feat: 完成 DOCX R4 元素级视觉门禁
建立 Chromium、Word 与 WPS 的可编辑内容、段落几何、页眉页脚、封面、媒体和逐页视觉比较链路。 支持封面独立分节、正文页码重启、主题页边距与横向纸张，并将自然分页差异降为诊断项。 修正代码块内边距和折叠表格单元格边框映射，14 套主题生产矩阵与六组布局视觉矩阵通过。
This commit is contained in:
SkyJourney
@@ -0,0 +1,452 @@
|
||||
import path from "node:path";
|
||||
import type { PreparedDocxMedia } from "@md-to-pdf/core";
|
||||
import {
|
||||
DRAWING_NAMESPACE,
|
||||
OFFICE_RELATIONSHIP_NAMESPACE,
|
||||
PACKAGE_RELATIONSHIP_NAMESPACE,
|
||||
WORD_NAMESPACE,
|
||||
firstDirectChild,
|
||||
parseXmlPart,
|
||||
type XmlElement
|
||||
} from "./ooxml.js";
|
||||
import {
|
||||
preparePandocMedia,
|
||||
type PandocMediaLayoutItem
|
||||
} from "./pandoc-media.js";
|
||||
import { readGeneratedDocxPackage } from "./reference-package.js";
|
||||
|
||||
const WORDPROCESSING_DRAWING_NAMESPACE =
|
||||
"http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing";
|
||||
const IMAGE_RELATIONSHIP_TYPE =
|
||||
"http://schemas.openxmlformats.org/officeDocument/2006/relationships/image";
|
||||
const EMUS_PER_TWIP = 635;
|
||||
const EMU_TOLERANCE = 20;
|
||||
const ASPECT_RATIO_TOLERANCE = 0.000_001;
|
||||
const PNG_SIGNATURE = Uint8Array.of(
|
||||
0x89,
|
||||
0x50,
|
||||
0x4e,
|
||||
0x47,
|
||||
0x0d,
|
||||
0x0a,
|
||||
0x1a,
|
||||
0x0a
|
||||
);
|
||||
|
||||
export interface DocxMediaAcceptancePage {
|
||||
widthTwips: number;
|
||||
heightTwips: number;
|
||||
marginsTwips: {
|
||||
top: number;
|
||||
right: number;
|
||||
bottom: number;
|
||||
left: number;
|
||||
};
|
||||
}
|
||||
|
||||
export interface DocxMediaAcceptanceExpectation {
|
||||
items: readonly PandocMediaLayoutItem[];
|
||||
maximumWidthEmu: number;
|
||||
maximumHeightEmu: number;
|
||||
}
|
||||
|
||||
export interface DocxMediaAcceptanceObservation {
|
||||
id: string;
|
||||
kind: PandocMediaLayoutItem["kind"];
|
||||
ordinal: number;
|
||||
altText: string;
|
||||
alignment: PandocMediaLayoutItem["alignment"];
|
||||
widthEmu: number;
|
||||
heightEmu: number;
|
||||
relationshipId: string;
|
||||
partName: string;
|
||||
}
|
||||
|
||||
export interface DocxMediaAcceptanceReport {
|
||||
expectedCount: number;
|
||||
drawingCount: number;
|
||||
inlineCount: number;
|
||||
anchorCount: number;
|
||||
maximumWidthEmu: number;
|
||||
maximumHeightEmu: number;
|
||||
observations: DocxMediaAcceptanceObservation[];
|
||||
checks: {
|
||||
exactCount: true;
|
||||
inlineLayout: true;
|
||||
dimensions: true;
|
||||
aspectRatio: true;
|
||||
alignment: true;
|
||||
relationships: true;
|
||||
pngMedia: true;
|
||||
noCroppingOrTransform: true;
|
||||
};
|
||||
}
|
||||
|
||||
interface Relationship {
|
||||
id: string;
|
||||
type: string;
|
||||
target: string;
|
||||
external: boolean;
|
||||
}
|
||||
|
||||
function fail(id: string, check: string, detail: string): never {
|
||||
throw new Error(
|
||||
`${id} 的 DOCX 媒体验收失败：${check}（${detail}）`
|
||||
);
|
||||
}
|
||||
|
||||
function numericAttribute(
|
||||
element: XmlElement,
|
||||
name: string,
|
||||
id: string,
|
||||
check: string
|
||||
) {
|
||||
const raw = element.getAttribute(name);
|
||||
const value = raw === null ? Number.NaN : Number(raw);
|
||||
if (!Number.isSafeInteger(value) || value <= 0) {
|
||||
fail(id, check, `${name}=${JSON.stringify(raw)}`);
|
||||
}
|
||||
return value;
|
||||
}
|
||||
|
||||
function withinTolerance(actual: number, expected: number) {
|
||||
return Math.abs(actual - expected) <= EMU_TOLERANCE;
|
||||
}
|
||||
|
||||
function paragraphForDrawing(
|
||||
drawing: XmlElement,
