datachain 0.1.13__py3-none-any.whl → 0.2.1__py3-none-any.whl

datachain/__init__.py

@@ -1,4 +0,0 @@
-try:
-    from ._version import version as __version__
-except ImportError:
-    __version__ = "UNKNOWN"

datachain/asyn.py

@@ -82,13 +82,13 @@ class AsyncMapper(Generic[InputT, ResultT]):
         for _i in range(self.workers):
             self.start_task(self.worker())
         try:
-            done, pending = await asyncio.wait(
+            done, _pending = await asyncio.wait(
                 self._tasks, return_when=asyncio.FIRST_COMPLETED
             )
             self.gather_exceptions(done)
             assert producer.done()
             join = self.start_task(self.work_queue.join())
-            done, pending = await asyncio.wait(
+            done, _pending = await asyncio.wait(
                 self._tasks, return_when=asyncio.FIRST_COMPLETED
             )
             self.gather_exceptions(done)
@@ -208,7 +208,7 @@ class OrderedMapper(AsyncMapper[InputT, ResultT]):
 
     async def _pop_result(self) -> Optional[ResultT]:
         if self.heap and self.heap[0][0] == self._next_yield:
-            i, out = heappop(self.heap)
+            _i, out = heappop(self.heap)
         else:
             self._getters[self._next_yield] = get_value = self.loop.create_future()
             out = await get_value

datachain/catalog/__init__.py

@@ -8,10 +8,10 @@ from .catalog import (
 from .loader import get_catalog
 
 __all__ = [
+    "QUERY_DATASET_PREFIX",
+    "QUERY_SCRIPT_CANCELED_EXIT_CODE",
+    "QUERY_SCRIPT_INVALID_LAST_STATEMENT_EXIT_CODE",
     "Catalog",
     "get_catalog",
     "parse_edatachain_file",
-    "QUERY_SCRIPT_INVALID_LAST_STATEMENT_EXIT_CODE",
-    "QUERY_SCRIPT_CANCELED_EXIT_CODE",
-    "QUERY_DATASET_PREFIX",
 ]

datachain/catalog/catalog.py

@@ -142,6 +142,7 @@ class QueryResult(NamedTuple):
     version: Optional[int]
     output: str
     preview: Optional[list[dict]]
+    metrics: dict[str, Any]
 
 
 class DatasetRowsFetcher(NodesThreadPool):
@@ -876,13 +877,11 @@ class Catalog:
             # so this is to improve performance
             return None
 
-        dsrc_all = []
+        dsrc_all: list[DataSource] = []
         for listing, file_path in enlisted_sources:
             nodes = listing.expand_path(file_path)
             dir_only = file_path.endswith("/")
-            for node in nodes:
-                dsrc_all.append(DataSource(listing, node, dir_only))
-
+            dsrc_all.extend(DataSource(listing, node, dir_only) for node in nodes)
         return dsrc_all
 
     def enlist_sources_grouped(
@@ -1997,6 +1996,7 @@ class Catalog:
             version=version,
             output=output,
             preview=exec_result.preview,
+            metrics=exec_result.metrics,
         )
 
     def run_query(
@@ -2068,8 +2068,8 @@ class Catalog:
                 "DATACHAIN_JOB_ID": job_id or "",
             },
         )
-        with subprocess.Popen(
-            [python_executable, "-c", query_script_compiled],  # noqa: S603
+        with subprocess.Popen(  # noqa: S603
+            [python_executable, "-c", query_script_compiled],
             env=envs,
             stdout=subprocess.PIPE if capture_output else None,
             stderr=subprocess.STDOUT if capture_output else None,

