datachain 0.3.16__py3-none-any.whl → 0.3.18__py3-none-any.whl

datachain/cache.py

@@ -1,56 +1,15 @@
-import hashlib
-import json
 import os
-from datetime import datetime
-from functools import partial
 from typing import TYPE_CHECKING, Optional
 
-import attrs
 from dvc_data.hashfile.db.local import LocalHashFileDB
 from dvc_objects.fs.local import LocalFileSystem
 from fsspec.callbacks import Callback, TqdmCallback
 
-from datachain.utils import TIME_ZERO
-
 from .progress import Tqdm
 
 if TYPE_CHECKING:
     from datachain.client import Client
-    from datachain.storage import StorageURI
-
-sha256 = partial(hashlib.sha256, usedforsecurity=False)
-
-
-@attrs.frozen
-class UniqueId:
-    storage: "StorageURI"
-    path: str
-    size: int
-    etag: str
-    version: str = ""
-    is_latest: bool = True
-    location: Optional[str] = None
-    last_modified: datetime = TIME_ZERO
-
-    def get_parsed_location(self) -> Optional[dict]:
-        if not self.location:
-            return None
-
-        loc_stack = (
-            json.loads(self.location)
-            if isinstance(self.location, str)
-            else self.location
-        )
-        if len(loc_stack) > 1:
-            raise NotImplementedError("Nested v-objects are not supported yet.")
-
-        return loc_stack[0]
-
-    def get_hash(self) -> str:
-        fingerprint = f"{self.storage}/{self.path}/{self.version}/{self.etag}"
-        if self.location:
-            fingerprint += f"/{self.location}"
-        return sha256(fingerprint.encode()).hexdigest()
+    from datachain.lib.file import File
 
 
 def try_scandir(path):
@@ -77,30 +36,30 @@ class DataChainCache:
     def tmp_dir(self):
         return self.odb.tmp_dir
 
-    def get_path(self, uid: UniqueId) -> Optional[str]:
-        if self.contains(uid):
-            return self.path_from_checksum(uid.get_hash())
+    def get_path(self, file: "File") -> Optional[str]:
+        if self.contains(file):
+            return self.path_from_checksum(file.get_hash())
         return None
 
-    def contains(self, uid: UniqueId) -> bool:
-        return self.odb.exists(uid.get_hash())
+    def contains(self, file: "File") -> bool:
+        return self.odb.exists(file.get_hash())
 
     def path_from_checksum(self, checksum: str) -> str:
         assert checksum
         return self.odb.oid_to_path(checksum)
 
-    def remove(self, uid: UniqueId) -> None:
-        self.odb.delete(uid.get_hash())
+    def remove(self, file: "File") -> None:
+        self.odb.delete(file.get_hash())
 
     async def download(
-        self, uid: UniqueId, client: "Client", callback: Optional[Callback] = None
+        self, file: "File", client: "Client", callback: Optional[Callback] = None
     ) -> None:
-        from_path = f"{uid.storage}/{uid.path}"
+        from_path = f"{file.source}/{file.path}"
         from dvc_objects.fs.utils import tmp_fname
 
         odb_fs = self.odb.fs
         tmp_info = odb_fs.join(self.odb.tmp_dir, tmp_fname())  # type: ignore[arg-type]
-        size = uid.size
+        size = file.size
         if size < 0:
             size = await client.get_size(from_path)
         cb = callback or TqdmCallback(
@@ -115,13 +74,13 @@ class DataChainCache:
                 cb.close()
 
         try:
-            oid = uid.get_hash()
+            oid = file.get_hash()
             self.odb.add(tmp_info, self.odb.fs, oid)
         finally:
             os.unlink(tmp_info)
 
-    def store_data(self, uid: UniqueId, contents: bytes) -> None:
-        checksum = uid.get_hash()
+    def store_data(self, file: "File", contents: bytes) -> None:
+        checksum = file.get_hash()
         dst = self.path_from_checksum(checksum)
         if not os.path.exists(dst):
             # Create the file only if it's not already in cache

datachain/catalog/catalog.py

@@ -34,7 +34,7 @@ import yaml
 from sqlalchemy import Column
 from tqdm import tqdm
 