|
||||
id: string
|
||||
) {
|
||||
let current = drawing.parentNode;
|
||||
while (current) {
|
||||
const element = current as XmlElement;
|
||||
if (
|
||||
element.nodeType === 1 &&
|
||||
element.namespaceURI === WORD_NAMESPACE &&
|
||||
element.localName === "p"
|
||||
) {
|
||||
return element;
|
||||
}
|
||||
current = current.parentNode;
|
||||
}
|
||||
fail(id, "paragraph", "Drawing 不在正文段落中");
|
||||
}
|
||||
|
||||
function parseRelationships(content: Uint8Array, id: string) {
|
||||
const document = parseXmlPart(
|
||||
content,
|
||||
"word/_rels/document.xml.rels"
|
||||
);
|
||||
return Array.from(
|
||||
document.getElementsByTagNameNS(
|
||||
PACKAGE_RELATIONSHIP_NAMESPACE,
|
||||
"Relationship"
|
||||
)
|
||||
).map(
|
||||
(element): Relationship => ({
|
||||
id: element.getAttribute("Id") ?? "",
|
||||
type: element.getAttribute("Type") ?? "",
|
||||
target: element.getAttribute("Target") ?? "",
|
||||
external: element.getAttribute("TargetMode") === "External"
|
||||
})
|
||||
).map((relationship) => {
|
||||
if (!relationship.id || !relationship.target) {
|
||||
fail(id, "relationships", "图片关系缺少 ID 或 Target");
|
||||
}
|
||||
return relationship;
|
||||
});
|
||||
}
|
||||
|
||||
function resolveImagePart(target: string, id: string) {
|
||||
const partName = path.posix.normalize(
|
||||
path.posix.join("word", target.replaceAll("\\", "/"))
|
||||
);
|
||||
if (!partName.startsWith("word/") || partName.includes("../")) {
|
||||
fail(id, "relationships", `图片关系越界：${target}`);
|
||||
}
|
||||
return partName;
|
||||
}
|
||||
|
||||
function hasPngSignature(content: Uint8Array) {
|
||||
return (
|
||||
content.byteLength >= PNG_SIGNATURE.byteLength &&
|
||||
PNG_SIGNATURE.every((value, index) => content[index] === value)
|
||||
);
|
||||
}
|
||||
|
||||
export function createDocxMediaAcceptanceExpectation(
|
||||
media: PreparedDocxMedia,
|
||||
page: DocxMediaAcceptancePage
|
||||
): DocxMediaAcceptanceExpectation {
|
||||
const contentWidthTwips =
|
||||
page.widthTwips -
|
||||
page.marginsTwips.left -
|
||||
page.marginsTwips.right;
|
||||
const contentHeightTwips =
|
||||
page.heightTwips -
|
||||
page.marginsTwips.top -
|
||||
page.marginsTwips.bottom;
|
||||
if (contentWidthTwips <= 0 || contentHeightTwips <= 0) {
|
||||
throw new Error("DOCX 媒体验收页面内容区无效");
|
||||
}
|
||||
return {
|
||||
items: preparePandocMedia(media).layout,
|
||||
maximumWidthEmu: contentWidthTwips * EMUS_PER_TWIP,
|
||||
maximumHeightEmu: contentHeightTwips * EMUS_PER_TWIP
|
||||
};
|
||||
}
|
||||
|
||||
export function inspectDocxMediaAcceptance(
|
||||
content: Uint8Array,
|
||||
expectation: DocxMediaAcceptanceExpectation,
|
||||
id = "document"
|
||||
): DocxMediaAcceptanceReport {
|
||||
const package_ = readGeneratedDocxPackage(content);
|
||||
const entries = package_.entries;
|
||||
const document = parseXmlPart(
|
||||
entries.get("word/document.xml")!,
|
||||
"word/document.xml"
|
||||
);
|
||||
const relationshipContent = entries.get(
|
||||
"word/_rels/document.xml.rels"
|
||||
);
|
||||
if (!relationshipContent) {
|
||||
fail(id, "relationships", "缺少 document.xml.rels");
|
||||
}
|
||||
const relationships = parseRelationships(relationshipContent, id);
|
||||
const relationshipsById = new Map(
|
||||
relationships.map((relationship) => [relationship.id, relationship])
|
||||
);
|
||||
if (relationshipsById.size !== relationships.length) {
|
||||
fail(id, "relationships", "存在重复关系 ID");
|
||||
}
|
||||
|
||||
const drawings = Array.from(
