feat: 完成 DOCX 通用结构映射与严格收口

2026-07-31 15:19:34 +08:00
parent 6e40374200
commit f3847f3e4b
31 changed files with 3175 additions and 64 deletions
@@ -0,0 +1,239 @@
+import { describe, expect, it } from "vitest";
+import type {
+  SemanticDocumentModel,
+  SemanticDocumentNode
+} from "@md-to-pdf/core";
+import type {
+  DocxResolvedStyleSlot,
+  DocxStyleSlotName,
+  DocxThemeTokenSet
+} from "@md-to-pdf/docx-theme-engine";
+import {
+  createPandocStructurePlan,
+  type PandocStructureBlock
+} from "../src/index.js";
+
+function slot(
+  name: DocxStyleSlotName,
+  hidden = false
+): DocxResolvedStyleSlot {
+  return {
+    slot: name,
+    source: "computed-css",
+    confidence: "exact",
+    style: {
+      fontCandidates: [],
+      ...(hidden ? { hidden: true } : {})
+    }
+  };
+}
+
+function tokens(
+  slots: DocxResolvedStyleSlot[] = []
+): DocxThemeTokenSet {
+  return {
+    schemaVersion: 1,
+    themeId: "test-theme",
+    themeFingerprint: "c".repeat(64),
+    mode: "auto-with-overrides",
+    basePreset: "technical",
+    slots,
+    diagnostics: []
+  };
+}
+
+function text(
+  role: Extract<SemanticDocumentNode, { kind: "text" }>["role"],
+  value: string,
+  label?: string
+): SemanticDocumentNode {
+  return {
+    kind: "text",
+    role,
+    text: value,
+    ...(label ? { label } : {})
+  };
+}
+
+function paragraphSegments(block: PandocStructureBlock): string[] {
+  if (block.kind === "paragraph") {
+    return block.segments;
+  }
+  if (block.kind === "container") {
+    return block.blocks.flatMap(paragraphSegments);
+  }
+  return [];
+}
+
+describe("Pandoc 语义结构投影", () => {
+  it("将公文分组投影为固定 Word 样式和可编辑段落", () => {
+    const model: SemanticDocumentModel = {
+      schemaVersion: 1,
+      profile: "official",
+      titlePolicy: {
+        metadataTitle: "suppress",
+        firstBodyHeading: "suppress"
+      },
+      regions: [
+        {
+          kind: "prefix",
+          nodes: [
+            {
+              kind: "group",
+              role: "official-masthead",
+              children: [
+                {
+                  kind: "group",
+                  role: "official-classification",
+                  children: [
+                    text("official-secrecy", "秘密"),
+                    text("official-urgency", "特急")
+                  ]
+                },
+                text("official-issuer", "示例单位"),
+                {
+                  kind: "group",
+                  role: "official-issue-row",
+                  children: [
+                    text("official-number", "示例〔2026〕1号"),
+                    text(
+                      "official-signatory",
+                      "张三",
+                      "签发人："
+                    )
+                  ]
+                }
+              ]
+            },
+            text("official-title", "关于开展工作的通知")
+          ]
+        }
+      ]
+    };
+
+    const plan = createPandocStructurePlan(model, tokens(), {
+      markerSeed: "official"
+    });
+
+    expect(plan.titlePolicy.firstBodyHeading).toBe("suppress");
+    expect(plan.prefix[0]).toMatchObject({
+      kind: "container",
+      styleId: "MdOfficialMasthead",
+      slot: "official-masthead",
+      startMarker:
+        "MD_TO_PDF_CONTAINER_START_official_1",
+      endMarker: "MD_TO_PDF_CONTAINER_END_official_2",
+      blocks: [
+        {
+          kind: "paragraph",
+          styleId: "MdOfficialClassification",
+          segments: ["秘密", "特急"],
+          separator: "tab"
+        },
+        {
+          kind: "paragraph",
+          styleId: "MdOfficialIssuer",
+          segments: ["示例单位"]
+        },
+        {
+          kind: "paragraph",
+          styleId: "MdOfficialIssueRow",
+          segments: ["示例〔2026〕1号", "签发人：张三"],
+          separator: "tab"
+        }
+      ]
+    });
+    expect(plan.prefix[1]).toMatchObject({
+      kind: "paragraph",
+      styleId: "MdOfficialTitle",
+      segments: ["关于开展工作的通知"]
+    });
+  });
+
+  it("按主题隐藏令牌过滤标书字段并保留封面分页", () => {
+    const model: SemanticDocumentModel = {
+      schemaVersion: 1,
+      profile: "tender",
+      titlePolicy: {
+        metadataTitle: "suppress",
+        firstBodyHeading: "suppress"
+      },
+      regions: [
+        {
+          kind: "cover",
+          section: {
+            headerFooter: "none",
+            pageNumber: "hidden",
+            breakAfter: "next-page",
+            followingPageNumberStart: 1
+          },
+          nodes: [
+            {
+              kind: "group",
+              role: "tender-cover",
+              children: [
+                text("tender-title", "投标文件"),
+                text("tender-bidder", "投标单位"),
+                text(
+                  "tender-representative",
+                  "授权代表"
+                ),
+                text("tender-date", "2026年7月")
+              ]
+            }
+          ]
+        }
+      ]
+    };
+
+    const plan = createPandocStructurePlan(
+      model,
+      tokens([
+        slot("tender-bidder", true),
+        slot("tender-representative", true)
+      ]),
+      { markerSeed: "tender" }
+    );
+
+    expect(plan.prefix.at(-1)).toEqual({
+      kind: "section-break",
+      marker: "MD_TO_PDF_SECTION_BREAK_tender_3",
+      headerFooter: "none",
+      pageNumber: "hidden",
+      followingPageNumberStart: 1
+    });
+    expect(plan.prefix[0]).toMatchObject({
+      kind: "container",
+      styleId: "MdTenderCover"
+    });
+    expect(paragraphSegments(plan.prefix[0]!)).toEqual([
+      "投标文件",
+      "2026年7月"
+    ]);
+  });
+
+  it("没有结构区域时只传递标题策略", () => {
+    const plan = createPandocStructurePlan(
+      {
+        schemaVersion: 1,
+        titlePolicy: {
+          metadataTitle: "emit",
+          firstBodyHeading: "keep"
+        },
+        regions: []
+      },
+      tokens(),
+      { markerSeed: "empty" }
+    );
+
+    expect(plan).toEqual({
+      schemaVersion: 1,
+      titlePolicy: {
+        metadataTitle: "emit",
+        firstBodyHeading: "keep"
+      },
+      prefix: [],
+      suffix: []
+    });
+  });
+});