feat: 建立统一语义文档模型
This commit is contained in:
SkyJourney
@@ -295,6 +295,14 @@ describe("共享应用服务", () => {
|
||||
resources: []
|
||||
});
|
||||
expect(prepared.document.metadata.title).toBe("DOCX 文档");
|
||||
expect(prepared.document.semanticDocument).toEqual({
|
||||
schemaVersion: 1,
|
||||
titlePolicy: {
|
||||
metadataTitle: "suppress",
|
||||
firstBodyHeading: "keep"
|
||||
},
|
||||
regions: []
|
||||
});
|
||||
expect(prepared.theme.manifest.id).toBe("test-theme");
|
||||
expect(prepared.theme.source).toBe("bundled");
|
||||
expect(prepared.theme.css).toContain(
|
||||
|
||||
@@ -59,6 +59,14 @@ const prepared: PreparedDocxExport = {
|
||||
keywords: [],
|
||||
language: "zh-CN"
|
||||
},
|
||||
semanticDocument: {
|
||||
schemaVersion: 1,
|
||||
titlePolicy: {
|
||||
metadataTitle: "suppress",
|
||||
firstBodyHeading: "keep"
|
||||
},
|
||||
regions: []
|
||||
},
|
||||
features: [],
|
||||
warnings: []
|
||||
},
|
||||
|
||||
@@ -70,6 +70,14 @@ const prepared: PreparedDocxExport = {
|
||||
keywords: [],
|
||||
language: "zh-CN"
|
||||
},
|
||||
semanticDocument: {
|
||||
schemaVersion: 1,
|
||||
titlePolicy: {
|
||||
metadataTitle: "suppress",
|
||||
firstBodyHeading: "keep"
|
||||
},
|
||||
regions: []
|
||||
},
|
||||
features: [],
|
||||
warnings: ["原始图片已降级"]
|
||||
},
|
||||
|
||||
@@ -1,5 +1,6 @@
|
||||
import type { ExportConfig } from "./export-config.js";
|
||||
import type { DocumentProfile } from "./document-profile.js";
|
||||
import type { SemanticDocumentModel } from "./semantic-document.js";
|
||||
import type { ThemeFeature } from "./theme.js";
|
||||
|
||||
export interface MarkdownDocumentMetadata {
|
||||
@@ -16,6 +17,7 @@ export interface RenderedMarkdownDocument {
|
||||
articleHtml: string;
|
||||
bodyHtml: string;
|
||||
metadata: MarkdownDocumentMetadata;
|
||||
semanticDocument: SemanticDocumentModel;
|
||||
features: ThemeFeature[];
|
||||
warnings: string[];
|
||||
}
|
||||
|
||||
@@ -4,4 +4,5 @@ export * from "./docx-style.js";
|
||||
export * from "./document-profile.js";
|
||||
export * from "./document-link.js";
|
||||
export * from "./export-config.js";
|
||||
export * from "./semantic-document.js";
|
||||
export * from "./theme.js";
|
||||
|
||||
@@ -0,0 +1,137 @@
|
||||
import { z } from "zod";
|
||||
import { documentProfileNameSchema } from "./document-profile.js";
|
||||
|
||||
export const semanticDocumentRoleSchema = z.enum([
|
||||
"official-masthead",
|
||||
"official-classification",
|
||||
"official-secrecy",
|
||||
"official-urgency",
|
||||
"official-issuer",
|
||||
"official-issue-row",
|
||||
"official-number",
|
||||
"official-signatory",
|
||||
"official-title",
|
||||
"official-signature",
|
||||
"official-signature-issuer",
|
||||
"official-signature-date",
|
||||
"official-edition",
|
||||
"official-copy-to",
|
||||
"official-printing-row",
|
||||
"official-printing-office",
|
||||
"official-printing-date",
|
||||
"briefing-masthead",
|
||||
"briefing-masthead-text",
|
||||
"briefing-meta",
|
||||
"briefing-issue",
|
||||
"briefing-publisher",
|
||||
"briefing-signatory",
|
||||
"briefing-date",
|
||||
"briefing-title",
|
||||
"briefing-contact",
|
||||
"project-report-cover",
|
||||
"project-report-project-name",
|
||||
"project-report-title",
|
||||
"project-report-version",
|
||||
"project-report-owner",
|
||||
"project-report-prepared-by",
|
||||
"project-report-date",
|
||||
"tender-cover",
|
||||
"tender-copy-mark",
|
||||
"tender-project-name",
|
||||
"tender-project-number",
|
||||
"tender-title",
|
||||
"tender-volume",
|
||||
"tender-bidder",
|
||||
"tender-representative",
|
||||
"tender-date"
|
||||
]);
|
||||
|
||||
export type SemanticDocumentRole = z.infer<
|
||||
typeof semanticDocumentRoleSchema
|
||||
>;
|
||||
|
||||
export interface SemanticDocumentTextNode {
|
||||
kind: "text";
|
||||
role: SemanticDocumentRole;
|
||||
text: string;
|
||||
label?: string | undefined;
|
||||
}
|
||||
|
||||
export interface SemanticDocumentGroupNode {
|
||||
kind: "group";
|
||||
role: SemanticDocumentRole;
|
||||
children: SemanticDocumentNode[];
|
||||
}
|
||||
|
||||
export type SemanticDocumentNode =
|
||||
| SemanticDocumentTextNode
|
||||
| SemanticDocumentGroupNode;
|
||||
|
||||
export const semanticDocumentNodeSchema: z.ZodType<
|
||||
SemanticDocumentNode
|
||||
> = z.lazy(() =>
|
||||
z.discriminatedUnion("kind", [
|
||||
z.object({
|
||||
kind: z.literal("text"),
|
||||
role: semanticDocumentRoleSchema,
|
||||