-from datachain.cache import DataChainCache, UniqueId
+from datachain.cache import DataChainCache
 from datachain.client import Client
 from datachain.config import get_remote_config, read_config
 from datachain.dataset import (
@@ -68,8 +68,6 @@ from datachain.utils import (
     DataChainDir,
     batched,
     datachain_paths_join,
-    import_object,
-    parse_params_string,
 )
 
 from .datasource import DataSource
@@ -621,13 +619,13 @@ class Catalog:
                 code_ast.body[-1:] = new_expressions
         return code_ast
 
-    def get_client(self, uri: StorageURI, **config: Any) -> Client:
+    def get_client(self, uri: str, **config: Any) -> Client:
         """
         Return the client corresponding to the given source `uri`.
         """
         config = config or self.client_config
         cls = Client.get_implementation(uri)
-        return cls.from_source(uri, self.cache, **config)
+        return cls.from_source(StorageURI(uri), self.cache, **config)
 
     def enlist_source(
         self,
@@ -843,7 +841,7 @@ class Catalog:
         from datachain.query import DatasetQuery
 
         def _row_to_node(d: dict[str, Any]) -> Node:
-            del d["source"]
+            del d["file__source"]
             return Node.from_dict(d)
 
         enlisted_sources: list[tuple[bool, bool, Any]] = []
@@ -1148,30 +1146,28 @@ class Catalog:
         if not sources:
             raise ValueError("Sources needs to be non empty list")
 
-        from datachain.query import DatasetQuery
+        from datachain.lib.dc import DataChain
+        from datachain.query.session import Session
+
+        session = Session.get(catalog=self, client_config=client_config)
 
-        dataset_queries = []
+        chains = []
         for source in sources:
             if source.startswith(DATASET_PREFIX):
-                dq = DatasetQuery(
-                    name=source[len(DATASET_PREFIX) :],
-                    catalog=self,
-                    client_config=client_config,
+                dc = DataChain.from_dataset(
+                    source[len(DATASET_PREFIX) :], session=session
                 )
             else:
-                dq = DatasetQuery(
-                    path=source,
-                    catalog=self,
-                    client_config=client_config,
-                    recursive=recursive,
+                dc = DataChain.from_storage(
+                    source, session=session, recursive=recursive
                 )
 
-            dataset_queries.append(dq)
+            chains.append(dc)
 
         # create union of all dataset queries created from sources
-        dq = reduce(lambda ds1, ds2: ds1.union(ds2), dataset_queries)
+        dc = reduce(lambda dc1, dc2: dc1.union(dc2), chains)
         try:
-            dq.save(name)
+            dc.save(name)
         except Exception as e:  # noqa: BLE001
             try:
                 ds = self.get_dataset(name)
@@ -1435,7 +1431,7 @@ class Catalog:
 
     def get_file_signals(
         self, dataset_name: str, dataset_version: int, row: RowDict
-    ) -> Optional[dict]:
+    ) -> Optional[RowDict]:
         """
         Function that returns file signals from dataset row.
         Note that signal names are without prefix, so if there was 'laion__file__source'
@@ -1452,7 +1448,7 @@ class Catalog:
 
         version = self.get_dataset(dataset_name).get_version(dataset_version)
 
-        file_signals_values = {}
+        file_signals_values = RowDict()
 
         schema = SignalSchema.deserialize(version.feature_schema)
         for file_signals in schema.get_signals(File):
@@ -1480,6 +1476,8 @@ class Catalog:
         use_cache: bool = True,
         **config: Any,
     ):
+        from datachain.lib.file import File
+
         file_signals = self.get_file_signals(dataset_name, dataset_version, row)
         if not file_signals:
             raise RuntimeError("Cannot open object without file signals")
@@ -1487,22 +1485,10 @@ class Catalog:
         config = config or self.client_config
         client = self.get_client(file_signals["source"], **config)
         return client.open_object(
-            self._get_row_uid(file_signals),  # type: ignore [arg-type]
+            File._from_row(file_signals),
             use_cache=use_cache,
         )
 
