Merge pull request #9 from hoxxep/trailing-slash

Error if invalid trailing slash

Author: lilith

.editorconfig

@@ -0,0 +1,13 @@
+root = true
+
+[*]
+charset = utf-8
+indent_size = 2
+indent_style = space
+end_of_line = lf
+insert_final_newline = true
+trim_trailing_whitespace = true
+max_line_length = 100
+
+[*.{yaml,yml,json,tf,tfvars}]
+indent_size = 2

.gitignore

@@ -131,4 +131,5 @@ dist
 node_modules/
 dist/
 broken-links.log
-tests/broken-links.log
+tests/broken-links.log
+.idea/

check-links.js

@@ -1,7 +1,7 @@
-import { parse } from 'node-html-parser';
+import {parse} from 'node-html-parser';
 import fs from 'fs';
 import fetch from 'node-fetch';
-import { URL, fileURLToPath } from 'url';
+import {URL, fileURLToPath} from 'url';
 import path from 'path';
 import pLimit from 'p-limit';
 
@@ -14,7 +14,8 @@ export async function checkLinksInHtml(
   distPath = '',
   astroConfigRedirects = {},
   logger,
-  checkExternalLinks = true
+  checkExternalLinks = true,
+  trailingSlash = 'ignore',
 ) {
   const root = parse(htmlContent);
   const linkElements = root.querySelectorAll('a[href]');
@@ -34,7 +35,6 @@ export async function checkLinksInHtml(
 
       let absoluteLink;
       try {
-        
         // Differentiate between absolute, domain-relative, and relative links
         if (/^https?:\/\//i.test(link) || /^:\/\//i.test(link)) {
           // Absolute URL
@@ -73,7 +73,6 @@ export async function checkLinksInHtml(
       }
 
       let isBroken = false;
-     
 
       if (fetchLink.startsWith('/') && distPath) {
         // Internal link in build mode, check if file exists
@@ -89,35 +88,41 @@ export async function checkLinksInHtml(
         if (!possiblePaths.some((p) => fs.existsSync(p))) {
           // console.log('Failed paths', possiblePaths);
           isBroken = true;
-          // Fall back to checking a redirect file if it exists. 
+          // Fall back to checking a redirect file if it exists.
+        }
 
+        // check trailing slash is correct on internal links
+        const re = /\/$|\.[a-z0-9]+$/;  // match trailing slash or file extension
+        if (trailingSlash === 'always' && !fetchLink.match(re)) {
+          isBroken = true;
+        } else if (trailingSlash === 'never' && fetchLink !== '/' && fetchLink.endsWith('/')) {
+          isBroken = true;
         }
-      } else  {
+      } else {
         // External link, check via HTTP request. Retry 3 times if ECONNRESET
         if (checkExternalLinks) {
           let retries = 0;
           while (retries < 3) {
             try {
-              const response = await fetch(fetchLink, { method: 'GET' });
+              const response = await fetch(fetchLink, {method: 'GET'});
               isBroken = !response.ok;
               if (isBroken) {
                 logger.error(`${response.status} Error fetching ${fetchLink}`);
               }
               break;
             } catch (error) {
               isBroken = true;
-              let statusCodeNumber = error.errno == 'ENOTFOUND' ? 404 : (error.errno);
+              let statusCodeNumber = error.errno === 'ENOTFOUND' ? 404 : (error.errno);
               logger.error(`${statusCodeNumber} error fetching ${fetchLink}`);
               if (error.errno === 'ECONNRESET') {
                 retries++;
                 continue;
               }
               break;
-              }
             }
           }
         }
-      
+      }
 
       // Cache the link's validity
       checkedLinks.set(fetchLink, !isBroken);
@@ -134,16 +139,13 @@ export async function checkLinksInHtml(
 
 function isValidUrl(url) {
   // Skip mailto:, tel:, javascript:, and empty links
-  if (
+  return !(
     url.startsWith('mailto:') ||
     url.startsWith('tel:') ||
     url.startsWith('javascript:') ||
     url.startsWith('#') ||
     url.trim() === ''
-  ) {
-    return false;
-  }
-  return true;
+  );
 }
 
 function normalizePath(p) {
@@ -177,9 +179,8 @@ function addBrokenLink(brokenLinksMap, documentPath, brokenLink, distPath) {
   // Normalize broken link for reporting
   let normalizedBrokenLink = brokenLink;
 
-
   if (!brokenLinksMap.has(normalizedBrokenLink)) {
     brokenLinksMap.set(normalizedBrokenLink, new Set());
   }
   brokenLinksMap.get(normalizedBrokenLink).add(documentPath);
-}
+}

index.js

@@ -1,7 +1,7 @@
-import { fileURLToPath } from 'url';
-import { join } from 'path';
+import {fileURLToPath} from 'url';
+import {join} from 'path';
 import fs from 'fs';
-import { checkLinksInHtml , normalizeHtmlFilePath } from './check-links.js';
+import {checkLinksInHtml, normalizeHtmlFilePath} from './check-links.js';
 import fastGlob from 'fast-glob';
 
 export default function astroBrokenLinksChecker(options = {}) {
@@ -10,21 +10,22 @@ export default function astroBrokenLinksChecker(options = {}) {
   const checkedLinks = new Map();
 
   return {
-    
     name: 'astro-broken-links-checker',
     hooks: {
-      'astro:config:setup': async ({ config }) => {
+      'astro:config:setup': async ({config}) => {
         //console.log('config.redirects', config.redirects);
         // save the redirects to the options
         options.astroConfigRedirects = config.redirects;
+
+        // use astro trailingSlash setting, falling back to astro default of 'ignore'
+        options.trailingSlash = config.trailingSlash || 'ignore';
       },
-      
-      'astro:build:done': async ({ dir, logger }) => {
 
+      'astro:build:done': async ({dir, logger}) => {
         const astroConfigRedirects = options.astroConfigRedirects;
         //console.log('astroConfigRedirects', astroConfigRedirects);
         const distPath = fileURLToPath(dir);
-        const htmlFiles = await fastGlob('**/*.html', { cwd: distPath });
+        const htmlFiles = await fastGlob('**/*.html', {cwd: distPath});
         logger.info(`Checking ${htmlFiles.length} html pages for broken links`);
         // start time
         const startTime = Date.now();
@@ -41,14 +42,22 @@ export default function astroBrokenLinksChecker(options = {}) {
             distPath,
             astroConfigRedirects,
             logger,
-            options.checkExternalLinks
+            options.checkExternalLinks,
+            options.trailingSlash,
           );
         });
+
         await Promise.all(checkHtmlPromises);
         logBrokenLinks(brokenLinksMap, logFilePath, logger);
+
         // end time
         const endTime = Date.now();
         logger.info(`Time to check links: ${endTime - startTime} ms`);
+
+        // stop the build if we have broken links and the option is set
+        if (options.throwError && brokenLinksMap.size > 0) {
+          throw new Error(`Broken links detected. Check the log file: ${logFilePath}`);
+        }
       },
     },
   };
@@ -74,5 +83,9 @@ function logBrokenLinks(brokenLinksMap, logFilePath, logger) {
     }
   } else {
     logger.info('No broken links detected.');
+    if (fs.existsSync(logFilePath)) {
+      logger.info('Removing old log file:', logFilePath);
+      fs.rmSync(logFilePath);
+    }
   }
-}
+}