fix: Improve file selector not to include unnecessary files into Webpub/EPUB

package.json

@@ -69,6 +69,7 @@
   "devDependencies": {
     "@hyrious/esbuild-plugin-commonjs": "^0.2.2",
     "@release-it/conventional-changelog": "^5.1.1",
+    "@types/adm-zip": "^0.5.5",
     "@types/archiver": "^5.3.2",
     "@types/command-exists": "1.2.0",
     "@types/debug": "^4.1.7",
@@ -85,6 +86,7 @@
     "@types/w3c-xmlserializer": "^2.0.2",
     "@types/whatwg-mimetype": "^3.0.0",
     "@vitest/coverage-v8": "^0.33.0",
+    "adm-zip": "^0.5.10",
     "esbuild": "^0.18.11",
     "file-type": "^16.5.3",
     "fs-extra": "^11.1.1",

src/build.ts

@@ -1,5 +1,4 @@
 import chalk from 'chalk';
-import { pathToFileURL } from 'node:url';
 import terminalLink from 'terminal-link';
 import { getExecutableBrowserPath } from './browser.js';
 import {
@@ -8,22 +7,15 @@ import {
   collectVivliostyleConfig,
   mergeConfig,
 } from './input/config.js';
-import { exportEpub } from './output/epub.js';
 import { buildPDF, buildPDFWithContainer } from './output/pdf.js';
-import {
-  copyWebPublicationAssets,
-  prepareWebPublicationDirectory,
-  retrieveWebbookEntry,
-  supplyWebPublicationManifestForWebbook,
-} from './output/webbook.js';
+import { buildWebPublication } from './output/webbook.js';
 import {
   checkOverwriteViolation,
   cleanupWorkspace,
   compile,
   copyAssets,
   prepareThemeDirectory,
 } from './processor/compile.js';
-import type { PublicationManifest } from './schema/publication.schema.js';
 import { teardownServer } from './server.js';
 import {
   checkContainerEnvironment,
@@ -34,7 +26,6 @@ import {
   setLogLevel,
   startLogging,
   upath,
-  useTmpDirectory,
 } from './util.js';
 
 export interface BuildCliFlags extends CliFlags {
@@ -122,66 +113,10 @@ export async function build(cliFlags: BuildCliFlags) {
           });
         }
       } else if (format === 'webpub' || format === 'epub') {
-        const { manifestPath, webbookEntryUrl } = config;
-        let outputDir: string;
-        if (format === 'webpub') {
-          outputDir = target.path;
-          await prepareWebPublicationDirectory({ outputDir });
-        } else if (format === 'epub') {
-          [outputDir] = await useTmpDirectory();
-        } else {
-          continue;
-        }
-
-        let entryContextUrl: string;
-        let entryHtmlFile: string | undefined;
-        let manifest: PublicationManifest;
-        if (manifestPath) {
-          entryContextUrl = pathToFileURL(manifestPath).href;
-          manifest = await copyWebPublicationAssets({
-            ...config,
-            input: config.workspaceDir,
-            outputDir,
-            manifestPath,
-          });
-          if (config.input.format === 'markdown') {
-            const entry = [manifest.readingOrder].flat()[0];
-            if (entry) {
-              entryHtmlFile = upath.join(
-                outputDir,
-                typeof entry === 'string' ? entry : entry.url,
-              );
-            }
-          }
-        } else if (webbookEntryUrl) {
-          const ret = await retrieveWebbookEntry({
-            webbookEntryUrl,
-            outputDir,
-          });
-          entryContextUrl = webbookEntryUrl;
-          entryHtmlFile = ret.entryHtmlFile;
-          manifest =
-            ret.manifest ||
-            (await supplyWebPublicationManifestForWebbook({
-              ...config,
-              entryHtmlFile: ret.entryHtmlFile,
-              outputDir,
-            }));
-        } else {
-          continue;
-        }
-
-        if (format === 'epub') {
-          await exportEpub({
-            webpubDir: outputDir,
-            entryHtmlFile,
-            entryContextUrl,
-            manifest,
-            target: target.path,
-            epubVersion: target.version,
-          });
-        }
-        output = target.path;
+        output = await buildWebPublication({
+          ...config,
+          target,
+        });
       }
       if (output) {
         const formattedOutput = chalk.bold.green(upath.relative(cwd, output));