-    def _get_row_uid(self, row: RowDict) -> UniqueId:
-        return UniqueId(
-            row["source"],
-            row["path"],
-            row["size"],
-            row["etag"],
-            row["version"],
-            row["is_latest"],
-            row["location"],
-            row["last_modified"],
-        )
-
     def ls(
         self,
         sources: list[str],
@@ -1731,26 +1717,6 @@ class Catalog:
             output, sources, client_config=client_config, recursive=recursive
         )
 
-    def apply_udf(
-        self,
-        udf_location: str,
-        source: str,
-        target_name: str,
-        parallel: Optional[int] = None,
-        params: Optional[str] = None,
-    ):
-        from datachain.query import DatasetQuery
-
-        if source.startswith(DATASET_PREFIX):
-            ds = DatasetQuery(name=source[len(DATASET_PREFIX) :], catalog=self)
-        else:
-            ds = DatasetQuery(path=source, catalog=self)
-        udf = import_object(udf_location)
-        if params:
-            args, kwargs = parse_params_string(params)
-            udf = udf(*args, **kwargs)
-        ds.add_signals(udf, parallel=parallel).save(target_name)
-
     def query(
         self,
         query_script: str,

datachain/cli.py

@@ -15,6 +15,7 @@ import shtab
 from datachain import utils
 from datachain.cli_utils import BooleanOptionalAction, CommaSeparatedArgs, KeyValueArgs
 from datachain.lib.dc import DataChain
+from datachain.telemetry import telemetry
 from datachain.utils import DataChainDir
 
 if TYPE_CHECKING:
@@ -494,27 +495,6 @@ def get_parser() -> ArgumentParser:  # noqa: PLR0915
         help="Query parameters",
     )
 
-    apply_udf_parser = subp.add_parser(
-        "apply-udf", parents=[parent_parser], description="Apply UDF"
-    )
-    apply_udf_parser.add_argument("udf", type=str, help="UDF location")
-    apply_udf_parser.add_argument("source", type=str, help="Source storage or dataset")
-    apply_udf_parser.add_argument("target", type=str, help="Target dataset name")
-    apply_udf_parser.add_argument(
-        "--parallel",
-        nargs="?",
-        type=int,
-        const=-1,
-        default=None,
-        metavar="N",
-        help=(
-            "Use multiprocessing to run the UDF with N worker processes. "
-            "N defaults to the CPU count."
-        ),
-    )
-    apply_udf_parser.add_argument(
-        "--udf-params", type=str, default=None, help="UDF class parameters"
-    )
     subp.add_parser(
         "clear-cache", parents=[parent_parser], description="Clear the local file cache"
     )
@@ -893,6 +873,7 @@ def main(argv: Optional[list[str]] = None) -> int:  # noqa: C901, PLR0912, PLR09
         # This also sets this environment variable for any subprocesses
         os.environ["DEBUG_SHOW_SQL_QUERIES"] = "True"
 
+    error = None
     try:
         catalog = get_catalog(client_config=client_config)
         if args.command == "cp":
@@ -1016,10 +997,6 @@ def main(argv: Optional[list[str]] = None) -> int:  # noqa: C901, PLR0912, PLR09
                 parallel=args.parallel,
                 params=args.param,
             )
-        elif args.command == "apply-udf":
-            catalog.apply_udf(
-                args.udf, args.source, args.target, args.parallel, args.udf_params
-            )
         elif args.command == "clear-cache":
             clear_cache(catalog)
         elif args.command == "gc":
@@ -1028,14 +1005,16 @@ def main(argv: Optional[list[str]] = None) -> int:  # noqa: C901, PLR0912, PLR09
             print(f"invalid command: {args.command}", file=sys.stderr)
             return 1
         return 0
-    except BrokenPipeError:
+    except BrokenPipeError as exc:
         # Python flushes standard streams on exit; redirect remaining output
         # to devnull to avoid another BrokenPipeError at shutdown
         # See: https://docs.python.org/3/library/signal.html#note-on-sigpipe
+        error = str(exc)
         devnull = os.open(os.devnull, os.O_WRONLY)
         os.dup2(devnull, sys.stdout.fileno())
         return 141  # 128 + 13 (SIGPIPE)
     except (KeyboardInterrupt, Exception) as exc:
+        error = str(exc)
         if isinstance(exc, KeyboardInterrupt):
             msg = "Operation cancelled by the user"
         else:
@@ -1053,3 +1032,5 @@ def main(argv: Optional[list[str]] = None) -> int:  # noqa: C901, PLR0912, PLR09
 
