriterConfig.output_language`→`Settings.writer`；`GenerationContext.output_language`+`to_vars().language_instruction`（zh/ja/auto 三档）；`writer_agent.py`【语言约束】改引 `{{language_instruction}}`；`context_builder.build_contexts` 透传；`orchestrator.generate` 加 `output_language` 参数；`run_trial.py` 加 `--output-language`。`tests/test_language_plumbing.py`（5 passed）。 +- **步骤 A（检测+强制）部分**：`src/genesis/writer/language.py` 已建（detect_script/has_kana/has_cjk/resolve_expected_language/find_language_violations）；`writer_agent.generate_chapter` 已接入期望语言推导 + 违规重试 + `max_retries` 默认 1→2。`tests/test_writer_language.py` 多数通过。 + +### 待修正（暂停点） +- **`language.py` 的 `resolve_expected_language` 逻辑需修正**：当前实现在标题为纯汉字（如「DB 設計」，无假名）时 `detect_script` 返回 "zh"，导致标题被误当作可靠信号、fallback 失效。 + 依据计划「显式 > **标题假名** > 规则文档主导脚本」，标题**仅当含假名（detect=="ja"）**才可信；纯汉字标题对中/日均可能，应跳过、改看 fallback。 + **修正**：将 `resolve_expected_language` 中标题分支改为 `if detect_script(title) == "ja": return "ja"`（不再用纯汉字标题推导 zh），随后才遍历 fallback_texts，最后返回 ""。 +- 修正后重跑 `tests/test_writer_language.py` 应全绿（已有 2 个用例因该逻辑失败：`test_resolve_falls_back_to_fallback_text`、`test_resolve_unverifiable_when_no_hint`）。 + +### 后续（未动） +- 步骤 A 通过后→ 步骤 B（`_format_impact` 标签本地化已预埋参数，待补 zh 标签集使其与 ja 实际不同 + 补测试）。 +- 步骤 C：QA 第 11 维度 `language_consistency` + `ChapterArtifact.expected_language` + `validator.validate_doc(expected_language=None)` + `DEFAULT_THRESHOLDS` 加项 + 同步修正 `test_eval_scorer.py`/`test_phase5_scorer.py` 总分断言（按评审 R1：不可验证=满分 1.0）。 +- 验收：全量 pytest ≥99% + zh/ja 双真实试运行 + 提交 + `_AI_USAGE_LOG.md` 追加。 + +### 已落盘但未提交 +- 上述步骤 0/1/A 的源码与测试改动**尚未 git commit**（暂停前未提交）。 + +## 评审记录（2026-08-24） + +### 核验通过的关键假设 +1. `config.py` 为 pydantic-settings + `extra="ignore"`，新增 `Settings.writer` 字段无 yaml 时取默认值，零风险 +2. 中文模板锚点 id（section:introduction/function_list/screen_list/report_list/db_design/if_definition/batch_list）已从真实日文模板解析确认，镜像后映射/注入器零改动 +3. `_format_impact` 无测试断言"受影响"字样，标签本地化安全；`[新規]/[変更]/[削除]/[警告]` 中日通用保留 +4. `WriterAgent.max_retries` 既有测试均显式传参（2 或 1），默认值 1→2 不破坏测试 +5. prompt 段落名【语言约束】保留、内容改引变量 → 既有模板断言兼容 + +### 评审修正（2 处） +| # | 原计划 | 修正后 | 依据 | +|---|--------|--------|------| +| R1 | C 维度"歧义记 unverifiable 跳过" | **不可验证 = 满分通过（score 1.0）**，维度恒参与均值；detail 注明 "unverifiable" | 实测 `test_eval_scorer.py L154` / `test_phase5_scorer.py L82` 断言 `total_score == 1.0`——若跳过维度或给非满分，既有用例总分断言即破裂；恒定维度数 + 满分兜底可让多数既有断言不变 | +| R2 | `ChapterArtifact.title: str = ""` 与 `expected_language` 两版表述并存 | 统一为 **仅加 `expected_language: str = ""`**，scorer 完全不自行推导 | 单一事实来源原则，避免 scorer 内再出现一套标题猜测逻辑 | + +### 评审补充（3 处说明） +| # | 说明 | +|---|------| +| N1 | `validate_doc` 增加**可选** kwarg `expected_language=None`（qa_loop.py L55/L69 与 test_phase5_report/test_phase5_validator 共 4 处调用方零破坏）；None 时该章按 unverifiable=满分处理 | +| N2 | `max_retries` 默认 2 后，orchestrator.py L72 与 qa_loop.py L28 两处真实管线在失败时会多一次 LLM 调用——与 QALoopController 外层循环叠加的成本上界 = 章 × 2 × QA轮次，可接受；实施时在 WriterAgent docstring 标注 | +| N3 | 运维注意事项写入 README/计划：zh 场景规则文档仍为日文 → auto 回落会判 ja，**中文输出必须显式 `--output-language zh`** | + +### 遗留风险（接受） +- auto 模式对纯汉字日文正式文档（无假名标题+无假名规则）无法证明语言——保守跳过强制，由 C 维度报告层呈现 unverifiable diff --git a/samples/概要设计书模板_中文.docx b/samples/概要设计书模板_中文.docx new file mode 100644 index 0000000000000000000000000000000000000000..3c66344fcfcc03965dd17c1c6d08bcf007b9628c GIT binary patch literal 35681 zcmZ^KWmH|s)-CP?cLIdq9w4~8ySpd21$PK;!QC~uI|p}n4({&myp!&}-*?u4^x{T`YLXA zrjEJ{uGUud3B0z;OsGNkZwP|nK?1omIdBLgcvoN8z>jiu&9BI+$)lmpPX^!lve6Mm zw?sai(CI{-U-Myt+j|4K)2vBiFEw!X=gJ1b5*A>VTOR7$=o!NzG;$N?KhO4I7dPp6 zrZ6^wQypiBmRv`1dgA!2z{)A@LK>V~ZRf?wBt)0PDZR$ghg8-A*>iZ9Ns`iWAC0_` zBXzXLZQNl9MXox)mv~y;my%nKR+^x2VT^1k{0ngK@ch3@)m&1s73(J_OsJ_sai8Rj zLvbBvc;$VPU1M+^U$rI?%lp)t6yqZS>%=gdy4cD;j?lAlmWT2~uU&pQ?>McWYErNK zBZ>NF)=QEazbUV*fIAB+Z04VNt)ILwkz0_6kJb3P2S!}(t&Jd6R;tB>lp2BDjxK-tdL&k*LxO9E zsSt``R&B}95PguXf#6p4rFN?I7Z$NpXrq!DZ+5VbVlR{Xg5s5egXr%hv>U`5Dt0F? z`k8a2H5q2$8dc`qUTp0h=A{pC+87)XI=}E|X)=`&4K~X+`h24Bt)Ls>++=>rk|}P6 zeL-~}Sj7d+NSFL*BHFXfX;f7x9~@d2z&@wtK0{j=9Qa^io9>WdLfG=dUGx z3K^F@{XEj5)g0EPn55feGCrsQt znzpB7Nn#cWrb9`#$W7A$CUOs2Ti;9GA>xbR-(-0ppC0ig6RYey|*%aq`3L`Pk&)mWW z*tC$~7O4&^D}5ukS;q?4_fk;S48^JjdPs(d$}zMO9*y*Nps*Gq6?9Yh7uL`C_IuqF zkR{5iMxzg|1}yp;)ClVIvOfn4(x{uY-v~a)Z*?}R!6c29=bF z!wUq%8Ma)AQwW+eR~gi@1PH&`1o8)^Dzh|Y{p@ztiqJ~YN2j4Hka;-(E8Ej^^9xrshLi&SzO^bM zW6abPpXtq)zup|D)64<}Mg$X3TyHOpE+ zFqqn}{Ki1wW85X#Y?=pKy zV3EjwPg@sT2NOon+;y@B?NNViPU^C@%WSxSb9K~~2TU)ZJ=(bQa#4X{l&0U!Qw4}VWGg`Wb6}KFtM)=feocciq&Hfbl3AUs)f>d?7D7F; z5m*Wm!kdJ)Vru})&9yvwO2rls3MPD4?-i19f$SJ^1uIa%bEyZr5-+!dt|s{z%NinY zLb+S=!F;Jp<2}6N@h6KCv7{O`9op&ImXs@t?JWE?4=T6AAL)Sa0yu$O>@HMoBl1xDm;> z(-^fzQ|%pBciY+rG-(N|V|EwMX2gXV!uGfD%cfk-jX1w>6CqVYK$xt$P;Xal8f%9} zAV|36D``iU!8mf66PQI?XMW|b*)~Siv_4oK7AEy7`(E#44L3+sGtd1-!|5HH-Yz}f zlC(cSEnvo!84%nC{Z?e_vD$bUzK^}Q?1)=36906$O@Q$n4BM{WQKT7w;Tn|wuwm+X zseZnnw^qt-PO*h6buydN_NBQ?&}2w1#G3r4jpDPDXoj;#<@$in`2gj*?gaM~aDqwZ z9K|@dIS65^XMgDMEi&Qk6h2}aa-3AfVXv;nKr7+8^ouGgzek+s>%?x>^UzQiVQMOU zs?!l4nuDU^-q+j++LG9_!_(wIcfgyI4L|f2t`8jhNXeJU z_%@$u{^k&UK&!6%wbT$I%@KhZIKh+%{=_+`*2qCA6LlYSKjbk z#i84Kt{)n0pBOX8+eh#fNNtxWr2Uo;`+A`5i59~uY^_!J8EmQhCfVd+Z%g#(Z7~D5 zafhtxbbjN{od}_*P|Iq^5nDaTD9><&Ga)c!Jm zLubWAt;v*nXYXh=-!s{@qIQ<7xo9F26=)-Ewz=UH@(3Z42=}R*lY+n*w`CwS&H*E= zF;pb8JiZiIrq&eWbj^rqeIs6SNBr+uZR1OT&H`HTM?fn+A!x<_YgYe#YWeTE{r3rG zwpy;|&j}`@D5TA#+XgWn;p-`cgzhIewFOQ}>;}e~%3lCp|3nH^>E2(L-qQCzS9aH? z+g4ZLpYU%qx%To{!7Hnn0Hf1MFP9F96fyaX`JyPAErH1lI5^joRk27<8Yx^9k+png zWuztNKSN3&C_fM2dv2HYZ;**b_k zP|t>Nrf5y*%BB2?)S6x-afdlx0B?D$%|0>o`gXx*mOBqZY_=Dx#p~K-`xMx(E<6`7 zpM8+uc0Ie_gm**WJNQsz38eJ;dHCz6yas<36_+9Lj^H zA2zpi3r{ASsib*$p(J1Yfl~eh2LRs3D=)gzK8ls+B=1;sD^tfm3+Sn?w&ONijW>Ts_ihpod7^6^YyEg4Rcx9HInavY z(k+pEJej*@=kw4dS;reDwG=kBk`=1}D%f&`RDTPY$@+ztZ1ywp`^!^X%OlI$S?t); zTiT{efUV8etx>JaeAao##JvbdL`TO!&_R%)++n^%bU;J<(*;UyCv4S z4S=t5z7=I6{CTsnoa6??dAqx5`MmU{4I=KjzV>*!=N?g6K!6?^Mg(W^`Nn4`BSP{C#1fMf9OwymYz zP0#b;^4h&NgTG?ydATvi@%fd1jgO8!oaBwwNB7j_Fl$-A@iFUBANZa-=HAij_RO?S zAG>p3^uP(|@RFNPJ4KnfWoT}9dq$mbc|K%F$?Cw{t1LL@FCZ>^*)wY7_W;T#4xQ?9 z9iE-1HKO>8tKT;!I(eEl&~`MWR`L_;hcB$ZU0%KzjqAC~Bo4ivdOn{Hm_PNkzO0=M zySym9(WX6^0wikO_Ok9S@80I|k){RJhqwS8^v69%Hb?@(;|fW^uW9=G%a8k~FS~~ny4s9p&CJ~c3j@3-O2M8< zPxc+FW&G@N#+sfpKPMRFuXSrIGeeYbI*K~?_w<)dfE)?hwd1oWZ^ssZkSw-R)3)|U z-nW)-uWyWLY0SP348O{6k`zf`fi*)Xee>+R^%%TV{wQ4G)62rQ@if4*y*#$fBA#jh zc!hARM5{d?XDc1Mn>j`xBoY(8!K~7*n%oZX?+pP4AZ}ipwP^HO{>rlNRSeSPfS~E2 zTZu_9WH>_|VZzZbGffWRJb1%?Ps!K(;(U8HUS+l&2>d|Y@Gx;~wvHr2FNgiU?M__j za*1rkmWIKGJO|DTgywV6Kn7zgt#;910{vG}0A;H! zbS>#GkiQ}7rDg4D{km>XOnvqZKqXXpfqV;mR zGI?~VNt|`<IvbUhyi)`nhF{*4F-Z{hdg3g% z-+p&JAXD}HxP9qpo*deExY}mreBT(wK5je!{MJCt$K_>WN{Yw=@fX+j4#+N?oqO$&QlZPvxnan~v2B~V1phGxyB z=8jRc?_ZUBab$p*&^rY4si%6vD)B3#?5FVyLXEu%*1>KQ)x!SP|zMCth9mc)%LUYw{*>cA*-5bYrXs5%`Z3m zN32uLU95yKneX?*h8s6yOFf-|wg6%nIel+_q!-tR^AkJqa7;*|%IUk6@3ARncXLR@ zYwP1I1`lrz2!1oFTrC_rJQbinh77cRjaXGn$wUe18n8qawE88MiWjRNarVgdObnFx zD6iS$O~Zvds>!jGo$)7H$~B^?vE{$1HO-RIi()wH#kN9N(P0h9&PAj()2dbLQ+6)a z8R!Pf>q*q}p_|P#e&t0$W0mzfoE%(0f%^uF#elV5TAcP{n6qc%Id7L3v#T3LdeY+0 zX-!GwV?~Zuk;)^|I=~#-0>kbu7gz|McqL0wA4z4t>sSgI$561c_52h&N=Aj<$5*kd z@B`CMon4w1$)}ecttzOF*_4RS!BQ&u8f`-3m}`nr_M1q>m`^zQaaRAHuRPgtH?=m5 zOFfHx)r8o5c$1Cb;mtmQa)wx)ze4%hd*pgGm{!+2V;}xzV2)0&+&hz)uifqU`vJ{c zy!)ZH)~vbC<=eJd$2!k~55M%}0*ak_TXIjurzQVc^Lx1rjS#QJ80GnaO;xl-E`me+DxwdlmJ{I}E7*2| zcT;r=j)3-qb4=Kxch?lg>7uffMHHcckSmV2KKySeB&9lB3tW9QmDj3(E1G~IM6<7i zz)X)%;WxwOOQIWOvGrtY&MS4tr>~Y-MH|Q?lE0 zen%>n^<`@_1f<<__En{5hyjTUl@q}}WThjp=Z|R?P;i+=dN}>eKNzcml>NRER##_d z_i=e_M#71(W#`y)Ug!(x_bc4>WDJNmo0n5! zr<7s?1D~9-s!sf7E<)}5sCT+4tp=={l)f1d`^|1v5D+ou-}2}+4*98kgWtVZUt02Y z32eoY&9UMKwhB3TowE&#tWH>$O53Y(ujPD{*8Co%6dfYIwozjCon?mN2SFSR=(&0 z6n%paYq+Azn(C^X|BwBYY;Iui3GP?UmXe_u64@xEh5UyAn+OX+ebsG^U-$zQ#nov& zW)V;6I&=FXzxXIA4wSnf_2;0>6Yq-|EpHcFiKJs`*sz<+(+NH0kHO?u(vgH`bNmeZ z@P74JMi4y!tj{NR0cCM7rfevs2I|=Uv6+a79@<4Txw{11?Ok4$G9CxC5@hp{SIoI{ zhe1ZR8ffOh*M}9R#3v2om>fbAm@334_5E0nz|f4jxv_ChqBjd?>#w+bf#Vga@aINa zOoK4T1`gi8x9F@)^!n<-^L21i*rSl3zn%66+~Cyg<>9IV`z>~VC{CnsZ!muZR~SuS zv*cS?oOyLNC<<~Fg;Upbw;-Nir zZynDPvrC5fn}oXP&H|Zfml|xy1X4~$PesRjn{7udG;-uc8smqwQ|7g^HCkiV?6EGq zB~N~{)P*qtFPt#P(hje(Cch_#nlZsS-Qzdj>Bg-_N_9PIoyrCFNlE2W6*LsP#7 zLFQIB3V;QvB+t05^l0JqP&zC2)aHG8@3dhWGdjOjLpq!#TCJug0G{qj9lKns`w_iL zPOUseoO&m#kf!fy`WU5P;T+*=@NF1!rIpljwk0)AZEVJ3udH?~=R5E=pLS3P3i*~p zpD`Ha@~+CUTJSYt7@cvU5v8?g#=ljg@7fw~ZKciYA-;?`H|Z_2j>wb|TYc!JUifTx zRsgP|UWSL?s=?=8k4NMC-D!^Ql4*3@uvmid$7=lC{^ZQ>$QO%97%iX71%#}I%TK{9 zQ5#wy_ef{OQptA;ky+tSNDz9^S*a^iy!vf~pnF5t=y2U@e~8#(4BS~2ooU7h9w1*R zAYpV81e4zsZsDU;!5CWd7%&Lg=4%SsF8RhWA)IuUKvDLoKZmXTL#Vh{VU5!r+8u52F(s}WH3m52!^hkcqZhuy0)6eq|}8(wa0 zVN@@VJqBV+#P$PHJ118zJG%o#p1Bj4gN%^vw%@)?c?*Fk>_xvJeEaXFe#kPxMSPUx z@JRym#K|wsc;>exs;s~!U!Wn=+dy89Ee(Hzw;J0MK$xk(Q2xmxS*7)R(ho`r z7oOq;h-^J+x}%Io$O*@%Fn7FT(PCXI7}Hyo?tYP3?P-f@DB&zcMoyl=P59es{M;bP z`fzti7ylmv0q)jdM}0U4pJqq2zX_j`@=}!+P4iRauEnZ_7tlUMxDRa#evmItnn(RA zQ=&U7-GY(NO*DH9roafOLd(%xdm^*IVHeWmO^{(myG^XqqI;bBHY@!B#lcmmy)LNO zQ0X*RZ#siy^)g!z$JhN5ItT9dYTik)l?u~FA0;7`8TBuxgz0}f(VriauZB*2L^-+z zE5^~62v=aV9#nWIqM*~QjC3#=2~ey^Mcy#Yk`Z2ltte!yJ_f1|TvEW3ok$8o?;yS85>iZ* z9K36r*`3-6B{x5Ke2rISwf(V%$ePV8hm52?m((SjEZbtd=-enj`IIu}JRo$qd*f1+ZTIR{Uap5%hfcu|M~v%LBf{K=Wm`7v8LU47>b#ORU$m zLiUCO_x?Fv&NAUARnIcvKh(w|@4w7H$boG7vdB!h?