From a1921e23f5a8b65541f9dfd4ddb6fed4744691cf Mon Sep 17 00:00:00 2001
From: spring-raining <harusamex.com@gmail.com>
Date: Wed, 17 Jan 2024 23:54:23 +0900
Subject: [PATCH] fix: Improve file selector not to include unnecessary files
 into Webpub/EPUB

fix #461
---
 src/output/webbook.ts    | 36 ++++++++++++----------
 src/processor/compile.ts | 65 ++++++++++++++++++++++++++++++++--------
 2 files changed, 74 insertions(+), 27 deletions(-)

diff --git a/src/output/webbook.ts b/src/output/webbook.ts
index 1552ad22..775504b6 100644
--- a/src/output/webbook.ts
+++ b/src/output/webbook.ts
@@ -2,7 +2,12 @@ import fs from 'node:fs';
 import MIMEType from 'whatwg-mimetype';
 import { MANIFEST_FILENAME } from '../const.js';
 import { MergedConfig, WebbookEntryConfig } from '../input/config.js';
-import { generateManifest, globAssetFiles } from '../processor/compile.js';
+import {
+  generateManifest,
+  getAutoGeneratedAssetPatterns,
+  getDefaultIgnorePatterns,
+  globAssetFiles,
+} from '../processor/compile.js';
 import {
   ResourceLoader,
   fetchLinkedPublicationManifest,
@@ -18,7 +23,6 @@ import {
   debug,
   logError,
   logUpdate,
-  pathContains,
   pathEquals,
   remove,
   safeGlob,
@@ -247,10 +251,14 @@ export async function copyWebPublicationAssets({
   exportAliases,
   outputs,
   copyAsset,
+  themesDir,
   manifestPath,
   input,
   outputDir,
-}: Pick<MergedConfig, 'exportAliases' | 'outputs' | 'copyAsset'> & {
+}: Pick<
+  MergedConfig,
+  'exportAliases' | 'outputs' | 'copyAsset' | 'themesDir'
+> & {
   input: string;
   outputDir: string;
   manifestPath: string;
@@ -261,21 +269,12 @@ export async function copyWebPublicationAssets({
       target: upath.relative(input, target),
     }))
     .filter(({ source }) => !source.startsWith('..'));
-  const ignore = [
-    // don't copy auto-generated assets
-    ...outputs.flatMap(({ format, path: p }) =>
-      !pathContains(input, p)
-        ? []
-        : format === 'webpub'
-        ? upath.join(upath.relative(input, p), '**')
-        : upath.relative(input, p),
-    ),
-  ];
   const allFiles = new Set([
     ...(await globAssetFiles({
       copyAsset,
       cwd: input,
-      ignore,
+      outputs,
+      themesDir,
     })),
     ...(await safeGlob(
       [
@@ -285,7 +284,14 @@ export async function copyWebPublicationAssets({
       {
         cwd: input,
         ignore: [
-          ...ignore,
+          ...getAutoGeneratedAssetPatterns({
+            cwd: input,
+            outputs,
+          }),
+          ...getDefaultIgnorePatterns({
+            cwd: input,
+            themesDir,
+          }),
           // only include dotfiles starting with `.vs-`
           '**/.!(vs-*)/**',
         ],
diff --git a/src/processor/compile.ts b/src/processor/compile.ts
index 5c14ae40..6d0df03e 100644
--- a/src/processor/compile.ts
+++ b/src/processor/compile.ts
@@ -362,22 +362,68 @@ export async function compile({
   }
 }
 
+export function getDefaultIgnorePatterns({
+  themesDir,
+  cwd,
+}: Pick<MergedConfig, 'themesDir'> & {
+  cwd: string;
+}): string[] {
+  const ignorePatterns = [
+    // ignore node_modules directory
+    '**/node_modules',
+  ];
+  if (pathContains(cwd, themesDir)) {
+    // ignore example files of theme packages
+    ignorePatterns.push(
+      `${upath.relative(cwd, themesDir)}/packages/*/example`,
+      `${upath.relative(cwd, themesDir)}/packages/*/*/example`,
+    );
+  }
+  return ignorePatterns;
+}
+
+export function getAutoGeneratedAssetPatterns({
+  outputs,
+  cwd,
+}: Pick<MergedConfig, 'outputs'> & {
+  cwd: string;
+}): string[] {
+  return outputs.flatMap(({ format, path: p }) =>
+    !pathContains(cwd, p)
+      ? []
+      : format === 'webpub'
+      ? upath.join(upath.relative(cwd, p), '**')
+      : upath.relative(cwd, p),
+  );
+}
+
 export async function globAssetFiles({
   copyAsset: { fileExtensions, includes, excludes },
+  outputs,
+  themesDir,
   cwd,
   ignore = [],
-}: Pick<MergedConfig, 'copyAsset'> & {
+}: Pick<MergedConfig, 'copyAsset' | 'outputs' | 'themesDir'> & {
   cwd: string;
   ignore?: string[];
 }): Promise<Set<string>> {
+  const ignorePatterns = [
+    ...ignore,
+    ...excludes,
+    ...getAutoGeneratedAssetPatterns({ outputs, cwd }),
+  ];
+  const weakIgnorePatterns = getDefaultIgnorePatterns({ themesDir, cwd });
+  debug('globAssetFiles > ignorePatterns', ignorePatterns);
+  debug('globAssetFiles > weakIgnorePatterns', weakIgnorePatterns);
+
   const assets = new Set([
     // Step 1: Glob files with an extension in `fileExtension`
-    // Ignore files in node_modules directory and files matched `excludes`
+    // Ignore files in node_modules directory, theme example files and files matched `excludes`
     ...(await safeGlob(
       fileExtensions.map((ext) => `**/*.${ext}`),
       {
         cwd,
-        ignore: [...ignore, ...excludes, '**/node_modules'],
+        ignore: [...ignorePatterns, ...weakIgnorePatterns],
         followSymbolicLinks: true,
         gitignore: false,
       },
@@ -386,7 +432,7 @@ export async function globAssetFiles({
     // Ignore only files matched `excludes`
     ...(await safeGlob(includes, {
       cwd,
-      ignore: [...ignore, ...excludes],
+      ignore: ignorePatterns,
       followSymbolicLinks: true,
       gitignore: false,
     })),
@@ -399,6 +445,7 @@ export async function copyAssets({
   workspaceDir,
   copyAsset,
   outputs,
+  themesDir,
 }: MergedConfig): Promise<void> {
   if (pathEquals(entryContextDir, workspaceDir)) {
     return;
@@ -407,15 +454,9 @@ export async function copyAssets({
   const assets = await globAssetFiles({
     copyAsset,
     cwd: entryContextDir,
+    outputs,
+    themesDir,
     ignore: [
-      // don't copy auto-generated assets
-      ...outputs.flatMap(({ format, path: p }) =>
-        !pathContains(entryContextDir, p)
-          ? []
-          : format === 'webpub'
-          ? upath.join(upath.relative(entryContextDir, p), '**')
-          : upath.relative(entryContextDir, p),
-      ),
       // don't copy workspace itself
       ...(relWorkspaceDir ? [upath.join(relWorkspaceDir, '**')] : []),
     ],