             pdb.post_mortem()
         return 1
+    finally:
+        telemetry.send_cli_call(args.command, error=error)

datachain/client/fsspec.py

@@ -3,7 +3,6 @@ import functools
 import logging
 import multiprocessing
 import os
-import posixpath
 import re
 import sys
 from abc import ABC, abstractmethod
@@ -26,8 +25,8 @@ from fsspec.asyn import get_loop, sync
 from fsspec.callbacks import DEFAULT_CALLBACK, Callback
 from tqdm import tqdm
 
-from datachain.cache import DataChainCache, UniqueId
-from datachain.client.fileslice import FileSlice, FileWrapper
+from datachain.cache import DataChainCache
+from datachain.client.fileslice import FileWrapper
 from datachain.error import ClientError as DataChainClientError
 from datachain.lib.file import File
 from datachain.nodes_fetcher import NodesFetcher
@@ -187,8 +186,8 @@ class Client(ABC):
     def url(self, path: str, expires: int = 3600, **kwargs) -> str:
         return self.fs.sign(self.get_full_path(path), expiration=expires, **kwargs)
 
-    async def get_current_etag(self, uid: UniqueId) -> str:
-        info = await self.fs._info(self.get_full_path(uid.path))
+    async def get_current_etag(self, file: "File") -> str:
+        info = await self.fs._info(self.get_full_path(file.path))
         return self.info_to_file(info, "").etag
 
     async def get_size(self, path: str) -> int:
@@ -317,7 +316,7 @@ class Client(ABC):
 
     def instantiate_object(
         self,
-        uid: UniqueId,
+        file: "File",
         dst: str,
         progress_bar: tqdm,
         force: bool = False,
@@ -328,10 +327,10 @@ class Client(ABC):
             else:
                 progress_bar.close()
                 raise FileExistsError(f"Path {dst} already exists")
-        self.do_instantiate_object(uid, dst)
+        self.do_instantiate_object(file, dst)
 
-    def do_instantiate_object(self, uid: "UniqueId", dst: str) -> None:
-        src = self.cache.get_path(uid)
+    def do_instantiate_object(self, file: "File", dst: str) -> None:
+        src = self.cache.get_path(file)
         assert src is not None
 
         try:
@@ -341,66 +340,33 @@ class Client(ABC):
             copy2(src, dst)
 
     def open_object(
-        self, uid: UniqueId, use_cache: bool = True, cb: Callback = DEFAULT_CALLBACK
+        self, file: File, use_cache: bool = True, cb: Callback = DEFAULT_CALLBACK
     ) -> BinaryIO:
         """Open a file, including files in tar archives."""
-        location = uid.get_parsed_location()
-        if use_cache and (cache_path := self.cache.get_path(uid)):
+        if use_cache and (cache_path := self.cache.get_path(file)):
             return open(cache_path, mode="rb")  # noqa: SIM115
-        if location and location["vtype"] == "tar":
-            return self._open_tar(uid, use_cache=True)
-        return FileWrapper(self.fs.open(self.get_full_path(uid.path)), cb)  # type: ignore[return-value]
-
-    def _open_tar(self, uid: UniqueId, use_cache: bool = True):
-        location = uid.get_parsed_location()
-        assert location
-
-        offset = location["offset"]
-        size = location["size"]
-        parent = location["parent"]
-
-        parent_uid = UniqueId(
-            parent["source"],
-            parent["path"],
-            parent["size"],
-            parent["etag"],
-            location=parent["location"],
-        )
-        f = self.open_object(parent_uid, use_cache=use_cache)
-        return FileSlice(f, offset, size, posixpath.basename(uid.path))
-
-    def download(self, uid: UniqueId, *, callback: Callback = DEFAULT_CALLBACK) -> None:
-        sync(get_loop(), functools.partial(self._download, uid, callback=callback))
-
-    async def _download(self, uid: UniqueId, *, callback: "Callback" = None) -> None:
-        if self.cache.contains(uid):
+        assert not file.location
+        return FileWrapper(self.fs.open(self.get_full_path(file.path)), cb)  # type: ignore[return-value]
+
+    def download(self, file: File, *, callback: Callback = DEFAULT_CALLBACK) -> None:
+        sync(get_loop(), functools.partial(self._download, file, callback=callback))
+
+    async def _download(self, file: File, *, callback: "Callback" = None) -> None:
+        if self.cache.contains(file):
             # Already in cache, so there's nothing to do.
             return
-        await self._put_in_cache(uid, callback=callback)
+        await self._put_in_cache(file, callback=callback)
 
