deriva-ml 1.13.1__py3-none-any.whl → 1.13.3__py3-none-any.whl

deriva_ml/dataset_aux_classes.py

@@ -2,8 +2,8 @@
 THis module defines the DataSet class with is used to manipulate n
 """
 
-from datetime import datetime
 from .deriva_definitions import RID
+
 from enum import Enum
 from pydantic import (
     BaseModel,
@@ -98,7 +98,7 @@ class DatasetHistory(BaseModel):
         version_rid (RID): The RID of the version record for the dataset in the Dataset_Version table.
         minid (str): The URL that represents the handle of the dataset bag.  This will be None if a MINID has not
                      been created yet.
-        timestamp (datetime): The timestamp of when the  dataset was created.
+        snapshot (str): Catalog snapshot ID of when the version record was created.
     """
 
     dataset_version: DatasetVersion
@@ -107,7 +107,7 @@ class DatasetHistory(BaseModel):
     execution_rid: Optional[RID] = None
     description: str = ""
     minid: Optional[str] = None
-    timestamp: Optional[datetime] = None
+    snapshot: Optional[str] = None
 
     model_config = ConfigDict(arbitrary_types_allowed=True)
 
@@ -128,12 +128,12 @@ class DatasetMinid(BaseModel):
     """
 
     dataset_version: DatasetVersion
-    metadata: dict[str, str | int]
-    minid: str = Field(alias="compact_uri")
+    metadata: dict[str, str | int] = {}
+    minid: str = Field(alias="compact_uri", default=None)
     bag_url: str = Field(alias="location")
-    identifier: str
-    landing_page: str
-    version_rid: RID = Field(alias="Dataset_RID")
+    identifier: Optional[str] = None
+    landing_page: Optional[str] = None
+    version_rid: RID = Field(alias="RID")
     checksum: str = Field(alias="checksums", default="")
 
     @computed_field
@@ -156,8 +156,8 @@ class DatasetMinid(BaseModel):
 
     @field_validator("bag_url", mode="before")
     @classmethod
-    def convert_location_to_str(cls, value: list[str]) -> str:
-        return value[0]
+    def convert_location_to_str(cls, value: list[str] | str) -> str:
+        return value[0] if isinstance(value, list) else value
 
     @field_validator("checksum", mode="before")
     @classmethod

deriva_ml/demo_catalog.py

@@ -1,18 +1,19 @@
 import atexit
-from importlib.metadata import version
 from importlib.resources import files
+import itertools
 import logging
 from random import randint, random
 from typing import Optional
-import itertools
+from tempfile import TemporaryDirectory
 
-from deriva.config.acl_config import AclConfig
-from deriva.core import DerivaServer
-from deriva.core import ErmrestCatalog, get_credential
+from deriva.core import DerivaServer, get_credential
+from deriva.core import ErmrestCatalog
 from deriva.core.datapath import DataPathException
 from deriva.core.ermrest_model import builtin_types, Schema, Table, Column
 from requests import HTTPError
+import subprocess
 
+from .schema_setup.annotations import catalog_annotation
 from deriva_ml import (
     DerivaML,
     ExecutionConfiguration,
@@ -23,8 +24,10 @@ from deriva_ml import (
     RID,
 )
 
-from deriva_ml.schema_setup.create_schema import initialize_ml_schema, create_ml_schema
-from deriva_ml.dataset import Dataset
+from deriva_ml.schema_setup.create_schema import (
+    initialize_ml_schema,
+    create_ml_schema,
+)
 
 TEST_DATASET_SIZE = 4
 
@@ -85,7 +88,7 @@ def create_demo_datasets(ml_instance: DerivaML) -> tuple[RID, list[RID], list[RI
 
     type_rid = ml_instance.add_term("Dataset_Type", "TestSet", description="A test")
     training_rid = ml_instance.add_term(
-        "Dataset_Type", "Training", description="A traing set"
+        "Dataset_Type", "Training", description="A training set"
     )
     testing_rid = ml_instance.add_term(
         "Dataset_Type", "Testing", description="A testing set"
@@ -98,32 +101,46 @@ def create_demo_datasets(ml_instance: DerivaML) -> tuple[RID, list[RID], list[RI
     )
     subject_rids = [i["RID"] for i in table_path.entities().fetch()]
 
