fix(routing): open the pathway a legacy news link points at

projects/pathway-browser/src/app/services/url-state.service.spec.ts

@@ -0,0 +1,69 @@
+/**
+ * Legacy pathway links in the URL fragment.
+ *
+ * The old browser addressed a pathway in the fragment, and this site's own news
+ * archive is full of those links: 770 spelled `#/R-HSA-1430728` and 86 spelled
+ * `#R-HSA-202733`, with no slash. The pattern used to require the slash, so the
+ * second kind matched nothing and opened the browser with no pathway in it.
+ *
+ * The cases below are taken from what is actually in the content, not invented.
+ */
+import { describe, expect, it } from 'vitest';
+import { FRAGMENT_PATTERN } from './url-state.service';
+
+/** What the subscriber does with a fragment, reduced to its decisions. */
+function route(fragment: string) {
+  const match = fragment.match(FRAGMENT_PATTERN);
+  if (!match?.groups) return { id: undefined, params: {} as Record<string, unknown> };
+  const params: Record<string, unknown> = {};
+  if (match.groups['params']) {
+    for (const [key, value] of match.groups['params'].split('&').map((p) => p.split('='))) {
+      params[key] = value || true;
+    }
+  }
+  return { id: match.groups['id'], params };
+}
+
+describe('a legacy pathway link in the fragment', () => {
+  it('opens the pathway when the link carries a slash', () => {
+    expect(route('/R-HSA-1430728').id).toBe('R-HSA-1430728');
+  });
+
+  it('opens the pathway when the link carries no slash', () => {
+    // 41 links in the news archive look like this, and every one of them used
+    // to land on an empty browser.
+    expect(route('R-HSA-202733').id).toBe('R-HSA-202733');
+    expect(route('R-HSA-913531').id).toBe('R-HSA-913531');
+  });
+
+  it('drops a stIdVersion rather than passing it on as a parameter', () => {
+    // 45 links carry one. It used to arrive as a query parameter named ".1".
+    for (const fragment of ['R-HSA-8853659.1', '/R-HSA-69231.4', 'R-HSA-3371497.12']) {
+      const { id, params } = route(fragment);
+      expect(id, fragment).toMatch(/^R-[A-Z]{3}-\d+$/);
+      expect(Object.keys(params), fragment).toEqual([]);
+    }
+  });
+
+  it('keeps the parameters an old link carries', () => {
+    const { id, params } = route('/R-HSA-8876384&PATH=R-HSA-1643685,R-HSA-5663205');
+    expect(id).toBe('R-HSA-8876384');
+    expect(params).toEqual({ PATH: 'R-HSA-1643685,R-HSA-5663205' });
+  });
+
+  it('leaves a fragment that is not a pathway alone', () => {
+    // A section to scroll to, and the old analysis-tool fragment. Matching
+    // these would turn them into a route to nowhere and a junk query
+    // parameter; they have to fall through untouched.
+    for (const fragment of ['introduction', 'summation', 'TOOL=AT', 'literature']) {
+      expect(route(fragment).id, fragment).toBeUndefined();
+      expect(Object.keys(route(fragment).params), fragment).toEqual([]);
+    }
+  });
+
+  it('does not take an identifier out of the middle of something else', () => {
+    for (const fragment of ['see-R-HSA-202733-here', 'XR-HSA-202733']) {
+      expect(route(fragment).id, fragment).toBeUndefined();
+    }
+  });
+});

projects/pathway-browser/src/app/services/url-state.service.ts

@@ -9,7 +9,24 @@ import type { Analysis } from '../model/analysis.model';
 import { UntilDestroy, untilDestroyed } from '@ngneat/until-destroy';
 import { toSignal } from '@angular/core/rxjs-interop';
 
-const FRAGMENT_PATTERN = /\/(?<id>R-[A-Z]{3}-\d+)?&?(?<params>.*)/;
+/**
+ * A legacy pathway link, as the old browser addressed one in the URL fragment.
+ *
+ * Both spellings are in the wild, because the old site produced both: 770 links
+ * in this site's own news archive say `#/R-HSA-1430728` and 86 say
+ * `#R-HSA-202733` with no slash. Requiring the slash meant the second kind
+ * matched nothing, so every one of them opened the browser with no pathway in
+ * it -- a blank page from a link in a release announcement.
+ *
+ * The id must be there for this to match at all. A fragment that is not a
+ * pathway reference -- `#introduction`, naming a section to scroll to -- has to
+ * fall through untouched rather than be read as a stale route.
+ *
+ * A trailing `.4` is a stIdVersion. The old links carry it, the content service
+ * does not want it, and it was previously parsed as a query parameter called
+ * ".4"; it is consumed and dropped here.
+ */
+export const FRAGMENT_PATTERN = /^\/?(?<id>R-[A-Z]{3}-\d+)(?:\.\d+)?(?:&(?<params>.*))?$/;
 
 export type UrlParam<T> = WritableSignal<T> & {
   otherTokens?: string[];