@5XCq$sF(r;Po&?RmL-?R?#$ zN96IjzFyJ)MX`i`o}B$^Vd}x+*K7R6+-A#Jyv&p2(ogZd@(MS{{WYkDNKo-XVYUM2`*Ee*e+FrOE7A5@oF5}rz0Y~o;lY4H37ofP#Vq* zdlcP21-AO9L7`wZ?pT5Nx1@H-`kpedoEgbl{4rB8Or@EZHP8B9eiX>Sd+E?&13{df0GOPKi`^ zgoiBQpM)1WCqw&+musZvqnPVcDX52xbeO`sTrNIgJ2rPX^AJGE>2|3p{-HNpXwu7* zEISlpZntlgOi7Auj1c~pe%jnP%5;&sbg66Ku!Vyv{BkhD6&v@1=&)&39ECKM!Os{+?;1Fnt$~CCaG@LnU+hBuLd{;T*c6sN1%hG z+aknI_lHPIiN`)r%Fl6VFlC{}?FSZVbw|UL0(Mkk=k&P}I7}!;O64I~uGgW zkNPmXnCmT-J#@8jmzo&bPbHFN|x-ZxxBsFoZ;X*PlLft!~>&zj@O;%Vs zz|;F$U*dF9G)P*-Ef$)88ag5E?PU+Oy-T!3xK}qyFAquNptP5(6@;;ugXH{n9U%MW zu-nTS>WIG2N>7%e%bW@_kFPn%ZN5pR*J4ef?~{$2vkj7ro0IItz1%-%2kG zB&$XA!~mIr+T+wK`k{^uq*`0)y_{n(c)gs+JLw;i9k;Z>Q9%tP(`V{_nK*Mg|1o=v za0Wip+K;P=NTMIBxG))x>5O$gcGm=QwOGadl=}I;W?PRM`$xTVZ#}YdJ+9c~+S4kzk z<y51-`^VQD*>7f&bj?&3_lBh9p@8JP>S zYTM0jI3m)AsES)Yl&I7MJuZ5g7X`^sP9xRhQG+F0;rH^Yg0@7&~uW8 zVWsJC>AG-~C^Sg16XV0aV0%t=w}K9JP|~a*cjPo=>6V)ez8-YsT?VAYpNa;sbHP&d zeFmYE(z^lVY)FY4me%NWXo%8Mw=D>O@7ua<_8R`=sXgewg2sZ#b)V$23`2t)aA2j| zn2=V6>(|tf6GPfm_%yaU4d4NU1YPLWRrxB@1smZX*7`tvHF7!W{_xfKi*F)`uRV%= zt^r|UBRg@WBE1Ibjnbn*C}IHlLl`fVG}Jjgb>DM5`WI0T!;}1R=UF0sqftGHb3{*hr~EOfvo-j&;9+~IbH${t^j$;ytGN+*?C@B z*^0J{&xCW|!VdOcHQ}d{1wijK=UJS$6;AfouSW}Y+Q5>J9+ipdBx8p(_l{s=)OTu$ zl)_`W*H@~#Sv}uIDUvBNYc1q?RwU{a&ntA*0)`FRu!lMY>QU+5wBSUBkg`pDdqz)6 z_0%cz6%C5k>q%L2&B~4zh}h~S$8ldX4^o6~kg~aT$r&@Uzc8u>rMB&1mOre{e4{gPCg% zwt7h8cE}p(_M3Tj=WCJq>VmmadSN{yNv=6ns#(u>e1IDE6f5L+{Om>K=cNTz=#Q8O zhc5JABERF8KGg@ySx>Wybx@`jqRK)Z$f*xW7nzSrC*)96a?2jK*<>8CmOFu?$qG-g z!ivpY)I)|A;t9-v^>r*R04iw0as=c%ENLx7W;Et1d4E{M6@Gc*onj6AjxXTSBBjSA zYphp-^wn+5T|(xVdS)YY=0NRDE3%E@Css&s)ckRSqcx|4~%BQ!vB2S1_y)w~$`wq`$9g zpE?il4Jq_9U;1;hZs<0Cq7XM6h#^!f#2By#-LrnR=LWbkdGHGrJtyxA@D>|oLdQXy zW#(h{EM@HFteUf$dX(8G7M(FB>s-|hgChlg6BUk`&y~0RgeHJ$Ww4L?Cjjx5JMj~lfzW$L9=3J~JvMg|+zcCuuX8j&04~{F0T-NqN_ZLL7K-vB zA(IuCdV?;gN~j<~R@@u2v%E~Yjs!HxF8rZ;5Fd{6e&R4#rigy_QE`-5Ll43M9SKUj zNcXxhXFx}SZp4RBqNU7YN>{&1nA9#;@NZv!;#h#PV8oSCtW0&cph3$^Fpd?jUT0F* z=Qg{|I{sqKPbVUXkLy2tpc(D3w!h#qg7{nr-Of~yz!kDoB{+rR z{M4f%*0VEk1=o)bz34VY1N=DyeE0v1#Ask9nDAS4G~hQxXfjM7&IJZ<9j~JN1%$pT zT)-v7xDgzqidG8oV^8X8Y?P6!;6cIFCNxeE&PAXg9HS0j5a9%jpR&uS#^M`vFJ18~ zB51RI#sO{CcZ}JT?%zFf0UOX2+e0wh(CK}sdc|v)8!$HC{mRQ`UJ*$d6Hpjknc95$ zbL3n%uiBP-i6bml)-e8(Nbcw7y)MEgk$*{ao}T`XL?z{;S+Wxr4{Vy6Nq;zPU3A}i zMLnEG8gpLwA}rRyf5*2$2++e1?iEPr7es5yx0k_$1CJUG;Yf|FM7a=ps+O)m_0|3& z#P|xmwr$tlR`=C_#NGu-L?$t0aCRGlJLBM2hg$TVR53(9W$30qxVbP9>MD1q4W)fr&pL+rR;RIB#Oz-s6^UgvrGcb*%|2_PvR zk%5AWTjrkOms{6V@jk2#zo=)>(L2+w{X^8aOV^$MU|0$J4%+%>fVihT9y#hdMs89k z%4+II(ULEKvk{96M?SrHeL(LaQ(;B1 zl+1~+mDvm2RCH1>51LT?2TtLykF(*VU`HA8g9UMY&tsFbU^WtyvqGgv_9(>|)iz-H zc*X`r$1hTxVJC1ol2Re>LNKR4{IkqIw3%gdJT%DE=T~iv&TJnwDMWW z>f{s9pska$*F$arjK z+6!hx1Q7z;+kcn=j+T~+hyH<3+t!jbL# z6A0v=Kt%oy!~j1i1{vj8H}m?q?WNAw(L0?AnwQ}%UH|wGirXKQ3)w#?IFS^Y%G)_a zEf8+k*~R|bV?9Pk;q1|2w3Ht{-V0td@fv!6#6&#bO3}yV{7QXM(n_{h1l~lx_bnGa zV_EQbF?I&>@8Aury^4Wr{bi{)~7RB z@Z8;FDI!&0_i5xPy$cafn_7cxvO5NkSD}$8twS57;DJg&A#CCTQ=gL8UPbWfiJO=F z5sL1xU>rk~IGZ)AWxjGGc7b5KNF=KwlyCEkf-Y^gN|u7Da*h1LZty|`=CPSr!yIQC zJL4+lZ2u#=H#Xg7udO->%3A3WUs_ol)WEY)UF#qDv; z&Tan#^=6S}x4bKo$^rt_CZ!56{$j@IakHW`UkyXLrhFXO<_6E7!#19DVad@tvYr-) z?u|)GwfW7xyE6Pka><6*YepuzHw5;rQdV;=ibA)%?(u@@&~fCYa<}}#&#VjF;tDKK zZE+0<=lH58gI%rEA4kqicFT963c_~~-L(>3^`w0Mgv&jf+Y{(nIZTVCwaV))bMtuV{C{u?Oqr#{oV z_ai9bg0aT)kboV?J=nYUorJPFsj`2Cwz+pD+Ydfd@J<$3?IUqt2wLrja**?HRGEB8N_UPg>!x42n>=lBB z&%VHS$Pwhn_8u8U(qe@U(NMva$Lm`|-YY&nZT=QEK|B8qQuiHnI7(C-Qn*G!?yC}Z zJXEA}h@j<60@;#iH#F4qPC>GRJbne8L?#dL{yu~_$SCbGJHpk9)D73H4UVROk6rfu z+Z5gQ6E%b*FgN6%IT(KT0*pNxAl2cz>`!l}h&dRfliPi_jRJ?j2Tsj6qHHK6?b_t+ z2v_ujHaNQMF*chp+=WY92EEbpF$k0Le#Nj?TS*Y$_-_}2aP9uW9rz2E`w#9>=-;?c z-2dQm*96<#VoVrq1#UDE1~$d8;K^>$?1K7lkq~-WyE&bEE0*=>UZ=iF{PA)4sZU$9G$g zg-|?6l7eH6lH)@~x|m)1MpLMrC$ocC02eF?2Lo|ba05Gm%>Rw5x#7Ckk)>Db7k9gl z$m;F+zV`r5hfVAp+$ZfzhN8J2lK&ZVT=lj6hZ=K7X>CvJLbu0iHgaxDl-5*$p61b% zu^xTUQJn@-g5^nO9q9i5rZJXO2ixYO+>bn49=>Jk+soWYq8laEN*v)0nvpS=DTRR6 z$6r5}%;VsRIWZFAj&SWR8DR!+5FJjNn23a$vAhU=ELg)nQgV`^WL6J(`;KJq!&53C zaYxVSZv>KFNgZ?aVKt_M5wXr_BeK||RxE`jnr`*!FUH~~e?TfziK$_tf`x@d3O6wH zRSXUOJ=*hU+n#7d$M{TPMukypHg2V&`#Ycsml1~+zOjyM(9jqiQ^r$<+#1f<;e#vKp9*BfpvY8lsgUG0{{rCupV+ME0O6 zC`@vw!CdGXnFiG`vlBMof$53GY2XRF(*GTZDJYP#zXCD!!&&0X@3y%wv%-ez!u0;V zpozyHVBd)ZqmP%28({zg*?E3~1OrwM`o9dUYIuSK+=yYMeuTVYF81BzF#`=e-Y>Y0 zxDhP1UX>LVper;!7-e)DbID&l24qwqf2k$Zegm^wAW5<>PA>yq1nAsp00vNTujm~g z)?uRU!3m*IqexCjC;?z$lc0FPlxhH*nTkj3VM^UJM)At_{#LJ>+fuTrfL!o@6vw~ zRKuXCG*bRVHB8EH=Ly?J;yg4=>cduk7^mm-yLXtRjP7($MZT}brTXUBwI=N#X$m<0iRia zB5Nw-{199?W->}g%$j={Ye!*AqW)}&K4If=>?Lc&tih^Ed&C6tm901MF!XU zNCS*@9fV*KM9r5DEb~F+`|6Rb+PPa&BUd;MB#?-R4FCkL;{+@{?gm7nfBIEgBh*=)v|1%A%Q9kE3s@;MO zXHNuy=ugCge@Dy*T@m*;=GpIm_zHdZU`a&4(fSMx&rVylX}YXSMx0>f`WS^7^}giw z{{HXOYXdj`8CqsAFvm(5F#La~UaOn`_%3Q@W@`N7pDb)(?6O#cXI7=bO?Va$X?2WB z(>_c!Fly43C$Z?9MO4TlV$noD_B8SJmc0{808-?%7-IUB=`(|>h^m^Z8PP^M@!S6L zjKzg`&1yl~?W;>(2TBJ4*>Rs)N=3(t&wbYV_BrJ`vh(w!JEy+&{Lrh54@K+w+di?+ z^XBcV&HFwL5E;N*KHqv@2};6l9e90vd~J1U?K8e&U4MP0*8P(kz5lw7lUC%CHjyS*&wwe0}#_;f7t@J`$x7X3&#dmml8ViHg|fv*qi?r&3|*lzi(?L>AiDFyJ>cL6<>BAoON%LR<`&2qjrtA zaALvU*!t8Q;KRf2aP9Ai^Zdl)@Ib-4GOj;!n`9Y4#_!|yGCgzjFk;)tfsZ~aG2gpO zK*pc5xc`{6({(@0#`@Ux+Rfg;Z^O2Aj9yt0rYLK4X@nZYL?Dx)IU^O`>bZ_aRYvM zU%pgscWi^Q#0lB}S?Ad1PtV3z@hj`R1;=A|ML_qYHDtYYO3-z5gY^C*d)|yC@Hraz z_jR-*DEknwTJRkD)*@ImPJaZmWW%bCj$ zqT-ecqBu8a3ly=)c3J2lZ-;MDb*R~qw`burS0nHf%& z2PZ&=4{$t-7#5Gwbw*X0g;<~ObjbTgqBHufY2AkTZh11v6#jtL-Q7F=d|KDhmb7sD zK4KTyd!=xs|7|7px;!UXU;SG2s!5YR?sVgRXF2nCj6PdHwD^Zvo=pg`UE!x)d;W(9 z*Tpyjj~fdL|Kd25L!0%Lo{&7)+@V{#LZY`6x|yV+YY8+vvI*>|=(&~Xo*8*IYBHQ{ z2Nxkbaw2(3;7***P;x7VKW&PgixIm^cY}1`^{IMIJ%^e0X_r|&(A!%R?VM&CO`qQc z$N17osID;rIPYnQK{SdTZX#Pp<>q1>R^i%c8dlMJGJ*Lq3|LIfg;`~+eAD0H^q%+< zW0uIb-FEKK)_6&|6a_|ceV#N1NFmcEEep?Dcs6O5hB47yB3-oHBr8bv1Q(|fwS^d0 z3tNLCtif!z&urM?h39exb!hjdbVIR0Qg|S{5o-0DJBHqgvZmlkq(LQX(1H~cvF%M*n zCjyVE`P&HU;UA-plell`8JwLeF?hw%=T233s3wq69pp*F3!ww@EIs+#2*UoyD8Guf z@tYfh2Slvs3GM>aZ6?U5><;A$@{|X8QvYMT`O65^UVTadFxf<34xw>uz_%KKWUuBK0B%S|i&mh(#VFD8w&rZfAoOfu{_*nf$zj{eJBA+s7z8Kc-(s7&>-YA@Yv&9le51m)?sW z88qK4b)N$kh8n$2(@KB5-oETLA8*(L#=Lg8(DqT}fI)dLB0DIR=N%M7*Nj7r*9(l5 zUJ?sK4DVaFT|@Fi`(td)9gY;RKQc4(+D`0)@VUx?34I0S&fw>Tl_U|)bJ4FKl@Pd3tsDl}< z%vU{JzZL|RN`vbtv8V5>C6>`YS>MA?tF4pJ!(@(IyKoW`gwV|zpE!f zuW={|(f!9>y(Ms=jA$Ge`SQ4Oe>dU4ZQbx{fhK<@WbFgxI;G ztZO5-k=MP)JN0O8=_}4Z41R-eDo^Gw=aB$?wiY<>n4I@ppPXsN^_hFj?YYe zdo`}WQ+D}aTzfYd#7`5$U7TJllujfJ+zB6tk zFcC#sPk6D|0kJe?#kWm2!lE7|No^PEDff42)lP%lyrdt3+@M`wyU`n!6Bo>Tl$g!3 zPN+gk?l+~^4y1Z@))J}+vX*Vue)Ulo-i|2$jumMsX*xgVh;K$YZru)TAInB)`rH_W z$#t7Q8PhO;W#EWWx8>XtW;zL`Y|Pd=zER}vG18W288ju;6kqcz_WXzKCj;(o=aL%( zO0!J;yNi)uiAmKF2_2emZW0puoHD@H0ngi=&++=g`}1i5pU4_@u?HO32m<4NtxytR zc{%5(X>4-YM&~ynA2B=iLTdB-@=R8zRwVBc-IgHM{qUO*70bxXQZT8*9>_7T>Vq7$ zATV^E5uI6K)^iGM01&7^J2hn%=DCNTJtf+)sS(x}{zm&hg?$B79of<~?jGFT-66PJ zaCdi?5Fl7^cXxM(Ai>>Tg9Ue&;QW`&%$u34_r9NA^o3rhcGa%F=k%#M-Ca@QT44BS zkygCYB@`6e@gh9xSEIdrYuZ0K^}L5*n(<#Md`gV-6rE<%LxU;tnU^m&52BMt8(58R zBZ%tQ@N1{+r@trPGnA<~8aW`y5=>fMuRY)%B^6~Qjfn-NZnQRhjLMp{h2{t^p1TbG zK1O1h#+LihisWHrBQXI7)cwInq37T$PkLkG)5W{Pq*k_@V~auVT-^~8OOB{woB7~K zEu1O+y0xOCr0xD_UB29f=(X6$1ZN#&J8r~<3K#E}@Q*ZEmhx!Lb@ze|k`Qz3Zt{rn zdR@gYPVI}fhab+0j!&>*H!oGN0u_4HPxrDAEZi>{(mT@O^*=^Tq^&+@ydTF?;^61P z-su@@bO@}c93Kk!TL^S>xK6|wb(XM`q(55AHGJo{^UVw zia^mm?G;|QGI&@@o`8>-6UaeFJltyqr}KyhYtCb+W?XM4nqWQF{#{?rMnC-E<`An& zI@IQ;i}Q268$agSPH()32Mwg4`^$Xxa1YnO=IEU5UC-vo^G8R#O2~uoW;)@XQR4W0 z`?RSI81bNGL}$t|2LY}#-K$oQ?!DEcfmW-ypI=bkZJEl12N)4&>47Fkr;B|-L5edK zknsmKoJJHdPfYMd!J08OmI)v+`i^LvoVJ~Pb+H&gR+Xo@z_#feA4v2 z*cs7XO<&=VBroqFD74D1=7e~`4ET2W!|Kdk(NA^P5Ev#qXch%ah)gF|dNd&FfvwBR zPUHJPO;8&@a_!h8a77~ zu>>98E3O&0sCS(X`Ld2EcOOIw%ji(L;bz+BTG4hf2{JWdmOa*p&lUMcIxR9@D6`|B za_@_-V+xv&%)CY$rC;Ukekw=So$^g+ml(Xj|8Q5bx55#)&?jl{DqsheI~6Hg)<+O} zY8)$jyGy*#TL+?X-z<|K(^@rxAil=rgC#m@mkmEy8j{-4vt}PS>r+cx{Uz7x>1~IN zt{hE6tAyofwc)MxX1w*iijpn|nbch(>TXi6ivntAGbLN2UMp~!a#K5(`@(_AQ7qy< zims)%zyWE+>!9$MI6@~i;>ktd=k5}k@3z49Y>+QL9k=JMi$H#3TxjnxX!qr-T4&_0 zZXB~ou-&FO$QnoM|f)qfZ!j5eCk^MCSu<KLJ62iggoE0jX13jR%riI?FF!E$c zc6Am1If299wTm&UJKuFlU`=2xIIx`6Qt~G(TwY={N2mTWrhn@JTUF%dBtnIhuFhw) z%l=6VBtkoD7&jwVowe|`5);U|BHvXbHtrcAimQp6ThRFjjbGa+rvb;Xc8uattWmICo|IBk} zKs!rAcT0Q1E{PRI@}~-RTLb=PkN1wo3#;De=i1fQ@}z?1!kzlwWjAP>d)DNY0ka9w z9>*(3zM7Yojw`St4fHG#4U1UQ8XP*~Km=xCjZg=DJ(OE;SbU&vQ~^e5hS>1Ar9F|t zvOshzwiMRq?io@Qwxe!ga%*$ugw%5XM#Ed#VrI0`a9-*9aNNnzI+?IF=bZ6yWixfq z)Jos4D%d_#%#5GrQbkMWX-h2W&Vw?OvHzr4c1B0J5bGC~Ki1pZ`GdzhIv3qLJlf z_%A5>idOu#Gdpw%A3tU3rSH6p6G$h*^By_4D=T#xw0Kcm0__9irlV@SFl}2mv3X0B zc(A+ubd|e$wf)LJHvf&)iYMjlwoZq8q5~^KcVx|blu!MBtj{(J@GO0jg_~47Y$5$l z5dJ|1VQNjh{G()P#5wlES9^sWkR#RH5yO5*l*5Z;hat{~v1P&DM`23*8D4?OfS4}^ zw`v*!qrti9ojTu?#fapXfELkx;PqkoyHheg#~#W8zF(@YUmCJrF-zZDD6WRAZWf*S zI35XjmUj0;rmZ_y*vb0F#Jr%1<$n)6LGAS2Sl>hu17<7NSAVU35rmXkk>*AN0b(Qt z^&4k`KVgi_O>Jpd1uSv+Dw;`3W{!V#zVxQJ$gfSCdg128CST-$5&wHr?YJN=B64S)iI8|ut{0*&=2n~Q#pk_8@z&a5IpU{X0;Qe5H4gmx?nMvA%l?4L$| zT<3I22-p+q4FPolGfg1A#j=&RLyg+^H(R>9d$-K<=7b(*GC6{`r|O&j$89yd9-XrU z-9Qfkz_yfUJ+ML0{C*^SPimRBa#wu|49kFwU&{Ii&9=HI2QJDr&3HTd(nbui$Lecj z;90j%cU2-8Q5~;3MM5Agsh8?88G13y@b{{;4i_2BEs4F$+Ki(w)QwyuGK_eUp{o7z z*#pz#o4fBWiqf8;`kw9+`HS7@RGt^J;QdmE6HuIa)m&-CdDVj806TA)!