datachain/catalog/loader.py

@@ -35,7 +35,7 @@ def get_id_generator() -> "AbstractIDGenerator":
         id_generator_obj = deserialize(id_generator_serialized)
         if not isinstance(id_generator_obj, AbstractIDGenerator):
             raise RuntimeError(
-                f"Deserialized ID generator is not an instance of AbstractIDGenerator: "
+                "Deserialized ID generator is not an instance of AbstractIDGenerator: "
                 f"{id_generator_obj}"
             )
         return id_generator_obj
@@ -67,7 +67,7 @@ def get_metastore(id_generator: Optional["AbstractIDGenerator"]) -> "AbstractMet
         metastore_obj = deserialize(metastore_serialized)
         if not isinstance(metastore_obj, AbstractMetastore):
             raise RuntimeError(
-                f"Deserialized Metastore is not an instance of AbstractMetastore: "
+                "Deserialized Metastore is not an instance of AbstractMetastore: "
                 f"{metastore_obj}"
             )
         return metastore_obj
@@ -101,7 +101,7 @@ def get_warehouse(id_generator: Optional["AbstractIDGenerator"]) -> "AbstractWar
         warehouse_obj = deserialize(warehouse_serialized)
         if not isinstance(warehouse_obj, AbstractWarehouse):
             raise RuntimeError(
-                f"Deserialized Warehouse is not an instance of AbstractWarehouse: "
+                "Deserialized Warehouse is not an instance of AbstractWarehouse: "
                 f"{warehouse_obj}"
             )
         return warehouse_obj

datachain/cli.py

@@ -5,13 +5,14 @@ import sys
 import traceback
 from argparse import SUPPRESS, Action, ArgumentParser, ArgumentTypeError, Namespace
 from collections.abc import Iterable, Iterator, Mapping, Sequence
+from importlib.metadata import PackageNotFoundError, version
 from itertools import chain
 from multiprocessing import freeze_support
 from typing import TYPE_CHECKING, Optional, Union
 
 import shtab
 
-from datachain import __version__, utils
+from datachain import utils
 from datachain.cli_utils import BooleanOptionalAction, CommaSeparatedArgs, KeyValueArgs
 from datachain.utils import DataChainDir
 
@@ -96,6 +97,12 @@ def add_show_args(parser: ArgumentParser) -> None:
 
 
 def get_parser() -> ArgumentParser:  # noqa: PLR0915
+    try:
+        __version__ = version("datachain")
+    except PackageNotFoundError:
+        # package is not installed
+        __version__ = "unknown"
+
     parser = ArgumentParser(
         description="DataChain: Wrangle unstructured AI data at scale", prog="datachain"
     )
@@ -845,6 +852,7 @@ def query(
         query=script_content,
         query_type=JobQueryType.PYTHON,
         python_version=python_version,
+        params=params,
     )
 
     try:
@@ -870,7 +878,7 @@ def query(
         )
         raise
 
-    catalog.metastore.set_job_status(job_id, JobStatus.COMPLETE)
+    catalog.metastore.set_job_status(job_id, JobStatus.COMPLETE, metrics=result.metrics)
 
     show_records(result.preview, collapse_columns=not no_collapse)
 

datachain/client/azure.py

@@ -1,10 +1,12 @@
+import posixpath
 from typing import Any
 
 from adlfs import AzureBlobFileSystem
+from tqdm import tqdm
 
 from datachain.node import Entry
 
-from .fsspec import DELIMITER, Client
+from .fsspec import DELIMITER, Client, ResultQueue
 
 
 class AzureClient(Client):
@@ -28,3 +30,37 @@ class AzureClient(Client):
             last_modified=v["last_modified"],
             size=v.get("size", ""),
         )
+
+    async def _fetch_flat(self, start_prefix: str, result_queue: ResultQueue) -> None:
+        prefix = start_prefix
+        if prefix:
+            prefix = prefix.lstrip(DELIMITER) + DELIMITER
+        found = False
+        try:
+            with tqdm(desc=f"Listing {self.uri}", unit=" objects") as pbar:
+                async with self.fs.service_client.get_container_client(
+                    container=self.name
+                ) as container_client:
+                    async for page in container_client.list_blobs(
+                        include=["metadata", "versions"], name_starts_with=prefix
+                    ).by_page():
+                        entries = []
+                        async for b in page:
+                            found = True
+                            if not self._is_valid_key(b["name"]):
+                                continue
+                            info = (await self.fs._details([b]))[0]
+                            full_path = info["name"]
+                            parent = posixpath.dirname(self.rel_path(full_path))
+                            entries.append(self.convert_info(info, parent))
+                        if entries:
+                            await result_queue.put(entries)
+                            pbar.update(len(entries))
+                    if not found:
+                        raise FileNotFoundError(
+                            f"Unable to resolve remote path: {prefix}"
+                        )
+        finally:
+            result_queue.put_nowait(None)
+
+    _fetch_default = _fetch_flat

