Read the About box and latest release from JSON

GitHub rebuilt the repository sidebar, so the selectors parseRepoInfo used
matched nothing and the About box rendered empty. Everything but the latest
release is in the page's embedded payload as sidebarAbout; the latest release
is only in the _sidebar JSON already fetched for languages, so it arrives with
that fragment and is moved beside the release count.

1 files changed, 43 insertions, 83 deletions

diff --git a/cgithub/src/scraper.ts b/cgithub/src/scraper.ts
index 223839f..4220dc4 100644
--- a/cgithub/src/scraper.ts
+++ b/cgithub/src/scraper.ts
@@ -143,7 +143,17 @@ interface CodeViewBlobRoute {
   richText: string | null;
 }
 
+interface SidebarAbout {
+  description?: string | null;
+  website?: string | null;
+  stargazerCount?: number;
+  forksCount?: number;
+  // A section is false when the repository has none of that thing.
+  sections?: { releases?: { releaseCount?: number } | false };
+}
+
 interface AppPayload {
+  sidebarAbout?: SidebarAbout;
   codeViewLayoutRoute?: CodeViewLayoutRoute;
   codeViewRepoRoute?: CodeViewRepoRoute;
   codeViewTreeRoute?: CodeViewTreeRoute;
@@ -230,8 +240,6 @@ export interface RepoInfo {
   stars: string | null;
   forks: string | null;
   numReleases?: number;
-  latestReleaseTag?: string;
-  latestReleaseDate?: string;
 }
 
 export interface GitHubRepo extends GitHubTree {
@@ -292,8 +300,14 @@ export interface Language {
   color: string | null;
 }
 
-export interface GitHubLanguages extends GitHubCommon {
+export interface LatestRelease {
+  tag: string;
+  publishedAt: string | null;
+}
+
+export interface GitHubSidebar extends GitHubCommon {
   languages: Language[];
+  latestRelease: LatestRelease | null;
 }
 
 export interface Ref {
@@ -425,81 +439,6 @@ function parsePayload(html: string): AppPayload | null {
   return parseEmbeddedPayload<AppPayload>(html);
 }
 
-function parseRepoInfo(html: string): RepoInfo {
-  const document = parseDocument(html);
-
-  const aboutSection = cssSelect.selectOne(".BorderGrid-cell h2:contains('About')", document);
-  let description: string | null = null;
-  let website: string | null = null;
-
-  if (aboutSection) {
-    const parent = aboutSection.parent;
-    if (parent) {
-      const descElem = cssSelect.selectOne("p.f4", parent);
-      if (descElem) {
-        description = textContent(descElem).trim();
-      }
-      const websiteElem = cssSelect.selectOne(
-        "a[rel='noopener noreferrer nofollow']",
-        parent,
-      ) as Element | null;
-      if (websiteElem) {
-        website = getAttributeValue(websiteElem, "href") || null;
-      }
-    }
-  }
-
-  const starsElem = cssSelect.selectOne("#repo-stars-counter-star", document) as Element | null;
-  const stars = starsElem
-    ? getAttributeValue(starsElem, "title") || textContent(starsElem).trim()
-    : null;
-
-  const forksElem = cssSelect.selectOne("#repo-network-counter", document) as Element | null;
-  const forks = forksElem
-    ? getAttributeValue(forksElem, "title") || textContent(forksElem).trim()
-    : null;
-
-  const releaseElem = cssSelect.selectOne(
-    "a[href$='/releases'] .Counter",
-    document,
-  ) as Element | null;
-  let numReleases: number | undefined;
-  let latestReleaseTag: string | undefined;
-  let latestReleaseDate: string | undefined;
-
-  if (releaseElem) {
-    const releaseText = getAttributeValue(releaseElem, "title") || textContent(releaseElem);
-    numReleases = parseInt(releaseText.replace(/,/g, ""), 10);
-
-    // Find the latest release tag and date if available
-    const cell = (releaseElem.parent as Element)?.parent?.parent as Element; // span -> a -> h2 -> cell
-    if (cell) {
-      const latestLink = cssSelect.selectOne("a[href*='/releases/tag/']", cell) as Element;
-      if (latestLink) {
-        const tagElem = cssSelect.selectOne(".css-truncate-target", latestLink);
-        if (tagElem) {
-          latestReleaseTag = textContent(tagElem).trim();
-        }
-
-        const timeElem = cssSelect.selectOne("relative-time", latestLink);
-        if (timeElem) {
-          latestReleaseDate = getAttributeValue(timeElem, "datetime") || undefined;
-        }
-      }
-    }
-  }
-
-  return {
-    description,
-    website,
-    stars,
-    forks,
-    numReleases,
-    latestReleaseTag,
-    latestReleaseDate,
-  };
-}
-
 function extractOverviewHtml(
   overviewFiles: OverviewFile[] = [],
 ): Record<string, string | null> | undefined {
@@ -539,7 +478,15 @@ export async function getGitHubRepo(owner: string, repo: string): Promise<GitHub
     throw new Error("Could not find tree data in embedded JSON");
   }
 
-  const info = parseRepoInfo(html);
+  const about = payload?.sidebarAbout;
+  const releases = about?.sections?.releases;
+  const info = {
+    description: about?.description ?? null,
+    website: about?.website ?? null,
+    stars: about?.stargazerCount?.toLocaleString("en-US") ?? null,
+    forks: about?.forksCount?.toLocaleString("en-US") ?? null,
+    numReleases: releases === false ? undefined : releases?.releaseCount,
+  };
   const overviewHtml = extractOverviewHtml(repoRoute.overview?.overviewFiles);
 
   return extractGitHub(payload, { items: repoRoute.tree.items, info, overviewHtml });
@@ -742,11 +689,16 @@ interface SidebarPayload {
       color?: string;
     }[];
   };
+  releases?: {
+    // The name is the release title, which need not be the tag; the path is.
+    latestRelease?: { name?: string; path: string; publishedAt?: string } | null;
+  };
 }
 
-// The repository sidebar (languages, contributors, ...) is not rendered into
-// the repo page at all; GitHub's own UI lazily fetches it as JSON from here.
-export async function getGitHubLanguages(owner: string, repo: string): Promise<GitHubLanguages> {
+// The repository sidebar (languages, releases, contributors, ...) is not
+// rendered into the repo page at all; GitHub's own UI lazily fetches it as
+// JSON from here.
+export async function getGitHubSidebar(owner: string, repo: string): Promise<GitHubSidebar> {
   const payload = await fetchGitHubJson<SidebarPayload>(`${owner}/${repo}/_sidebar`);
 
   const languages: Language[] = [];
@@ -758,7 +710,15 @@ export async function getGitHubLanguages(owner: string, repo: string): Promise<G
     });
   }
 
-  return { repo: { owner, name: repo }, languages };
+  const release = payload.releases?.latestRelease;
+  const latestRelease = release
+    ? {
+        tag: decodeURIComponent(release.path.split("/releases/tag/")[1] ?? release.name ?? ""),
+        publishedAt: release.publishedAt ?? null,
+      }
+    : null;
+
+  return { repo: { owner, name: repo }, languages, latestRelease };
 }
 
 interface BranchRefEntry {