text: z.string().min(1).max(1000),
|
||||
label: z.string().min(1).max(100).optional()
|
||||
}),
|
||||
z.object({
|
||||
kind: z.literal("group"),
|
||||
role: semanticDocumentRoleSchema,
|
||||
children: z.array(semanticDocumentNodeSchema).min(1).max(30)
|
||||
})
|
||||
])
|
||||
);
|
||||
|
||||
export const semanticDocumentRegionKindSchema = z.enum([
|
||||
"cover",
|
||||
"prefix",
|
||||
"suffix"
|
||||
]);
|
||||
|
||||
export type SemanticDocumentRegionKind = z.infer<
|
||||
typeof semanticDocumentRegionKindSchema
|
||||
>;
|
||||
|
||||
export const semanticDocumentSectionIntentSchema = z.object({
|
||||
headerFooter: z.literal("none"),
|
||||
pageNumber: z.literal("hidden"),
|
||||
breakAfter: z.literal("next-page"),
|
||||
followingPageNumberStart: z.number().int().min(0).max(100000)
|
||||
});
|
||||
|
||||
export type SemanticDocumentSectionIntent = z.infer<
|
||||
typeof semanticDocumentSectionIntentSchema
|
||||
>;
|
||||
|
||||
export const semanticDocumentRegionSchema = z.object({
|
||||
kind: semanticDocumentRegionKindSchema,
|
||||
nodes: z.array(semanticDocumentNodeSchema).min(1).max(30),
|
||||
section: semanticDocumentSectionIntentSchema.optional()
|
||||
});
|
||||
|
||||
export type SemanticDocumentRegion = z.infer<
|
||||
typeof semanticDocumentRegionSchema
|
||||
>;
|
||||
|
||||
export const semanticDocumentTitlePolicySchema = z.object({
|
||||
metadataTitle: z.enum(["emit", "suppress"]),
|
||||
firstBodyHeading: z.enum(["keep", "suppress"])
|
||||
});
|
||||
|
||||
export type SemanticDocumentTitlePolicy = z.infer<
|
||||
typeof semanticDocumentTitlePolicySchema
|
||||
>;
|
||||
|
||||
export const semanticDocumentModelSchema = z.object({
|
||||
schemaVersion: z.literal(1),
|
||||
profile: documentProfileNameSchema.optional(),
|
||||
titlePolicy: semanticDocumentTitlePolicySchema,
|
||||
regions: z.array(semanticDocumentRegionSchema).max(10)
|
||||
});
|
||||
|
||||
export type SemanticDocumentModel = z.infer<
|
||||
typeof semanticDocumentModelSchema
|
||||
>;
|
||||
@@ -18,6 +18,14 @@ describe("分页文档载荷", () => {
|
||||
keywords: [],
|
||||
language: "zh-CN"
|
||||
},
|
||||
semanticDocument: {
|
||||
schemaVersion: 1,
|
||||
titlePolicy: {
|
||||
metadataTitle: "suppress",
|
||||
firstBodyHeading: "keep"
|
||||
},
|
||||
regions: []
|
||||
},
|
||||
features: ["table"],
|
||||
warnings: []
|
||||
};
|
||||
|
||||
@@ -27,6 +27,7 @@
|
||||
"@mdit/plugin-tasklist": "^1.0.1",
|
||||
"@md-to-pdf/core": "0.1.0",
|
||||
"@md-to-pdf/markdown-echarts": "0.1.0",
|
||||
"@md-to-pdf/semantic-document": "0.1.0",
|
||||
"gray-matter": "^4.0.3",
|
||||
"highlight.js": "^11.11.1",
|
||||
"markdown-it": "^14.3.0",
|
||||
|
||||
@@ -1,280 +1,226 @@
|
||||
import type {
|
||||
DocumentProfile,
|
||||
MarkdownDocumentMetadata
|
||||
SemanticDocumentModel,
|
||||
SemanticDocumentNode,
|
||||
SemanticDocumentRole
|
||||
} from "@md-to-pdf/core";
|
||||
|
||||
export interface RenderedDocumentStructure {
|
||||
profile: DocumentProfile["profile"];
|
||||
profile: NonNullable<SemanticDocumentModel["profile"]>;
|
||||
prefixHtml: string;
|
||||
suffixHtml: string;
|
||||
}
|
||||
|
||||
interface HtmlRole {
|
||||
tagName: string;
|
||||
className: string;
|
||||
}
|
||||
|
||||
const htmlRoles: Readonly<Record<SemanticDocumentRole, HtmlRole>> = {
|
||||
"official-masthead": {
|
||||
tagName: "header",
|
||||
className: "doc-masthead doc-masthead-official"
|
||||
},
|
||||
"official-classification": {
|
||||
tagName: "div",
|
||||
className: "doc-classification"
|
||||
},
|
||||
"official-secrecy": {
|
||||
tagName: "span",
|
||||
className: "doc-secrecy"
|
||||
},
|
||||
"official-urgency": {
|
||||
tagName: "span",
|
||||
className: "doc-urgency"
|
||||
},
|
||||
"official-issuer": {
|
||||
tagName: "p",
|
||||
className: "doc-issuer"
|
||||
},
|
||||
"official-issue-row": {
|
||||
tagName: "div",
|
||||
className: "doc-issue-row"
|
||||
},
|
||||
"official-number": {
|
||||
tagName: "span",
|
||||
className: "doc-number"
|
||||
},
|
||||
"official-signatory": {
|
||||
tagName: "span",
|
||||
className: "doc-signatory"
|
||||
},
|
||||
"official-title": {
|
||||
tagName: "h1",
|
||||
className: "doc-title"
|
||||
},
|
||||
"official-signature": {
|
||||
tagName: "section",
|
||||
className: "doc-signature"
|
||||
},
|
||||
"official-signature-issuer": {
|
||||
tagName: "p",
|
||||
className: "doc-signature-issuer"
|
||||
},
|
||||
"official-signature-date": {
|
||||
tagName: "time",
|
||||
