d909d1bd1b
实现侧： - 上下文：任务范围拆分为 范围校验/范围授权；索引按可发现口径重建、索引新鲜度改对称差；依赖校验统一快照漂移说明。 - 知识方法：方法与材料读取口径统一；超限方法材料按可选省略，核对路径不再二次计费；删除无合同的读时重算。 - 任务运行：新增 context.usage/tool.denied 事件类型；连接池常驻并在装配生命周期内开关；调用结算与核对分列。 - 效果评测/审校修订/交付连载/作者经验/作品规划：凭据冻结、标定消费、导出补证、事实引文核对等收尾修复。 - 资源加载：能力正文不再夹带索引用的导航注记（该注记此前进入角色与技能的模型提示）。 - 元数据：受保护骨架与代码保护属性对齐；字段校验与内置结构口径同步。 - 基础设施：环境预检进入装配生命周期；数据库连接运行期字段不参与相等比较；索引指纹归一化 jsonb 浮点。 - 删除被替代实现：7 份旧提示词模板与空壳 资料来源 读取器。 用例侧： - 用例身份与导航元信息迁移；夹具补生命周期、同库暴露与模板封存； - 本轮定向修复：方法材料省略、事实引文、迁移回执、额度与暂停用例、慢用例超时预算等。
402 lines
18 KiB
Python
402 lines
18 KiB
Python
"""固定定稿格式的文本保真、成品核对及标准包结构。"""
|
||
|
||
import copy
|
||
import hashlib
|
||
from io import BytesIO
|
||
from uuid import UUID
|
||
from xml.etree import ElementTree as ET
|
||
from zipfile import ZIP_STORED, ZipFile
|
||
|
||
import pytest
|
||
|
||
from muse.交付连载.导出格式 import EPUB, Markdown, Word, 导出, 校验, 纯文本
|
||
from muse.交付连载.模型 import 交付错误
|
||
|
||
格式 = ("txt", "md", "docx", "epub")
|
||
读取器 = {"txt": 纯文本.读取, "md": Markdown.读取, "docx": Word.读取, "epub": EPUB.读取}
|
||
|
||
|
||
@pytest.fixture
|
||
def 固定定稿():
|
||
paragraphs = [
|
||
" 原文标记：夜色里，青禾握着旧钥匙。 ",
|
||
"",
|
||
"",
|
||
"半角 空格\t制表\r单CR\n单LF\r\n双字符换行。",
|
||
"e\u0301 ≠ é；家庭👨👩👧👦；😀；汉字𠮷。",
|
||
"# 标题 *不是强调* `代码` [链接](https://example.invalid) <script> </script>",
|
||
'章节 {"paragraph_bytes":[999]}\n<!-- muse-chapter-end -->',
|
||
"\t \u00a0 ",
|
||
"",
|
||
]
|
||
return {
|
||
"delivery_id": UUID("00000000-0000-4000-8000-000000000027"),
|
||
"manifest_hash": "a" * 64,
|
||
"manifest": {
|
||
"materials": {
|
||
"title": '雾港 #1 <>& "题名"',
|
||
"author_name": "青禾 & 海风",
|
||
"introduction": "简介首行\r\n第二行 连续空白\t结尾。",
|
||
"synopsis": "梗概与正文独立。",
|
||
"characters": "青禾：守门人。\n海风：旅人。",
|
||
}
|
||
},
|
||
"chapters": [
|
||
{
|
||
"chapter_id": 'c-2<&"',
|
||
"title": "第一章 # 标题 <>& *正文*",
|
||
"paragraphs": paragraphs,
|
||
"text": "\n".join(paragraphs),
|
||
},
|
||
{
|
||
"chapter_id": "c-1",
|
||
"title": "第二章",
|
||
"paragraphs": ["", "最后一盏灯。", ""],
|
||
"text": "\n最后一盏灯。\n",
|
||
},
|
||
],
|
||
}
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d001",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["四格式可见正文重读保留段落与Unicode"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
@pytest.mark.parametrize("format", 格式)
|
||
def test_四格式可见正文重读保留段落与Unicode__27d001(固定定稿, format):
|
||
data = 导出(format, 固定定稿)
|
||
actual = 纯文本.读取(data, 固定定稿) if format == "txt" else 读取器[format](data)
|
||
expected_chapters = copy.deepcopy(固定定稿["chapters"])
|
||
if format == "txt":
|
||
for chapter in expected_chapters:
|
||
chapter["paragraphs"] = None
|
||
chapter.pop("chapter_id")
|
||
assert actual["chapters"] == expected_chapters
|
||
assert actual["materials"] == 固定定稿["manifest"]["materials"]
|
||
proof = 校验(format, data, 固定定稿)
|
||
assert proof["verified"] is True
|
||
assert proof["sha256"] == hashlib.sha256(data).hexdigest()
|
||
assert proof["chapter_ids"] == (None if format == "txt" else ['c-2<&"', "c-1"])
|
||
assert proof["chapters"][0]["paragraph_count"] == (None if format == "txt" else 9)
|
||
assert (
|
||
proof["chapters"][0]["text_sha256"]
|
||
== hashlib.sha256(固定定稿["chapters"][0]["text"].encode("utf-8")).hexdigest()
|
||
)
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d002",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["同一定稿重复导出字节相同且不修改输入"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
@pytest.mark.parametrize("format", 格式)
|
||
def test_同一定稿重复导出字节相同且不修改输入__27d002(固定定稿, format):
|
||
before = copy.deepcopy(固定定稿)