-    def put_in_cache(self, uid: UniqueId, *, callback: "Callback" = None) -> None:
-        sync(get_loop(), functools.partial(self._put_in_cache, uid, callback=callback))
+    def put_in_cache(self, file: File, *, callback: "Callback" = None) -> None:
+        sync(get_loop(), functools.partial(self._put_in_cache, file, callback=callback))
 
-    async def _put_in_cache(
-        self, uid: UniqueId, *, callback: "Callback" = None
-    ) -> None:
-        location = uid.get_parsed_location()
-        if location and location["vtype"] == "tar":
-            loop = asyncio.get_running_loop()
-            await loop.run_in_executor(
-                None, functools.partial(self._download_from_tar, uid, callback=callback)
-            )
-            return
-        if uid.etag:
-            etag = await self.get_current_etag(uid)
-            if uid.etag != etag:
+    async def _put_in_cache(self, file: File, *, callback: "Callback" = None) -> None:
+        assert not file.location
+        if file.etag:
+            etag = await self.get_current_etag(file)
+            if file.etag != etag:
                 raise FileNotFoundError(
-                    f"Invalid etag for {uid.storage}/{uid.path}: "
-                    f"expected {uid.etag}, got {etag}"
+                    f"Invalid etag for {file.source}/{file.path}: "
+                    f"expected {file.etag}, got {etag}"
                 )
-        await self.cache.download(uid, self, callback=callback)
-
-    def _download_from_tar(self, uid, *, callback: "Callback" = None):
-        with self._open_tar(uid, use_cache=False) as f:
-            contents = f.read()
-        self.cache.store_data(uid, contents)
+        await self.cache.download(file, self, callback=callback)

datachain/client/local.py

@@ -7,7 +7,6 @@ from urllib.parse import urlparse
 
 from fsspec.implementations.local import LocalFileSystem
 
-from datachain.cache import UniqueId
 from datachain.lib.file import File
 from datachain.storage import StorageURI
 
@@ -114,8 +113,8 @@ class FileClient(Client):
             use_symlinks=use_symlinks,
         )
 
-    async def get_current_etag(self, uid: UniqueId) -> str:
-        info = self.fs.info(self.get_full_path(uid.path))
+    async def get_current_etag(self, file: "File") -> str:
+        info = self.fs.info(self.get_full_path(file.path))
         return self.info_to_file(info, "").etag
 
     async def get_size(self, path: str) -> int:

