fix(writer): 防静默丢章告警 + 占位符正则宽容化(大小写/全角冒号)

This commit is contained in:
lhl
2026-08-23 14:47:14 +08:00
parent 48a9624723
commit da33df92e1
10 changed files with 119 additions and 9 deletions
+2 -1
View File
@@ -108,4 +108,5 @@
| 2026-08-13 23:20 | 测试验证 | P5-T10 引擎工厂接线:补 engine=None 委托工厂测试(orchestrator/qa_loop),全量 296 passed / 99.04% | tests/test_phase5_writer_orchestrator.py; tests/test_phase5_qa_loop.py | hy3-free |
| 2026-08-13 23:35 | 测试验证 | P5-T10 门禁诊断:修复 WriterGenerationError 吞掉底层 LLM 错误(如 401 详情),补透传测试 | src/genesis/writer/writer_agent.py; tests/test_phase5_writer_agent.py | hy3-free |
| 2026-08-13 23:50 | Agent 实现 | 修复 HttpLLMClient 在同步门禁中多次 asyncio.run 复用已关闭事件循环致 Event loop is closedchat 改为每次调用新建 client(保留 async with 协议)| src/genesis/inference/client.py | hy3-free |
| 2026-08-13 23:58 | Agent 实现 | P5-T10 语言对齐:WRITER_PROMPT_TEMPLATE 增加「正文语言须与章节标题一致」约束(模板日文则输出日文)| src/genesis/writer/writer_agent.py; tests/test_phase5_writer_agent.py | hy3-free |
| 2026-08-13 23:58 | Agent 实现 | P5-T10 语言对齐:WRITER_PROMPT_TEMPLATE 增加「正文语言须与章节标题一致」约束(模板日文则输出日文)| src/genesis/writer/writer_agent.py; tests/test_phase5_writer_agent.py | hy3-free |
| 2026-08-23 | Agent 实现 | 7 章模板只注入 2/3/5 章缺陷修复(用户圈定两项):①正则宽容化——word_template_parser.PLACEHOLDER_RE 键名大小写不敏感+支持全角冒号(解析归一为小写 section:id);template_mapper._SECTION_RE 与 docx_injector._SECTION_RE 加 IGNORECASE/全角冒号容忍;②防静默丢章——orchestrator 新增 _warn_unanchored(ctxs),对无 {{section:id}} 锚点章打 WARNING 列章名,generate 与 qa_loop._build 调用;TDD RED6 failed)→ GREEN(子集 25 passed)→ 全量 304 passed / 99.19% 覆盖,fail_under=99 达标;端到端冒烟 PASS{{Section:1}}/{{section2}} 注入 + 无锚点章告警) | src/genesis/parsers/word_template_parser.py; src/genesis/writer/template_mapper.py; src/genesis/writer/docx_injector.py; src/genesis/writer/orchestrator.py; src/genesis/qa/qa_loop.py; tests/test_word_template_parser.py; tests/test_phase5_template_mapper.py; tests/test_docx_injector.py; tests/test_phase5_writer_orchestrator.py; _AI_USAGE_LOG.md | deepseek-v4-flash |
+8 -4
View File
@@ -9,8 +9,12 @@ from docx.oxml.ns import qn
from genesis.data_models import ChapterMarker, ParsedTemplate
from genesis.parsers._word_common import heading_level
# 统一占位符正则:{{键名}} 或 {{键名:章节名}}spec §3.2
PLACEHOLDER_RE = re.compile(r"\{\{([a-z][a-z0-9_]*)(?::([a-z][a-z0-9_]*))?\}\}")
# 统一占位符正则:{{键名}} 或 {{键名:章节名}}spec §3.2
# 宽容:键名大小写不敏感、分隔符支持半角(:)/全角(:)冒号;解析时归一为小写键名 + 半角冒号。
PLACEHOLDER_RE = re.compile(
r"\{\{([A-Za-z][A-Za-z0-9_]*)(?:[:]([^}]+?))?\}\}",
re.IGNORECASE,
)
class WordTemplateParser:
@@ -37,9 +41,9 @@ class WordTemplateParser:
for m in PLACEHOLDER_RE.finditer(text):
if m.group(2):
key = f"{m.group(1)}:{m.group(2)}"
key = f"{m.group(1).lower()}:{m.group(2).strip()}"
else:
key = m.group(1)
key = m.group(1).lower()
placeholders[key] = text
sections.append(ChapterMarker(type="placeholder", name=key, level=0))
+2 -1
View File
@@ -11,7 +11,7 @@ from genesis.qa.validator import QAValidator
from genesis.writer.context_builder import build_contexts
from genesis.writer.docx_injector import Block, DocxInjector
from genesis.writer.models import ChapterContent
from genesis.writer.orchestrator import _section_id_of