datachain/client/fsspec.py

@@ -202,7 +202,7 @@ class Client(ABC):
         try:
             impl = getattr(self, f"_fetch_{method}")
         except AttributeError:
-            raise ValueError("Unknown indexing method '{method}'") from None
+            raise ValueError(f"Unknown indexing method '{method}'") from None
         result_queue: ResultQueue = asyncio.Queue()
         loop = get_loop()
         main_task = loop.create_task(impl(start_prefix, result_queue))

datachain/client/local.py

@@ -135,7 +135,7 @@ class FileClient(Client):
         return posixpath.relpath(path, self.name)
 
     def get_full_path(self, rel_path):
-        full_path = Path(self.name, rel_path).as_uri()
+        full_path = Path(self.name, rel_path).as_posix()
         if rel_path.endswith("/") or not rel_path:
             full_path += "/"
         return full_path

datachain/data_storage/__init__.py

@@ -5,8 +5,8 @@ from .warehouse import AbstractWarehouse
 
 __all__ = [
     "AbstractDBIDGenerator",
-    "AbstractIDGenerator",
     "AbstractDBMetastore",
+    "AbstractIDGenerator",
     "AbstractMetastore",
     "AbstractWarehouse",
     "JobQueryType",

datachain/data_storage/metastore.py

@@ -385,6 +385,7 @@ class AbstractMetastore(ABC, Serializable):
         query_type: JobQueryType = JobQueryType.PYTHON,
         workers: int = 1,
         python_version: Optional[str] = None,
+        params: Optional[dict[str, str]] = None,
     ) -> str:
         """
         Creates a new job.
@@ -398,6 +399,7 @@ class AbstractMetastore(ABC, Serializable):
         status: JobStatus,
         error_message: Optional[str] = None,
         error_stack: Optional[str] = None,
+        metrics: Optional[dict[str, Any]] = None,
     ) -> None:
         """Set the status of the given job."""
 
@@ -1165,9 +1167,7 @@ class AbstractDBMetastore(AbstractMetastore):
         return dataset_version
 
     def _parse_dataset(self, rows) -> Optional[DatasetRecord]:
-        versions = []
-        for r in rows:
-            versions.append(self.dataset_class.parse(*r))
+        versions = [self.dataset_class.parse(*r) for r in rows]
         if not versions:
             return None
         return reduce(lambda ds, version: ds.merge_versions(version), versions)
@@ -1463,6 +1463,8 @@ class AbstractDBMetastore(AbstractMetastore):
             Column("python_version", Text, nullable=True),
             Column("error_message", Text, nullable=False, default=""),
             Column("error_stack", Text, nullable=False, default=""),
+            Column("params", JSON, nullable=False),
+            Column("metrics", JSON, nullable=False),
         ]
 
     @cached_property
@@ -1489,6 +1491,7 @@ class AbstractDBMetastore(AbstractMetastore):
         query_type: JobQueryType = JobQueryType.PYTHON,
         workers: int = 1,
         python_version: Optional[str] = None,
+        params: Optional[dict[str, str]] = None,
         conn: Optional[Any] = None,
     ) -> str:
         """
@@ -1508,6 +1511,8 @@ class AbstractDBMetastore(AbstractMetastore):
                 python_version=python_version,
                 error_message="",
                 error_stack="",
+                params=json.dumps(params or {}),
+                metrics=json.dumps({}),
             ),
             conn=conn,
         )
@@ -1519,6 +1524,7 @@ class AbstractDBMetastore(AbstractMetastore):
         status: JobStatus,
         error_message: Optional[str] = None,
         error_stack: Optional[str] = None,
+        metrics: Optional[dict[str, Any]] = None,
         conn: Optional[Any] = None,
     ) -> None:
         """Set the status of the given job."""
@@ -1529,6 +1535,8 @@ class AbstractDBMetastore(AbstractMetastore):
             values["error_message"] = error_message
         if error_stack:
             values["error_stack"] = error_stack
+        if metrics:
+            values["metrics"] = json.dumps(metrics)
         self.db.execute(
             self._jobs_update(self._jobs.c.id == job_id).values(**values),
             conn=conn,