|
||
first = 导出(format, 固定定稿)
|
||
assert first == 导出(format, 固定定稿)
|
||
assert 固定定稿 == before
|
||
|
||
|
||
def _修改条目(data: bytes, name: str, modify) -> bytes:
|
||
output = BytesIO()
|
||
with ZipFile(BytesIO(data)) as source, ZipFile(output, "w") as target:
|
||
for info in source.infolist():
|
||
value = source.read(info.filename)
|
||
target.writestr(info, modify(value) if info.filename == name else value)
|
||
return output.getvalue()
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d003",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["篡改实际正文而保留身份必须被拒绝"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
@pytest.mark.parametrize("format", 格式)
|
||
def test_篡改实际正文而保留身份必须被拒绝__27d003(固定定稿, format):
|
||
data = 导出(format, 固定定稿)
|
||
|
||
def replace(value):
|
||
assert "原文标记".encode() in value
|
||
return value.replace("原文标记".encode(), "篡改标记".encode())
|
||
|
||
if format in ("docx", "epub"):
|
||
name = "word/document.xml" if format == "docx" else "EPUB/chapter-000001.xhtml"
|
||
damaged = _修改条目(data, name, replace)
|
||
else:
|
||
damaged = replace(data)
|
||
with pytest.raises(交付错误) as error:
|
||
校验(format, damaged, 固定定稿)
|
||
assert error.value.错误码 == "EXPORT_VERIFICATION_FAILED"
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d004",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["同一清单哈希不能掩盖成品顺序缺章段落或材料变化"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
@pytest.mark.parametrize("format", 格式)
|
||
@pytest.mark.parametrize("mutation", ("reorder", "missing", "paragraph", "materials"))
|
||
def test_同一清单哈希不能掩盖成品顺序缺章段落或材料变化__27d004(固定定稿, format, mutation):
|
||
changed = copy.deepcopy(固定定稿)
|
||
if mutation == "reorder":
|
||
changed["chapters"].reverse()
|
||
elif mutation == "missing":
|
||
changed["chapters"].pop()
|
||
elif mutation == "paragraph":
|
||
# 可见文本相同，段落边界不同；不能靠整章文本哈希替代段落核对。
|
||
changed["chapters"][0]["paragraphs"] = [changed["chapters"][0]["text"]]
|
||
else:
|
||
changed["manifest"]["materials"]["synopsis"] = "材料被换成另一版本。"
|
||
data = 导出(format, changed)
|
||
if format == "txt" and mutation == "paragraph":
|
||
assert 校验(format, data, 固定定稿)["verified"]
|
||
return
|
||
with pytest.raises(交付错误) as error:
|
||
校验(format, data, 固定定稿)
|
||
assert error.value.错误码 == "EXPORT_VERIFICATION_FAILED"
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d005",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["XML格式拒绝无法表示字符而非静默丢弃"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
@pytest.mark.parametrize("format", ("md", "docx", "epub"))
|
||
@pytest.mark.parametrize("char", ("\x00", "\x0b", "\uffff"))
|
||
def test_XML格式拒绝无法表示字符而非静默丢弃__27d005(固定定稿, format, char):
|
||
固定定稿["chapters"][0]["paragraphs"][0] += char
|
||
固定定稿["chapters"][0]["text"] = "\n".join(固定定稿["chapters"][0]["paragraphs"])
|
||
with pytest.raises(交付错误) as error:
|
||
导出(format, 固定定稿)
|
||
assert error.value.错误码 == "EXPORT_INVALID_TEXT"
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d006",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["TXT保留XML不支持的合法UTF8控制字符"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
def test_TXT保留XML不支持的合法UTF8控制字符__27d006(固定定稿):
|
||
固定定稿["chapters"][0]["paragraphs"][0] += "\x00\x0b\uffff"
|
||
固定定稿["chapters"][0]["text"] = "\n".join(固定定稿["chapters"][0]["paragraphs"])
|
||
actual = 纯文本.读取(导出("txt", 固定定稿), 固定定稿)
|
||
assert actual["chapters"][0]["text"] == 固定定稿["chapters"][0]["text"]