-    dataset_rids = []
-    for r in subject_rids[0:4]:
-        d = ml_instance.create_dataset(
-            type=[type_rid.name, "Testing"],
-            description=f"Dataset {r}",
-            version=DatasetVersion(1, 0, 0),
-        )
-        ml_instance.add_dataset_members(d, [r])
-        dataset_rids.append(d)
-
-    nested_datasets = []
-    for i in range(0, 4, 2):
-        nested_dataset = ml_instance.create_dataset(
-            type=[type_rid.name, "Training"],
-            description=f"Nested Dataset {i}",
-            version=DatasetVersion(1, 0, 0),
-        )
-        ml_instance.add_dataset_members(nested_dataset, dataset_rids[i : i + 2])
-        nested_datasets.append(nested_dataset)
+    ml_instance.add_term(
+        MLVocab.workflow_type,
+        "Create Dataset Workflow",
+        description="A Workflow that creates a new dataset.",
+    )
+    dataset_workflow = ml_instance.create_workflow(
+        name="API Workflow", workflow_type="Create Dataset Workflow"
+    )
 
-    double_nested_dataset = ml_instance.create_dataset(
-        type_rid.name,
-        description="Double nested dataset",
-        version=DatasetVersion(1, 0, 0),
+    dataset_execution = ml_instance.create_execution(
+        ExecutionConfiguration(workflow=dataset_workflow, description="Create Dataset")
     )
-    ml_instance.add_dataset_members(double_nested_dataset, nested_datasets)
+
+    with dataset_execution.execute() as exe:
+        dataset_rids = []
+        for r in subject_rids[0:4]:
+            d = exe.create_dataset(
+                dataset_types=[type_rid.name, "Testing"],
+                description=f"Dataset {r}",
+                version=DatasetVersion(1, 0, 0),
+            )
+            ml_instance.add_dataset_members(d, [r])
+            dataset_rids.append(d)
+
+        nested_datasets = []
+        for i in range(0, 4, 2):
+            nested_dataset = exe.create_dataset(
+                dataset_types=[type_rid.name, "Training"],
+                description=f"Nested Dataset {i}",
+                version=DatasetVersion(1, 0, 0),
+            )
+            exe.add_dataset_members(nested_dataset, dataset_rids[i : i + 2])
+            nested_datasets.append(nested_dataset)
+
+        double_nested_dataset = exe.create_dataset(
+            dataset_types=type_rid.name,
+            description="Double nested dataset",
+            version=DatasetVersion(1, 0, 0),
+        )
+        exe.add_dataset_members(double_nested_dataset, nested_datasets)
     return double_nested_dataset, nested_datasets, dataset_rids
 
 
@@ -251,14 +268,13 @@ def create_domain_schema(ml_instance: DerivaML, sname: str) -> None:
     :return:
     """
 
-    # Make sure that we have a ml schema
     _ = ml_instance.model.schemas["deriva-ml"]
 
     if ml_instance.model.schemas.get(sname):
         # Clean out any old junk....
         ml_instance.model.schemas[sname].drop()
 
-    domain_schema = ml_instance.model.model.create_schema(
+    domain_schema = ml_instance.model.create_schema(
         Schema.define(sname, annotations={"name_style": {"underline_space": True}})
     )
     subject_table = domain_schema.create_table(
@@ -266,6 +282,8 @@ def create_domain_schema(ml_instance: DerivaML, sname: str) -> None:
     )
     ml_instance.create_asset("Image", referenced_tables=[subject_table])
 