^{TpeUE() zvWHAhNy>F&%|+8$JJ#>f92A-1Pu0vK5Cv6Y2SzrxrZz$sjg@!8@5A$oO?&)y7d%jN z9_o<4?kju05b+j!*(iG>MLj!wV4{|8J={Bt+!@?EjNa(nJB*UNsk?=mL!7G;xv67- z-bs^zT@HVU5W{&uhg;22s(4_q1}@brJskLcJoerH&`#$@+`&9Zbu18auIeYH(2W_1 z^-Rf2t6|XLg4oUDM*$I?RgGONqjS%`xRXs1$pZ1JXesieGM3 zIxg6%TsYZB8A`3I4}b0}cLam9%Rg_z-pM|1H0}Ca_Lg~7o+ckUojVUIR{9n3euN;Z(!m;2k77z-PA;)ePUNz^H8g0j2<1Mn1++v$4O>pJ`(rj? z4q*VbgyB;=QI?8UF_3=}&}_9<9z-fiFb3h%?MQ_Ml~26SH#FCD5(z7gboA}e$TZ_Q z8DKRjjt&H%QfQ{OEHkm2^@9rpilz04XYNhaS0-5@=Bv)eJUMj-^jH!#L&xQX#|Atu-Qo9voni_=+jRl5EO2_R!oj!_A@`j9NmIF#pnT6ZABf_#%;kh;SU zF#}hKB~#~hdhQHYEgoQPkeULk^oBXQ^1=;NBCF2nXog*w!ncbeQ)lYqE7QXY3RBNK zhRZ!8BjVSqBl^A`y`^W--Y0L=dniKKDXZ?i`M08zL?4RBV2RD+aNvl-I z{a}c1p!ZobuBt#d0W%V)PHxf+Vo=)~NoYK^Y1KG$T}oY~o4)>OEpc|1x%;|m?VVcT zYH=GEdvD9XTyXW2Hg(W&{5){%m$0sVeA}G#?40s7_gz7k%(vnpD64kk8M)-=u6 z3Zu*D29MdT*{xDO5W^!w&cOpq)hS>*;7IZEYk)^Ys?%3~{AGcx8#KPwyNmkNBCH;p zC;}fHCu9DRB5YRjI!#{`5*sb8SpA<1Zg&-z5nSIozUa)Q!3^=pp?nxYjwT-tPVc_F z>Qs|J`*7txMby8@I(?G=vS6ISF`oKK{l;jV2DKoTDk)<1%t%Zw54Yh;HlLV35U3lbFSruu>D$bFfWr^KRW)?ju&<^f$bVtxjwPE4;ZaC;XkhAYcp zzOx)T)-E;k$UJr|H9?z1gMAcRw=R&Lt_r!n_Q(`|HNLCdSAzsx{W;mCZ45gz|4{vq z(dt^_J3^h)&2>;AMVjg4nVpnBF|eqX^B%=9m-`g0hqL}r(K-`Yt;5V%+7NrDrpI!1tqGFp8C zAShCbO%7M4u{kS2)Ipd=pz7Zk&EK$FeS*<2DCsGm%9BSQ#jYpIQx0?#+j9&%s=FLnVHx$U9g&O5L31Rzik8L@TWEK8`39TS{;(XjAk8=1 z`Nsgn`}ekW-zU$cdy||TgS9DPRmvncR->&-%Onh*{s7HQYU_N zF>O3hve^Ej(@@sCBVGNGdSn(8T2F@-k|CmBjm2j8y~!+RZ8cywb z0l0>j1{Cp=mr`aAK(g_Ep&qUx2QH=bqjoH3k{6{bF~pqcXNmgcn<9Wlri;R_QcqQ! z8|+S*9ok)zYX{ktPkt3kmr3pDG9ZKon9pQ~YCkh6hywHS6(1lgD|pdc$%|qs#~4NU zQd>%*`7y$o0bhWy_)@9?`)m&xB4nA=kmr(W2 z5WG`C1-yVO6<-i-$7~u*C0-DXYaX*WH&J;i>RDk7lr1txSz!|}Xox0#p$;7chb^Ui z(e0CDgv)3_HP}2;**=Io&=m#Df=&T{MxE%2zTgwgUPF~q@<(A5=f-=WEiL>)bJz{E zHAZG&d{>;?E4^71iVv=W8WVz`tnm5eYY9K@g7ufoPo)6YD+O`@z!a~~uLnfX{`9@* zy@rSYnu)w9!1txaWGdm$`iV4d3-z5c*`X*5AsN&tJmbWhREd&Dqext#4LeO>Mx8R| zL(8THRofLT@+z18DW9lpt`JdDA-9eND)9KAg<*7r|4;5nKW7Bj+yDN&_*(bD3obKbbg1Obx5OUm>=jO1LT zr?{0&wL)(-P?i)~ClVVITl^rY@H)E>L{v5L=+Zx-1Sm-G=vLe7ZzL0s8f*YjRoeqX zGMa1QtaL<7z`-~S?cxAonf@g@T*r2XPAz5ru(ld|YZzE@5pq{`&T9L^II#!7V-NY0 zbo>N-w?yfWi)-8X&+z zO85rMIGr17;P*uTp-!nv_#f(iH1plmE#klq>D69Jd;e;8y z_W?$a@lnDTD zwiN%zo28kD<{P~V$Fm|-W&QuwxZd?nY{p({91^bDezfF0-E21z`JW1bWB|@|gAF2t zj0~RTn3=~*JH5vhy-ruvCHt)2hA@nu{Dd&fr_O^oOuzV5vrra4-v#6~kA|&jhPyHy zW!hY7RwC(1Xuc$ruEIg74%c9HaBTr|E+8B)F$*fu-@vQyP`|;&whyM)ZCP4Xn?b!M zwOk#!u%JX*wa>8y*x47y4`Dp%*r}k1u%KF{psS2ztz5>RIklR1_cep!-6k*(mZ`VD zp_VWKe(^l+oa48tbW>?99bDkR8QxryF9JRH_)|w?p~0Mln?Avy#0aaQoc}UxR7z}J9ma?ceTs!$$Zm)=JFHDay0|j^ zeS{3lXU`m(VE$n3x?q0&vBOqVWI*dKuyo=VpF6)(OwlLccp)#_<4^3FZoRcQ7mz=& zFCcm6F}$C;3WNCxw@@|_0Us}uOLu;6N%Lj}4Cx@k7a*^|J3sVo%frWZtqe3(0TA^! zD>_rztj2!fLURIeFCbT$LhnE+1D=pHF$3#-)IXBE0BMuFWDh)SM)7l^UIcx%2l|iw z(icI$LRiN4}KWJumf%>dIoQ|H|Nn+j*B#~OZ^px+o2 z1oJn&S?_ZPx{0V0!LL3=i?3ok+gC)4V9UiQkLB(n|4S8MY= zV#Jc<{@_e@Ft&(fsUStj3}P0Jz}XK|hvSs#eX_OnIVG5=L8U*5e5Atxy&z|dhw`ol z)vg4*QS1lnt)mi>FPIxb=WhL_$j; z6eJxjhHUxFfX(r2++d`}sOq6zI(Vl&#AMkCaGxLw(%??!&<=YwP}fR9h`%wr88Me= zohW50U9f4e>5PaIqjQjKgL(TVR_@!@OhHj&rV(kfa14bs_KY&71bEL-5H1|j*ecqX zjXrR0v-CUYDeG|&$57KWc~(8kW?@!g)->YOqV}_IGs7;feJmnxo0FWZA;qPbs!PCD zZhWBmX}dYThtdebU>DHTiq6yu1Ug~rH`W-Wv0WL`BJ+<>qE5O~LI5BfqEtzk!?(YL zS%2=3el_ygR$aeRVQnBv{l<{Gx|kAawyp1Dep+d63kr)grjsfWZ0bZFtHk=vKUYH_ z#_5ex-vsMBWnVw@K)`l7B54i38p2VzgiM(FNS1ms)tmD0-w{0C5KLr$A?%<-OC<^X zWkqIW#znQQI6!TZd89Q1=JwLt)N#wo7L3032HSk?oTuy+Z zh_?`ox`|l@=UdXV%He5a1Bkd1Cfg0f4DmOd5PrxG{Pge8n2$AI_$h`nAWqndl9uBY#& z$g#-kqRO$r*Oi=Tjb;40dFX6tFFmNrwt@n9)~=aM^#|D2VR-wwwDY@aV+vynDe1jd zh)7ACWIr>*$%|7f@f3ui&Jipli>^(1+f%-H`mlE7gYil zWpv=kHRF4Vf?@A~d4{`9ds*}EW`JF4kkd5FUT*}2f`~i@1t#?Z_V`SX|5_?@kxl?AIO-_t@Xf2OZb9FU_JIyqt!C!P)~~l0m7To-K%KsD>@s~ z0M5f!e{!CujE0C>=Au&hMqBUIg&`+~2$=LSyUVj8?&a%^OQQDc@{kpSgYWr&=OS8* zAI?F)l``4Kje><}^Fnc1dd;a~n z*{-tYpRv&Q{0luH06RBMif2dJP0EkCmX(UF@4$a^^abz2`xCB448HWZTNBzPXpPig ztevI-tkeHwJuZ116nTckLi;W~ZV}P6Ic+hG9(r6-YkVlkxzVm#8i8P=QQ|U0$si`U zL4r^Np30>ZKCOvNO+1kt#u$5v#X~5wod)V!DXGA!DMO^{Gw^hS>4|dOd<=URBO!7^ zK6#wH22s0`SSiey!&TKsq+`aPt@&r zS<~>7jqs--GrS>-18mzp0w6AMuAM_bLkO-0=De6+Mg-PThT2?8D@x1q!Z=bBClC!< z9QY9ydCJg1I!dv`ePw|*qjwWB`U9c}4e~DtvCAPwM@y*;x4_sL&hSDu@PpRG!Gow} z=OqCtM1qaI-1}waF43K$xx|d+dw;=voy5dEVPF>1ocG&^40tajt=7mXNTPKT55`{o zh&i^B9wFkTtZFVKF1b%$E6v(<*irGv`DJl{liDQJ4}m0Vfzd1c4myt8ga9pN7P__O zTJXHT{akDp47-p;+#3ZD2Aut50 zi&(H~iLfuCp@Z#U=o$}%>c)ENq)vsfGaAO*5AVEXK5yv9q{8La&^jJ2W}ZZw9UZcS zSsRrb*9|geMwsDWXfpYV0^)IC+UY|57~*MCU(rN9DD))^^r7_FZfA5X_m~sNUJL~E zk|9Fu5#oWh*;>hd7~0x|n%FVow}^h*st*kP4=aM+tT6l23OjXDt5Ya|6@XD9RjZ?n z9+y@zPLMTGwB_yYi@~_M+Z55SOscgS`2>%7sRL%X!W#L#(kPw8Ixy%|#1}~F#s_}- zF`fo7qmuYCW-;v;fox0LrT;rd8Y|gVvA12MSH(>je}@*@ON^ffOebXJ-prvtLh^G^pA*<*_5Z ztm$=~_vsyCG7@2sZW`$l8Y za5=Zfo41CLK>o$7?wmH+ZBSYPdCj=4Y9j5vD5VJg9H#h3@)$y~Syq*I^PyryKhm1H z-2~+V6%C8GRk#cWtd4DrFW)HDoo<0HIxz^)?2+iCPpm` z3!Q96@RbklaYY-cPzcB7s zlsJCiR~!`JjTZ9d##QSYzOLL55)@>*`o?%$Rn>BxhBSIKjW;>UAfE=$^wN!@2(7RO z0q3q$N~UPJJi|Z*^0STJ29*(Fn5eC=8(Y&dGiNX&n*LZ670$UC9K1jQ=W`a4BuUIJoWuIa*un8i?4a9%0Ylp|z?3 zIi}r*iM)J{ZW65aD10q4e)hGYjENGV$~pZMu<~SG+jnt6{ek)Xv5fagK>LEK%^GXw^c<=lCbi+Uik4Z7V3JcOX_d=gC9M-7#19 z0_G6$wy5!u`(dUQqV?KFtIOBO$}D|&yt9(^-48Epsv#@c4>}>L2-Zac2q#@u)3QWVwI|30Wf7+gfeSTqUj0nmeiy)7__wIRm zv9NY6bzE+^gHlAm(_=N*RhqW`U~WP_`UEp8{i?t^H7Lr90k8cET%co}e4oc;5YrR+ zl!%}utA1Slse2snBcDM?4@#H~A_^MW6WNvChp>e&ickOh8+p*=;(Z9&z{T$Xgzov7udjS3QI+)m`Ft-zQEu+7s)>FKzr` z;F|8w#a7lsyk`y9b7C2@?|so`t{mBE!%%D-3iy31jiZpo)A_4N6PqkTOpzoU{tqbxVceG$&lD(jOlv$mLc1hO@6+a z%8Bh+=+6QhmBk~a@|~&>+a)x7YaC$%LILKV7-cbAHbHXKe`=(qf<5B(2J?WskPqY# zE$;+j?ui}gBMOkmGlOOgMQjfB%7{hDgmPIvJ*w0CEVDp%3=N4Vjo59tkv`8d}z`if=Q={zl^S6oi z-zrHxa%DT20E)jo!T?I4{N*)MK(Rw*14HXy72sAAM? zSh5}oE>2Pkuf$CKNtBY1I^>XPDGG+*{mGFfFjkleq7#{gaVz zs(kbymU^0hrPZJhjCCk=jw9`-^7_ts)P&P-AV+dN+o_Z>`XepjEbkpWeM0Nvr;3V; zIZgfBRgEc&>z5uY3vsmy^RcRX%V(#RW)%xOLR9^+dyMtY12Q{hDotATt$3YViJW)s}Y~H*p_--T)TFW3K8DTX1lVAo+Wf%axxHnE*5|uMiRjK zOeq-M154j^1P^pK_R*uK3>=7Ku5)=yWn?RX*Bv&)$*rM!-!kxubL^g$bIk3Yp$GcO zMz)IwrJRJ~!LqC5mQ*(C=Q_cecWw|v7D2n0M<=CIi>8Yt3>FY2A zx`Nq#L7| zRXSDUB%`>eD7#BnKRs5h!mcE!_37out9kjm9ba}2S72yq2p}=>AmWEis_FDf?3)#1 zBBMZ~7ucw8%sIFv2oHRokrMzy8jESHM$ ztCDRLW0KNT%gnJ3qv~)xFdZm@#RRy-$$onFubQ`^Usa#bRN^S7^0l+Jc4oKKII-a5 zL@?C|nnoQvHLS>}Aq15&m!dsT)|xW~RR2uvH%=}1_-$2_I@e;`JT=Y&y$p7&!5#Sn zi}(z6$OnTb!@iyd=x@)>74S~^x(3zSJ1+Q<0nwjok(tGW_Mm)KwJ$nPA^N%IQ7b9;qoIQAcpT?Cua$>EZKxKkOk+4BK0QJXmN)fvM+yy8}8o z2U0JQTEW=!zx_;Tv5$R4$~gIhc`@)3esL99o)Waah6@-G%)pjvbN)PLW44kZFf^X% z^v>6JaWnmEGbLgto{bpyN8#NMHPUc1K6pOQigny6sNXRI`4A*ZVk4vA)xJ_Og5#U2 z6Fo)U-}9mLR*4TAsRMTtvrj1th3N;7WB&; z7tO5a=6&B#@j9I7fo9)IIS~mrP`Y9fj$vivftUDV%7e}%Msonqh?ncr)4@d}17EFd zBe}%A;lyoYiI26~k^M$&R9;C%PqdBEuHH?3S?0&NnYE%!_6F*)_P7)s8ot+viH}7C zZTjCW>;x!RRY*cP3YLJ#ih55_q~r=}0C^XL+?@5CLXHHXV;EWw5Mm+}ZBMcIjw9h< zo3g}Y7JLeLEfgx4=K6|Krqf`IoL)9#W;)lvCH;=GLe`%C`k)!7$?@LM$o-Xk@x4DByKaQGfefT)lX+ z3OZXPTmxEZ!?v}vk5Xb?#^;6fwZMtkof-DC1MtY&BgSm=9X}%L^+x^=TG>htCLJ$P z(pxjLoX#C32Xe8xP@OIQ@Kyg|N3?GvQLB>cXdh_Ky$T&7PeSgUEU7L&Hwg2P2AgROV_@2q#eI5xU(pcvqLb}W zPO3(9RWpeU1KT|_@R)q-%%nFtV@%||l78v3&?zUe7`|9u;aTSNkHQhd}k!?I(r{r&-+G_%Org2!jb_b_DQ zX!+Jhsi*fX@uc1DHgkpswgFaX1Z#rMH4jAViusNe4)faIpu|n1tt!4KTV-&})R!0Q zi3H7D%;lk52%e5Gwt*hW1r(!VYeu7LJ zDHV+j>BQH=75;x;=Q4e`7fJxvHweI#^{vXZk&BbFosGJUHG_q-iOsL8T*YV?%oj#X zAV8&JrjJWhW_ZN;RDS!YMCy|+<9c}3k0xn3b_BCD&d&(z&AJ(6c2HB<7}i-B;XVtC zIQ%fTpS4{n#0(YN)zEpNz6u(5` zM74`47TKfEsx8le@Od*Rm-BMmdHx6HERW4~(sPkz*B9BzMvFQ^%A|-KXor->+hZ5fe;SfN6Q)Mq7uF`4sBNIFeLN zdI4%}z^0HIQI-wzYQ`j0-~CTt8F=UY2l$DBWdaVFLOaku(I zd{!Gzx)Rp|@iS#yQq$>+N@gruP41$`u5dwHotlW{pSQXL%ar@=WcIo>>ajgFvbphS zei$95!pD6Tw##}5eE)Ca4c}xwYyt=y0*GV&i})MmU$vEgo!5-$Hd{YNR591gZo>;O zhfF*Z-5A-i#T&SFu7}7X6`D(E7AY$j>>y+^eQuh3 zESQIjH3&$lNe7aqS_l|-$o!(|_b{X2uDBT)8=C@gDl~r0o!tZ(X^#EXID$@aCt#*? zrHysF2*U3XM{nDXv%jw9c5x<(C=9oNXUwTpd?sd0LbxuVf2u0{P}@T!>$jW!RWRZ7 zwo5vGPH>+5v6QQ$|EBWcRyr!0?qJ~_g4cFC#OB)w?JLfwI1aKC#w&}~$72IyxnamV zC+UOd-{Oqu$)FHtKtd)2FrWKRoMG!?V`$2vY`54^pU>>IYJel+da=KJbV!ej1-O42I9UNR3PC=+o#*9JUWApzY7lu9%MMHl z8O;JE&4X%YuvVlt2rQUM5ZdgE%?_tQui-MCRJICZdyV#T&5hEq21tN_GY2FqWOw;K z)VJ^mpL{Q8oA>ol0`(hr>d4YWo+O_}IDr+zvDP)f&d)U5lNe(`Lgt}J+~lf%4r&D>#;4tI`RimYV{?>2oqX(!i3>S4V8}-+-EBoaDW~4 z;+)&hheps#w)D&%zW9Mmt;U?kUUHBH|ERsR6&q;pL>8SQq^H_=PIVzGN~_wZ7%OS| zc2xfyQAf*Ln{%|7dGIqZtlPU!#m)J_afNk45c+31eGLY9VIYDYazbCwB>hj6C#ydBD>o zvH491cCRyx6uObv~yfW{tbj<_#-<2?g==sae3 zuygY%cTcYG&6S{l4ClS|L%pGynBXk^SqB z-6@=P%^cry!7Et8b9lAKutSz?wB{(mfe? zxWyty>DvE!_pousfW!oDc3GX?y# zdcdl4#w7vqb?PAZyNvt8w}Xtj8;^r4;{(9UI~(!PojY4UE_y#{jh9slQ~`ONc$@6`Se@Zl{+|Z)(H*-I($-l=Nsd zQLcfo?H!>l`VTV1ojtj&wePtxOnAtSFARnZE?}xFSTcYUf$dO25}Vze+{VU|%|$jB z+)Cj<0beg?`1;T$`?GL2G;bT#eBLFqI4oxVmQQw3xsnxWK*-)`k z8TCC(SvDdMZ1PGMjy`GzeqRkZSw3siPNpW0x0ohR=29|r!DO?pSoH~BS3}C;C;4(? zQ|gX=t1i0_a981d9dL5<$0)_q;3XUSA{#y*CLk_MC^V;6PM;@K+fq6ZMtmd)Hp+qq zg?9ThDY-ErAYtIH6Zj=7!CJIv%rJE{w(-I;}e`X+4j<^ zI9V71b-Wvw>Ry`2)+u&jdm0B+C84`H!