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d007",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["孤立代理码和正文段落不一致明确失败"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
@pytest.mark.parametrize("format", 格式)
|
||
def test_孤立代理码和正文段落不一致明确失败__27d007(固定定稿, format):
|
||
固定定稿["chapters"][0]["paragraphs"][0] += "\ud800"
|
||
with pytest.raises(交付错误) as mismatch:
|
||
导出(format, 固定定稿)
|
||
assert mismatch.value.错误码 == "EXPORT_INVALID_TEXT"
|
||
固定定稿["chapters"][0]["text"] = "\n".join(固定定稿["chapters"][0]["paragraphs"])
|
||
with pytest.raises(交付错误) as surrogate:
|
||
导出(format, 固定定稿)
|
||
assert surrogate.value.错误码 == "EXPORT_INVALID_TEXT"
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d008",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["截断成品明确失败"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
@pytest.mark.parametrize("format", 格式)
|
||
def test_截断成品明确失败__27d008(固定定稿, format):
|
||
data = 导出(format, 固定定稿)
|
||
with pytest.raises(交付错误) as error:
|
||
校验(format, data[: len(data) // 2], 固定定稿)
|
||
assert error.value.错误码 == "EXPORT_VERIFICATION_FAILED"
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d009",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["DOCX真实页面与文本节点满足阅读结构"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
def test_DOCX真实页面与文本节点满足阅读结构__27d009(固定定稿):
|
||
data = 导出("docx", 固定定稿)
|
||
ns = {"w": "http://schemas.openxmlformats.org/wordprocessingml/2006/main"}
|
||
w = "{" + ns["w"] + "}"
|
||
with ZipFile(BytesIO(data)) as archive:
|
||
document = ET.fromstring(archive.read("word/document.xml"))
|
||
styles = ET.fromstring(archive.read("word/styles.xml"))
|
||
assert document.find("w:body/w:sectPr/w:pgSz", ns).attrib == {
|
||
w + "w": "11906",
|
||
w + "h": "16838",
|
||
}
|
||
chapter_style = styles.find("w:style[@w:styleId='Heading1']", ns)
|
||
assert chapter_style.find("w:pPr/w:pageBreakBefore", ns) is not None
|
||
assert chapter_style.find("w:pPr/w:outlineLvl", ns).get(w + "val") == "0"
|
||
assert styles.find("w:docDefaults/w:rPrDefault/w:rPr/w:rFonts", ns).get(w + "eastAsia")
|
||
body = document.findall("w:body/w:sdt/w:sdtContent", ns)[5:]
|
||
assert [len(b.findall("w:p", ns)) - 1 for b in body] == [9, 3]
|
||
assert document.findall(".//w:br[@w:type='textWrapping']", ns)
|
||
assert document.findall(".//w:br", ns)
|
||
assert not document.findall(".//w:vanish", ns)
|
||
assert all(i.date_time == (1980, 1, 1, 0, 0, 0) for i in archive.infolist())
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d010",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["EPUB真实mimetype包关系目录与spine一致"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
def test_EPUB真实mimetype包关系目录与spine一致__27d010(固定定稿):
|
||
data = 导出("epub", 固定定稿)
|
||
ns = {"o": "http://www.idpf.org/2007/opf", "x": "http://www.w3.org/1999/xhtml"}
|
||
with ZipFile(BytesIO(data)) as archive:
|
||
first = archive.infolist()[0]
|
||
assert (first.filename, first.compress_type, first.extra) == ("mimetype", ZIP_STORED, b"")
|
||
assert archive.read("mimetype") == b"application/epub+zip"
|
||
package = ET.fromstring(archive.read("EPUB/package.opf"))
|
||
items = {n.get("id"): n.get("href") for n in package.findall("o:manifest/o:item", ns)}
|
||
order = [n.get("idref") for n in package.findall("o:spine/o:itemref", ns)]
|
||
nav = ET.fromstring(archive.read("EPUB/nav.xhtml"))
|
||
links = [n.get("href") for n in nav.findall(".//x:nav/x:ol/x:li/x:a", ns)]
|
||
assert links == [items[key] for key in order]
|
||
assert order == ["materials", "chapter-000001", "chapter-000002"]
|
||
