From 89ac31780bbcc447549b2678b7c8d8214a556e46 Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Sun, 18 Oct 2026 20:40:23 +0200
Subject: [PATCH] fix: stop when a profile does not parse

generate_db wrote a database without the aliases of a profile it could not read, and the MAME and FBNeo refreshes passed without it. Each now names the file and exits non-zero.
---
 scripts/generate_db.py                | 27 ++++++++++++++++++++++++---
 scripts/scraper/fbneo_hash_scraper.py |  6 ++++--
 scripts/scraper/mame_hash_scraper.py  |  6 ++++--
 tests/test_db_aliases.py              | 25 +++++++++++++++++++++++++
 tests/test_hash_merge.py              | 22 ++++++++++++++++++++++
 5 files changed, 79 insertions(+), 7 deletions(-)

diff --git a/scripts/generate_db.py b/scripts/generate_db.py
index fc3864c9..803cc494 100644
--- a/scripts/generate_db.py
+++ b/scripts/generate_db.py
@@ -375,7 +375,13 @@ def main() -> int:
     if preserved:
         print(f"  Preserved {preserved} large file entries from existing database")
 
-    platform_aliases = _collect_all_aliases(files)
+    try:
+        platform_aliases = _collect_all_aliases(files)
+    except UnreadableSource as exc:
+        for source in exc.sources:
+            print(f"Error: cannot read {source}", file=sys.stderr)
+        print(f"Error: {args.output} not written", file=sys.stderr)
+        return 1
     for sha1, name_list in platform_aliases.items():
         for alias_entry in name_list:
             if sha1 not in aliases:
@@ -440,6 +446,14 @@ def _own_aliases(file_entry: dict, primary_names: set[str]) -> list[str]:
     ]
 
 
+class UnreadableSource(Exception):
+    """Platform or profile files that do not parse, named one per line."""
+
+    def __init__(self, sources: list[str]) -> None:
+        super().__init__("\n".join(sources))
+        self.sources = sources
+
+
 def _collect_all_aliases(files: dict) -> dict:
     """Collect alternate filenames from platform YAMLs, core-info, and known aliases.
 
@@ -464,6 +478,10 @@ def _collect_all_aliases(files: dict) -> dict:
         if name not in existing:
             aliases[matched_sha1].append({"name": name, "path": ""})
 
+    # A source that does not parse is reported and the run refused: written
+    # without it, the database lost every alias the file proves and the
+    # command exited 0 (dosbox-x's wqy fonts answered not_found).
+    unreadable: list[str] = []
     platforms_dir = Path("platforms")
     if platforms_dir.is_dir():
         for platform_name in list_registered_platforms(
@@ -474,7 +492,7 @@ def _collect_all_aliases(files: dict) -> dict:
                 with open(config_file) as f:
                     config = yaml_load(f) or {}
             except (yaml.YAMLError, OSError) as e:
-                print(f"Warning: {config_file.name}: {e}", file=sys.stderr)
+                unreadable.append(f"{config_file}: {e}")
                 continue
 
             for sys_id, system in config.get("systems", {}).items():
@@ -513,7 +531,8 @@ def _collect_all_aliases(files: dict) -> dict:
             try:
                 with open(emu_file) as f:
                     emu_config = yaml_load(f) or {}
-            except (yaml.YAMLError, OSError):
+            except (yaml.YAMLError, OSError) as e:
+                unreadable.append(f"{emu_file}: {e}")
                 continue
             # A profile whose entries name each other (dosbox-x opens
             # MT32_CONTROL.ROM and CM32L_CONTROL.ROM alike, so each entry
@@ -602,6 +621,8 @@ def _collect_all_aliases(files: dict) -> dict:
         for name in group:
             _add_alias(name, matched_sha1)
 
