fix: emulator JSON-LD names every repository read

The structured data of an emulator page published one repository and one commit. A profile whose builds live in two repositories, or a port with an upstream_commit, now lists each repository with its own pin.

scripts/generate_site.py

@@ -1498,21 +1498,30 @@ def emulator_structured_data(profile: dict) -> dict:
     the core loads, their hashes, the source read or when.
     """
     data: dict = {}
-    source = profile.get("source")
-    if isinstance(source, dict):
-        source = next(iter(source.values()), "")
-    commit = profile.get("source_commit")
-    if isinstance(commit, dict):
-        commit = next(iter(commit.values()), "")
-    if source:
-        about = {
-            "@type": "SoftwareSourceCode",
-            "name": str(profile.get("emulator", "")),
-            "codeRepository": str(source),
-        }
-        if commit:
-            about["version"] = str(commit)
-        data["about"] = about
+    # Every repository the profile reads, each with its own pin: xemu's
+    # libretro build lives in another repository at another commit, and a
+    # port's upstream_commit was never published.
+    about = []
+    seen: set[str] = set()
+    for field in ("source", "upstream"):
+        urls = profile.get(field)
+        pins = profile.get(f"{field}_commit")
+        for mode, url in (urls.items() if isinstance(urls, dict) else [("", urls)]):
+            if not url or not str(url).startswith("http") or str(url) in seen:
+                continue
+            seen.add(str(url))
+            pin = pins.get(mode) if isinstance(pins, dict) else pins
+            entry = {
+                "@type": "SoftwareSourceCode",
+                "name": str(profile.get("emulator", "")),
+                "codeRepository": str(url),
+                "description": " ".join(filter(None, [mode, field])),
+            }
+            if pin:
+                entry["version"] = str(pin)
+            about.append(entry)
+    if about:
+        data["about"] = about[0] if len(about) == 1 else about
     if profile.get("profiled_date"):
         data["dateModified"] = str(profile["profiled_date"])
     parts = []

tests/test_audit_regressions.py

@@ -4125,6 +4125,22 @@ class EmulatorPagesStateTheirDataForMachines(unittest.TestCase):
         self.assertEqual(data["hasPart"][0]["identifier"], ["sha1:" + "a" * 40, "md5:490f666e"])
         self.assertNotIn("identifier", data["hasPart"][1])
 
+    def test_every_repository_with_its_own_pin(self):
+        """xemu's libretro build lives in another repository at another
+        commit, and a port's upstream_commit was never published."""
+        from scripts.generate_site import emulator_structured_data
+
+        data = emulator_structured_data({
+            "emulator": "xemu",
+            "source": {"standalone": "https://github.com/xemu-project/xemu",
+                       "libretro": "https://github.com/LibretroAdmin/xemu-libretro"},
+            "source_commit": {"standalone": "f9b14039", "libretro": "d9804db7"},
+            "upstream": "https://github.com/mborgerson/xemu", "upstream_commit": "aa11",
+        })
+        self.assertEqual(
+            [(a["codeRepository"].rsplit("/", 1)[1], a.get("version")) for a in data["about"]],
+            [("xemu", "f9b14039"), ("xemu-libretro", "d9804db7"), ("xemu", "aa11")])
+
 
 
 class AManifestHashesAFileOnce(unittest.TestCase):