|
||||
document.getElementsByTagNameNS(WORD_NAMESPACE, "drawing")
|
||||
);
|
||||
const inlineCount = document.getElementsByTagNameNS(
|
||||
WORDPROCESSING_DRAWING_NAMESPACE,
|
||||
"inline"
|
||||
).length;
|
||||
const anchorCount = document.getElementsByTagNameNS(
|
||||
WORDPROCESSING_DRAWING_NAMESPACE,
|
||||
"anchor"
|
||||
).length;
|
||||
if (
|
||||
drawings.length !== expectation.items.length ||
|
||||
inlineCount !== expectation.items.length ||
|
||||
anchorCount !== 0
|
||||
) {
|
||||
fail(
|
||||
id,
|
||||
"inline-layout",
|
||||
`expected=${expectation.items.length}, drawings=${drawings.length}, inline=${inlineCount}, anchor=${anchorCount}`
|
||||
);
|
||||
}
|
||||
|
||||
const usedRelationshipIds = new Set<string>();
|
||||
const usedParts = new Set<string>();
|
||||
const observations: DocxMediaAcceptanceObservation[] = [];
|
||||
for (const [index, item] of expectation.items.entries()) {
|
||||
const drawing = drawings[index]!;
|
||||
const inline = drawing.getElementsByTagNameNS(
|
||||
WORDPROCESSING_DRAWING_NAMESPACE,
|
||||
"inline"
|
||||
)[0];
|
||||
if (!inline) {
|
||||
fail(id, "inline-layout", `${item.id} 缺少 wp:inline`);
|
||||
}
|
||||
for (const attribute of ["distT", "distB", "distL", "distR"]) {
|
||||
if (inline.getAttribute(attribute) !== "0") {
|
||||
fail(id, "inline-layout", `${item.id} 的 ${attribute} 非零`);
|
||||
}
|
||||
}
|
||||
const extent = firstDirectChild(
|
||||
inline,
|
||||
WORDPROCESSING_DRAWING_NAMESPACE,
|
||||
"extent"
|
||||
);
|
||||
const documentProperties = firstDirectChild(
|
||||
inline,
|
||||
WORDPROCESSING_DRAWING_NAMESPACE,
|
||||
"docPr"
|
||||
);
|
||||
if (!extent || !documentProperties) {
|
||||
fail(id, "drawing-properties", `${item.id} 的属性不完整`);
|
||||
}
|
||||
const widthEmu = numericAttribute(
|
||||
extent,
|
||||
"cx",
|
||||
id,
|
||||
"dimensions"
|
||||
);
|
||||
const heightEmu = numericAttribute(
|
||||
extent,
|
||||
"cy",
|
||||
id,
|
||||
"dimensions"
|
||||
);
|
||||
if (
|
||||
!withinTolerance(widthEmu, item.widthEmu) ||
|
||||
!withinTolerance(heightEmu, item.heightEmu) ||
|
||||
widthEmu > expectation.maximumWidthEmu + EMU_TOLERANCE ||
|
||||
heightEmu > expectation.maximumHeightEmu + EMU_TOLERANCE
|
||||
) {
|
||||
fail(
|
||||
id,
|
||||
"dimensions",
|
||||
`${item.id}=${widthEmu}x${heightEmu}, expected=${item.widthEmu}x${item.heightEmu}`
|
||||
);
|
||||
}
|
||||
const ratioDelta = Math.abs(
|
||||
widthEmu / heightEmu - item.widthEmu / item.heightEmu
|
||||
) / (item.widthEmu / item.heightEmu);
|
||||
if (ratioDelta > ASPECT_RATIO_TOLERANCE) {
|
||||
fail(id, "aspect-ratio", `${item.id} delta=${ratioDelta}`);
|
||||
}
|
||||
|
||||
const transforms = Array.from(
|
||||
drawing.getElementsByTagNameNS(DRAWING_NAMESPACE, "xfrm")
|
||||
);
|
||||
const transformedExtent =
|
||||
transforms.length === 1
|
||||
? firstDirectChild(
|
||||
transforms[0]!,
|
||||
DRAWING_NAMESPACE,
|
||||
"ext"
|
||||
)
|
||||
: undefined;
|
||||
if (!transformedExtent) {
|
||||
fail(id, "drawing-transform", `${item.id} 缺少唯一 a:xfrm/a:ext`);
|
||||
}
|
||||
if (
|
||||
numericAttribute(transformedExtent, "cx", id, "dimensions") !==
|
||||
widthEmu ||
|
||||
numericAttribute(transformedExtent, "cy", id, "dimensions") !==
|
||||
heightEmu
|
||||
) {
|
||||
fail(id, "dimensions", `${item.id} 的双层尺寸不一致`);
|
||||
}
|
||||
const transform = transforms[0]!;
|
||||
if (
|
||||
transform.hasAttribute("rot") ||
|
||||
transform.hasAttribute("flipH") ||
|
||||
transform.hasAttribute("flipV") ||