+    catalog_annotation(ml_instance.model)
+
 
 def destroy_demo_catalog(catalog):
     catalog.delete_ermrest_catalog(really=True)
@@ -280,43 +298,47 @@ def create_demo_catalog(
     create_datasets=False,
     on_exit_delete=True,
 ) -> ErmrestCatalog:
-    credentials = get_credential(hostname)
-    server = DerivaServer("https", hostname, credentials=credentials)
+    credential = get_credential(hostname)
+
+    server = DerivaServer("https", hostname, credentials=credential)
     test_catalog = server.create_ermrest_catalog()
+    model = test_catalog.getCatalogModel()
+    model.configure_baseline_catalog()
+    policy_file = files("deriva_ml.schema_setup").joinpath("policy.json")
+    subprocess.run(
+        [
+            "deriva-acl-config",
+            "--host",
+            test_catalog.deriva_server.server,
+            "--config-file",
+            policy_file,
+            test_catalog.catalog_id,
+        ]
+    )
+
     if on_exit_delete:
         atexit.register(destroy_demo_catalog, test_catalog)
-    model = test_catalog.getCatalogModel()
 
     try:
-        create_ml_schema(model, project_name=project_name)
-        deriva_ml = DerivaML(
-            hostname=hostname,
-            catalog_id=test_catalog.catalog_id,
-            project_name=project_name,
-            domain_schema=domain_schema,
-            logging_level=logging.WARN,
-        )
-        create_domain_schema(deriva_ml, domain_schema)
-        working_dir = deriva_ml.working_dir
-        dataset_table = deriva_ml.dataset_table
-        dataset_table.annotations.update(
-            Dataset(
-                deriva_ml.model,
-                cache_dir=deriva_ml.cache_dir,
-                working_dir=deriva_ml.working_dir,
-            )._generate_dataset_annotations()
-        )
-        deriva_ml.model.apply()
-        policy_file = files("deriva_ml.schema_setup").joinpath("policy.json")
-        AclConfig(
-            hostname, test_catalog.catalog_id, policy_file, credentials=credentials
-        )
-        if populate or create_features or create_datasets:
-            populate_demo_catalog(deriva_ml, domain_schema)
-            if create_features:
-                create_demo_features(deriva_ml)
-            if create_datasets:
-                create_demo_datasets(deriva_ml)
+        with TemporaryDirectory() as tmpdir:
+            create_ml_schema(test_catalog, project_name=project_name)
+            deriva_ml = DerivaML(
+                hostname=hostname,
+                catalog_id=test_catalog.catalog_id,
+                project_name=project_name,
+                domain_schema=domain_schema,
+                logging_level=logging.WARN,
+                working_dir=tmpdir,
+                credential=credential,
+            )
+            create_domain_schema(deriva_ml, domain_schema)
+
+            if populate or create_features or create_datasets:
+                populate_demo_catalog(deriva_ml, domain_schema)
+                if create_features:
+                    create_demo_features(deriva_ml)
+                if create_datasets:
+                    create_demo_datasets(deriva_ml)
 
     except Exception:
         # on failure, delete catalog and re-raise exception
@@ -332,6 +354,7 @@ class DemoML(DerivaML):
         catalog_id,
         cache_dir: Optional[str] = None,
         working_dir: Optional[str] = None,
+        use_minid=True,
     ):
         super().__init__(
             hostname=hostname,
@@ -339,5 +362,5 @@ class DemoML(DerivaML):
             project_name="ml-test",
             cache_dir=cache_dir,
             working_dir=working_dir,
-            model_version=version(__name__.split(".")[0]),
+            use_minid=use_minid,
         )

deriva_ml/deriva_definitions.py

@@ -2,12 +2,16 @@
 Shared definitions that are used in different DerivaML modules.
 """
 