-{62Y(BI;e1&&-t@D^ZwlUkxx-^j4kDxh+ z>WK9N6lxcmraoBt*`*3MK|E|bQ|jzw`Ln%-8ajTBv-n$8FoTCW=|I&sQi}>WTeu&o zzt)=Le^j`f>vs(8S&^m5Iw2KE4*~CmhW45j{sPOV?F)?%pMyybmfaZz@rfu84&`%n zcBk~Iz`It@^HLSrSy^W6xqWqF_6(I;ENg5U|Bx}mpEW6? z!?fkmALM#fi)Qy4#>p;eg)w7?wR|>jGz^?-*RV}O7F_m+t`(4#*#By<~KR$~?;;IKUbEH)*(%lB*88$9bWy&)5Ds5U^)Gvk*4@S>LrV@3P z-6G2z5UTX(4nlQ7W9giCWp#fmH1xCcRCNrJ6VlXjX@;>PCv8jIO3b3eOB_qINhGLB zbF5OzP1~1!0A3Y!otZGm_1M>aAX=4M&ps)~@Y%leiW%Rw_k# zMNQJ5q~~Y)*CFlO@*Q@@`#JdxtdPkUumXq_o-PJIZV02fxDHs%fQ5Czz85uhOGzJ( zQEbB!F3Lv}JhNqEB29=8C-nnAM06&=#|*A(5y#MI0lQ-_l8Cn;)zbZB2XfjyS0k{{ zCxs=Bf?$jbmsl+5g)99IQaulges$HDT6MX2MhRmhlrA{mFx5m?JUd7gf9{Rg*zhTK zc}$SlLto~dz`TN?zdq@#1o0yy3ZY><(rYz}FSA0C+|dF*Z!O|`NFl;_nD9_yBOo^) zDAZ9(1ndO0d>o#Mk_nB(m=5=z-TJA%_qPZRxLo{&*}57tRj-qy20eBCkt1yDY+jeE zN`j&?CaV0=c$25z9sJ;KZ>$E-LA_bjRxeLGmp6+ zKm2T5_gr^64T8mBcz#&kS%1B|VyF!BAnbar%%999qZdeu5VHii z&i(u@)L$ubC?L;pES(C)f|xkKbafbzaeIqG z{`&C&90LN<5(XrLO>CX@l-=x2oOIr1YQF=xkxL6a0-CM?^2o4%NdrE8fO`QDe@ys) z&GUa33=V`p0p1^@s6 literal 0 HcmV?d00001 diff --git a/scripts/make_zh_template.py b/scripts/make_zh_template.py new file mode 100644 index 0000000..4eb78bf --- /dev/null +++ b/scripts/make_zh_template.py @@ -0,0 +1,82 @@ +"""生成中文镜像模板（步骤 0）。 + +镜像既有日文概要设计书模板的结构与锚点 id（section:xxx 原样保留）， +仅将 H1/H2 标题与封面字段翻译为中文，产物 samples/概要设计书模板_中文.docx 入库。 + +锚点段落文本保持不变是硬约束：Writer 子系统（template_mapper / docx_injector） +依赖这些占位符定位章节，任何改动都会破坏映射与注入。 +""" +from __future__ import annotations + +import shutil +from pathlib import Path + +from docx import Document + +# H1/H2 标题日文 → 中文（保持编号前缀一一对应） +HEADING_TRANSLATIONS: dict[str, str] = { + "1. はじめに": "1. 前言", + "2. 機能一覧": "2. 功能一览", + "2.1 機能一覧表": "2.1 功能一览表", + "2.2 機能詳細": "2.2 功能详细", + "3. 画面一覧": "3. 画面一览", + "3.1 画面遷移図": "3.1 画面迁移图", + "3.2 画面一覧表": "3.2 画面一览表", + "4. 帳票一覧": "4. 报表一览", + "5. DB設計": "5. DB设计", + "5.1 テーブル一覧": "5.1 表一览", + "5.2 ER図": "5.2 ER图", + "6. IF定義": "6. 接口定义", + "7. バッチ一覧": "7. 批处理一览", +} + +# 封面字段（Normal 段落）日文 → 中文 +COVER_TRANSLATIONS: dict[str, str] = { + "概要設計書": "概要设计书", + "ドキュメント名: {{doc_title}}": "文档名: {{doc_title}}", + "バージョン: {{version}}": "版本: {{version}}", + "作成日: {{created_at}}": "创建日期: {{created_at}}", + "テンプレート開始": "模板开始", +} + + +def _translate(text: str) -> str: + """锚点段落（{{section:...}}）原样保留；其余按字典翻译，未命中则原样。""" + if text.startswith("{{section:"): + return text + return HEADING_TRANSLATIONS.get(text, COVER_TRANSLATIONS.get(text, text)) + + +def build_zh_template(src: str, dst: str) -> str: + """镜像 src（日文模板）为 dst（中文模板），返回 dst 路径。""" + src_path = Path(src) + dst_path = Path(dst) + if not src_path.exists(): + raise FileNotFoundError(f"源模板不存在: {src_path}") + dst_path.parent.mkdir(parents=True, exist_ok=True) + # 先物理复制（保留样式/编号定义等所有非文本属性），再逐段改写文本 + shutil.copyfile(str(src_path), str(dst_path)) + doc = Document(str(dst_path)) + for para in doc.paragraphs: + new_text = _translate(para.text) + if new_text != para.text: + para.text = new_text + doc.save(str(dst_path)) + return str(dst_path) + + +def main() -> None: + import argparse + + parser = argparse.ArgumentParser(description="生成中文镜像模板") + default_src = str(Path(__file__).resolve().parents[1] / "samples" / "概要設計書テンプレート.docx") + default_dst = str(Path(__file__).resolve().parents[1] / "samples" / "概要设计书模板_中文.docx") + parser.add_argument("--src", default=default_src, help="源日文模板") + parser.add_argument("--dst", default=default_dst, help="输出中文模板") + args = parser.parse_args() + out = build_zh_template(args.src, args.dst) + print(f"中文模板已生成: {out}") + + +if __name__ == "__main__": + main() diff --git a/scripts/run_trial.py b/scripts/run_trial.py index 6cc60e0..2346463 100644 --- a/scripts/run_trial.py +++ b/scripts/run_trial.py @@ -53,6 +53,8 @@ def _parse_args(argv: list[str] | None) -> argparse.Namespace: p.add_argument("--existing-system", default=str(ROOT / "samples" / "existing-system"), help="既有系统源码目录（追加/改修场景）") p.add_argument("--language", default="java", help="既有系统开发语言（默认 java，None=自动探测）") + p.add_argument("--output-language", default="auto", + choices=["auto", "zh", "ja"], help="输出语言：auto=与标题一致，zh=简体中文，ja=日文") p.add_argument("--samples-dir", default=str(ROOT / "samples"), help="样本资产根目录") p.add_argument("--output", default=str(ROOT / "output" / "output.docx"), help="概要设计书输出路径") p.add_argument("--impact-report", default=str(ROOT / "output" / "impact-report.json"), @@ -86,6 +88,7 @@ def main(argv: list[str] | None = None) -> dict: samples_dir=args.samples_dir, engine=engine, template_path=str(Path(args.template)), + output_language=args.output_language, ) report = ss.impact_report diff --git a/src/genesis/config.py b/src/genesis/config.py index 7e11b8d..ef79b2e 100644 --- a/src/genesis/config.py +++ b/src/genesis/config.py @@ -2,7 +2,7 @@ from __future__ import annotations import os from pathlib import Path -from typing import Any +from typing import Any, Literal import yaml from pydantic import BaseModel, Field @@ -127,6 +127,19 @@ class RagConfig(BaseModel): rerank: RerankConfig = Field(default_factory=RerankConfig) +class WriterConfig(BaseModel): + """Writer 子系统配置（步骤 1：输出语言参数）。 + + output_language: 生成概要设计书正文的自然语言 + - "auto"：与章节标题所用语言保持一致（默认，向后兼容既有日文文档） + - "zh"：强制简体中文 + - "ja"：强制日文 + 表格数据始终照抄源 Excel 原文（不翻译），见 design.md §7.2。 + """ + + output_language: Literal["auto", "zh", "ja"] = "auto" + + # ---------- 加载辅助 ---------- def _expand_env(data: Any) -> Any: @@ -192,6 +205,7 @@ class Settings(BaseSettings): app: AppConfig = Field(default_factory=AppConfig) inference: InferenceConfig = Field(default_factory=InferenceConfig) rag: RagConfig = Field(default_factory=RagConfig) + writer: WriterConfig = Field(default_factory=WriterConfig) @classmethod def from_dir(cls, config_dir: Path | str) -> "Settings": diff --git a/src/genesis/eval/scorer.py b/src/genesis/eval/scorer.py index 2ea7bc5..99c8dcc 100644 --- a/src/genesis/eval/scorer.py +++ b/src/genesis/eval/scorer.py @@ -34,6 +34,9 @@ class ChapterArtifact: text: str source_uris: list[str] template_sections_expected: list[str] + expected_language: str = "" # 期望输出语言（"zh"/"ja"；空=不可验证，维度记满分） + # 块级 (type, text) 列表：供语言一致性维度排除 heading/table（照抄源/跟随模板） + blocks: list[tuple[str, str]] = field(default_factory=list) # 逐章评估通过阈值（基于逐章总分） @@ -56,6 +59,7 @@ DEFAULT_THRESHOLDS: dict[str, float] = { "traceability": 1.0, "placeholder_residue": 1.0, "chapter_completeness": 1.0, + "language_consistency": 1.0, } @@ -75,6 +79,7 @@ class ChapterScorer: dimensions.append(self._traceability(chapters, source)) dimensions.append(self._placeholder_residue(chapters)) dimensions.append(self._completeness(chapters)) + dimensions.append(self._language_consistency(chapters)) # LLM 语义维度钩子（每个章节独立评，取该维度平均） for name, fn in self.llm_evaluators.items(): @@ -144,6 +149,45 @@ class ChapterScorer: detail = " | ".join(s.detail for s in scores) if scores else "no chapters" return DimensionScore(name, round(avg, 4), all(s.passed for s in scores), detail) + # ---------- 语言一致性维度（步骤 C，确定性）---------- + + def _language_consistency(self, chapters: list[ChapterArtifact]) -> DimensionScore: + """第 11 维度：输出语言与期望语言一致（确定性，脚本可验证）。 + + 与 writer.language 共用单一检测事实来源。期望语言为空（auto/不可验证） + → 记满分 1.0 通过（评审 R1：不拉低总分，避免误伤既有断言）。 + 仅检正文块（heading/table 不检，表格照抄源、标题跟随模板）。 + """ + from genesis.writer.language import find_language_violations + + if not chapters: + return DimensionScore("language_consistency", 1.0, True, "no chapters") + per: list[DimensionScore] = [] + for ch in chapters: + expected = ch.expected_language + if not expected: + per.append(DimensionScore( + "language_consistency", 1.0, True, "unverifiable (no expected language)")) + continue + from genesis.writer.models import ContentBlock + if ch.blocks: + # 优先用块级信息（可排除 heading/table） + blocks = [ContentBlock(block_id=str(i), type=t, text=tx) + for i, (t, tx) in enumerate(ch.blocks)] + else: + # 回退：整段正文作为单个 paragraph 块 + blocks = [ContentBlock(block_id="0", type="paragraph", text=ch.text or "")] + viol = find_language_violations(blocks, expected) + score = 0.0 if viol else 1.0 + passed = score >= self.thresholds["language_consistency"] + per.append(DimensionScore( + "language_consistency", score, passed, + f"期望 {expected}，违规 {len(viol)} 处" if viol else f"期望 {expected}，一致")) + avg = sum(p.score for p in per) / len(per) + passed = all(p.passed for p in per) + detail = " | ".join(p.detail for p in per) + return DimensionScore("language_consistency", round(avg, 4), passed, detail) + # ---------- 逐章评估 ---------- def _score_chapter(self, chapter: ChapterArtifact, source: StructuredSource) -> EvalReport: @@ -156,6 +200,7 @@ class ChapterScorer: self._traceability([chapter], source), self._placeholder_residue([chapter]), self._adequacy(chapter), + self._language_consistency([chapter]), ] total = sum(d.score for d in dims) / len(dims) if dims else 0.0 failed = (not all(d.passed for d in dims)) or total < PASS_THRESHOLD diff --git a/src/genesis/qa/qa_loop.py b/src/genesis/qa/qa_loop.py index 30bec49..799514d 100644 --- a/src/genesis/qa/qa_loop.py +++ b/src/genesis/qa/qa_loop.py @@ -21,8 +21,8 @@ class QALoop: def __init__(self, max_rounds: int = DEFAULT_MAX_QA_ROUNDS) -> None: self.controller = QALoopController(max_rounds=max_rounds) - def _build(self, structured_source, samples_dir, engine, prompt_registry, template_path, output_path, session_id, only_ids=None, prev=None): - ctxs = build_contexts(structured_source, samples_dir) + def _build(self, structured_source, samples_dir, engine, prompt_registry, template_path, output_path, session_id, only_ids=None, prev=None, output_language: str = "auto"): + ctxs = build_contexts(structured_source, samples_dir, output_language=output_language) _warn_unanchored(ctxs) state = WriterState([c.chapter_id for c in ctxs]) agent = WriterAgent(session_id=session_id, engine=engine, prompt_registry=prompt_registry, state=state) @@ -47,12 +47,14 @@ class QALoop: doc.save(output_path) return [contents_map[cid] for cid in order] - def run(self, structured_source, output_path, session_id="writer", samples_dir="samples", engine=None, prompt_registry=None, template_path=None) -> QAReport: + def run(self, structured_source, output_path, session_id="writer", samples_dir="samples", engine=None, prompt_registry=None, template_path=None, output_language: str = "auto") -> QAReport: engine = engine or build_inference_engine() prompt_registry = prompt_registry or PromptRegistry() validator = QAValidator() - contents = self._build(structured_source, samples_dir, engine, prompt_registry, template_path, output_path, session_id) - report = validator.validate_doc(contents, structured_source) + # auto 不可推导期望语言 → 语言维度记满分（unverifiable）；zh/ja 显式强制 + expected = output_language if output_language in ("zh", "ja") else "" + contents = self._build(structured_source, samples_dir, engine, prompt_registry, template_path, output_path, session_id, output_language=output_language) + report = validator.validate_doc(contents, structured_source, expected_language=expected) while self.controller.can_continue() and report.failed_chapters: self.controller.advance() contents = self._build( @@ -65,6 +67,7 @@ class QALoop: session_id, only_ids=set(report.failed_chapters), prev={c.chapter_id: c for c in contents}, + output_language=output_language, ) - report = validator.validate_doc(contents, structured_source) + report = validator.validate_doc(contents, structured_source, expected_language=expected) return report diff --git a/src/genesis/qa/validator.py b/src/genesis/qa/validator.py index 0482f4a..973630d 100644 --- a/src/genesis/qa/validator.py +++ b/src/genesis/qa/validator.py @@ -10,8 +10,12 @@ class QAValidator: def __init__(self, scorer: ChapterScorer | None = None) -> None: self.scorer = scorer or ChapterScorer() - def _to_artifact(self, content: ChapterContent) -> ChapterArtifact: - """把 ChapterContent 转换为评分器所需的 ChapterArtifact（聚合正文与来源 URI）。""" + def _to_artifact(self, content: ChapterContent, expected_language: str = "") -> ChapterArtifact: + """把 ChapterContent 转换为评分器所需的 ChapterArtifact（聚合正文与来源 URI）。 + + expected_language：期望输出语言（"zh"/"ja"；空=不可验证，维度记满分）。 + blocks 保留 (type, text) 供语言维度排除 heading/table。 + """ text = "".join(b.text or "" for b in content.blocks) source_uris: list[str] = [] for b in content.blocks: @@ -21,18 +25,26 @@ class QAValidator: text=text, source_uris=source_uris, template_sections_expected=[], + expected_language=expected_language, + blocks=[(b.type, b.text or "") for b in content.blocks], ) - def validate_chapter(self, content: ChapterContent, structured_source) -> tuple[bool, EvalReport]: - report = self.scorer.score([self._to_artifact(content)], structured_source) + def validate_chapter( + self, content: ChapterContent, structured_source, expected_language: str = "" + ) -> tuple[bool, EvalReport]: + report = self.scorer.score([self._to_artifact(content, expected_language)], structured_source) passed = content.chapter_id not in report.failed_chapters return passed, report - def validate_document(self, contents: list[ChapterContent], structured_source) -> list[tuple[bool, EvalReport]]: - return [self.validate_chapter(c, structured_source) for c in contents] + def validate_document( + self, contents: list[ChapterContent], structured_source, expected_language: str = "" + ) -> list[tuple[bool, EvalReport]]: + return [self.validate_chapter(c, structured_source, expected_language) for c in contents] - def validate_doc(self, contents: list[ChapterContent], structured_source) -> QAReport: - results = self.validate_document(contents, structured_source) + def validate_doc( + self, contents: list[ChapterContent], structured_source, expected_language: str = "" + ) -> QAReport: + results = self.validate_document(contents, structured_source, expected_language) passed_flags = [ok for ok, _ in results] reports = [rep for _, rep in results] failed = [c.chapter_id for c, ok in zip(contents, passed_flags) if not ok] diff --git a/src/genesis/writer/context_builder.py b/src/genesis/writer/context_builder.py index f95e65a..06a1c4a 100644 --- a/src/genesis/writer/context_builder.py +++ b/src/genesis/writer/context_builder.py @@ -7,7 +7,11 @@ from genesis.writer.models import ChapterSpec, GenerationContext from genesis.writer.template_mapper import map_template -def build_contexts(structured_source: StructuredSource, samples_dir: str = "samples") -> list[GenerationContext]: +def build_contexts( + structured_source: StructuredSource, + samples_dir: str = "samples", + output_language: str = "auto", +) -> list[GenerationContext]: specs: list[ChapterSpec] = map_template(structured_source.template) rag = CannedRagService(samples_dir=samples_dir) out: list[GenerationContext] = [] @@ -27,6 +31,7 @@ def build_contexts(structured_source: StructuredSource, samples_dir: str = "samp template_styles=set(used), prior_state=None, impact_report=getattr(structured_source, "impact_report", None), + output_language=output_language, ) ) return out diff --git a/src/genesis/writer/language.py b/src/genesis/writer/language.py new file mode 100644 index 0000000..e83ef84 --- /dev/null +++ b/src/genesis/writer/language.py @@ -0,0 +1,104 @@ +"""输出语言确定性检测与强制（步骤 A）。 + +设计要点： +- 日文标题多为纯汉字（如「DB設計」无假名），仅凭标题无法判定期望语言； + 故 resolve_expected_language 采用两级推导：显式 > 标题假名 > 规则文档主导脚本。 +- 检测仅基于「是否含日文假名」：CJK 汉字零假名视为中文（日文不可能不含假名地 + 使用汉字），反之中日混排含假名判日文。这是确定可机器验证的唯一稳健信号。 +- find_language_violations 仅检正文类块（paragraph/note/list）；heading 跟随模板、 + table 照抄源 Excel 原文，二者均不检（design.md §7.2 内容准确性/可追溯性）。 +- 短文本（<12 字）不误杀（如专有术语），阈值见 MIN_VIOLATION_LEN。 +""" +from __future__ import annotations + +from genesis.writer.models import ContentBlock + +# 日文假名 Unicode 区间 +_HIRAGANA = (0x3040, 0x309F) +_KATAKANA = (0x30A0, 0x30FF) +# 中日韩统一表意文字（CJK 汉字） +_CJK = (0x4E00, 0x9FFF) + +# 受检的正文块类型（heading/table 不检） +_CHECKED_BLOCK_TYPES = {"paragraph", "note", "list"} +# 触发违规判定的最小正文长度（防短术语误杀） +MIN_VIOLATION_LEN = 12 + + +def _in_range(ch: str, lo: int, hi: int) -> bool: + cp = ord(ch) + return lo <= cp <= hi + + +def has_kana(text: str) -> bool: + """文本是否含日文假名（平假名/片假名）。""" + return any(_in_range(c, *_HIRAGANA) or _in_range(c, *_KATAKANA) for c in text) + + +def has_cjk(text: str) -> bool: + """文本是否含 CJK 汉字。""" + return any(_in_range(c, *_CJK) for c in text) + + +def detect_script(text: str) -> str | None: + """检测文本主导自然语言。 + + 含假名 → "ja"；含 CJK 汉字但零假名 → "zh"；二者皆非（纯 ASCII 等）→ None。 + """ + if not text: + return None + if has_kana(text): + return "ja" + if has_cjk(text): + return "zh" + return None + + +def resolve_expected_language( + explicit: str, + title: str = "", + fallback_texts: tuple[str, ...] | list[str] = (), +) -> str: + """推导本章期望输出语言（单一事实来源，A 的重试校验与 C 的 QA 维度共用）。 + + - explicit 为 "zh"/"ja" → 直接采用（用户显式选择优先） + - 否则看标题是否含假名（仅假名可可靠判为日文；纯汉字标题对中/日均可能，不可信） + - 否则看 fallback_texts（如影响调查书/章节数据，通常日文）的主导脚本 + - 均无法推导 → 返回 ""（不可验证，交由上层按 unverifiable 处理） + """ + if explicit in ("zh", "ja"): + return explicit + # 标题仅当含假名时可靠指示日文；纯汉字/ASCII 标题跳过，改看 fallback + if has_kana(title or ""): + return "ja" + for text in fallback_texts: + s = detect_script(text or "") + if s: + return s + return "" + + +def find_language_violations(blocks: list[ContentBlock], expected_language: str) -> list[str]: + """返回违规正文块文本片段（期望语言非空时才有意义）。 + + 违规判定： + - 期望 "ja"：正文块含 CJK 汉字且零假名（即纯中文）且长度 ≥ 阈值 + - 期望 "zh"：正文块含日文假名 + heading/table 块始终跳过（跟随模板 / 照抄源）。 + """ + if expected_language not in ("zh", "ja"): + return [] + violations: list[str] = [] + for b in blocks: + if b.type not in _CHECKED_BLOCK_TYPES: + continue + text = b.text or "" + if len(text) < MIN_VIOLATION_LEN: + continue + if expected_language == "ja": + if has_cjk(text) and not has_kana(text): + violations.append(text) + else: # zh + if has_kana(text): + violations.append(text) + return violations diff --git a/src/genesis/writer/models.py b/src/genesis/writer/models.py index 6342161..c2128ed 100644 --- a/src/genesis/writer/models.py +++ b/src/genesis/writer/models.py @@ -109,6 +109,19 @@ class GenerationContext: template_styles: set[str] prior_state: object | None = None # WriterState，避免循环 import 用 object impact_report: object | None = None # ImpactReport 影响调查书（生成主上下文） + output_language: str = "auto" # "auto" | "zh" | "ja"（步骤 1：用户可选输出语言） + + def _language_instruction(self) -> str: + """根据 output_language 生成【语言约束】段的具体指令（步骤 1）。""" + if self.output_language == "zh": + return "必须使用简体中文撰写（标题、正文与所有说明一律中文）。" + if self.output_language == "ja": + return "必ず日本語で記述すること（タイトル・本文・すべての説明は日本語）。" + # auto：沿用与标题语言一致的旧语义（向后兼容既有日文文档） + return ( + f"必须与章节标题「{self.title}」所用自然语言保持一致：" + "标题为日文则用日文撰写，为中文则用中文撰写，依此类推。" + ) def to_vars(self) -> dict: """返回供 prompt 渲染的变量字典。""" @@ -125,6 +138,7 @@ class GenerationContext: f"- {h}" for h in (getattr(tm, "sub_headings", None) or []) ), "prior_state": str(self.prior_state) if self.prior_state is not None else "", + "language_instruction": self._language_instruction(), "data": _format_chapter_data( self.structured_source, CHAPTER_SHEET_TYPES.get(self.chapter_id, []), @@ -132,6 +146,7 @@ class GenerationContext: "impact": _format_impact( self.impact_report, CHAPTER_IMPACT_ELEMENT.get(self.chapter_id, None), + self.output_language, ), } @@ -177,11 +192,30 @@ def _format_chapter_data(structured_source: object | None, sheet_types: list[She return "\n".join(lines).rstrip() -def _format_impact(report: object | None, element_type: ElementType | None = None) -> str: +# 影响调查标签本地化（步骤 B）：auto/ja 默认日文，zh 中文 +# 注：方括号标记 [..] 保留（中日通用），仅标签词本地化 +_IMPACT_LABELS: dict[str, dict[str, str]] = { + "zh": { + "new": "新建", "modified": "变更", "deleted": "删除", "warning": "警告", + "affected": "受影响", + }, + "ja": { + "new": "新規", "modified": "変更", "deleted": "削除", "warning": "警告", + "affected": "受影响", + }, +} + + +def _format_impact( + report: object | None, + element_type: ElementType | None = None, + output_language: str = "auto", +) -> str: """将影响调查书格式化为 prompt 可读文本（无报告/无分析时为空串）。 element_type 非 None 时仅保留该类型要素（章节级定向，design.md §6.8 ①）； 警告始终保留（不依赖要素类型）。 + output_language 控制标签语言（步骤 B）；auto 回落到 ja 标签。 """ if report is None: return "" @@ -192,6 +226,7 @@ def _format_impact(report: object | None, element_type: ElementType | None = Non def keep(el) -> bool: return element_type is None or el.element_type == element_type.value + lab = _IMPACT_LABELS.get(output_language, _IMPACT_LABELS["ja"]) summary = getattr(report, "summary", {}) or {} lines = [f"project_type={getattr(ca, 'project_type', '')}"] lines.append( @@ -204,15 +239,15 @@ def _format_impact(report: object | None, element_type: ElementType | None = Non ) for el in getattr(ca, "new_elements", []) or []: if keep(el): - lines.append(f"[新規] {el.element_id} {el.element_type} {el.name}") + lines.append(f"[{lab['new']}] {el.element_id} {el.element_type} {el.name}") for el in getattr(ca, "modified_elements", []) or []: if keep(el): impacted = ", ".join(el.impacted_existing) or "-" - lines.append(f"[変更] {el.element_id} {el.element_type} {el.name} → 受影响: {impacted}") + lines.append(f"[{lab['modified']}] {el.element_id} {el.element_type} {el.name} → {lab['affected']}: {impacted}") for el in getattr(ca, "deleted_elements", []) or []: if keep(el): impacted = ", ".join(el.impacted_existing) or "-" - lines.append(f"[削除] {el.element_id} {el.element_type} {el.name} → 受影响: {impacted}") + lines.append(f"[{lab['deleted']}] {el.element_id} {el.element_type} {el.name} → {lab['affected']}: {impacted}") for w in getattr(ca, "warnings", []) or []: - lines.append(f"[警告] {w.element_id}: {w.issue}") + lines.append(f"[{lab['warning']}] {w.element_id}: {w.issue}") return "\n".join(lines) diff --git a/src/genesis/writer/orchestrator.py b/src/genesis/writer/orchestrator.py index 1995407..19bedbd 100644 --- a/src/genesis/writer/orchestrator.py +++ b/src/genesis/writer/orchestrator.py @@ -56,6 +56,7 @@ class WriteOrchestrator: template_path: str | None = None, impact_report=None, meta: dict | None = None, + output_language: str = "auto", ) -> list[ChapterContent]: engine = engine or build_inference_engine() prompt_registry = prompt_registry or PromptRegistry() @@ -66,7 +67,7 @@ class WriteOrchestrator: if impact_report is not None: # 回填 structured_source，便于 QA/日志/后续下载 structured_source.impact_report = impact_report - ctxs = build_contexts(structured_source, samples_dir) + ctxs = build_contexts(structured_source, samples_dir, output_language=output_language) _warn_unanchored(ctxs) state = WriterState([c.chapter_id for c in ctxs]) agent = WriterAgent(session_id=session_id, engine=engine, prompt_registry=prompt_registry, state=state) diff --git a/src/genesis/writer/writer_agent.py b/src/genesis/writer/writer_agent.py index 5287229..687da73 100644 --- a/src/genesis/writer/writer_agent.py +++ b/src/genesis/writer/writer_agent.py @@ -13,6 +13,10 @@ from genesis.inference.types import Prompt, StructuredResult from genesis.writer.models import ChapterContent, GenerationContext from genesis.writer.writer_state import WriterState from genesis.writer.exceptions import WriterGenerationError +from genesis.writer.language import ( + find_language_violations, + resolve_expected_language, +) WRITER_PROMPT_TEMPLATE = ( @@ -33,10 +37,9 @@ WRITER_PROMPT_TEMPLATE = ( "严格以「参考资料（本章对应数据）」中的要件定义数据和「影响调查上下文」为核心依据；" "禁止输出与本章无关的系统整体架构、通用设计说明等内容，禁止套用其他章节的主题。" "若本章数据为空，则基于规则与影响调查上下文简要撰写，不得虚构数据。\n" - "【语言约束】章节正文（所有 block 的 text 字段）所使用的自然语言，" - "必须与章节标题「{{title}}」所用语言保持一致：标题为日文则用日文撰写，" - "为中文则用中文撰写，依此类推。" -) + "【语言约束】章节正文（所有 block 的 text 字段）所使用的自然语言：{{language_instruction}}\n" + ) + CHAPTER_OUTPUT_SCHEMA = { "type": "object", @@ -70,7 +73,7 @@ class WriterAgent: engine: InferenceEngine, prompt_registry: PromptRegistry, state: WriterState, - max_retries: int = 1, + max_retries: int = 2, ) -> None: self.session_id = session_id self.engine = engine @@ -132,6 +135,15 @@ class WriterAgent: def generate_chapter(self, context: GenerationContext) -> ChapterContent: self._chunk_source(context.structured_source) # 分块可用性验证（真实拼回留待后续） + + # 步骤 A：推导本章期望输出语言（仅依赖 context，循环前置，稳定） + v = context.to_vars() + expected = resolve_expected_language( + explicit=context.output_language, + title=context.title, + fallback_texts=[v.get("impact", "") or "", v.get("data", "") or ""], + ) + last_err: Exception | None = None for _ in range(max(1, self.max_retries)): try: @@ -148,6 +160,15 @@ class WriterAgent: # heading 块（prompt 约束为尽力而为，此处程序化强制） if not (getattr(context.template_marker, "sub_headings", None) or []): content.blocks = [b for b in content.blocks if b.type != "heading"] + # 步骤 A：输出语言一致性强制（期望语言可推导时，违规按失败重试） + if expected: + viol = find_language_violations(content.blocks, expected) + if viol: + last_err = WriterGenerationError( + f"章节 {context.chapter_id} 语言不一致（期望 {expected}，" + f"发现 {len(viol)} 处违规正文）" + ) + continue self.state.record_success(content) return content raise WriterGenerationError(f"章节 {context.chapter_id} 重试耗尽: {last_err}") diff --git a/tests/test_language_coverage.py b/tests/test_language_coverage.py new file mode 100644 index 0000000..ffb7df7 --- /dev/null +++ b/tests/test_language_coverage.py @@ -0,0 +1,102 @@ +"""步骤 A/C 边界覆盖测试（恢复覆盖率至 ≥99%）。""" +from __future__ import annotations + +from pathlib import Path +from types import SimpleNamespace + +from genesis.writer.language import find_language_violations +from genesis.writer.models import ChapterContent, ContentBlock, GenerationContext, ChapterSpec +from genesis.writer.writer_agent import WriterAgent +from genesis.writer.writer_state import WriterState +from genesis.writer.exceptions import WriterGenerationError +from genesis.qa.qa_loop import QALoop +from genesis.data_models import StructuredSource +from genesis.parsers.word_template_parser import WordTemplateParser + + +class _Reg: + @staticmethod + def get_or_create(name, template): + return SimpleNamespace(name=name, version="1", template=template) + + +def test_find_violations_unexpected_language_code_returns_empty(): + blocks = [ContentBlock(block_id="0", type="paragraph", text="中文正文内容充分长度足够")] + # 非 zh/ja 的期望语言 → 不强制，返回空 + assert find_language_violations(blocks, "xx") == [] + + +def test_writer_agent_zh_expected_rejects_japanese(): + class JpEngine: + def chat_structured(self, *, session_id, prompt, variables, schema, retry_count=2): + return SimpleNamespace( + data={"title": variables["title"], + "blocks": [{"type": "paragraph", + "text": "本機能は注文処理を行う画面であり、詳細は以下の通り。"}]}, + status="ok", + ) + agent = WriterAgent(session_id="s", engine=JpEngine(), prompt_registry=_Reg(), + state=WriterState(["db_design"]), max_retries=1) + ctx = GenerationContext(chapter_id="db_design", title="DB 設計", + template_marker=ChapterSpec(chapter_id="db_design", title="DB 設計"), + structured_source=None, write_rules=[], design_rules=[], + template_styles=set(), output_language="zh") + try: + agent.generate_chapter(ctx) + assert False, "应抛 WriterGenerationError" + except WriterGenerationError: + pass + + +def test_qa_loop_passes_with_explicit_zh_on_chinese_fake(tmp_path): + """显式 zh + 中文 Fake → 语言维度通过（验证透传链路）。""" + template = Path("samples/phase5-slice/template.docx") + if not template.exists(): + import pytest + pytest.skip("样本模板缺失") + parsed = WordTemplateParser().parse(str(template)) + ss = StructuredSource(template=parsed, tables=[], rule_docs=[], image_analyses=[], + existing_system=None, comments=[]) + out = tmp_path / "o.docx" + + class ZhEngine: + def chat_structured(self, *, session_id, prompt, variables, schema, retry_count=2): + return SimpleNamespace( + data={"title": variables["title"], + "blocks": [{"type": "paragraph", + "text": "由 FakeLLM 生成的充分说明内容，满足写入规则要求。"