|
||||
drawing.getElementsByTagNameNS(DRAWING_NAMESPACE, "srcRect")
|
||||
.length > 0
|
||||
) {
|
||||
fail(id, "cropping-transform", `${item.id} 存在裁切、旋转或翻转`);
|
||||
}
|
||||
const locks = drawing.getElementsByTagNameNS(
|
||||
DRAWING_NAMESPACE,
|
||||
"graphicFrameLocks"
|
||||
);
|
||||
if (
|
||||
locks.length !== 1 ||
|
||||
locks[0]!.getAttribute("noChangeAspect") !== "1"
|
||||
) {
|
||||
fail(id, "aspect-lock", `${item.id} 未锁定宽高比`);
|
||||
}
|
||||
if (
|
||||
documentProperties.getAttribute("descr") !== item.altText ||
|
||||
(documentProperties.getAttribute("title") ?? "").includes(
|
||||
"mdtp-media:"
|
||||
)
|
||||
) {
|
||||
fail(id, "alternative-text", `${item.id} 的替代文本或标题无效`);
|
||||
}
|
||||
|
||||
const paragraph = paragraphForDrawing(drawing, id);
|
||||
const properties = firstDirectChild(
|
||||
paragraph,
|
||||
WORD_NAMESPACE,
|
||||
"pPr"
|
||||
);
|
||||
const justification = properties
|
||||
? firstDirectChild(properties, WORD_NAMESPACE, "jc")
|
||||
: undefined;
|
||||
const alignment = justification?.getAttributeNS(
|
||||
WORD_NAMESPACE,
|
||||
"val"
|
||||
);
|
||||
if (alignment !== item.alignment) {
|
||||
fail(
|
||||
id,
|
||||
"alignment",
|
||||
`${item.id}=${alignment ?? "missing"}, expected=${item.alignment}`
|
||||
);
|
||||
}
|
||||
|
||||
const blips = Array.from(
|
||||
drawing.getElementsByTagNameNS(DRAWING_NAMESPACE, "blip")
|
||||
);
|
||||
const relationshipId =
|
||||
blips.length === 1
|
||||
? blips[0]!.getAttributeNS(
|
||||
OFFICE_RELATIONSHIP_NAMESPACE,
|
||||
"embed"
|
||||
)
|
||||
: null;
|
||||
if (!relationshipId || usedRelationshipIds.has(relationshipId)) {
|
||||
fail(id, "relationships", `${item.id} 的图片关系缺失或重复`);
|
||||
}
|
||||
const relationship = relationshipsById.get(relationshipId);
|
||||
if (
|
||||
!relationship ||
|
||||
relationship.type !== IMAGE_RELATIONSHIP_TYPE ||
|
||||
relationship.external
|
||||
) {
|
||||
fail(id, "relationships", `${item.id} 的图片关系类型无效`);
|
||||
}
|
||||
const partName = resolveImagePart(relationship.target, id);
|
||||
const image = entries.get(partName);
|
||||
if (
|
||||
!partName.toLowerCase().endsWith(".png") ||
|
||||
!image ||
|
||||
!hasPngSignature(image) ||
|
||||
usedParts.has(partName)
|
||||
) {
|
||||
fail(id, "png-media", `${item.id} 未绑定唯一有效 PNG`);
|
||||
}
|
||||
usedRelationshipIds.add(relationshipId);
|
||||
usedParts.add(partName);
|
||||
observations.push({
|
||||
id: item.id,
|
||||
kind: item.kind,
|
||||
ordinal: item.ordinal,
|
||||
altText: item.altText,
|
||||
alignment: item.alignment,
|
||||
widthEmu,
|
||||
heightEmu,
|
||||
relationshipId,
|
||||
partName
|
||||
});
|
||||
}
|
||||
|
||||
const imageRelationships = relationships.filter(
|
||||
(relationship) => relationship.type === IMAGE_RELATIONSHIP_TYPE
|
||||
);
|
||||
if (imageRelationships.length !== usedRelationshipIds.size) {
|
||||
fail(id, "relationships", "存在未使用或额外的正文图片关系");
|
||||
}
|
||||
|
||||
return {
|
||||
expectedCount: expectation.items.length,
|
||||
drawingCount: drawings.length,
|
||||
inlineCount,
|
||||
anchorCount,
|
||||
maximumWidthEmu: expectation.maximumWidthEmu,
|
||||
maximumHeightEmu: expectation.maximumHeightEmu,
|
||||
observations,
|
||||
checks: {
|
||||
exactCount: true,
|
||||
inlineLayout: true,
|
||||
dimensions: true,
|
||||
aspectRatio: true,
|
||||
alignment: true,
|
||||
relationships: true,
|
||||
pngMedia: true,
|
||||
noCroppingOrTransform: true
|
||||
}
|
||||
};
|
||||
}
|
||||
Reference in New Issue
Block a user