paragraphs = ET.fromstring(archive.read("EPUB/chapter-000001.xhtml")).findall(
|
||
"x:body/x:section/x:p", ns
|
||
)
|
||
assert [p.text or "" for p in paragraphs] == 固定定稿["chapters"][0]["paragraphs"]
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d011",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["字体隐藏或EPUB关系被破坏不能冒充文本校验通过"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
@pytest.mark.parametrize("format", ("docx", "epub"))
|
||
def test_字体隐藏或EPUB关系被破坏不能冒充文本校验通过__27d011(固定定稿, format):
|
||
data = 导出(format, 固定定稿)
|
||
if format == "docx":
|
||
data = _修改条目(
|
||
data, "word/styles.xml", lambda b: b.replace(b"<w:rPr>", b"<w:rPr><w:vanish/>")
|
||
)
|
||
else:
|
||
data = _修改条目(
|
||
data,
|
||
"EPUB/package.opf",
|
||
lambda b: b.replace(b'properties="nav"', b'properties="cover-image"'),
|
||
)
|
||
with pytest.raises(交付错误) as error:
|
||
校验(format, data, 固定定稿)
|
||
assert error.value.错误码 == "EXPORT_VERIFICATION_FAILED"
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d012",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["未知格式明确拒绝"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
def test_未知格式明确拒绝__27d012(固定定稿):
|
||
with pytest.raises(交付错误) as error:
|
||
导出("pdf", 固定定稿)
|
||
assert error.value.错误码 == "EXPORT_FORMAT_UNSUPPORTED"
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d013",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["改变渲染而未改变文本节点的篡改也被拒绝"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
@pytest.mark.parametrize("format", ("md", "epub"))
|
||
def test_改变渲染而未改变文本节点的篡改也被拒绝__27d013(固定定稿, format):
|
||
data = 导出(format, 固定定稿)
|
||
if format == "md":
|
||
data = data.replace(b"*", b"*")
|
||
else:
|
||
data = _修改条目(
|
||
data,
|
||
"EPUB/chapter-000001.xhtml",
|
||
lambda b: b.replace(b"</p>", "</p>额外正文".encode(), 1),
|
||
)
|
||
with pytest.raises(交付错误) as error:
|
||
校验(format, data, 固定定稿)
|
||
assert error.value.错误码 == "EXPORT_VERIFICATION_FAILED"
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d014",
|
||
environment="离线",
|
||
given="本作者完整目录、确切正文版本与独立隔离数据",
|
||
when="通过定稿、S02导出、跨章候选或连载反馈实际入口执行并检查失败恢复",
|
||
then=["TXT是直接交稿文本且不包含工程头"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
def test_TXT是直接交稿文本且不包含工程头__27d014(固定定稿):
|
||
data = 导出("txt", 固定定稿)
|
||
assert data.startswith('雾港 #1 <>& "题名"\n\n青禾 & 海风\n\n简介\n'.encode())
|
||
assert not data.startswith(b"Muse")
|
||
assert b'"delivery_id"' not in data and "UTF-8字节数".encode() not in data
|
||
assert 固定定稿["chapters"][0]["text"].encode() in data
|
||
assert data.endswith((固定定稿["chapters"][1]["text"] + "\n").encode())
|
||
|
||
|
||
@pytest.mark.case_id(
|
||
"NC-w27-27d015",
|
||
environment="离线",
|
||
given="同一可见TXT及不同外部清单身份",
|
||
when="核对实际文本而不编码的身份改变",
|
||
then=["身份标为not_encoded，回执不报告虚构的章节身份"],
|
||
contract="docs/系统架构/新版设计/模块设计/B08-交付连载.md",
|
||
)
|
||
def test_TXT不把外部清单身份冒充文件内验证结果__27d015(固定定稿):
|
||
data = 导出("txt", 固定定稿)
|
||
changed = copy.deepcopy(固定定稿)
|
||
changed["chapters"][0]["chapter_id"] = "another-chapter"
|
||
changed["delivery_id"] = UUID("00000000-0000-4000-8000-000000000028")
|
||
changed["manifest_hash"] = "b" * 64
|
||
proof = 校验("txt", data, changed)
|
||
assert proof["verified"] is True # 只证明相同的可见文稿。
|
||
assert proof["identities"] == "not_encoded" and proof["chapter_ids"] is None
|
||
assert "identities" not in proof["verified_scope"]
|
||
assert all("chapter_id" not in row for row in proof["chapters"])
|
||
actual = 纯文本.读取(data, changed)
|
||
assert "delivery_id" not in actual and "manifest_hash" not in actual
|