feat: 建立统一语义文档模型

packages/core/src/document.ts

@@ -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[];
 }

packages/core/src/index.ts

@@ -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";

packages/core/src/semantic-document.ts

@@ -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
+>;

packages/core/tests/document.test.ts

@@ -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: []
     };