from genesis.writer.orchestrator import _section_id_of, _warn_unanchored
from genesis.writer.renderer import render_chapter_blocks
from genesis.writer.writer_agent import WriterAgent
from genesis.writer.writer_state import WriterState
@@ -23,6 +23,7 @@ class QALoop:
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)
_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)
contents_map = dict(prev) if prev else {}
+3 -2
View File
@@ -21,7 +21,8 @@ from docx.document import Document as DocxDocument
from docx.oxml.ns import qn
from docx.text.paragraph import Paragraph
_SECTION_RE = re.compile(r"\{\{section:([^}]+)\}\}")
# 宽容:docx 正文锚点可能写为 {{Section:id}} / {{sectionid}}(大小写/全角冒号)
_SECTION_RE = re.compile(r"\{\{\s*section\s*[:]\s*([^}]+?)\s*\}\}", re.IGNORECASE)
_INLINE_RE = re.compile(r"\{\{([^}]+)\}\}")
@@ -63,7 +64,7 @@ class DocxInjector:
m = _SECTION_RE.search(para.text)
if not m:
continue
section_id = m.group(1)
section_id = m.group(1).strip()
blocks = sections.get(section_id)
if blocks is None:
# 未提供该章节内容 → 保留占位符段落,交由残留检查报错
+18
View File
@@ -1,6 +1,7 @@
"""Writer 编排:上下文装配 → 逐章生成 → 渲染 → docx 注入(Phase 5 垂直切片)。"""
from __future__ import annotations
import logging
from pathlib import Path
from genesis.data_models import StructuredSource
@@ -13,6 +14,8 @@ from genesis.writer.renderer import render_chapter_blocks
from genesis.writer.writer_agent import WriterAgent
from genesis.writer.writer_state import WriterState
_LOGGER = logging.getLogger(__name__)
def _section_id_of(placeholder: str | None) -> str | None:
if not placeholder or not placeholder.startswith("section:"):
@@ -20,6 +23,20 @@ def _section_id_of(placeholder: str | None) -> str | None:
return placeholder[len("section:"):]
def _warn_unanchored(ctxs) -> None:
"""防静默丢章:对缺少 {{section:<id>}} 锚点的章节打显式告警。
管线会对模板中每个 Heading 都生成内容,但只有带锚点的章才会注入 docx;
无锚点章生成后会被丢弃。此函数将其从「静默丢弃」变为「可见告警」。
"""
unanchored = [ctx.title for ctx in ctxs if not ctx.template_marker.section_placeholder]
if unanchored:
_LOGGER.warning(
"章节已生成但模板缺少 {{section:<id>}} 锚点,内容未注入(静默丢弃):%s",
", ".join(unanchored),
)
class WriteOrchestrator:
def generate(
self,
@@ -34,6 +51,7 @@ class WriteOrchestrator:
engine = engine or build_inference_engine()
prompt_registry = prompt_registry or PromptRegistry()
ctxs = build_contexts(structured_source, samples_dir)
_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)
+1 -1
View File
@@ -11,7 +11,7 @@ import re
from genesis.data_models import ParsedTemplate
from genesis.writer.models import ChapterSpec
_SECTION_RE = re.compile(r"^section:(.+)$")
_SECTION_RE = re.compile(r"^section:(.+)$", re.IGNORECASE)
def map_template(parsed: ParsedTemplate) -> list[ChapterSpec]:
+22
View File
@@ -48,6 +48,28 @@ def test_section_placeholder_replaced(tmp_path):
assert "以下がDB表定义です。" in full_text
def test_section_placeholder_case_insensitive(tmp_path):
# 宽容:docx 中锚点为 {{Section:id}}(大写键)也能注入
tpl = _make_template(tmp_path, "{{Section:db_design}}")
inj = DocxInjector(tpl)
out = inj.inject({"db_design": _section_blocks()}, {"doc_title": "概要設計書"})
full_text = "\n".join(p.text for p in out.paragraphs)
