cache migrate: add utility for migrating local cache files to 3.x

Author: pmrowla

dvc/cachemgr.py

@@ -1,10 +1,13 @@
 import os
-from typing import Optional, Tuple
+from typing import TYPE_CHECKING, Optional, Tuple
 
 from dvc.fs import GitFileSystem, Schemes
 from dvc_data.hashfile.db import get_odb
 from dvc_data.hashfile.hash import DEFAULT_ALGORITHM
 
+if TYPE_CHECKING:
+    from dvc.repo import Repo
+
 LEGACY_HASH_NAMES = {"md5-dos2unix", "params"}
 
 
@@ -99,3 +102,37 @@ def local_cache_dir(self) -> str:
         (i.e. `dvc cache dir`).
         """
         return self.legacy.path
+
+
+def migrate_2_to_3(repo: "Repo", dry: bool = False):
+    """Migrate legacy 2.x objects to 3.x cache.
+
+    Legacy 'md5-dos2unix' objects will be re-hashed with 'md5', added to 3.x cache,
+    and then a link from the legacy 2.x location to the 3.x location will be created.
+    """
+    from dvc.fs.callbacks import TqdmCallback
+    from dvc.ui import ui
+    from dvc_data.hashfile.db.migrate import migrate, prepare
+
+    src = repo.cache.legacy
+    dest = repo.cache.local
+    if dry:
+        oids = list(src._list_oids())  # pylint: disable=protected-access
+        ui.write(
+            f"{len(oids)} files will be re-hashed and migrated to the DVC 3.0 cache "
+            "location."
+        )
+        return
+
+    with TqdmCallback(
+        desc="Computing DVC 3.0 hashes",
+        unit="files",
+    ) as cb:
+        migration = prepare(src, dest, callback=cb)
+
+    with TqdmCallback(
+        desc="Migrating to DVC 3.0 cache",
+        unit="files",
+    ) as cb:
+        count = migrate(migration, callback=cb)
+    ui.write(f"Migrated {count} files to DVC 3.0 cache location.")

dvc/commands/cache.py

@@ -2,6 +2,7 @@
 import os
 
 from dvc.cli import completion
+from dvc.cli.command import CmdBase
 from dvc.cli.utils import append_doc_link, fix_subparsers
 from dvc.commands.config import CmdConfig
 from dvc.ui import ui
@@ -36,6 +37,14 @@ def run(self):
         return 0
 
 
+class CmdCacheMigrate(CmdBase):
+    def run(self):
+        from dvc.cachemgr import migrate_2_to_3
+
+        migrate_2_to_3(self.repo, dry=self.args.dry)
+        return 0
+
+
 def add_parser(subparsers, parent_parser):
     from dvc.commands.config import parent_config_parser
 
@@ -54,8 +63,6 @@ def add_parser(subparsers, parent_parser):
         help="Use `dvc cache CMD --help` for command-specific help.",
     )
 
-    fix_subparsers(cache_subparsers)
-
     parent_cache_config_parser = argparse.ArgumentParser(
         add_help=False, parents=[parent_config_parser]
     )
@@ -86,3 +93,23 @@ def add_parser(subparsers, parent_parser):
         nargs="?",
     ).complete = completion.DIR
     cache_dir_parser.set_defaults(func=CmdCacheDir)
+
+    CACHE_MIGRATE_HELP = "Migrate cached files to the DVC 3.0 cache location."
+    cache_migrate_parser = cache_subparsers.add_parser(
+        "migrate",
+        parents=[parent_parser],
+        description=append_doc_link(CACHE_HELP, "cache/migrate"),
+        help=CACHE_MIGRATE_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    cache_migrate_parser.add_argument(
+        "--dry",
+        help=(
+            "Only print actions which would be taken without actually migrating "
+            "any data."
+        ),
+        action="store_true",
+    )
+    cache_migrate_parser.set_defaults(func=CmdCacheMigrate)
+
+    fix_subparsers(cache_subparsers)