fix: normalize trailing slashes in standardizeURL

__tests__/common/standardizeURL.ts

@@ -1,4 +1,9 @@
-import { domainAllowedSearchParams, standardizeURL } from '../../src/common';
+import {
+  domainAllowedSearchParams,
+  getUrlDedupVariants,
+  getUrlTrailingSlashVariants,
+  standardizeURL,
+} from '../../src/common';
 
 describe('standardizeURL', () => {
   it('should keep url without query', () => {
@@ -13,6 +18,14 @@ describe('standardizeURL', () => {
     expect(canonicalUrl).toBe('https://test.com/posts/1');
   });
 
+  it('should preserve a trailing slash on the submitted url', () => {
+    const { url, canonicalUrl } = standardizeURL(
+      'https://github.com/versity/versitygw/',
+    );
+    expect(url).toBe('https://github.com/versity/versitygw/');
+    expect(canonicalUrl).toBe('https://github.com/versity/versitygw/');
+  });
+
   it('should clean query params', () => {
     const { url, canonicalUrl } = standardizeURL(
       'https://test.com/posts/1?utm_source=google',
@@ -51,3 +64,60 @@ describe('standardizeURL', () => {
     });
   });
 });
+
+describe('getUrlTrailingSlashVariants', () => {
+  it('should return both slash forms for a url without a trailing slash', () => {
+    expect(
+      getUrlTrailingSlashVariants('https://github.com/versity/versitygw'),
+    ).toEqual([
+      'https://github.com/versity/versitygw',
+      'https://github.com/versity/versitygw/',
+    ]);
+  });
+
+  it('should return both slash forms for a url with a trailing slash', () => {
+    expect(
+      getUrlTrailingSlashVariants('https://github.com/versity/versitygw/'),
+    ).toEqual([
+      'https://github.com/versity/versitygw',
+      'https://github.com/versity/versitygw/',
+    ]);
+  });
+
+  it('should collapse multiple trailing slashes', () => {
+    expect(
+      getUrlTrailingSlashVariants('https://github.com/versity/versitygw//'),
+    ).toEqual([
+      'https://github.com/versity/versitygw',
+      'https://github.com/versity/versitygw/',
+    ]);
+  });
+
+  it('should preserve query params on both forms', () => {
+    expect(
+      getUrlTrailingSlashVariants('https://test.com/posts/1/?sk=google'),
+    ).toEqual([
+      'https://test.com/posts/1?sk=google',
+      'https://test.com/posts/1/?sk=google',
+    ]);
+  });
+
+  it('should not toggle a root url', () => {
+    expect(getUrlTrailingSlashVariants('https://github.com/')).toEqual([
+      'https://github.com/',
+    ]);
+  });
+});
+
+describe('getUrlDedupVariants', () => {
+  it('should cross www and trailing-slash variants', () => {
+    expect(
+      getUrlDedupVariants('https://github.com/versity/versitygw/'),
+    ).toEqual([
+      'https://github.com/versity/versitygw',
+      'https://github.com/versity/versitygw/',
+      'https://www.github.com/versity/versitygw',
+      'https://www.github.com/versity/versitygw/',
+    ]);
+  });
+});

src/common/links.ts

@@ -97,6 +97,37 @@ export const getUrlWwwVariants = (url: string): string[] => {
   return [url, www ? `${scheme}${rest}` : `${scheme}www.${rest}`];
 };
 
+// The same article is often reachable with or without a trailing slash.
+// Return the url alongside its trailing-slash-toggled form so dedup lookups
+// can match either without rewriting the url we persist. Root-only urls
+// (`https://host/`) are left untouched.
+export const getUrlTrailingSlashVariants = (url: string): string[] => {
+  const [path, params] = url.split('?');
+  const trimmed = path.replace(/^(https?:\/\/[^/]+\/.+?)\/+$/i, '$1');
+
+  // No real (non-root) path to toggle.
+  if (!/^https?:\/\/[^/]+\/.+/i.test(trimmed)) {
+    return [url];
+  }
+
+  const suffix = params !== undefined ? `?${params}` : '';
+  return [`${trimmed}${suffix}`, `${trimmed}/${suffix}`];
+};
+
+// All url variants we treat as the same article for dedup lookups: every
+// www.-toggled form crossed with its trailing-slash-toggled form.
+export const getUrlDedupVariants = (url: string): string[] => {
+  const variants = new Set<string>();
+
+  for (const wwwVariant of getUrlWwwVariants(url)) {
+    for (const slashVariant of getUrlTrailingSlashVariants(wwwVariant)) {
+      variants.add(slashVariant);
+    }
+  }
+
+  return [...variants];
+};
+
 export const standardizeURL = (
   inputUrl: string,
 ): { url: string; canonicalUrl: string } => {

src/common/post.ts

@@ -25,7 +25,7 @@ import {
   WelcomePost,
 } from '../entity';
 import { ForbiddenError, ValidationError } from 'apollo-server-errors';
-import { getUrlWwwVariants, isValidHttpUrl, standardizeURL } from './links';
+import { getUrlDedupVariants, isValidHttpUrl, standardizeURL } from './links';
 import {
   findMarkdownTag,
   markdown,
@@ -939,8 +939,8 @@ export const getExistingPost = async (
     .createQueryBuilder(Post, 'post')
     .select(['post.id', 'post.deleted', 'post.visible'])
     .where([
-      { canonicalUrl: In(getUrlWwwVariants(canonicalUrl)) },
-      { url: In(getUrlWwwVariants(url)) },
+      { canonicalUrl: In(getUrlDedupVariants(canonicalUrl)) },
+      { url: In(getUrlDedupVariants(url)) },
     ])
     .getOne();
 
@@ -989,8 +989,8 @@ export const findPostByUrl = async <T extends keyof ArticlePost>(
     ]);
   } else {
     queryBuilder = queryBuilder.andWhere([
-      { canonicalUrl: In(getUrlWwwVariants(canonicalUrl)) },
-      { url: In(getUrlWwwVariants(cleanUrl)) },
+      { canonicalUrl: In(getUrlDedupVariants(canonicalUrl)) },
+      { url: In(getUrlDedupVariants(cleanUrl)) },
     ]);
   }