assert "{{Section:db_design}}" not in full_text
assert "3.1 テーブル一覧" in full_text
def test_section_placeholder_fullwidth_colon(tmp_path):
# 宽容:docx 中锚点为 {{sectionid}}(全角冒号)也能注入
tpl = _make_template(tmp_path, "{{sectiondb_design}}")
inj = DocxInjector(tpl)
out = inj.inject({"db_design": _section_blocks()}, {"doc_title": "概要設計書"})
full_text = "\n".join(p.text for p in out.paragraphs)
assert "{{sectiondb_design}}" not in full_text
assert "3.1 テーブル一覧" in full_text
# ---------- 行内占位符替换 ----------
def test_inline_meta_replaced(tmp_path):
+12
View File
@@ -30,3 +30,15 @@ def test_map_template_falls_back_without_placeholder():
specs = map_template(parsed)
assert specs[0].chapter_id == "chapter_1"
assert specs[0].section_placeholder is None
def test_map_template_section_re_case_insensitive():
# 宽容:占位符键名大小写不敏感(解析端已归一,防御性双保险)
sections = [
ChapterMarker(type="heading", name="画面一覧", level=1),
ChapterMarker(type="placeholder", name="Section:2", level=0),
]
parsed = ParsedTemplate(file_name="t.docx", sections=sections, placeholders={}, styles={})
specs = map_template(parsed)
assert specs[0].chapter_id == "2"
assert specs[0].section_placeholder == "Section:2"
+25
View File
@@ -83,3 +83,28 @@ def test_generate_delegates_to_factory_when_engine_none(monkeypatch, tmp_path):
assert len(contents) == 1
loaded = Document(str(out))
assert "自动生成的内容" in "\n".join(p.text for p in loaded.paragraphs)
def test_generate_warns_on_unanchored_heading(tmp_path, caplog):
# 防静默丢章:无 {{section:id}} 锚点的章 → 打 WARNING 并列出章名
tpl = tmp_path / "tpl.docx"
out = tmp_path / "out.docx"
doc = Document()
doc.add_paragraph("附録", style="Heading 1")
doc.save(str(tpl))
parsed = ParsedTemplate(
file_name=str(tpl),
sections=[ChapterMarker(type="heading", name="附録", level=1)],
placeholders={},
styles={"defined": ["Heading 1"], "used": ["Heading 1"]},
)
orch = WriteOrchestrator()
with caplog.at_level("WARNING", logger="genesis.writer.orchestrator"):
orch.generate(
SimpleNamespace(template=parsed),
str(out),
samples_dir="nonexistent_dir_xyz",
engine=FakeEngine(),
template_path=str(tpl),
)
assert any("附録" in r.message for r in caplog.records)
+26
View File
@@ -104,3 +104,29 @@ def test_parse_styles_collected(tmp_path):
assert "Heading 1" in result.styles["used"]
assert "Normal" in result.styles["used"]
assert "Heading 1" in result.styles["defined"]
def test_parse_extracts_placeholder_case_insensitive(tmp_path):
# 宽容:键名大小写不敏感 → 归一为小写 section:id
doc = new_document()
doc.add_paragraph("{{Section:2}}")
path = save_document(tmp_path, doc)
result = WordTemplateParser().parse(path)
assert result.placeholders == {"section:2": "{{Section:2}}"}
ph = [s for s in result.sections if s.type == "placeholder"]
assert [s.name for s in ph] == ["section:2"]
def test_parse_extracts_placeholder_fullwidth_colon(tmp_path):
# 宽容:全角冒号(:)也视为分隔符 → 归一为半角 section:id
doc = new_document()
doc.add_paragraph("{{section2}}")
path = save_document(tmp_path, doc)
result = WordTemplateParser().parse(path)
assert result.placeholders == {"section:2": "{{section2}}"}
ph = [s for s in result.sections if s.type == "placeholder"]
assert [s.name for s in ph] == ["section:2"]