feat: record the commit a branch scrape read

Six platforms are read from a branch: System.dat, batocera-systems.json, checkBIOS.sh, the MiSTer BiosDB, the RetroPie scriptmodules and the RetroDECK component manifests. Their YAML named the branch, so the native export patched whatever the branch had become since the scrape. Each scraper now resolves the branch head once, reads every file at that commit and writes it into source. The export follows a recorded tarball and a recorded tree as it already followed a raw file URL.

tests/test_e2e.py

@@ -5497,7 +5497,7 @@ struct BurnDriver BurnDrvneogeo = {
 
         from scraper.misterfpga_scraper import Scraper
 
-        scraper = Scraper()
+        scraper = Scraper(url="https://example.test/bios_db.json.zip")
         scraper._raw_data = json.dumps(self.MISTER_DB)
         return scraper
 

tests/test_retrodeck_scraper_errors.py

@@ -34,6 +34,8 @@ class _Response(io.BytesIO):
 def _opener(failure):
     def urlopen(req, timeout=0):
         url = req.full_url
+        if "/commits/" in url:
+            return _Response(json.dumps({"sha": "a" * 40}).encode())
         if "api.github.com" in url:
             return _Response(json.dumps(TREE).encode())
         if "pcsx2" in url and failure is not None:
@@ -60,6 +62,20 @@ class ManifestErrors(unittest.TestCase):
         error = urllib.error.HTTPError("u", 404, "not found", {}, None)
         self.assertEqual([name for name, _ in self.fetch(error)], ["ppsspp"])
 
+    def test_the_tree_and_files_are_read_at_one_commit(self):
+        seen = []
+        opener = _opener(None)
+
+        def recording(req, timeout=0):
+            seen.append(req.full_url)
+            return opener(req, timeout)
+
+        with mock.patch("urllib.request.urlopen", recording):
+            Scraper()._fetch_remote_manifests()
+        reads = [url for url in seen if "/commits/" not in url]
+        self.assertTrue(reads)
+        self.assertTrue(all("a" * 40 in url for url in reads), reads)
+
     def test_a_truncated_tree_stops_the_scrape(self):
         TREE["truncated"] = True
         self.addCleanup(TREE.pop, "truncated")

tests/test_scraper_contract.py

@@ -11,6 +11,8 @@ exercised is a promise to contributors backed by nothing.
 from __future__ import annotations
 
 import importlib
+import io
+import json
 import subprocess
 import sys
 import tempfile
@@ -246,7 +248,9 @@ class RetroPiePackageList(unittest.TestCase):
     def _scraper(self, modules: dict[str, str]):
         from scraper.retropie_scraper import Scraper
 
-        scraper = Scraper()
+        # An explicit archive URL: the tests read a local archive, never the
+        # branch head.
+        scraper = Scraper(url="https://codeload.github.com/RetroPie/RetroPie-Setup/tar.gz/" + "0" * 40)
         scraper.module_ids(self._archive(modules))
         return scraper
 
@@ -502,6 +506,11 @@ class UnreadableReleaseStopsTheScrape(unittest.TestCase):
             "scraper.romm_scraper",
             "scraper.batocera_scraper",
             "scraper.recalbox_scraper",
+            "scraper.libretro_scraper",
+            "scraper.retrobat_scraper",
+            "scraper.emudeck_scraper",
+            "scraper.misterfpga_scraper",
+            "scraper.retropie_scraper",
         ):
             module = importlib.import_module(name)
             with self.subTest(scraper=name), mock.patch(
@@ -509,6 +518,52 @@ class UnreadableReleaseStopsTheScrape(unittest.TestCase):
             ), self.assertRaises(RuntimeError):
                 module.Scraper()
 
+    def test_a_branch_read_records_its_commit(self):
+        """System.dat, batocera-systems.json, checkBIOS.sh, the BiosDB and the
+        RetroPie scriptmodules are read from a branch. The YAML named the
+        branch, so the exporter patched whatever it had become since."""
+        sha = "0123456789abcdef0123456789abcdef01234567"
+
+        class Answer(io.BytesIO):
+            def __enter__(self):
+                return self
+
+            def __exit__(self, *_exc):
+                return False
+
+        def answer(request, timeout=None):
+            return Answer(json.dumps({"sha": sha}).encode())
+
+        for name in (
+            "scraper.libretro_scraper",
+            "scraper.retrobat_scraper",
+            "scraper.emudeck_scraper",
+            "scraper.misterfpga_scraper",
+            "scraper.retropie_scraper",
+        ):
+            module = importlib.import_module(name)
+            with self.subTest(scraper=name), mock.patch("urllib.request.urlopen", answer):
+                self.assertRegex(module.Scraper().url, f"/{sha}(/|$)")
+
+    def test_the_export_reads_the_recorded_tarball(self):
+        from export_native import _tarball_at
+
+        branch = "https://codeload.github.com/RetroPie/RetroPie-Setup/tar.gz/refs/heads/master"
+        pinned = "https://codeload.github.com/RetroPie/RetroPie-Setup/tar.gz/" + "a" * 40
+        self.assertEqual(_tarball_at(branch, pinned), pinned)
+        other = "https://codeload.github.com/Other/Repo/tar.gz/" + "b" * 40
+        self.assertEqual(_tarball_at(branch, other), "")
+
+    def test_the_export_reads_every_component_at_the_recorded_tree(self):
+        from export_native import pinned_base
+
+        sha = "c" * 40
+        base = pinned_base(
+            {"pcsx2/component_manifest.json": "unused"},
+            {"source": f"https://github.com/RetroDECK/components/tree/{sha}"},
+        )
+        self.assertEqual(base, f"https://raw.githubusercontent.com/RetroDECK/components/{sha}/")
+
     def test_a_release_lookup_raises(self):
         with mock.patch("urllib.request.urlopen", self._refuse), self.assertRaises(
             RuntimeError