+    if unreadable:
+        raise UnreadableSource(unreadable)
     return aliases
 
 
diff --git a/scripts/scraper/fbneo_hash_scraper.py b/scripts/scraper/fbneo_hash_scraper.py
index 4b6099ef..f21fdd75 100644
--- a/scripts/scraper/fbneo_hash_scraper.py
+++ b/scripts/scraper/fbneo_hash_scraper.py
@@ -204,8 +204,10 @@ def _find_fbneo_profiles() -> list[Path]:
     for path in source_yaml_paths(EMULATORS_DIR):
         try:
             data = yaml_load(path.read_text(encoding="utf-8"))
-        except (yaml.YAMLError, OSError):
-            continue
+        except (yaml.YAMLError, OSError) as exc:
+            # Skipped, a broken fbneo.yml left the list and the refresh
+            # reported "no matching emulator profiles found" with exit 0.
+            raise RuntimeError(f"cannot read {path}: {exc}") from exc
         if not data or not isinstance(data, dict):
             continue
         upstream = data.get("upstream", "")
diff --git a/scripts/scraper/mame_hash_scraper.py b/scripts/scraper/mame_hash_scraper.py
index f530f302..076e01ef 100644
--- a/scripts/scraper/mame_hash_scraper.py
+++ b/scripts/scraper/mame_hash_scraper.py
@@ -216,8 +216,10 @@ def _find_mame_profiles() -> list[Path]:
                 and source.rstrip("/") in _MAME_TREES
             ):
                 profiles.append(path)
-        except (yaml.YAMLError, OSError):
-            continue
+        except (yaml.YAMLError, OSError) as exc:
+            # Skipped, a broken mame.yml left the list and the refresh
+            # reported success without touching it.
+            raise RuntimeError(f"cannot read {path}: {exc}") from exc
     return profiles
 
 
diff --git a/tests/test_db_aliases.py b/tests/test_db_aliases.py
index f8136433..3d2af69f 100644
--- a/tests/test_db_aliases.py
+++ b/tests/test_db_aliases.py
@@ -128,5 +128,30 @@ class CrossNamingStaysInItsProfile(unittest.TestCase):
         self.assertEqual([a["name"] for a in aliases["mt"]], ["MT32_CONTROL.ROM"])
         self.assertEqual([a["name"] for a in aliases["cm"]], ["CM32L_CONTROL.ROM"])
 
+class AnUnreadableSourceStopsTheRun(unittest.TestCase):
+    """A profile with a YAML error was skipped in silence: the database was
+    written without its aliases and the command exited 0."""
+
+    def test_the_file_is_named_and_nothing_collected(self):
+        import os
+        import shutil
+        import tempfile
+
+        import generate_db
+
+        tmp_root = Path(__file__).resolve().parents[1] / "tmp"
+        tmp_root.mkdir(exist_ok=True)
+        root = Path(tempfile.mkdtemp(dir=tmp_root))
+        self.addCleanup(shutil.rmtree, root, True)
+        previous = os.getcwd()
+        os.chdir(root)
+        self.addCleanup(os.chdir, previous)
+        Path("emulators").mkdir()
+        Path("emulators/broken.yml").write_text("files: [\n  - name: a\n")
+        with self.assertRaises(generate_db.UnreadableSource) as caught:
+            generate_db._collect_all_aliases({})
+        self.assertIn("broken.yml", str(caught.exception))
+
+
 if __name__ == "__main__":
     unittest.main()
diff --git a/tests/test_hash_merge.py b/tests/test_hash_merge.py
index d4dbd69c..c7a79a62 100644
--- a/tests/test_hash_merge.py
+++ b/tests/test_hash_merge.py
@@ -774,5 +774,27 @@ class OneCloneDirectoryPerRun(unittest.TestCase):
                 self.assertTrue(clone.name.endswith(f"-{os.getpid()}"))
 
 
+class ABrokenProfileStopsTheRefresh(unittest.TestCase):
+    """A profile with a YAML error left the list in silence: the refresh
+    printed 'no matching emulator profiles found' and exited 0."""
+
+    def test_both_scrapers_name_it(self):
+        from unittest import mock
+
+        from scripts.scraper import fbneo_hash_scraper, mame_hash_scraper
+
+        with tempfile.TemporaryDirectory() as directory:
+            Path(directory, "fbneo.yml").write_text("files: [\n  - name: a\n")
+            for module, attr, finder in (
+                (fbneo_hash_scraper, "EMULATORS_DIR", "_find_fbneo_profiles"),
+                (mame_hash_scraper, "_EMULATORS_DIR", "_find_mame_profiles"),
+            ):
+                with self.subTest(scraper=module.__name__), \
+                        mock.patch.object(module, attr, Path(directory)), \
+                        self.assertRaises(RuntimeError) as caught:
+                    getattr(module, finder)()
+                self.assertIn("fbneo.yml", str(caught.exception))
+
+
 if __name__ == "__main__":
     unittest.main()