+from __future__ import annotations
+
 import warnings
 from datetime import date
 from enum import Enum
-from typing import Any, Iterable, Optional, Annotated
+from pathlib import Path
+from typing import Any, Iterable, Optional, Annotated, Generator
 
 import deriva.core.ermrest_model as em
+import deriva.core.utils.hash_utils as hash_utils
 from urllib.parse import urlparse
 from deriva.core.ermrest_model import builtin_types
 from pydantic import (
@@ -136,11 +140,14 @@ class FileSpec(BaseModel):
     @field_validator("url")
     @classmethod
     def validate_file_url(cls, v):
+        """Examine the provided URL.  If it's a local path, convert it into a tag URL."""
         url_parts = urlparse(v)
         if url_parts.scheme == "tag":
+            # Already a tag URL, so just return it.
             return v
-        elif not url_parts.scheme:
-            return f"tag://{gethostname()},{date.today()}:file://{v}"
+        elif (not url_parts.scheme) or url_parts.scheme == "file":
+            # There is no scheme part tof the URL, or it is a file URL, so it is a local file path, so convert to a tag URL.
+            return f"tag://{gethostname()},{date.today()}:file://{url_parts.path}"
         else:
             raise ValidationError("url is not a file URL")
 
@@ -153,6 +160,38 @@ class FileSpec(BaseModel):
             "Length": self.length,
         }
 
+    @staticmethod
+    def create_filespecs(
+        path: Path | str, description: str
+    ) -> Generator["FileSpec", None, None]:
+        """Given a file or directory, generate the sequence of corresponding FileSpecs sutable to create a File table
+
+        Arguments:
+            path: Path to the file or directory.
+            description: The description of the file(s)
+
+        Returns:
+            An iterable of FileSpecs for each file in the directory.
+        """
+        path = Path(path)
+
+        def list_all_files(p) -> list[Path]:
+            return (
+                (f for f in Path(p).rglob("*") if f.is_file()) if path.is_dir() else [p]
+            )
+
+        def create_spec(p: Path, description: str) -> FileSpec:
+            hashes = hash_utils.compute_file_hashes(p, hashes=["md5", "sha256"])
+            md5 = hashes["md5"][0]
+            return FileSpec(
+                length=path.stat().st_size,
+                md5=md5,
+                description=description,
+                url=p.as_posix(),
+            )
+
+        return (create_spec(file, description) for file in list_all_files(path))
+
 
 class VocabularyTerm(BaseModel):
     """An entry in a vocabulary table.
@@ -162,7 +201,7 @@ class VocabularyTerm(BaseModel):
        synonyms: List of alternative names for the term
        id: CURI identifier for the term
        uri: Unique URI for the term.
-       description: A description of the meaning of the term
+       description: A description of the term meaning
        rid: Resource identifier assigned to the term
 
     Args:

deriva_ml/deriva_ml_base.py

@@ -51,6 +51,7 @@ from .deriva_definitions import (
     FileSpec,
     TableDefinition,
 )
+from .schema_setup.annotations import asset_annotation
 
 try:
     from icecream import ic
@@ -82,9 +83,10 @@ class DerivaML(Dataset):
         project_name: Optional[str] = None,
         cache_dir: Optional[str] = None,
         working_dir: Optional[str] = None,
-        model_version: str = "1",
         ml_schema: str = ML_SCHEMA,
         logging_level=logging.INFO,
+        credential=None,
+        use_minid=True,
     ):
         """Create and initialize a DerivaML instance.
 
@@ -93,13 +95,14 @@ class DerivaML(Dataset):
 
         Args:
             hostname: Hostname of the Deriva server.
-            catalog_id: Catalog ID. Either and identifier, or a catalog name.
-            domain_schema: Schema name for domain specific tables and relationships.
+            catalog_id: Catalog ID. Either an identifier or a catalog name.
+            domain_schema: Schema name for domain-specific tables and relationships.
+            project_name: Project name. Defaults to name of domain schema.
             cache_dir: Directory path for caching data downloaded from the Deriva server as bdbag.
             working_dir: Directory path for storing data used by or generated by any computations.
-            model_version: A string that indicates the version model.  Typically passed in via
+            use_minid: Use the MINID serice when downloading dataset bags.
         """
-        self.credential = get_credential(hostname)
+        self.credential = credential or get_credential(hostname)
         server = DerivaServer(
             "https",
             hostname,
@@ -119,21 +122,20 @@ class DerivaML(Dataset):
         ) / default_workdir
 