className: "doc-signature-date"
|
||||
},
|
||||
"official-edition": {
|
||||
tagName: "footer",
|
||||
className: "doc-edition"
|
||||
},
|
||||
"official-copy-to": {
|
||||
tagName: "p",
|
||||
className: "doc-copy-to"
|
||||
},
|
||||
"official-printing-row": {
|
||||
tagName: "div",
|
||||
className: "doc-printing-row"
|
||||
},
|
||||
"official-printing-office": {
|
||||
tagName: "span",
|
||||
className: "doc-printing-office"
|
||||
},
|
||||
"official-printing-date": {
|
||||
tagName: "time",
|
||||
className: "doc-printing-date"
|
||||
},
|
||||
"briefing-masthead": {
|
||||
tagName: "header",
|
||||
className: "doc-masthead doc-masthead-briefing"
|
||||
},
|
||||
"briefing-masthead-text": {
|
||||
tagName: "p",
|
||||
className: "doc-briefing-masthead"
|
||||
},
|
||||
"briefing-meta": {
|
||||
tagName: "div",
|
||||
className: "doc-briefing-meta"
|
||||
},
|
||||
"briefing-issue": {
|
||||
tagName: "span",
|
||||
className: "doc-briefing-issue"
|
||||
},
|
||||
"briefing-publisher": {
|
||||
tagName: "span",
|
||||
className: "doc-briefing-publisher"
|
||||
},
|
||||
"briefing-signatory": {
|
||||
tagName: "span",
|
||||
className: "doc-briefing-signatory"
|
||||
},
|
||||
"briefing-date": {
|
||||
tagName: "time",
|
||||
className: "doc-briefing-date"
|
||||
},
|
||||
"briefing-title": {
|
||||
tagName: "h1",
|
||||
className: "doc-title"
|
||||
},
|
||||
"briefing-contact": {
|
||||
tagName: "footer",
|
||||
className: "doc-briefing-contact"
|
||||
},
|
||||
"project-report-cover": {
|
||||
tagName: "header",
|
||||
className: "doc-cover doc-cover-project-report"
|
||||
},
|
||||
"project-report-project-name": {
|
||||
tagName: "p",
|
||||
className: "doc-cover-project-name"
|
||||
},
|
||||
"project-report-title": {
|
||||
tagName: "h1",
|
||||
className: "doc-cover-title"
|
||||
},
|
||||
"project-report-version": {
|
||||
tagName: "p",
|
||||
className: "doc-cover-version"
|
||||
},
|
||||
"project-report-owner": {
|
||||
tagName: "p",
|
||||
className: "doc-cover-owner"
|
||||
},
|
||||
"project-report-prepared-by": {
|
||||
tagName: "p",
|
||||
className: "doc-cover-prepared-by"
|
||||
},
|
||||
"project-report-date": {
|
||||
tagName: "time",
|
||||
className: "doc-cover-date"
|
||||
},
|
||||
"tender-cover": {
|
||||
tagName: "header",
|
||||
className: "doc-cover doc-cover-tender"
|
||||
},
|
||||
"tender-copy-mark": {
|
||||
tagName: "p",
|
||||
className: "doc-cover-copy-mark"
|
||||
},
|
||||
"tender-project-name": {
|
||||
tagName: "p",
|
||||
className: "doc-cover-project-name"
|
||||
},
|
||||
"tender-project-number": {
|
||||
tagName: "p",
|
||||
className: "doc-cover-project-number"
|
||||
},
|
||||
"tender-title": {
|
||||
tagName: "h1",
|
||||
className: "doc-cover-title"
|
||||
},
|
||||
"tender-volume": {
|
||||
tagName: "p",
|
||||
className: "doc-cover-volume"
|
||||
},
|
||||
"tender-bidder": {
|
||||
tagName: "p",
|
||||
className: "doc-cover-bidder"
|
||||
},
|
||||
"tender-representative": {
|
||||
tagName: "p",
|
||||
className: "doc-cover-representative"
|
||||
},
|
||||
"tender-date": {
|
||||
tagName: "time",
|
||||
className: "doc-cover-date"
|
||||
}
|
||||
};
|
||||
|
||||
function renderNode(node: SemanticDocumentNode): string {
|
||||
const role = htmlRoles[node.role];
|
||||
const content =
|
||||
node.kind === "group"
|
||||
? node.children.map(renderNode).join("")
|
||||
: `${node.label ? `<span>${escapeHtml(node.label)}</span>` : ""}${escapeHtml(node.text)}`;
|
||||
return `<${role.tagName} class="${role.className}">${content}</${role.tagName}>`;
|
||||
}
|
||||
|
||||
export function renderDocumentStructure(
|
||||
metadata: MarkdownDocumentMetadata
|
||||
model: SemanticDocumentModel
|
||||
): RenderedDocumentStructure | undefined {
|
||||
const document = metadata.document;
|
||||
if (!document) {
|
||||
if (!model.profile) {
|
||||
return undefined;
|
||||
}
|
||||
|
||||
switch (document.profile) {
|
||||
case "official":
|
||||
return renderOfficialStructure(metadata.title, document);
|
||||
case "briefing":
|
||||
return renderBriefingStructure(metadata.title, document);
|
||||
case "project-report":
|
||||
return renderProjectReportStructure(metadata.title, document);
|
||||
case "tender":
|
||||
return renderTenderStructure(metadata.title, document);
|
||||
}
|
||||
}
|
||||
|
||||
function renderOfficialStructure(
|
||||
title: string,
|
||||
document: Extract<DocumentProfile, { profile: "official" }>
|
||||
): RenderedDocumentStructure {
