Speed up startup by improving optimizing database opening

Do not open all shards for a SymbolIndex if we want to just get the
binary dir of it.

* bdx/cli.py (guess_index_path_for_directory): Do not open the
SymbolIndex, instead call the new static `get_binary_dir' function to
try to get the binary directory for the index at given path.
* bdx/index.py (DB_SUBDIR): New helper var.
(SymbolIndex.open): Use it.
(SymbolIndex.open_shard): Use it and remove the `readonly' argument as
shards should only be used for writing.
(SymbolIndex.get_binary_dir): New staticmethod.

Author: mkcms

bdx/cli.py

@@ -100,15 +100,12 @@ def guess_index_path_for_directory(directory: str | Path) -> Path:
 
     for index_path in cache_dir.iterdir():
         try:
-            with SymbolIndex.open(index_path, readonly=True) as index:
-                binary_dir = index.binary_dir()
-                if binary_dir is None:
-                    continue
-                if binary_dir == directory:
-                    return index_path
-                if binary_dir in directory.parents:
-                    return index_path
-        except SymbolIndex.Error:
+            binary_dir = SymbolIndex.get_binary_dir(index_path)
+            if binary_dir == directory:
+                return index_path
+            if binary_dir in directory.parents:
+                return index_path
+        except Exception:
             pass
 
     return SymbolIndex.default_path(directory)

bdx/index.py

@@ -507,6 +507,9 @@ def index_document(self, document: xapian.Document, **data: str):
                 handler.index(document, val)
 
 
+DB_SUBDIR = "db"
+
+
 class SymbolIndex:
     """Easy interface for a xapian database, with schema support."""
 
@@ -644,7 +647,9 @@ def open(directory: Path | str, readonly: bool = False) -> "SymbolIndex":
 
         """
         index = SymbolIndex(
-            Path(directory) / "db", readonly=readonly, is_shard=False
+            Path(directory) / DB_SUBDIR,
+            readonly=readonly,
+            is_shard=False,
         )
 
         debug("Opened index: {}", index.path)
@@ -655,13 +660,13 @@ def open(directory: Path | str, readonly: bool = False) -> "SymbolIndex":
         return index
 
     @staticmethod
-    def open_shard(
-        directory: Path | str, readonly: bool = False
-    ) -> "SymbolIndex":
+    def open_shard(directory: Path | str) -> "SymbolIndex":
         """Open a writable shard for index in given directory."""
-        for path in SymbolIndex.generate_shard_paths(Path(directory) / "db"):
+        for path in SymbolIndex.generate_shard_paths(
+            Path(directory) / DB_SUBDIR
+        ):
             try:
-                return SymbolIndex(path, readonly=readonly, is_shard=True)
+                return SymbolIndex(path, readonly=False, is_shard=True)
             except Exception:
                 pass
 
@@ -681,6 +686,16 @@ def generate_shard_paths(directory: Path | str) -> Iterator[Path]:
             yield directory.parent / f"{directory.name}.{i:0>3}"
             i += 1
 
+    @staticmethod
+    def get_binary_dir(index_path: Path | str) -> Optional[Path]:
+        """Try to get the saved binary directory for given index path."""
+        with SymbolIndex(
+            Path(index_path) / DB_SUBDIR,
+            readonly=True,
+            is_shard=True,
+        ) as index:
+            return index.binary_dir()
+
     def shards(self) -> Iterator[Path]:
         """Yield the shards of this database."""
         for x in self.generate_shard_paths(self.path):