         self.working_dir.mkdir(parents=True, exist_ok=True)
-        self.cache_dir = (
-            Path(cache_dir) if cache_dir else Path.home() / "deriva-ml" / "cache"
-        )
+        self.cache_dir = Path(cache_dir) if cache_dir else self.working_dir / "cache"
 
         self.cache_dir.mkdir(parents=True, exist_ok=True)
 
         # Initialize dataset class.
-        super().__init__(self.model, self.cache_dir, self.working_dir)
+        super().__init__(
+            self.model, self.cache_dir, self.working_dir, use_minid=use_minid
+        )
         self._logger = logging.getLogger("deriva_ml")
         self._logger.setLevel(logging_level)
 
         self.host_name = hostname
         self.catalog_id = catalog_id
         self.ml_schema = ml_schema
-        self.version = model_version
         self.configuration = None
         self._execution: Optional[Execution] = None
         self.domain_schema = self.model.domain_schema
@@ -150,11 +152,6 @@ class DerivaML(Dataset):
         deriva_logger = logging.getLogger("deriva")
         deriva_logger.setLevel(logging_level)
 
-        if "dirty" in self.version:
-            logging.info(
-                f"Loading dirty model.  Consider commiting and tagging: {self.version}"
-            )
-
     def __del__(self):
         try:
             if self._execution and self._execution.status != Status.completed:
@@ -438,6 +435,8 @@ class DerivaML(Dataset):
             )
         )
         atable.create_reference(self.model.name_to_table("Asset_Role"))
+
+        asset_annotation(asset_table)
         return asset_table
 
     # @validate_call(config=ConfigDict(arbitrary_types_allowed=True))
@@ -820,6 +819,8 @@ class DerivaML(Dataset):
     ) -> Iterable[RID]:
         """Add a new file to the File table in the catalog.
 
+        The input is an iterator of FileSpec objects which provide the MD5 checksum, length, and URL.
+
         Args:
             file_types: One or more file types.  Must be a term from the File_Type controlled vocabulary.
             files: A sequence of file specifications that describe the files to add.
@@ -841,7 +842,6 @@ class DerivaML(Dataset):
                     return True
             return False
 
-        # Create the entry for the new dataset_table and get its RID.
         file_types = [file_types] if isinstance(file_types, str) else file_types
         pb = self._model.catalog.getPathBuilder()
         for file_type in file_types:
@@ -868,18 +868,12 @@ class DerivaML(Dataset):
 
         if execution_rid:
             # Get the name of the association table between file_table and execution.
-            exec_table = next(
-                self._model.schemas[self._ml_schema]
-                .tables["Execution"]
-                .find_associations()
-            ).name
-            pb.schemas[self._ml_schema].tables[exec_table].insert(
+            pb.schemas[self._ml_schema].File_Execution.insert(
                 [
                     {"File": file_rid, "Execution": execution_rid}
                     for file_rid in file_rids
                 ]
             )
-
         return file_rids
 
     def list_files(
@@ -890,9 +884,10 @@ class DerivaML(Dataset):
         file_path = ml_path.File
         type_path = ml_path.File_File_Type
 
-        # Get a list of all the dataset_type values associated with this dataset_table.
-        path = file_path.link(type_path)
-        path = path.attributes(
+        path = file_path.link(
+            type_path, on=file_path.RID == type_path.File, join_type="left"
+        )
+        path = path.File.attributes(
             path.File.RID,
             path.File.URL,
             path.File.MD5,
@@ -902,9 +897,9 @@ class DerivaML(Dataset):
         )
         file_map = {}
         for f in path.fetch():
-            file_map.setdefault(f["RID"], f | {"File_Types": []})["File_Types"].append(
-                f["File_Type"]
-            )
+            entry = file_map.setdefault(f["RID"], {**f, "File_Types": []})
+            if ft := f.get("File_Type"):  # assign-and-test in one go
+                entry["File_Types"].append(ft)
 