|
||||
const classification = joinParts(
|
||||
[
|
||||
renderText("span", "doc-secrecy", document.secrecy),
|
||||
renderText("span", "doc-urgency", document.urgency)
|
||||
],
|
||||
"doc-classification"
|
||||
);
|
||||
const issueRow = joinParts(
|
||||
[
|
||||
renderText("span", "doc-number", document.number),
|
||||
renderLabeledText(
|
||||
"span",
|
||||
"doc-signatory",
|
||||
"签发人：",
|
||||
document.signatory
|
||||
)
|
||||
],
|
||||
"doc-issue-row"
|
||||
);
|
||||
const masthead = joinParts(
|
||||
[
|
||||
classification,
|
||||
renderText("p", "doc-issuer", document.issuer),
|
||||
issueRow
|
||||
],
|
||||
"doc-masthead doc-masthead-official",
|
||||
"header"
|
||||
);
|
||||
const signature = joinParts(
|
||||
[
|
||||
renderText("p", "doc-signature-issuer", document.issuer),
|
||||
renderTime("doc-signature-date", document.date)
|
||||
],
|
||||
"doc-signature",
|
||||
"section"
|
||||
);
|
||||
const edition = joinParts(
|
||||
[
|
||||
document.copyTo?.length
|
||||
? `<p class="doc-copy-to"><span>抄送：</span>${escapeHtml(document.copyTo.join("、"))}</p>`
|
||||
: "",
|
||||
joinParts(
|
||||
[
|
||||
renderText(
|
||||
"span",
|
||||
"doc-printing-office",
|
||||
document.printingOffice
|
||||
),
|
||||
document.date
|
||||
? `<time class="doc-printing-date">${escapeHtml(document.date)}印发</time>`
|
||||
: ""
|
||||
],
|
||||
"doc-printing-row"
|
||||
)
|
||||
],
|
||||
"doc-edition",
|
||||
"footer"
|
||||
);
|
||||
|
||||
const prefixHtml = model.regions
|
||||
.filter(
|
||||
(region) =>
|
||||
region.kind === "cover" || region.kind === "prefix"
|
||||
)
|
||||
.flatMap((region) => region.nodes)
|
||||
.map(renderNode)
|
||||
.join("");
|
||||
const suffixHtml = model.regions
|
||||
.filter((region) => region.kind === "suffix")
|
||||
.flatMap((region) => region.nodes)
|
||||
.map(renderNode)
|
||||
.join("");
|
||||
return {
|
||||
profile: document.profile,
|
||||
prefixHtml:
|
||||
masthead + renderText("h1", "doc-title", title),
|
||||
suffixHtml: signature + edition
|
||||
profile: model.profile,
|
||||
prefixHtml,
|
||||
suffixHtml
|
||||
};
|
||||
}
|
||||
|
||||
function renderBriefingStructure(
|
||||
title: string,
|
||||
document: Extract<DocumentProfile, { profile: "briefing" }>
|
||||
): RenderedDocumentStructure {
|
||||
const details = joinParts(
|
||||
[
|
||||
renderText("span", "doc-briefing-issue", document.issue),
|
||||
renderText(
|
||||
"span",
|
||||
"doc-briefing-publisher",
|
||||
document.publisher
|
||||
),
|
||||
renderLabeledText(
|
||||
"span",
|
||||
"doc-briefing-signatory",
|
||||
"签发：",
|
||||
document.signatory
|
||||
),
|
||||
renderTime("doc-briefing-date", document.date)
|
||||
],
|
||||
"doc-briefing-meta"
|
||||
);
|
||||
const masthead = joinParts(
|
||||
[
|
||||
renderText(
|
||||
"p",
|
||||
"doc-briefing-masthead",
|
||||
document.masthead
|
||||
),
|
||||
details
|
||||
],
|
||||
"doc-masthead doc-masthead-briefing",
|
||||
"header"
|
||||
);
|
||||
const contact = renderText(
|
||||
"footer",
|
||||
"doc-briefing-contact",
|
||||
document.contact
|
||||
);
|
||||
|
||||
return {
|
||||
profile: document.profile,
|
||||
prefixHtml:
|
||||
masthead + renderText("h1", "doc-title", title),
|
||||
suffixHtml: contact
|
||||
};
|
||||
}
|
||||
|
||||
function renderProjectReportStructure(
|
||||
title: string,
|
||||
document: Extract<
|
||||
DocumentProfile,
|
||||
{ profile: "project-report" }
|
||||
>
|
||||
): RenderedDocumentStructure {
|
||||
const cover = joinParts(
|
||||
[
|
||||
renderText(
|
||||
"p",
|
||||
"doc-cover-project-name",
|
||||
document.projectName
|
||||
),
|
||||
renderText(
|
||||
"h1",
|
||||
"doc-cover-title",
|
||||
document.documentType || title
|
||||
),
|
||||
renderText("p", "doc-cover-version", document.version),
|
||||
renderLabeledText(
|
||||
"p",
|
||||
"doc-cover-owner",
|
||||
"建设单位：",
|
||||
document.owner
|
||||
),
|
||||
renderLabeledText(
|
||||
"p",
|
||||
"doc-cover-prepared-by",
|
||||
"编制单位：",
|
||||
document.preparedBy
|
||||
),
|
||||
renderTime("doc-cover-date", document.date)
|
||||
],
|
||||
"doc-cover doc-cover-project-report",
|
||||
"header"
|
||||
);
|
||||
|
||||
return {
|
||||
profile: document.profile,
|
||||
prefixHtml: cover,
|
||||
suffixHtml: ""
|
||||
};
|
||||
}
|
||||
|
||||
function renderTenderStructure(
|
||||
title: string,
|
||||
document: Extract<DocumentProfile, { profile: "tender" }>
|
||||
): RenderedDocumentStructure {
|
||||
const cover = joinParts(
