feat(import): add universal manuscript import pipeline (.docx, .odt, .epub, .md, .txt, .html, .rtf, .json)

packages/exporters/src/importer.test.ts

@@ -0,0 +1,173 @@
+import { describe, expect, it } from "vitest";
+import {
+  exportProjectData,
+  importManuscriptFile,
+  type ExportProject,
+} from "./index.js";
+
+const SAMPLE_PROJECT: ExportProject = {
+  meta: {
+    title: "The Harbor Ledger",
+    author: "Austin Breinholt",
+    language: "en",
+  },
+  books: [
+    {
+      id: "book-1",
+      projectId: "proj-import-1",
+      title: "Book One",
+      order: 0,
+      chapters: [
+        {
+          id: "ch-1",
+          bookId: "book-1",
+          projectId: "proj-import-1",
+          title: "Chapter 1: Fog Over the Pier",
+          order: 0,
+          rev: 1,
+          meta: { wordCount: 15, scenes: [], updatedAt: "2026-01-01T00:00:00.000Z" },
+          doc: {
+            type: "doc",
+            content: [
+              {
+                type: "heading",
+                attrs: { level: 1 },
+                content: [{ type: "text", text: "Chapter 1: Fog Over the Pier" }],
+              },
+              {
+                type: "paragraph",
+                content: [
+                  { type: "text", text: "The lantern swung in the " },
+                  { type: "text", text: "harbor wind", marks: [{ type: "bold" }] },
+                  { type: "text", text: " as Mara opened the ledger." },
+                ],
+              },
+            ],
+          },
+        },
+        {
+          id: "ch-2",
+          bookId: "book-1",
+          projectId: "proj-import-1",
+          title: "Chapter 2: The Bell Tower",
+          order: 1,
+          rev: 1,
+          meta: { wordCount: 6, scenes: [], updatedAt: "2026-01-01T00:00:00.000Z" },
+          doc: {
+            type: "doc",
+            content: [
+              {
+                type: "heading",
+                attrs: { level: 1 },
+                content: [{ type: "text", text: "Chapter 2: The Bell Tower" }],
+              },
+              {
+                type: "paragraph",
+                content: [
+                  { type: "text", text: "Twelve strokes echoed across the water." },
+                ],
+              },
+            ],
+          },
+        },
+      ],
+    },
+  ],
+};
+
+describe("Universal Manuscript Importer (importer.ts)", () => {
+  it("imports plain text (.txt) and splits chapters on Chapter headings", async () => {
+    const txt = [
+      "Chapter 1: The Arrival",
+      "",
+      "The train pulled into the station at dusk.",
+      "",
+      "* * *",
+      "",
+      "She stepped onto the wet platform.",
+      "",
+      "Chapter 2: The Departure",
+      "",
+      "By morning, the tracks were empty.",
+    ].join("\n");
+
+    const res = await importManuscriptFile("arrival-story.txt", txt);
+    expect(res.format).toBe("txt");
+    expect(res.chapters).toHaveLength(2);
+    expect(res.chapters[0]?.title).toBe("Chapter 1: The Arrival");
+    expect(res.chapters[1]?.title).toBe("Chapter 2: The Departure");
+    expect(res.totalWords).toBeGreaterThan(10);
+  });
+
+  it("imports Markdown (.md) with frontmatter, inline marks, and chapters", async () => {
+    const md = [
+      "---",
+      "title: Glass Harbor",
+      "author: Schwitzz",
+      "---",
+      "# Chapter One",
+      "",
+      "He read the **bold entry** and *italic margin note*.",
+      "",
+      "# Chapter Two",
+      "",
+      "The second chapter begins here.",
+    ].join("\n");
+
+    const res = await importManuscriptFile("glass-harbor.md", md);
+    expect(res.format).toBe("md");
+    expect(res.title).toBe("Glass Harbor");
+    expect(res.author).toBe("Schwitzz");
+    expect(res.chapters).toHaveLength(2);
+    expect(res.chapters[0]?.title).toBe("Chapter One");
+    expect(res.chapters[1]?.title).toBe("Chapter Two");
+  });
+
+  it("round-trips DOCX (Word / OnlyOffice) into PubScript chapters", async () => {
+    const docxBytes = (await exportProjectData(SAMPLE_PROJECT, "docx", {
+      generatedAt: "2026-01-01T00:00:00.000Z",
+    })) as Uint8Array;
+
+    const imported = await importManuscriptFile("the-harbor-ledger.docx", docxBytes);
+    expect(imported.format).toBe("docx");
+    expect(imported.chapters.length).toBeGreaterThanOrEqual(2);
+    expect(imported.totalWords).toBeGreaterThan(10);
+  });
+
+  it("round-trips ODT (OpenOffice / OnlyOffice / LibreOffice) into PubScript chapters", async () => {
+    const odtBytes = (await exportProjectData(SAMPLE_PROJECT, "odt", {
+      generatedAt: "2026-01-01T00:00:00.000Z",
+    })) as Uint8Array;
+
+    const imported = await importManuscriptFile("the-harbor-ledger.odt", odtBytes);
+    expect(imported.format).toBe("odt");
+    expect(imported.title).toBe("The Harbor Ledger");
+    expect(imported.chapters.length).toBeGreaterThanOrEqual(2);
+    expect(imported.totalWords).toBeGreaterThan(10);
+  });
+
+  it("round-trips EPUB into PubScript chapters", async () => {
+    const epubBytes = (await exportProjectData(SAMPLE_PROJECT, "epub", {
+      generatedAt: "2026-01-01T00:00:00.000Z",
+    })) as Uint8Array;
+
+    const imported = await importManuscriptFile("the-harbor-ledger.epub", epubBytes);
+    expect(imported.format).toBe("epub");
+    expect(imported.title).toBe("The Harbor Ledger");
+    expect(imported.chapters).toHaveLength(2);
+    expect(imported.totalWords).toBeGreaterThan(10);
+  });
+
+  it("imports RTF (.rtf) and HTML (.html) manuscripts", async () => {
+    const rtf = "{\\rtf1\\ansi{\\fonttbl\\f0 Times;}\\pard Chapter 1: Dawn\\par The sun rose over the ridge.\\par}";
+    const rtfRes = await importManuscriptFile("dawn.rtf", rtf);
+    expect(rtfRes.format).toBe("rtf");
+    expect(rtfRes.totalWords).toBeGreaterThan(4);
+
+    const html = "<html><head><title>Sea Wall</title></head><body><h1>Chapter 1</h1><p>Waves crashed against the <strong>stone</strong>.</p></body></html>";
+    const htmlRes = await importManuscriptFile("seawall.html", html);
+    expect(htmlRes.format).toBe("html");
+    expect(htmlRes.title).toBe("Sea Wall");
+    expect(htmlRes.chapters[0]?.title).toBe("Chapter 1");
+  });
+});

packages/exporters/src/index.ts

@@ -132,3 +132,4 @@ export { exportEpub } from "./epub.js";
 export { exportDocx } from "./docx.js";
 export { exportOdt } from "./odt.js";
 export { exportPdf } from "./pdf.js";
+export * from "./importer.js";