}]}, + status="ok", + ) + loop = QALoop(max_rounds=1) + report = loop.run(ss, str(out), samples_dir="nonexistent_dir_xyz", + engine=ZhEngine(), template_path=str(template), output_language="zh") + assert report.passed is True + + +def test_qa_loop_fails_with_explicit_ja_on_chinese_fake(tmp_path): + """显式 ja + 中文 Fake → 语言维度判失败。""" + template = Path("samples/phase5-slice/template.docx") + if not template.exists(): + import pytest + pytest.skip("样本模板缺失") + parsed = WordTemplateParser().parse(str(template)) + ss = StructuredSource(template=parsed, tables=[], rule_docs=[], image_analyses=[], + existing_system=None, comments=[]) + out = tmp_path / "o.docx" + + class ZhEngine: + def chat_structured(self, *, session_id, prompt, variables, schema, retry_count=2): + return SimpleNamespace( + data={"title": variables["title"], + "blocks": [{"type": "paragraph", + "text": "由 FakeLLM 生成的充分说明内容，满足写入规则要求。"}]}, + status="ok", + ) + loop = QALoop(max_rounds=1) + # 显式 ja + 中文 Fake：writer 生成阶段即因语言不一致硬失败（设计语义） + try: + loop.run(ss, str(out), samples_dir="nonexistent_dir_xyz", + engine=ZhEngine(), template_path=str(template), output_language="ja") + assert False, "应抛出 WriterGenerationError" + except WriterGenerationError: + pass diff --git a/tests/test_language_plumbing.py b/tests/test_language_plumbing.py new file mode 100644 index 0000000..7dcf274 --- /dev/null +++ b/tests/test_language_plumbing.py @@ -0,0 +1,93 @@ +"""步骤 1：语言参数贯通测试（config → GenerationContext → to_vars → context_builder → run_trial）。""" +from __future__ import annotations + +from genesis.data_models import ( + CellValue, ExcelTable, ParsedTemplate, Provenance, SheetType, StructuredSource, +) +from genesis.writer.context_builder import build_contexts +from genesis.writer.models import ChapterSpec, GenerationContext + + +def _src() -> StructuredSource: + return StructuredSource( + tables=[], + template=ParsedTemplate("t.docx", [__import__("genesis.data_models", fromlist=["ChapterMarker"]).ChapterMarker("heading", "1. はじめに", 1)], {}, {"used": []}), + rule_docs=[], image_analyses=[], existing_system=None, comments=[], + ) + + +def test_generation_context_default_output_language_is_auto(): + ctx = GenerationContext( + chapter_id="db_design", title="DB 設計", + template_marker=ChapterSpec(chapter_id="db_design", title="DB 設計"), + structured_source=None, write_rules=[], design_rules=[], template_styles=set(), + ) + assert ctx.output_language == "auto" + + +def test_to_vars_language_instruction_explicit_zh(): + ctx = GenerationContext( + chapter_id="db_design", title="DB 设計", + template_marker=ChapterSpec(chapter_id="db_design", title="DB 设計"), + structured_source=None, write_rules=[], design_rules=[], template_styles=set(), + output_language="zh", + ) + v = ctx.to_vars() + assert "language_instruction" in v + assert "简体中文" in v["language_instruction"] + + +def test_to_vars_language_instruction_explicit_ja(): + ctx = GenerationContext( + chapter_id="db_design", title="DB 設計", + template_marker=ChapterSpec(chapter_id="db_design", title="DB 設計"), + structured_source=None, write_rules=[], design_rules=[], template_styles=set(), + output_language="ja", + ) + v = ctx.to_vars() + assert "日本語" in v["language_instruction"] + + +def test_to_vars_language_instruction_auto_follows_title(): + ctx = GenerationContext( + chapter_id="db_design", title="DB 設計", + template_marker=ChapterSpec(chapter_id="db_design", title="DB 設計"), + structured_source=None, write_rules=[], design_rules=[], template_styles=set(), + output_language="auto", + ) + v = ctx.to_vars() + assert "标题" in v["language_instruction"] and "一致" in v["language_instruction"] + + +def test_context_builder_passes_output_language(): + src = _src() + ctxs = build_contexts(src, output_language="zh") + assert all(c.output_language == "zh" for c in ctxs) + # 回归：默认仍为 auto + ctxs2 = build_contexts(src) + assert all(c.output_language == "auto" for c in ctxs2) + + +def test_format_impact_zh_labels_localized(): + """步骤 B：output_language=zh 时影响调查标签本地化为中文，且不影响 ja 默认。""" + from genesis.data_models import ( + ChangeAnalysis, ChangeElement, ChangeType, ImpactReport, ImpactWarning, + ) + from genesis.writer.models import _format_impact + + ca = ChangeAnalysis( + project_type="enhancement", + new_elements=[ChangeElement("F001", "機能", "止损风控", ChangeType.NEW)], + modified_elements=[ChangeElement("F002", "機能", "下单流程", ChangeType.MODIFIED)], + deleted_elements=[ChangeElement("F030", "機能", "旧功能", ChangeType.DELETED)], + unchanged_elements=[], + warnings=[ImpactWarning("F030", "缺少既存対応")], + ) + report = ImpactReport(metadata={}, change_analysis=ca, summary={}) + + zh = _format_impact(report, None, "zh") + assert "[新建]" in zh and "[变更]" in zh and "[删除]" in zh and "[警告]" in zh + assert "受影响" in zh + # ja / auto 回落仍用日文标签 + ja = _format_impact(report, None, "ja") + assert "[新規]" in ja and "[削除]" in ja diff --git a/tests/test_phase5_e2e.py b/tests/test_phase5_e2e.py index 67ea653..bce5ced 100644 --- a/tests/test_phase5_e2e.py +++ b/tests/test_phase5_e2e.py @@ -12,13 +12,14 @@ from genesis.qa.qa_loop import QALoop class FakeEngine: def chat_structured(self, *, session_id, prompt, variables, schema, retry_count=2): cid = variables.get("title", "x") + # 日文模板章节（标题含假名）→ 返回日文正文，满足语言一致性强制 return SimpleNamespace( data={ "title": cid, "blocks": [ - {"type": "heading", "level": 2, "text": f"{cid} 小节"}, - {"type": "paragraph", "text": "由 FakeLLM 生成的充分说明内容，满足写入规则要求。"}, - {"type": "table", "caption": "示例表", "rows": [["列1", "列2"], ["值1", "值2"]]}, + {"type": "heading", "level": 2, "text": f"{cid} 小節"}, + {"type": "paragraph", "text": "本機能はFakeLLMにより生成された十分な説明内容であり、書込規則を満たす。"}, + {"type": "table", "caption": "示例表", "rows": [["列1", "列2"], ["値1", "値2"]]}, ], }, status="ok", @@ -50,5 +51,5 @@ def test_phase5_e2e_fake_llm(tmp_path): assert report.passed is True loaded = Document(str(out)) text = _all_text(loaded) - assert "由 FakeLLM 生成的充分说明内容" in text + assert "本機能はFakeLLMにより生成された十分な説明内容" in text assert "示例表" in text diff --git a/tests/test_scorer_language.py b/tests/test_scorer_language.py new file mode 100644 index 0000000..12216cd --- /dev/null +++ b/tests/test_scorer_language.py @@ -0,0 +1,92 @@ +"""步骤 C：QA 第 11 维度 language_consistency 测试。 + +评审 R1：不可验证（expected_language 为空/auto）→ 维度满分 1.0 通过，不拉低总分， +使既有 test_eval_scorer / test_phase5_scorer 的 total_score==1.0 断言继续成立。 +""" +from __future__ import annotations + +from genesis.data_models import ( + CellValue, ExcelTable, ParsedTemplate, Provenance, SheetType, StructuredSource, +) +from genesis.eval.scorer import ChapterArtifact, ChapterScorer, EvalReport + + +def _dim(report: EvalReport, name: str): + for d in report.dimensions: + if d.name == name: + return d + raise AssertionError(f"维度未找到: {name}") + + +def _source() -> StructuredSource: + cell = CellValue( + value="登録", + provenance=Provenance(file_name="f.xlsx", sheet_name="機能一覧", row=3, column="C", column_header="x"), + ) + table = ExcelTable( + name="機能一覧", detected_type=SheetType.FUNCTION, + extraction_method="structured", headers=["v"], rows=[{"v": cell}], + ) + return StructuredSource( + tables=[table], + template=ParsedTemplate(file_name="t.docx", sections=[], placeholders={}, styles={}), + rule_docs=[], image_analyses=[], existing_system=None, comments=[], + ) + + +def test_language_dim_unverifiable_is_full_score(): + """expected_language 为空（auto/不可验证）→ 维度满分通过，不拉低总分。""" + src = _source() + artifact = ChapterArtifact( + chapter_id="ch3", text="本機能は注文処理を行う画面である。", + source_uris=[], template_sections_expected=["ch3"], + expected_language="", + ) + report = ChapterScorer().score([artifact], src) + dim = _dim(report, "language_consistency") + assert dim.score == 1.0 + assert dim.passed is True + # 既有总分断言仍成立（全部确定性维度满分） + assert report.total_score == 1.0 + + +def test_language_dim_ja_expected_but_chinese_fails(): + src = _source() + artifact = ChapterArtifact( + chapter_id="ch3", text="这是一段纯中文的章节正文内容，应当判定为语言违规。", + source_uris=[], template_sections_expected=["ch3"], + expected_language="ja", + ) + report = ChapterScorer().score([artifact], src) + dim = _dim(report, "language_consistency") + assert dim.score == 0.0 + assert dim.passed is False + assert report.total_score < 1.0 + + +def test_language_dim_zh_expected_japanese_fails(): + src = _source() + artifact = ChapterArtifact( + chapter_id="ch3", text="本機能は注文処理を行う画面であり、詳細は以下の通り。", + source_uris=[], template_sections_expected=["ch3"], + expected_language="zh", + ) + report = ChapterScorer().score([artifact], src) + dim = _dim(report, "language_consistency") + assert dim.score == 0.0 + + +def test_language_dim_threshold_in_defaults(): + assert ChapterScorer().thresholds["language_consistency"] == 1.0 + + +def test_dimension_count_increases_by_one(): + src = _source() + artifact = ChapterArtifact( + chapter_id="ch3", text="正常内容", source_uris=[], template_sections_expected=["ch3"], + ) + before = len(ChapterScorer(thresholds={}, llm_evaluators={}).score([artifact], src).dimensions) + # 默认 scorer 含 language_consistency；对比一个不含该维度的基线不可行， + # 此处仅断言维度中包含 language_consistency 名称 + report = ChapterScorer().score([artifact], src) + assert any(d.name == "language_consistency" for d in report.dimensions) diff --git a/tests/test_writer_language.py b/tests/test_writer_language.py new file mode 100644 index 0000000..a39f89b --- /dev/null +++ b/tests/test_writer_language.py @@ -0,0 +1,172 @@ +"""步骤 A：确定性语言检测 + 输出语言强制测试。 + +- detect_script：含假名→ja，CJK 零假名→zh，否则 None +- resolve_expected_language：显式 > 标题假名 > 规则文档主导脚本 > ""（不可验证） +- find_language_violations：仅检正文块（heading/table 不检），长度≥12 防误杀 +- WriterAgent.generate_chapter：期望语言非空时违规即重试，耗尽→硬失败 +""" +from __future__ import annotations + +from genesis.writer.language import ( + detect_script, + find_language_violations, + resolve_expected_language, +) +from genesis.writer.models import ChapterContent, ContentBlock, GenerationContext, ChapterSpec +from genesis.writer.writer_agent import WriterAgent +from genesis.writer.writer_state import WriterState +from genesis.writer.exceptions import WriterGenerationError + + +# ---------- detect_script ---------- + +def test_detect_script_kana_is_ja(): + assert detect_script("機能一覧の説明を記述します。") == "ja" + + +def test_detect_script_cjk_no_kana_is_zh(): + assert detect_script("功能一览的说明内容。") == "zh" + + +def test_detect_script_ascii_is_none(): + assert detect_script("Hello world 123") is None + + +def test_detect_script_mixed_cjk_kana_is_ja(): + # 含假名即判 ja，即使混有汉字（中文不可能含日文假名） + assert detect_script("DB設計の概要を説明する。") == "ja" + + +# ---------- resolve_expected_language ---------- + +def test_resolve_explicit_overrides(): + assert resolve_expected_language("zh", "DB 設計", ["日本語の影響"]) == "zh" + assert resolve_expected_language("ja", "DB 设計", ["中文影响"]) == "ja" + + +def test_resolve_falls_back_to_title_kana(): + # 标题含假名→ja（注意：纯汉字日文标题如「DB設計」无假名，会落入 fallback） + assert resolve_expected_language("auto", "機能一覧の説明", []) == "ja" + + +def test_resolve_falls_back_to_fallback_text(): + # 标题纯汉字无假名（中文式）→ 看 fallback（日文影响书）→ ja + assert resolve_expected_language("auto", "DB 設計", ["影響調査の結果（日本語）"]) == "ja" + + +def test_resolve_unverifiable_when_no_hint(): + assert resolve_expected_language("auto", "DB 設計", ["123 abc"]) == "" + + +# ---------- find_language_violations ---------- + +def _blocks(*specs): + out = [] + for i, (t, text) in enumerate(specs): + out.append(ContentBlock(block_id=str(i), type=t, text=text)) + return out + + +def test_violation_ja_expected_but_pure_chinese_paragraph(): + blocks = _blocks(("paragraph", "这是一段纯中文的章节正文内容。")) + viol = find_language_violations(blocks, "ja") + assert viol # 期望日文却为中文 → 违规 + + +def test_no_violation_ja_expected_japanese_paragraph(): + blocks = _blocks(("paragraph", "本機能は注文処理を行う画面であり、詳細は以下の通り。")) + assert not find_language_violations(blocks, "ja") + + +def test_heading_and_table_blocks_excluded(): + # heading 与 table（照抄原文）即使含中文也不算违规 + blocks = _blocks( + ("heading", "機能一覧表"), + ("table", "機能ID 機能名"), # 表格不检 + ("paragraph", "本機能は注文処理を行う。"), + ) + assert not find_language_violations(blocks, "ja") + + +def test_short_chinese_not_flagged_under_threshold(): + # 长度<12 的短术语不误杀 + blocks = _blocks(("paragraph", "中文术语")) + assert not find_language_violations(blocks, "ja") + + +def test_violation_zh_expected_but_kana_present(): + blocks = _blocks(("paragraph", "本機能は注文処理を行う画面である。")) + assert find_language_violations(blocks, "zh") + + +# ---------- WriterAgent 强制 ---------- + +class MixedLanguageEngine: + """返回中英混杂（期望日文时违规）的章节内容。""" + def __init__(self, texts): + self._texts = list(texts) + self._i = 0 + def chat_structured(self, *, session_id, prompt, variables, schema, retry_count=2): + from types import SimpleNamespace + text = self._texts[min(self._i, len(self._texts) - 1)] + self._i += 1 + return SimpleNamespace( + data={"title": variables["title"], "blocks": [{"type": "paragraph", "text": text}]}, + status="ok", + ) + + +class FakePromptRegistry: + @staticmethod + def get_or_create(name, template): + from types import SimpleNamespace + return SimpleNamespace(name=name, version="1", template=template) + + +def _ctx_with_lang(cid, title, lang): + return GenerationContext( + chapter_id=cid, title=title, + template_marker=ChapterSpec(chapter_id=cid, title=title), + structured_source=None, write_rules=[], design_rules=[], + template_styles=set(), output_language=lang, + ) + + +def test_generate_chapter_retries_on_language_violation_then_passes(): + # 第一次返回中文（ja 期望→违规），第二次返回日文（通过） + engine = MixedLanguageEngine([ + "这是一段纯中文的章节正文内容，应当被判定为语言违规。", + "本機能は注文処理を行う画面であり、詳細は以下の通り記述する。", + ]) + agent = WriterAgent(session_id="s", engine=engine, + prompt_registry=FakePromptRegistry(), + state=WriterState(["db_design"]), max_retries=3) + content = agent.generate_chapter(_ctx_with_lang("db_design", "DB 設計", "ja")) + assert content is not None + assert "本機能" in content.blocks[0].text + + +def test_generate_chapter_hard_fails_when_violation_persists(): + # 始终返回中文（ja 期望），重试耗尽→硬失败 + engine = MixedLanguageEngine([ + "这是一段纯中文的章节正文内容，应当被判定为语言违规。", + "还是一段纯中文的章节正文内容，依旧违规。", + ]) + agent = WriterAgent(session_id="s", engine=engine, + prompt_registry=FakePromptRegistry(), + state=WriterState(["db_design"]), max_retries=2) + try: + agent.generate_chapter(_ctx_with_lang("db_design", "DB 設計", "ja")) + assert False, "应抛出 WriterGenerationError" + except WriterGenerationError as e: + assert "语言" in str(e) or "language" in str(e).lower() + + +def test_generate_chapter_auto_unverifiable_not_enforced(): + # auto 且无法推导期望语言（标题纯汉字无假名、无日文 fallback）→ 不强制，直接通过 + engine = MixedLanguageEngine(["这是中文正文但 auto 不可验证所以放行。"]) + agent = WriterAgent(session_id="s", engine=engine, + prompt_registry=FakePromptRegistry(), + state=WriterState(["db_design"]), max_retries=1) + content = agent.generate_chapter(_ctx_with_lang("db_design", "DB 設計", "auto")) + assert content is not None diff --git a/tests/test_zh_template.py b/tests/test_zh_template.py new file mode 100644 index 0000000..a3f1948 --- /dev/null +++ b/tests/test_zh_template.py @@ -0,0 +1,77 @@ +"""中文样本模板镜像测试（步骤 0）。 + +断言 zh 模板与 ja 模板章节结构一致：7 个 H1 章、锚点 id（section:xxx）逐一对应， +且锚点段落文本原样保留（映射/注入器零改动）。 +""" +from __future__ import annotations + +import shutil +from pathlib import Path + +import pytest +from docx import Document + +from genesis.parsers.word_template_parser import WordTemplateParser +from genesis.writer.template_mapper import map_template + +_SAMPLES = Path(r"D:\00_project\Genesis\samples") +_JA = _SAMPLES / "概要設計書テンプレート.docx" +_ZH = _SAMPLES / "概要设计书模板_中文.docx" + + +EXPECTED_ANCHORS = [ + "introduction", "function_list", "screen_list", "report_list", + "db_design", "if_definition", "batch_list", +] + + +@pytest.fixture +def zh_template(tmp_path): + """生成 zh 模板到临时目录并返回路径（不污染 samples/）。""" + from scripts.make_zh_template import build_zh_template + dst = tmp_path / "概要设计书模板_中文.docx" + build_zh_template(str(_JA), str(dst)) + return dst + + +def _parse(path: Path): + return WordTemplateParser().parse(str(path)) + + +def test_zh_template_has_same_seven_chapter_anchors(zh_template): + ja = _parse(_JA) + zh = _parse(zh_template) + + ja_ids = [ph for ph in ja.placeholders if ph.startswith("section:")] + zh_ids = [ph for ph in zh.placeholders if ph.startswith("section:")] + + assert sorted(zh_ids) == sorted(ja_ids) + assert zh_ids == [f"section:{a}" for a in EXPECTED_ANCHORS] + + +def test_zh_template_anchor_paragraphs_unchanged(zh_template): + """锚点段落文本在 zh 模板中必须与 ja 完全一致（映射/注入器依赖它）。""" + ja_doc = Document(str(_JA)) + zh_doc = Document(str(zh_template)) + ja_anchors = [p.text for p in ja_doc.paragraphs if p.text.startswith("{{section:")] + zh_anchors = [p.text for p in zh_doc.paragraphs if p.text.startswith("{{section:")] + assert zh_anchors == ja_anchors + + +def test_zh_template_headings_translated(zh_template): + zh_doc = Document(str(zh_template)) + texts = {p.text for p in zh_doc.paragraphs if p.style.name.startswith("Heading")} + assert "1. 前言" in texts + assert "2. 功能一览" in texts + assert "5. DB设计" in texts + assert "7. 批处理一览" in texts + + +def test_zh_template_maps_to_same_seven_chapters(zh_template): + """template_mapper 对 zh 模板产出 7 个章节，锚点与 ja 对应。""" + pt = _parse(zh_template) + specs = map_template(pt) + assert len(specs) == 7 + assert [s.section_placeholder for s in specs] == [ + f"section:{a}" for a in EXPECTED_ANCHORS + ]