|
||||
[
|
||||
renderText("p", "doc-cover-copy-mark", document.copyMark),
|
||||
renderText(
|
||||
"p",
|
||||
"doc-cover-project-name",
|
||||
document.projectName
|
||||
),
|
||||
renderText(
|
||||
"p",
|
||||
"doc-cover-project-number",
|
||||
document.projectNumber
|
||||
),
|
||||
renderText("h1", "doc-cover-title", title),
|
||||
renderText("p", "doc-cover-volume", document.volume),
|
||||
renderLabeledText(
|
||||
"p",
|
||||
"doc-cover-bidder",
|
||||
"投标人：",
|
||||
document.bidder
|
||||
),
|
||||
renderLabeledText(
|
||||
"p",
|
||||
"doc-cover-representative",
|
||||
"法定代表人或授权代表：",
|
||||
document.representative
|
||||
),
|
||||
renderTime("doc-cover-date", document.date)
|
||||
],
|
||||
"doc-cover doc-cover-tender",
|
||||
"header"
|
||||
);
|
||||
|
||||
return {
|
||||
profile: document.profile,
|
||||
prefixHtml: cover,
|
||||
suffixHtml: ""
|
||||
};
|
||||
}
|
||||
|
||||
function renderText(
|
||||
tagName: string,
|
||||
className: string,
|
||||
value: string | undefined
|
||||
) {
|
||||
return value
|
||||
? `<${tagName} class="${className}">${escapeHtml(value)}</${tagName}>`
|
||||
: "";
|
||||
}
|
||||
|
||||
function renderLabeledText(
|
||||
tagName: string,
|
||||
className: string,
|
||||
label: string,
|
||||
value: string | undefined
|
||||
) {
|
||||
return value
|
||||
? `<${tagName} class="${className}"><span>${label}</span>${escapeHtml(value)}</${tagName}>`
|
||||
: "";
|
||||
}
|
||||
|
||||
function renderTime(className: string, value: string | undefined) {
|
||||
return value
|
||||
? `<time class="${className}">${escapeHtml(value)}</time>`
|
||||
: "";
|
||||
}
|
||||
|
||||
function joinParts(
|
||||
parts: readonly string[],
|
||||
className: string,
|
||||
tagName = "div"
|
||||
) {
|
||||
const content = parts.filter(Boolean).join("");
|
||||
return content
|
||||
? `<${tagName} class="${className}">${content}</${tagName}>`
|
||||
: "";
|
||||
}
|
||||
|
||||
function escapeHtml(value: string) {
|
||||
return value.replace(
|
||||
/[&<>"']/gu,
|
||||
|
||||
@@ -13,6 +13,7 @@ import {
|
||||
documentProfileSchema,
|
||||
isSafeDocumentLink
|
||||
} from "@md-to-pdf/core";
|
||||
import { createSemanticDocumentModel } from "@md-to-pdf/semantic-document";
|
||||
import {
|
||||
markdownItECharts,
|
||||
type MarkdownItEChartsErrorContext
|
||||
@@ -348,12 +349,17 @@ export function renderMarkdown(
|
||||
};
|
||||
const rendered = markdown.render(parsed.content, environment);
|
||||
const bodyHtml = sanitizeHtml(rendered, safeHtmlOptions);
|
||||
const firstBodyHeading = extractFirstHeading(parsed.content);
|
||||
const metadata = normalizeMetadata(
|
||||
parsed.data,
|
||||
options.language ?? "zh-CN",
|
||||
extractFirstHeading(parsed.content)
|
||||
firstBodyHeading
|
||||
);
|
||||
const structure = renderDocumentStructure(metadata);
|
||||
const semanticDocument = createSemanticDocumentModel({
|
||||
metadata,
|
||||
firstBodyHeading
|
||||
});
|
||||
const structure = renderDocumentStructure(semanticDocument);
|
||||
const profileAttribute = structure
|
||||
? ` data-document-profile="${structure.profile}"`
|
||||
: "";
|
||||
@@ -364,6 +370,7 @@ export function renderMarkdown(
|
||||
articleHtml,
|
||||
bodyHtml,
|
||||
metadata,
|
||||
semanticDocument,
|
||||
features: detectFeatures(bodyHtml),
|
||||
warnings
|
||||
};
|
||||
|
||||
@@ -32,6 +32,14 @@ const answer = 42;
|
||||
expect(result.bodyHtml).toContain("hljs-keyword");
|
||||
expect(result.bodyHtml).toContain("footnote");
|
||||
expect(result.metadata.title).toBe("示例文档");
|
||||
expect(result.semanticDocument).toEqual({
|
||||
schemaVersion: 1,
|
||||
titlePolicy: {
|
||||
metadataTitle: "suppress",
|
||||
firstBodyHeading: "keep"
|
||||
},
|
||||
regions: []
|
||||
});
|
||||
expect(result.features).toEqual(
|
||||
expect.arrayContaining(["code", "table", "task-list", "footnote"])
|
||||
);
|
||||
@@ -152,6 +160,8 @@ document:
|
||||
expect(result.articleHtml).toContain(
|
||||
'<footer class="doc-edition">'
|
||||
);
|
||||
expect(result.semanticDocument.profile).toBe("official");
|
||||
expect(result.semanticDocument.regions).toHaveLength(2);
|
||||
expect(result.bodyHtml).not.toContain("doc-masthead");
|
||||
});
|
||||
|
||||
|
||||
@@ -0,0 +1,15 @@
|
||||
# @md-to-pdf/semantic-document
|
||||
|
||||
统一语义文档模型构建器。该包把规范化 Front Matter 和正文标题信息转换为
|
||||
主题无关、渲染器无关的结构树，供 HTML/PDF 与 DOCX 共用。
|
||||
|
||||
当前支持：
|
||||
|
||||
- 普通文档标题去重策略；
|
||||
- 政企公文版头、文号、签发人、落款和版记；
|
||||