datachain/data_storage/metastore.py

@@ -297,39 +297,6 @@ class AbstractMetastore(ABC, Serializable):
     #
     # Dataset dependencies
     #
-
-    def add_dependency(
-        self,
-        dependency: DatasetDependency,
-        source_dataset_name: str,
-        source_dataset_version: int,
-    ) -> None:
-        """Add dependency to dataset or storage."""
-        if dependency.is_dataset:
-            self.add_dataset_dependency(
-                source_dataset_name,
-                source_dataset_version,
-                dependency.dataset_name,
-                int(dependency.version),
-            )
-        else:
-            self.add_storage_dependency(
-                source_dataset_name,
-                source_dataset_version,
-                StorageURI(dependency.name),
-                dependency.version,
-            )
-
-    @abstractmethod
-    def add_storage_dependency(
-        self,
-        source_dataset_name: str,
-        source_dataset_version: int,
-        storage_uri: StorageURI,
-        storage_timestamp_str: Optional[str] = None,
-    ) -> None:
-        """Adds storage dependency to dataset."""
-
     @abstractmethod
     def add_dataset_dependency(
         self,
@@ -1268,32 +1235,6 @@ class AbstractDBMetastore(AbstractMetastore):
     #
     # Dataset dependencies
     #
-
-    def _insert_dataset_dependency(self, data: dict[str, Any]) -> None:
-        """Method for inserting dependencies."""
-        self.db.execute(self._datasets_dependencies_insert().values(**data))
-
-    def add_storage_dependency(
-        self,
-        source_dataset_name: str,
-        source_dataset_version: int,
-        storage_uri: StorageURI,
-        storage_timestamp_str: Optional[str] = None,
-    ) -> None:
-        source_dataset = self.get_dataset(source_dataset_name)
-        storage = self.get_storage(storage_uri)
-
-        self._insert_dataset_dependency(
-            {
-                "source_dataset_id": source_dataset.id,
-                "source_dataset_version_id": (
-                    source_dataset.get_version(source_dataset_version).id
-                ),
-                "bucket_id": storage.id,
-                "bucket_version": storage_timestamp_str,
-            }
-        )
-
     def add_dataset_dependency(
         self,
         source_dataset_name: str,
@@ -1305,15 +1246,15 @@ class AbstractDBMetastore(AbstractMetastore):
         source_dataset = self.get_dataset(source_dataset_name)
         dataset = self.get_dataset(dataset_name)
 
-        self._insert_dataset_dependency(
-            {
-                "source_dataset_id": source_dataset.id,
-                "source_dataset_version_id": (
+        self.db.execute(
+            self._datasets_dependencies_insert().values(
+                source_dataset_id=source_dataset.id,
+                source_dataset_version_id=(
                     source_dataset.get_version(source_dataset_version).id
                 ),
-                "dataset_id": dataset.id,
-                "dataset_version_id": dataset.get_version(dataset_version).id,
-            }
+                dataset_id=dataset.id,
+                dataset_version_id=dataset.get_version(dataset_version).id,
+            )
         )
 
     def update_dataset_dependency_source(

datachain/data_storage/sqlite.py

@@ -651,11 +651,14 @@ class SQLiteWarehouse(AbstractWarehouse):
         self, dataset: DatasetRecord, version: int
     ) -> list[StorageURI]:
         dr = self.dataset_rows(dataset, version)
-        query = dr.select(dr.c.source).distinct()
+        query = dr.select(dr.c.file__source).distinct()
         cur = self.db.cursor()
         cur.row_factory = sqlite3.Row  # type: ignore[assignment]
 
-        return [StorageURI(row["source"]) for row in self.db.execute(query, cursor=cur)]
+        return [
+            StorageURI(row["file__source"])
+            for row in self.db.execute(query, cursor=cur)
+        ]
 
     def merge_dataset_rows(
         self,

datachain/data_storage/warehouse.py

@@ -942,28 +942,6 @@ class AbstractWarehouse(ABC, Serializable):
                 self.db.drop_table(Table(name, self.db.metadata), if_exists=True)
                 pbar.update(1)
 
-    def changed_query(
-        self,
-        source_query: sa.sql.selectable.Select,
-        target_query: sa.sql.selectable.Select,
-    ) -> sa.sql.selectable.Select:
-        sq = source_query.alias("source_query")
-        tq = target_query.alias("target_query")
-
-        source_target_join = sa.join(
-            sq, tq, (sq.c.source == tq.c.source) & (sq.c.path == tq.c.path)
-        )
-
-        return (
-            select(*sq.c)
-            .select_from(source_target_join)
-            .where(
-                (sq.c.last_modified > tq.c.last_modified)
-                & (sq.c.is_latest == true())
-                & (tq.c.is_latest == true())
-            )
-        )
-
 
 def _random_string(length: int) -> str:
     return "".join(

datachain/lib/arrow.py

@@ -49,7 +49,8 @@ class ArrowGenerator(Generator):
 
     def process(self, file: File):
         if file._caching_enabled:
-            path = file.get_local_path(download=True)
+            file.ensure_cached()
+            path = file.get_local_path()
             ds = dataset(path, schema=self.input_schema, **self.kwargs)
         elif self.nrows:
             path = _nrows_file(file, self.nrows)