initial commit

src/archipelago_tester/pipeline/download/asset_matcher.py

@@ -0,0 +1,75 @@
+"""Choosing which asset in a release belongs to this game."""
+
+import re
+
+from archipelago_tester.core.model.name import Name
+
+
+class AssetMatcher:
+    """Matches release assets and release titles to one sheet row."""
+
+    def __init__(self, config, game_name):
+        self.config = config
+        self.game = Name(game_name)
+
+    def named(self, assets, exact):
+        """The assets whose filename names this game."""
+        matched = []
+        for asset in assets:
+            name = Name(asset["name"].rsplit(".", 1)[0]).slug
+            if (name == self.game.slug) if exact else (name in self.game.slug):
+                matched.append(asset)
+        return matched
+
+    def select(self, assets):
+        """The assets to download for this game."""
+        if len(assets) <= 1:
+            return assets
+        for matches in (self.named(assets, True), self.named(assets, False)):
+            if len(matches) == 1:
+                return matches
+        return assets
+
+    def title_slug(self, release):
+        """A release's own title, reduced to the game it is for."""
+        title = release.get("name") or release.get("tag_name") or ""
+        title = re.sub(r"\bv?\d+(?:\.\d+)*\b", " ", title)
+        title = re.sub(r"\bapworld\b", " ", title, flags=re.IGNORECASE)
+        return Name(title).slug
+
+    def uninformative(self):
+        """Titles that name no game at all."""
+        return set(self.config.value(
+            "downloads", "uninformative_release_slugs", ()))
+
+    def claimed_by(self, title, sheet_names):
+        """The other sheet row this title names, or None.
+
+        Exact equality only. This is not "the title looks a bit off" -
+        it is positive evidence that a DIFFERENT row is named, which is
+        the only thing strong enough to skip a release the link points
+        at. "Mega Man X2" is exactly a row and is skipped when resolving
+        "Mega Man X"; "Mega Man X1" is nobody's row and is accepted.
+        """
+        if not sheet_names or not self.game.text:
+            return None
+        if not title or title in self.uninformative():
+            return None
+        if title == self.game.slug:
+            return None
+        for other in sheet_names:
+            slug = Name(other).slug
+            if slug != self.game.slug and slug == title:
+                return other
+        return None
+
+    def release_claimed(self, release, sheet_names):
+        return self.claimed_by(self.title_slug(release), sheet_names)
+
+    def identity_claimed(self, found_games, sheet_names):
+        """Whether a downloaded file identifies as another row's game."""
+        for found in found_games or []:
+            other = self.claimed_by(Name(found).slug, sheet_names)
+            if other:
+                return other
+        return None