src/output/epub.ts

@@ -121,22 +121,22 @@ const appendManifestProperty = (entry: ManifestEntry, newProperty: string) => {
 export async function exportEpub({
   webpubDir,
   entryHtmlFile,
-  entryContextUrl,
   manifest,
+  relManifestPath,
   target,
   epubVersion,
 }: {
   webpubDir: string;
   entryHtmlFile?: string;
-  entryContextUrl?: string;
   manifest: PublicationManifest;
+  relManifestPath?: string;
   target: string;
   epubVersion: '3.0';
 }) {
   debug('Export EPUB', {
     webpubDir,
-    entryContextUrl,
     entryHtmlFile,
+    relManifestPath,
     target,
     epubVersion,
   });
@@ -221,9 +221,7 @@ export async function exportEpub({
   const htmlFiles = Object.keys(manifestItem).filter((url) =>
     /\.html?$/.test(url),
   );
-  let tocHtml = htmlFiles.find(
-    (f) => f === (tocResource?.url || entryHtmlRelPath),
-  );
+  let tocHtml = htmlFiles.find((f) => f === tocResource?.url);
   const readingOrder = [manifest.readingOrder || entryHtmlRelPath]
     .flat()
     .flatMap((v) => (v ? (typeof v === 'string' ? { url: v } : v) : []));
@@ -233,7 +231,8 @@ export async function exportEpub({
         'No table of contents document was found. for EPUB output, we recommend to enable `toc` option in your Vivliostyle config file to generate a table of contents document.',
       ),
     );
-    tocHtml = readingOrder[0].url;
+    tocHtml =
+      htmlFiles.find((f) => f === entryHtmlRelPath) || readingOrder[0].url;
   }
   const spineItems = readingOrder.map<SpineEntry>(({ url }) => ({
     href: changeExtname(url, '.xhtml'),
@@ -267,8 +266,6 @@ export async function exportEpub({
       parseResult = await transpileHtmlToXhtml({
         target,
         contextDir: upath.join(tmpDir, 'EPUB'),
-        entryContextUrl:
-          entryContextUrl || pathToFileURL(upath.join(webpubDir, '/')).href,
         landmarks,
         isTocHtml,
         isPagelistHtml: target === (pageListResource?.url || entryHtmlRelPath),
@@ -316,7 +313,7 @@ export async function exportEpub({
     await processHtml(target, false);
   }
 
-  let { tocParseTree, linkedPubManifest } = tocProcessResult;
+  let { tocParseTree } = tocProcessResult;
   if (!tocParseTree) {
     tocParseTree = await supplyTocNavElement({
       tocHtml,
@@ -326,9 +323,9 @@ export async function exportEpub({
       docLanguages,
     });
   }
-  if (linkedPubManifest) {
-    await remove(upath.join(tmpDir, 'EPUB', linkedPubManifest));
-    delete manifestItem[linkedPubManifest];
+  if (relManifestPath) {
+    await remove(upath.join(tmpDir, 'EPUB', relManifestPath));
+    delete manifestItem[relManifestPath];
   }
 
   // EPUB/toc.ncx
@@ -377,22 +374,19 @@ export async function exportEpub({
 async function transpileHtmlToXhtml({
   target,
   contextDir,
-  entryContextUrl,
   landmarks,
   isTocHtml,
   isPagelistHtml,
 }: {
   target: string;
   contextDir: string;
-  entryContextUrl: string;
   landmarks: LandmarkEntry[];
   isTocHtml: boolean;
   isPagelistHtml: boolean;
 }): Promise<{
   dom: JSDOM;
   tocParseTree?: TocResourceTreeRoot;
   pageListParseTree?: PageListResourceTreeRoot;
-  linkedPubManifest?: string;
   hasMathmlContent: boolean;
   hasRemoteResources: boolean;
   hasScriptedContent: boolean;
@@ -424,7 +418,6 @@ async function transpileHtmlToXhtml({
 
   let tocParseTree: TocResourceTreeRoot | undefined;
   let pageListParseTree: PageListResourceTreeRoot | undefined;
-  let linkedPubManifest: string | undefined;
 
   if (isTocHtml) {
     if (!document.querySelector('[epub:type="toc"]')) {
@@ -473,16 +466,6 @@ async function transpileHtmlToXhtml({
         if (scriptEl?.getAttribute('type') === 'application/ld+json') {
           scriptEl.parentNode?.removeChild(scriptEl);
         }
-      } else {
-        const entryUrl = entryContextUrl;
-        const publicationUrl = new URL(href, new URL(target, entryUrl)).href;
-        const rootUrl = /^https?:/i.test(entryUrl)
-          ? new URL('/', entryUrl).href
-          : new URL('.', entryUrl).href;
-        const relPublicationPath = upath.relative(rootUrl, publicationUrl);
-        if (!relPublicationPath.startsWith('..')) {
-          linkedPubManifest = relPublicationPath;
-        }
       }
       publicationLinkEl.parentNode?.removeChild(publicationLinkEl);
     }
@@ -505,7 +488,6 @@ async function transpileHtmlToXhtml({
     dom,
     tocParseTree,
     pageListParseTree,
-    linkedPubManifest,
     // FIXME: Yes, I recognize this implementation is inadequate.
     hasMathmlContent: !!document.querySelector('math'),
     hasRemoteResources: !!document.querySelector(