fix: update_cucumber script should not fail on manually created releases (#396)

The script was assuming that `target_commitish` in the releases API
would generally be the hash of the release tag, other than for an edge
case on v29.0.0 which had been patched manually in the code.

In fact, `target_commitish` will only be a hash if the tag was created
before the release (e.g. by cucumber's automated release actions). If
the release is created manually in the github UI with a not-yet-existing
tag then `target_commitish` will instead be the branch name it was
created from. In that case, the only way to get the hash is to
separately look up the tag name in the tags API.

This was causing the script to fail due to two recent manual releases on
the upstream repo.

I have fixed this to now look up tag details if they're missing from the
releases API. As part of this, I have refactored to only look up
releases that are `>=` the one we are currently pinned to. This means we
will only need to look up tags for manual releases until we update past
them.

Author: acoulton

bin/update_cucumber

@@ -36,7 +36,7 @@ class {
         $currentVersion = $this->findComposerGherkinVersion($composerFile);
         echo "Current local version is {$currentVersion['hash']} (tagged {$currentVersion['tag_name']})\n";
 
-        $releases = $this->listGitHubReleases('cucumber/gherkin');
+        $releases = $this->listGitHubReleasesSince('cucumber/gherkin', $currentVersion['tag_name']);
         echo "Latest upstream version is {$releases[0]['hash']} (tagged {$releases[0]['tag_name']})\n";
 
         // We only want to bump by one version at a time so that any test failures are easier to trace
@@ -112,14 +112,12 @@ class {
     /**
      * @phpstan-return non-empty-list<TGitHubReleaseArray>
      */
-    private function listGitHubReleases(string $repo): array
+    private function listGitHubReleasesSince(string $repo, string $currentVersionTag): array
     {
         // GitHub requires API requests to send a user_agent header for tracing
         ini_set('user_agent', 'https://github.com/Behat/Gherkin updater');
         $releases = Behat\Gherkin\Filesystem::readJsonFileArray('https://api.github.com/repos/' . $repo . '/releases');
 
-        assert($releases !== [], 'github should have returned at least one release');
-
         // Sanity check and simplify the GitHub API response structure to reduce assertions and type issues elsewhere
         $releases = array_map(
             static function (mixed $release) {
@@ -138,19 +136,8 @@ class {
                     'github release JSON should match expected structure',
                 );
 
-                // Edge case: they created the 29.0.0 release manually and for some reason it reports a target_commitish
-                // of `main` rather than the hash of the v29.0.0 tag. We could look up the tag in the API, but not worth
-                // it just for this one so hardcoded the value instead.
-                if ($release['tag_name'] === 'v29.0.0') {
-                    $release['target_commitish'] = 'c7d527d49d67ae39fe5a91aabcc5578c23ef1c5c';
-                }
-
-                // Protect against any future releases that don't properly target a tag SHA.
-                assert(
-                    (bool) preg_match('/^[a-z0-9]{40}$/', $release['target_commitish']),
-                    'target_commitish for ' . $release['tag_name'] . ' should be a hash',
-                );
-
+                // Note: `target_commitish` may not actually be a hash, but we will look this up later once we know if
+                // we actually need this release. See populateMissingTagHashes below.
                 return [
                     'tag_name' => $release['tag_name'],
                     'hash' => $release['target_commitish'],
@@ -161,13 +148,92 @@ class {
             $releases,
         );
 
+        // Remove older releases - we only need to validate the current release & possibly bump forwards.
+        $releases = array_values(array_filter(
+            $releases,
+            static fn (array $release) => version_compare($release['tag_name'], $currentVersionTag, '>=')
+        ));
+        assert($releases !== [], 'Should have found at least one current or newer release');
+
+        // The github releases API only contains a tag/commit hash for releases that were created from an existing tag.
+        // We may need to look up hashes for releases where the release & tag were created manually from a branch.
+        $releases = $this->populateMissingTagHashes($repo, $releases);
+
         // Releases API is in order of github release ID, this may not be version order (e.g. major releases may be
         // interspersed with point releases to older versions). So sort it most recent -> oldest by version number.
         usort($releases, static fn (array $a, array $b) => version_compare($b['tag_name'], $a['tag_name']));
 
         return $releases;
     }
 
+    /**
+     * @phpstan-param non-empty-list<TGitHubReleaseArray> $releases
+     *
+     * @phpstan-return non-empty-list<TGitHubReleaseArray>
+     */
+    private function populateMissingTagHashes(string $repo, array $releases): array
+    {
+        if (array_filter($releases, $this->hasValidHash(...)) === $releases) {
+            // All releases we're interested in have a hash already, nothing to do
+            return $releases;
+        }
+
+        echo "Some releases do not have hashes, looking up tags\n";
+        $tags = Behat\Gherkin\Filesystem::readJsonFileArray('https://api.github.com/repos/' . $repo . '/tags');
+        $tagMap = [];
+        foreach ($tags as $tag) {
+            assert(
+                is_array($tag)
+                && is_string($tag['name'] ?? null)
+                && is_array($tag['commit'] ?? null)
+                && is_string($tag['commit']['sha'] ?? null),
+                'github tag JSON should match expected structure',
+            );
+            $tagMap[$tag['name']] = $tag['commit']['sha'];
+        }
+
+        return array_map(
+            function (array $release) use ($tagMap) {
+                if ($this->hasValidHash($release)) {
+                    return $release;
+                }
+
+                $originalHash = $release['hash'];
+                $tagHash = $tagMap[$release['tag_name']] ?? null;
+                if ($tagHash !== null) {
+                    $release['hash'] = $tagHash;
+                }
+
+                assert(
+                    $this->hasValidHash($release),
+                    sprintf(
+                        'github tags API should provide valid hash for %s with release target of %s',
+                        $release['tag_name'],
+                        $release['hash']
+                    )
+                );
+
+                echo sprintf(
+                    " - Identified %s (target: %s) as %s\n",
+                    $release['tag_name'],
+                    $originalHash,
+                    $release['hash']
+                );
+
+                return $release;
+            },
+            $releases,
+        );
+    }
+
+    /**
+     * @param TGitHubReleaseArray $release
+     */
+    private function hasValidHash(array $release): bool
+    {
+        return preg_match('/^[a-z0-9]{40}$/', $release['hash']) === 1;
+    }
+
     /**
      * @phpstan-param non-empty-list<TGitHubReleaseArray> $releases
      * @phpstan-param TCurrentVersionArray $currentVersion