fix(core): fix handling of Swc html minifier warnings (#10581)

packages/docusaurus-bundler/src/minifyHtml.ts

@@ -5,7 +5,6 @@
  * LICENSE file in the root directory of this source tree.
  */
 
-import logger from '@docusaurus/logger';
 import {minify as terserHtmlMinifier} from 'html-minifier-terser';
 import {importSwcHtmlMinifier} from './importFaster';
 import type {DocusaurusConfig} from '@docusaurus/types';
@@ -13,12 +12,17 @@ import type {DocusaurusConfig} from '@docusaurus/types';
 // Historical env variable
 const SkipHtmlMinification = process.env.SKIP_HTML_MINIFICATION === 'true';
 
+export type HtmlMinifierResult = {
+  code: string;
+  warnings: string[];
+};
+
 export type HtmlMinifier = {
-  minify: (html: string) => Promise<string>;
+  minify: (html: string) => Promise<HtmlMinifierResult>;
 };
 
 const NoopMinifier: HtmlMinifier = {
-  minify: async (html: string) => html,
+  minify: async (html: string) => ({code: html, warnings: []}),
 };
 
 type SiteConfigSlice = {
@@ -50,7 +54,7 @@ async function getTerserMinifier(): Promise<HtmlMinifier> {
   return {
     minify: async function minifyHtmlWithTerser(html) {
       try {
-        return await terserHtmlMinifier(html, {
+        const code = await terserHtmlMinifier(html, {
           removeComments: false,
           removeRedundantAttributes: true,
           removeEmptyAttributes: true,
@@ -59,6 +63,7 @@ async function getTerserMinifier(): Promise<HtmlMinifier> {
           useShortDoctype: true,
           minifyJS: true,
         });
+        return {code, warnings: []};
       } catch (err) {
         throw new Error(`HTML minification failed (Terser)`, {
           cause: err as Error,
@@ -95,49 +100,16 @@ async function getSwcMinifier(): Promise<HtmlMinifier> {
           minifyCss: true,
         });
 
-        // Escape hatch because SWC is quite aggressive to report errors
+        const warnings = (result.errors ?? []).map((diagnostic) => {
-        // TODO figure out what to do with these errors: throw or swallow?
+          return `[HTML minifier diagnostic - ${diagnostic.level}] ${
-        //  See https://github.com/facebook/docusaurus/pull/10554
-        //  See https://github.com/swc-project/swc/discussions/9616#discussioncomment-10846201
-        const ignoreSwcMinifierErrors =
-          process.env.DOCUSAURUS_IGNORE_SWC_HTML_MINIFIER_ERRORS === 'true';
-        if (!ignoreSwcMinifierErrors && result.errors) {
-          const ignoredErrors: string[] = [
-            // TODO Docusaurus seems to emit NULL chars, and minifier detects it
-            //  see https://github.com/facebook/docusaurus/issues/9985
-            'Unexpected null character',
-          ];
-          result.errors = result.errors.filter(
-            (diagnostic) => !ignoredErrors.includes(diagnostic.message),
-          );
-          if (result.errors.length) {
-            throw new Error(
-              `HTML minification diagnostic errors:
-- ${result.errors
-                .map(
-                  (diagnostic) =>
-                    `[${diagnostic.level}] ${
             diagnostic.message
-                    } - ${JSON.stringify(diagnostic.span)}`,
+          } - ${JSON.stringify(diagnostic.span)}`;
-                )
+        });
-                .join('\n- ')}
+
-Note: please report the problem to the Docusaurus team
+        return {
-In the meantime, you can skip this error with ${logger.code(
+          code: result.code,
-                'DOCUSAURUS_IGNORE_SWC_HTML_MINIFIER_ERRORS=true',
+          warnings,
-              )}`,
+        };
-            );
-          }
-          /*
-          if (result.errors.length) {
-            throw new AggregateError(
-              result.errors.map(
-                (diagnostic) => new Error(JSON.stringify(diagnostic, null, 2)),
-              ),
-            );
-          }
-           */
-        }
-        return result.code;
       } catch (err) {
         throw new Error(`HTML minification failed (SWC)`, {
           cause: err as Error,

packages/docusaurus/src/ssg.ts

@@ -110,6 +110,57 @@ function pathnameToFilename({
   return `${outputFileName}.html`;
 }
 