- 工作简报版头、期号、发布单位、签发和联系信息；
|
||||
- 项目报告独立封面；
|
||||
- 标书独立封面；
|
||||
- 封面无页眉页脚、不显示页码、正文重新从 1 编号的分节意图。
|
||||
|
||||
本包不生成 HTML、不调用 Pandoc、不处理 OOXML，也不包含主题 ID 分支。
|
||||
@@ -0,0 +1,29 @@
|
||||
{
|
||||
"name": "@md-to-pdf/semantic-document",
|
||||
"version": "0.1.0",
|
||||
"private": true,
|
||||
"type": "module",
|
||||
"main": "./dist/index.js",
|
||||
"types": "./dist/index.d.ts",
|
||||
"exports": {
|
||||
".": {
|
||||
"types": "./dist/index.d.ts",
|
||||
"import": "./dist/index.js"
|
||||
}
|
||||
},
|
||||
"files": [
|
||||
"dist"
|
||||
],
|
||||
"scripts": {
|
||||
"dev": "tsc -p tsconfig.json --watch --preserveWatchOutput",
|
||||
"build": "tsc -p tsconfig.json",
|
||||
"test": "vitest run",
|
||||
"typecheck": "tsc -p tsconfig.json --noEmit --pretty false"
|
||||
},
|
||||
"dependencies": {
|
||||
"@md-to-pdf/core": "0.1.0"
|
||||
},
|
||||
"devDependencies": {
|
||||
"vitest": "^4.1.10"
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1 @@
|
||||
export * from "./model-builder.js";
|
||||
@@ -0,0 +1,268 @@
|
||||
import {
|
||||
semanticDocumentModelSchema,
|
||||
type DocumentProfile,
|
||||
type MarkdownDocumentMetadata,
|
||||
type SemanticDocumentGroupNode,
|
||||
type SemanticDocumentModel,
|
||||
type SemanticDocumentNode,
|
||||
type SemanticDocumentRegion,
|
||||
type SemanticDocumentRole,
|
||||
type SemanticDocumentTextNode
|
||||
} from "@md-to-pdf/core";
|
||||
|
||||
export interface CreateSemanticDocumentModelOptions {
|
||||
metadata: MarkdownDocumentMetadata;
|
||||
firstBodyHeading?: string;
|
||||
}
|
||||
|
||||
function normalizeTitle(value: string | undefined): string {
|
||||
return (value ?? "")
|
||||
.normalize("NFKC")
|
||||
.replace(/\s+/gu, " ")
|
||||
.trim();
|
||||
}
|
||||
|
||||
function text(
|
||||
role: SemanticDocumentRole,
|
||||
value: string | undefined,
|
||||
label?: string
|
||||
): SemanticDocumentTextNode | undefined {
|
||||
const normalized = value?.trim();
|
||||
if (!normalized) {
|
||||
return undefined;
|
||||
}
|
||||
return {
|
||||
kind: "text",
|
||||
role,
|
||||
text: normalized,
|
||||
...(label ? { label } : {})
|
||||
};
|
||||
}
|
||||
|
||||
function group(
|
||||
role: SemanticDocumentRole,
|
||||
children: readonly (SemanticDocumentNode | undefined)[]
|
||||
): SemanticDocumentGroupNode | undefined {
|
||||
const resolved = children.filter(
|
||||
(child): child is SemanticDocumentNode => child !== undefined
|
||||
);
|
||||
return resolved.length
|
||||
? {
|
||||
kind: "group",
|
||||
role,
|
||||
children: resolved
|
||||
}
|
||||
: undefined;
|
||||
}
|
||||
|
||||
function region(
|
||||
kind: SemanticDocumentRegion["kind"],
|
||||
nodes: readonly (SemanticDocumentNode | undefined)[],
|
||||
cover = false
|
||||
): SemanticDocumentRegion | undefined {
|
||||
const resolved = nodes.filter(
|
||||
(node): node is SemanticDocumentNode => node !== undefined
|
||||
);
|
||||
if (!resolved.length) {
|
||||
return undefined;
|
||||
}
|
||||
return {
|
||||
kind,
|
||||
nodes: resolved,
|
||||
...(cover
|
||||
? {
|
||||
section: {
|
||||
headerFooter: "none",
|
||||
pageNumber: "hidden",
|
||||
breakAfter: "next-page",
|
||||
followingPageNumberStart: 1
|
||||
} as const
|
||||
}
|
||||
: {})
|
||||
};
|
||||
}
|
||||
|
||||
function officialRegions(
|
||||
title: string,
|
||||
document: Extract<DocumentProfile, { profile: "official" }>
|
||||
): SemanticDocumentRegion[] {
|
||||
const classification = group("official-classification", [
|
||||
text("official-secrecy", document.secrecy),
|
||||
text("official-urgency", document.urgency)
|
||||
]);
|
||||
const issueRow = group("official-issue-row", [
|
||||
text("official-number", document.number),
|
||||
text("official-signatory", document.signatory, "签发人：")
|
||||
]);
|
||||
const masthead = group("official-masthead", [
|
||||
classification,
|
||||
text("official-issuer", document.issuer),
|
||||
issueRow
|
||||
]);
|
||||
const signature = group("official-signature", [
|
||||
text("official-signature-issuer", document.issuer),
|
||||
text("official-signature-date", document.date)
|
||||
]);
|
||||
const printingRow = group("official-printing-row", [
|
||||
text("official-printing-office", document.printingOffice),
|
||||
text(
|
||||
"official-printing-date",
|
||||
document.date ? `${document.date}印发` : undefined
|
||||
)
|
||||
]);
|
||||
const edition = group("official-edition", [
|
||||
text(
|
||||
"official-copy-to",
|
||||
document.copyTo?.join("、"),
|