datachain/data_storage/schema.py

@@ -31,11 +31,10 @@ def dedup_columns(columns: Iterable[sa.Column]) -> list[sa.Column]:
     """
     c_set: dict[str, sa.Column] = {}
     for c in columns:
-        if ec := c_set.get(c.name, None):
+        if (ec := c_set.get(c.name, None)) is not None:
             if str(ec.type) != str(c.type):
                 raise ValueError(
-                    f"conflicting types for column {c.name}:"
-                    f"{c.type!s} and {ec.type!s}"
+                    f"conflicting types for column {c.name}:{c.type!s} and {ec.type!s}"
                 )
             continue
         c_set[c.name] = c
@@ -172,8 +171,8 @@ class DataTable:
     ):
         # copy columns, since re-using the same objects from another table
         # may raise an error
-        columns = [cls.copy_column(c) for c in columns if c.name != "id"]
-        columns = [sa.Column("id", Int, primary_key=True), *columns]
+        columns = cls.sys_columns() + [cls.copy_column(c) for c in columns]
+        columns = dedup_columns(columns)
 
         if metadata is None:
             metadata = sa.MetaData()
@@ -231,10 +230,17 @@ class DataTable:
     def delete(self):
         return self.apply_conditions(self.table.delete())
 
+    @staticmethod
+    def sys_columns():
+        return [
+            sa.Column("id", Int, primary_key=True),
+            sa.Column("random", Int64, nullable=False, default=f.random()),
+        ]
+
     @classmethod
     def file_columns(cls) -> list[sa.Column]:
         return [
-            sa.Column("id", Int, primary_key=True),
+            *cls.sys_columns(),
             sa.Column("vtype", String, nullable=False, index=True),
             sa.Column("dir_type", Int, index=True),
             sa.Column("parent", String, index=True),
@@ -246,7 +252,6 @@ class DataTable:
             sa.Column("size", Int64, nullable=False, index=True),
             sa.Column("owner_name", String),
             sa.Column("owner_id", String),
-            sa.Column("random", Int64, nullable=False),
             sa.Column("location", JSON),
             sa.Column("source", String, nullable=False),
         ]

datachain/data_storage/sqlite.py

@@ -33,6 +33,7 @@ from datachain.data_storage.schema import (
 from datachain.dataset import DatasetRecord
 from datachain.error import DataChainError
 from datachain.sql.sqlite import create_user_defined_sql_functions, sqlite_dialect
+from datachain.sql.sqlite.base import load_usearch_extension
 from datachain.sql.types import SQLType
 from datachain.storage import StorageURI
 from datachain.utils import DataChainDir
@@ -114,6 +115,8 @@ class SQLiteDatabaseEngine(DatabaseEngine):
             if os.environ.get("DEBUG_SHOW_SQL_QUERIES"):
                 db.set_trace_callback(print)
 
+            load_usearch_extension(db)
+
             return cls(engine, MetaData(), db, db_file)
         except RuntimeError:
             raise DataChainError("Can't connect to SQLite DB") from None

datachain/data_storage/warehouse.py

@@ -95,14 +95,14 @@ class AbstractWarehouse(ABC, Serializable):
 
         exc = None
         try:
-            if col_python_type == list and value_type in (list, tuple, set):
+            if col_python_type is list and value_type in (list, tuple, set):
                 if len(val) == 0:
                     return []
                 item_python_type = self.python_type(col_type.item_type)
-                if item_python_type != list:
+                if item_python_type is not list:
                     if isinstance(val[0], item_python_type):
                         return val
-                    if item_python_type == float and isinstance(val[0], int):
+                    if item_python_type is float and isinstance(val[0], int):
                         return [float(i) for i in val]
                 # Optimization: Reuse these values for each function call within the
                 # list comprehension.
@@ -114,18 +114,18 @@ class AbstractWarehouse(ABC, Serializable):
                 )
                 return [self.convert_type(i, *item_type_info) for i in val]
             # Special use case with JSON type as we save it as string
-            if col_python_type == dict or col_type_name == "JSON":
-                if value_type == str:
+            if col_python_type is dict or col_type_name == "JSON":
+                if value_type is str:
                     return val
                 if value_type in (dict, list):
                     return json.dumps(val)
                 raise ValueError(
-                    f"Cannot convert value {val!r} with type" f"{value_type} to JSON"
+                    f"Cannot convert value {val!r} with type {value_type} to JSON"
                 )
 