+export function printSSGWarnings(
+  results: {
+    pathname: string;
+    warnings: string[];
+  }[],
+): void {
+  // Escape hatch because SWC is quite aggressive to report errors
+  // See https://github.com/facebook/docusaurus/pull/10554
+  // See https://github.com/swc-project/swc/discussions/9616#discussioncomment-10846201
+  if (process.env.DOCUSAURUS_IGNORE_SSG_WARNINGS === 'true') {
+    return;
+  }
+
+  const ignoredWarnings: string[] = [
+    // TODO React/Docusaurus emit NULL chars, and minifier detects it
+    //  see https://github.com/facebook/docusaurus/issues/9985
+    'Unexpected null character',
+  ];
+
+  const keepWarning = (warning: string) => {
+    return !ignoredWarnings.some((iw) => warning.includes(iw));
+  };
+
+  const resultsWithWarnings = results
+    .map((result) => {
+      return {
+        ...result,
+        warnings: result.warnings.filter(keepWarning),
+      };
+    })
+    .filter((result) => result.warnings.length > 0);
+
+  if (resultsWithWarnings.length) {
+    const message = `Docusaurus static site generation process emitted warnings for ${
+      resultsWithWarnings.length
+    } path${resultsWithWarnings.length ? 's' : ''}
+This is non-critical and can be disabled with DOCUSAURUS_IGNORE_SSG_WARNINGS=true
+Troubleshooting guide: https://github.com/facebook/docusaurus/discussions/10580
+
+- ${resultsWithWarnings
+      .map(
+        (result) => `${logger.path(result.pathname)}:
+  - ${result.warnings.join('\n  - ')}
+`,
+      )
+      .join('\n- ')}`;
+
+    logger.warn(message);
+  }
+}
+
 export async function generateStaticFiles({
   pathnames,
   renderer,
@@ -121,8 +172,18 @@ export async function generateStaticFiles({
   params: SSGParams;
   htmlMinifier: HtmlMinifier;
 }): Promise<{collectedData: SiteCollectedData}> {
-  type SSGSuccess = {pathname: string; error: null; result: AppRenderResult};
+  type SSGSuccess = {
-  type SSGError = {pathname: string; error: Error; result: null};
+    pathname: string;
+    error: null;
+    result: AppRenderResult;
+    warnings: string[];
+  };
+  type SSGError = {
+    pathname: string;
+    error: Error;
+    result: null;
+    warnings: string[];
+  };
   type SSGResult = SSGSuccess | SSGError;
 
   // Note that we catch all async errors on purpose
@@ -136,15 +197,27 @@ export async function generateStaticFiles({
         params,
         htmlMinifier,
       }).then(
-        (result) => ({pathname, result, error: null}),
+        (result) => ({
-        (error) => ({pathname, result: null, error: error as Error}),
+          pathname,
+          result,
+          error: null,
+          warnings: result.warnings,
+        }),
+        (error) => ({
+          pathname,
+          result: null,
+          error: error as Error,
+          warnings: [],
+        }),
       ),
     {concurrency: Concurrency},
   );
 
+  printSSGWarnings(results);
+
   const [allSSGErrors, allSSGSuccesses] = _.partition(
     results,
-    (r): r is SSGError => !!r.error,
+    (result): result is SSGError => !!result.error,
   );
 
   if (allSSGErrors.length > 0) {
@@ -179,7 +252,7 @@ async function generateStaticFile({
   renderer: AppRenderer;
   params: SSGParams;
   htmlMinifier: HtmlMinifier;
-}) {
+}): Promise<AppRenderResult & {warnings: string[]}> {
   try {
     // This only renders the app HTML
     const result = await renderer({
@@ -190,13 +263,17 @@ async function generateStaticFile({
       params,
       result,
     });
-    const content = await htmlMinifier.minify(fullPageHtml);
+    const minifierResult = await htmlMinifier.minify(fullPageHtml);
     await writeStaticFile({
       pathname,
-      content,
+      content: minifierResult.code,
       params,
     });
-    return result;
+    return {
+      ...result,
+      // As of today, only the html minifier can emit SSG warnings
+      warnings: minifierResult.warnings,
+    };
   } catch (errorUnknown) {
     const error = errorUnknown as Error;
     const tips = getSSGErrorTips(error);