         # Now get rid of the File_Type key and return the result
         return [(f, f.pop("File_Type"))[0] for f in file_map.values()]
@@ -974,7 +969,7 @@ class DerivaML(Dataset):
     ) -> Workflow:
         """Identify current executing program and return a workflow RID for it
 
-        Determine the notebook or script that is currently being executed. Assume that  this is
+        Determine the notebook or script that is currently being executed. Assume that this is
         being executed from a cloned GitHub repository.  Determine the remote repository name for
         this object.  Then either retrieve an existing workflow for this executable or create
         a new one.
@@ -983,6 +978,9 @@ class DerivaML(Dataset):
             name: The name of the workflow.
             workflow_type: The type of the workflow.
             description: The description of the workflow.
+
+        Returns:
+            A workflow object.
         """
         # Make sure type is correct.
         self.lookup_term(MLVocab.workflow_type, workflow_type)
@@ -1001,6 +999,9 @@ class DerivaML(Dataset):
         1. The datasets specified in the configuration are downloaded and placed in the cache-dir. If a version is
         not specified in the configuration, then a new minor version number is created for the dataset and downloaded.
 
+        2. If any execution assets are provided in the configuration, they are downloaded and placed in the working directory.
+
+
         Args:
             configuration: ExecutionConfiguration:
             dry_run: Do not create an execution record or upload results.

deriva_ml/deriva_model.py

@@ -21,7 +21,7 @@ from .deriva_definitions import (
 
 from collections import Counter
 from pydantic import validate_call, ConfigDict
-from typing import Iterable, Optional
+from typing import Iterable, Optional, Any
 
 
 class DerivaModel:
@@ -61,7 +61,7 @@ class DerivaModel:
         self.schemas = self.model.schemas
 
         self.ml_schema = ml_schema
-        builtin_schemas = ["public", self.ml_schema, "www"]
+        builtin_schemas = ["public", self.ml_schema, "www", "WWW"]
         try:
             self.domain_schema = (
                 domain_schema
@@ -73,6 +73,11 @@ class DerivaModel:
             # No domain schema defined.
             self.domain_schema = domain_schema
 
+    @property
+    def chaise_config(self) -> dict[str, Any]:
+        """Return the chaise configuration."""
+        return self.model.chaise_config
+
     def __getattr__(self, name):
         # Called only if `name` is not found in Manager.  Delegate attributes to model class.
         return getattr(self.model, name)
@@ -115,7 +120,12 @@ class DerivaModel:
         return vocab_columns.issubset({c.name.upper() for c in table.columns})
 
     def is_association(
-        self, table_name: str | Table, unqualified: bool = True, pure: bool = True
+        self,
+        table_name: str | Table,
+        unqualified: bool = True,
+        pure: bool = True,
+        min_arity: int = 2,
+        max_arity: int = 2,
     ) -> bool | set | int:
         """Check the specified table to see if it is an association table.
 
@@ -130,7 +140,9 @@ class DerivaModel:
 
         """
         table = self.name_to_table(table_name)
-        return table.is_association(unqualified=unqualified, pure=pure)
+        return table.is_association(
+            unqualified=unqualified, pure=pure, min_arity=min_arity, max_arity=max_arity
+        )
 
     def find_association(self, table1: Table | str, table2: Table | str) -> Table:
         """Given two tables, return an association table that connects the two.
@@ -302,7 +314,7 @@ class DerivaModel:
     ) -> list[list[Table]]:
         """Recursively walk over the domain schema graph and extend the current path.
 
-        Walk a schema graph and return a list all  the paths through the graph.
+        Walk a schema graph and return a list all the paths through the graph.
 
         Args:
             path: Source path so far