||||
"抄送："
|
||||
),
|
||||
printingRow
|
||||
]);
|
||||
return [
|
||||
region("prefix", [
|
||||
masthead,
|
||||
text("official-title", title)
|
||||
]),
|
||||
region("suffix", [signature, edition])
|
||||
].filter(
|
||||
(item): item is SemanticDocumentRegion => item !== undefined
|
||||
);
|
||||
}
|
||||
|
||||
function briefingRegions(
|
||||
title: string,
|
||||
document: Extract<DocumentProfile, { profile: "briefing" }>
|
||||
): SemanticDocumentRegion[] {
|
||||
const details = group("briefing-meta", [
|
||||
text("briefing-issue", document.issue),
|
||||
text("briefing-publisher", document.publisher),
|
||||
text("briefing-signatory", document.signatory, "签发："),
|
||||
text("briefing-date", document.date)
|
||||
]);
|
||||
const masthead = group("briefing-masthead", [
|
||||
text("briefing-masthead-text", document.masthead),
|
||||
details
|
||||
]);
|
||||
return [
|
||||
region("prefix", [
|
||||
masthead,
|
||||
text("briefing-title", title)
|
||||
]),
|
||||
region("suffix", [
|
||||
text("briefing-contact", document.contact)
|
||||
])
|
||||
].filter(
|
||||
(item): item is SemanticDocumentRegion => item !== undefined
|
||||
);
|
||||
}
|
||||
|
||||
function projectReportRegions(
|
||||
title: string,
|
||||
document: Extract<
|
||||
DocumentProfile,
|
||||
{ profile: "project-report" }
|
||||
>
|
||||
): SemanticDocumentRegion[] {
|
||||
const cover = group("project-report-cover", [
|
||||
text(
|
||||
"project-report-project-name",
|
||||
document.projectName
|
||||
),
|
||||
text(
|
||||
"project-report-title",
|
||||
document.documentType || title
|
||||
),
|
||||
text("project-report-version", document.version),
|
||||
text("project-report-owner", document.owner, "建设单位："),
|
||||
text(
|
||||
"project-report-prepared-by",
|
||||
document.preparedBy,
|
||||
"编制单位："
|
||||
),
|
||||
text("project-report-date", document.date)
|
||||
]);
|
||||
const coverRegion = region("cover", [cover], true);
|
||||
return coverRegion ? [coverRegion] : [];
|
||||
}
|
||||
|
||||
function tenderRegions(
|
||||
title: string,
|
||||
document: Extract<DocumentProfile, { profile: "tender" }>
|
||||
): SemanticDocumentRegion[] {
|
||||
const cover = group("tender-cover", [
|
||||
text("tender-copy-mark", document.copyMark),
|
||||
text("tender-project-name", document.projectName),
|
||||
text("tender-project-number", document.projectNumber),
|
||||
text("tender-title", title),
|
||||
text("tender-volume", document.volume),
|
||||
text("tender-bidder", document.bidder, "投标人："),
|
||||
text(
|
||||
"tender-representative",
|
||||
document.representative,
|
||||
"法定代表人或授权代表："
|
||||
),
|
||||
text("tender-date", document.date)
|
||||
]);
|
||||
const coverRegion = region("cover", [cover], true);
|
||||
return coverRegion ? [coverRegion] : [];
|
||||
}
|
||||
|
||||
function createRegions(
|
||||
metadata: MarkdownDocumentMetadata
|
||||
): SemanticDocumentRegion[] {
|
||||
const document = metadata.document;
|
||||
if (!document) {
|
||||
return [];
|
||||
}
|
||||
switch (document.profile) {
|
||||
case "official":
|
||||
return officialRegions(metadata.title, document);
|
||||
case "briefing":
|
||||
return briefingRegions(metadata.title, document);
|
||||
case "project-report":
|
||||
return projectReportRegions(metadata.title, document);
|
||||
case "tender":
|
||||
return tenderRegions(metadata.title, document);
|
||||
}
|
||||
}
|
||||
|
||||
function emittedTitle(metadata: MarkdownDocumentMetadata): string {
|
||||
const document = metadata.document;
|
||||
if (!document) {
|
||||
return "";
|
||||
}
|
||||
return document.profile === "project-report"
|
||||
? document.documentType || metadata.title
|
||||
: metadata.title;
|
||||
}
|
||||
|
||||
export function createSemanticDocumentModel(
|
||||
options: CreateSemanticDocumentModelOptions
|
||||
): SemanticDocumentModel {
|
||||
const { metadata } = options;
|
||||
const firstHeading = normalizeTitle(options.firstBodyHeading);
|
||||
const metadataTitle = normalizeTitle(metadata.title);
|
||||
const structuralTitle = normalizeTitle(emittedTitle(metadata));
|
||||
const hasStructure = metadata.document !== undefined;
|
||||
const duplicatedMetadataTitle =
|
||||
Boolean(metadataTitle) && metadataTitle === firstHeading;
|
||||
const duplicatedStructuralTitle =
|
||||
Boolean(structuralTitle) && structuralTitle === firstHeading;
|
||||
return semanticDocumentModelSchema.parse({
|
||||
schemaVersion: 1,
|
||||
...(metadata.document
|
||||