             if isinstance(val, col_python_type):
                 return val
-            if col_python_type == float and isinstance(val, int):
+            if col_python_type is float and isinstance(val, int):
                 return float(val)
         except Exception as e:  # noqa: BLE001
             exc = e
@@ -335,6 +335,7 @@ class AbstractWarehouse(ABC, Serializable):
             return select_query
         if recursive:
             root = False
+            where = self.path_expr(dr).op("GLOB")(path)
             if not path or path == "/":
                 # root of the bucket, e.g s3://bucket/ -> getting all the nodes
                 # in the bucket
@@ -344,14 +345,18 @@ class AbstractWarehouse(ABC, Serializable):
                 # not a root and not a explicit glob, so it's pointing to some directory
                 # and we are adding a proper glob syntax for it
                 # e.g s3://bucket/dir1 -> s3://bucket/dir1/*
-                path = path.rstrip("/") + "/*"
+                dir_path = path.rstrip("/") + "/*"
+                where = where | self.path_expr(dr).op("GLOB")(dir_path)
 
             if not root:
                 # not a root, so running glob query
-                select_query = select_query.where(self.path_expr(dr).op("GLOB")(path))
+                select_query = select_query.where(where)
+
         else:
             parent = self.get_node_by_path(dr, path.lstrip("/").rstrip("/*"))
-            select_query = select_query.where(dr.c.parent == parent.path)
+            select_query = select_query.where(
+                (dr.c.parent == parent.path) | (self.path_expr(dr) == path)
+            )
         return select_query
 
     def rename_dataset_table(
@@ -493,7 +498,10 @@ class AbstractWarehouse(ABC, Serializable):
         This gets nodes based on the provided query, and should be used sparingly,
         as it will be slow on any OLAP database systems.
         """
-        return (Node(*row) for row in self.db.execute(query))
+        columns = [c.name for c in query.columns]
+        for row in self.db.execute(query):
+            d = dict(zip(columns, row))
+            yield Node(**d)
 
     def get_dirs_by_parent_path(
         self,
@@ -570,14 +578,12 @@ class AbstractWarehouse(ABC, Serializable):
         matched_paths: list[list[str]] = [[]]
         for curr_name in path_list[:-1]:
             if glob.has_magic(curr_name):
-                new_paths = []
+                new_paths: list[list[str]] = []
                 for path in matched_paths:
                     nodes = self._get_nodes_by_glob_path_pattern(
                         dataset_rows, path, curr_name
                     )
-                    for node in nodes:
-                        if node.is_container:
-                            new_paths.append([*path, node.name or ""])
+                    new_paths.extend([*path, n.name] for n in nodes if n.is_container)
                 matched_paths = new_paths
             else:
                 for path in matched_paths:
@@ -772,7 +778,7 @@ class AbstractWarehouse(ABC, Serializable):
         self,
         dataset_rows: "DataTable",
         parent_path: str,
-        fields: Optional[Iterable[str]] = None,
+        fields: Optional[Sequence[str]] = None,
         type: Optional[str] = None,
         conds=None,
         order_by: Optional[Union[str, list[str]]] = None,
@@ -794,9 +800,9 @@ class AbstractWarehouse(ABC, Serializable):
         else:
             conds.append(path != "")
 
-        if fields is None:
-            fields = [c.name for c in dr.file_columns()]
-        columns = [getattr(q.c, f) for f in fields]
+        columns = q.c
+        if fields:
+            columns = [getattr(columns, f) for f in fields]
 
         query = sa.select(*columns)
         query = query.where(*conds)
@@ -833,19 +839,16 @@ class AbstractWarehouse(ABC, Serializable):
 
         prefix_len = len(node.path)
 
-        def make_node_with_path(row):
-            sub_node = Node(*row)
-            return NodeWithPath(
-                sub_node, sub_node.path[prefix_len:].lstrip("/").split("/")
-            )
+        def make_node_with_path(node: Node) -> NodeWithPath:
+            return NodeWithPath(node, node.path[prefix_len:].lstrip("/").split("/"))
 
-        return map(make_node_with_path, self.db.execute(query))
+        return map(make_node_with_path, self.get_nodes(query))
 
     def find(
         self,
         dataset_rows: "DataTable",
         node: Node,
-        fields: Iterable[str],
+        fields: Sequence[str],
         type=None,
         conds=None,
         order_by=None,
@@ -890,11 +893,9 @@ class AbstractWarehouse(ABC, Serializable):
     def is_temp_table_name(self, name: str) -> bool:
         """Returns if the given table name refers to a temporary
         or no longer needed table."""
-        if name.startswith(
+        return name.startswith(
             (self.TMP_TABLE_NAME_PREFIX, self.UDF_TABLE_NAME_PREFIX, "ds_shadow_")
-        ) or name.endswith("_shadow"):
-            return True
-        return False
+        ) or name.endswith("_shadow")
 
     def get_temp_table_names(self) -> list[str]:
         return [

datachain/dataset.py

@@ -405,9 +405,7 @@ class DatasetRecord:
         Checks if a number can be a valid next latest version for dataset.
         The only rule is that it cannot be lower than current latest version
         """
-        if self.latest_version and self.latest_version >= version:
-            return False
-        return True
+        return not (self.latest_version and self.latest_version >= version)
 
     def get_version(self, version: int) -> DatasetVersion:
         if not self.has_version(version):

datachain/lib/arrow.py

@@ -0,0 +1,85 @@
+import re
+from typing import TYPE_CHECKING, Optional
+
+from pyarrow.dataset import dataset
+
+from datachain.lib.feature import Feature
+from datachain.lib.file import File
+
+if TYPE_CHECKING:
+    import pyarrow as pa
+
+
+class Source(Feature):
+    """File source info for tables."""
+
+    file: File
+    index: int
+
+
+class ArrowGenerator:
+    def __init__(self, schema: Optional["pa.Schema"] = None, **kwargs):
+        """
+        Generator for getting rows from tabular files.
+
+        Parameters:
+
+        schema : Optional pyarrow schema for validation.
+        kwargs: Parameters to pass to pyarrow.dataset.dataset.
+        """
+        self.schema = schema
+        self.kwargs = kwargs
+
+    def __call__(self, file: File):
+        path = file.get_path()
+        ds = dataset(path, filesystem=file.get_fs(), schema=self.schema, **self.kwargs)
+        index = 0
+        for record_batch in ds.to_batches():
+            for record in record_batch.to_pylist():
+                source = Source(file=file, index=index)
+                yield [source, *record.values()]
+                index += 1
+
+
+def schema_to_output(schema: "pa.Schema"):
+    """Generate UDF output schema from pyarrow schema."""
+    default_column = 0
+    output = {"source": Source}
+    for field in schema:
+        column = field.name.lower()
+        column = re.sub("[^0-9a-z_]+", "", column)
+        if not column:
+            column = f"c{default_column}"
+            default_column += 1
+        output[column] = _arrow_type_mapper(field.type)  # type: ignore[assignment]
+
+    return output
+
+
+def _arrow_type_mapper(col_type: "pa.DataType") -> type:  # noqa: PLR0911
+    """Convert pyarrow types to basic types."""
+    from datetime import datetime
+
+    import pyarrow as pa
+
+    if pa.types.is_timestamp(col_type):
+        return datetime
+    if pa.types.is_binary(col_type):
+        return bytes
+    if pa.types.is_floating(col_type):
+        return float
+    if pa.types.is_integer(col_type):
+        return int
+    if pa.types.is_boolean(col_type):
+        return bool
+    if pa.types.is_date(col_type):
+        return datetime
+    if pa.types.is_string(col_type) or pa.types.is_large_string(col_type):
+        return str
+    if pa.types.is_list(col_type):
+        return list[_arrow_type_mapper(col_type.value_type)]  # type: ignore[misc]
+    if pa.types.is_struct(col_type) or pa.types.is_map(col_type):
+        return dict
+    if isinstance(col_type, pa.lib.DictionaryType):
+        return _arrow_type_mapper(col_type.value_type)  # type: ignore[return-value]
+    raise TypeError(f"{col_type!r} datatypes not supported")