? { profile: metadata.document.profile }
|
||||
: {}),
|
||||
titlePolicy: {
|
||||
metadataTitle:
|
||||
hasStructure || duplicatedMetadataTitle
|
||||
? "suppress"
|
||||
: "emit",
|
||||
firstBodyHeading:
|
||||
hasStructure && duplicatedStructuralTitle
|
||||
? "suppress"
|
||||
: "keep"
|
||||
},
|
||||
regions: createRegions(metadata)
|
||||
});
|
||||
}
|
||||
@@ -0,0 +1,139 @@
|
||||
import {
|
||||
semanticDocumentModelSchema,
|
||||
type MarkdownDocumentMetadata
|
||||
} from "@md-to-pdf/core";
|
||||
import { describe, expect, it } from "vitest";
|
||||
import { createSemanticDocumentModel } from "../src/index.js";
|
||||
|
||||
function metadata(
|
||||
input: Partial<MarkdownDocumentMetadata> = {}
|
||||
): MarkdownDocumentMetadata {
|
||||
return {
|
||||
title: "",
|
||||
author: "",
|
||||
subject: "",
|
||||
keywords: [],
|
||||
language: "zh-CN",
|
||||
...input
|
||||
};
|
||||
}
|
||||
|
||||
describe("统一语义文档模型", () => {
|
||||
it("为普通文档记录标题去重策略但不创建结构区域", () => {
|
||||
const model = createSemanticDocumentModel({
|
||||
metadata: metadata({ title: "项目报告" }),
|
||||
firstBodyHeading: " 项目报告 "
|
||||
});
|
||||
expect(model).toEqual({
|
||||
schemaVersion: 1,
|
||||
titlePolicy: {
|
||||
metadataTitle: "suppress",
|
||||
firstBodyHeading: "keep"
|
||||
},
|
||||
regions: []
|
||||
});
|
||||
});
|
||||
|
||||
it("完整表达公文版头、落款和版记的稳定顺序", () => {
|
||||
const model = createSemanticDocumentModel({
|
||||
metadata: metadata({
|
||||
title: "关于推进项目建设的通知",
|
||||
document: {
|
||||
profile: "official",
|
||||
issuer: "示例市人民政府办公室",
|
||||
number: "示例办发〔2026〕12号",
|
||||
secrecy: "内部",
|
||||
urgency: "加急",
|
||||
signatory: "张三",
|
||||
date: "2026年7月30日",
|
||||
copyTo: ["市发展改革委", "市财政局"],
|
||||
printingOffice: "示例市人民政府办公室"
|
||||
}
|
||||
})
|
||||
});
|
||||
expect(model.profile).toBe("official");
|
||||
expect(model.regions.map((item) => item.kind)).toEqual([
|
||||
"prefix",
|
||||
"suffix"
|
||||
]);
|
||||
expect(JSON.stringify(model)).toContain("official-masthead");
|
||||
expect(JSON.stringify(model)).toContain("抄送：");
|
||||
expect(JSON.stringify(model)).toContain("2026年7月30日印发");
|
||||
expect(() => semanticDocumentModelSchema.parse(model))
|
||||
.not.toThrow();
|
||||
});
|
||||
|
||||
it("表达简报前后结构并省略空字段", () => {
|
||||
const model = createSemanticDocumentModel({
|
||||
metadata: metadata({
|
||||
title: "工作简报",
|
||||
document: {
|
||||
profile: "briefing",
|
||||
masthead: "项目工作简报",
|
||||
issue: "第12期",
|
||||
contact: "联系人：张三"
|
||||
}
|
||||
})
|
||||
});
|
||||
expect(model.regions).toHaveLength(2);
|
||||
expect(JSON.stringify(model)).not.toContain(
|
||||
"briefing-signatory"
|
||||
);
|
||||
expect(JSON.stringify(model)).toContain("briefing-contact");
|
||||
});
|
||||
|
||||
it.each([
|
||||
[
|
||||
"project-report",
|
||||
{
|
||||
profile: "project-report",
|
||||
projectName: "智慧园区建设项目",
|
||||
documentType: "可行性研究报告",
|
||||
owner: "示例集团"
|
||||
}
|
||||
],
|
||||
[
|
||||
"tender",
|
||||
{
|
||||
profile: "tender",
|
||||
copyMark: "正本",
|
||||
projectName: "数据中心建设项目",
|
||||
bidder: "示例科技有限公司"
|
||||
}
|
||||
]
|
||||
] as const)("为 %s 创建独立封面分节意图", (_profile, document) => {
|
||||
const model = createSemanticDocumentModel({
|
||||
metadata: metadata({
|
||||
title: "项目文件",
|
||||
document
|
||||
})
|
||||
});
|
||||
expect(model.regions).toHaveLength(1);
|
||||
expect(model.regions[0]).toMatchObject({
|
||||
kind: "cover",
|
||||
section: {
|
||||
headerFooter: "none",
|
||||
pageNumber: "hidden",
|
||||
breakAfter: "next-page",
|
||||
followingPageNumberStart: 1
|
||||
}
|
||||
});
|
||||
});
|
||||
|
||||
it("结构标题与首个 H1 一致时记录正文标题抑制", () => {
|
||||
const model = createSemanticDocumentModel({
|
||||
metadata: metadata({
|
||||
title: "投标文件",
|
||||
document: {
|
||||
profile: "tender",
|
||||
projectName: "示例项目"
|
||||
}
|
||||
}),
|
||||
firstBodyHeading: "投标文件"
|
||||
});
|
||||
expect(model.titlePolicy).toEqual({
|
||||
metadataTitle: "suppress",
|
||||
firstBodyHeading: "suppress"
|
||||
});
|
||||
});
|
||||
});
|
||||
@@ -0,0 +1,16 @@
|
||||
{
|
||||
"extends": "../../tsconfig.base.json",
|
||||
"compilerOptions": {
|
||||
"target": "ES2022",
|
||||
"module": "NodeNext",
|
||||
"moduleResolution": "NodeNext",
|
||||
"rootDir": "src",
|
||||
"outDir": "dist",
|
||||
"declaration": true,
|
||||
"declarationMap": true,
|
||||
"sourceMap": true
|
||||
},
|
||||
"include": [
|
||||
"src"
|
||||
]
|
||||
}
|
||||
Reference in New Issue
Block a user