datamasque-python 1.1.7__tar.gz → 1.1.8__tar.gz

{datamasque_python-1.1.7 → datamasque_python-1.1.8}/HISTORY.rst

@@ -2,6 +2,18 @@
 History
 =======
 
+1.1.8 (2026-07-27)
+------------------
+
+* Added typed Safe Data Preview support:
+
+  * ``safe_data_preview`` on ``InDataDiscoveryConfig`` (via ``SafeDataPreviewOptions``) to configure or disable it.
+  * ``safe_data_preview`` on schema-discovery result columns and file-discovery locators, typed by ``kind``.
+
+* Added ``row_count`` to ``TableConstraints`` in schema-discovery ``table_metadata``.
+* Added table-reference management APIs (``list_table_references``, ``create_table_reference``, and friends),
+  along with the ``TableReference`` and ``TableReferenceOptions`` models.
+
 1.1.7 (2026-07-14)
 ------------------
 

{datamasque_python-1.1.7 → datamasque_python-1.1.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datamasque-python
-Version: 1.1.7
+Version: 1.1.8
 Summary: Official Python client for the DataMasque data-masking API.
 Project-URL: Homepage, https://datamasque.com/
 Project-URL: Documentation, https://datamasque-python.readthedocs.io/

{datamasque_python-1.1.7 → datamasque_python-1.1.8}/datamasque/client/__init__.py

@@ -116,6 +116,38 @@ from datamasque.client.models.runs import (
     RunInfo,
     UnfinishedRun,
 )
+from datamasque.client.models.safe_data_preview import (
+    BooleanPreview,
+    BooleanStatistics,
+    ColumnKind,
+    ColumnPreview,
+    CommonStatistics,
+    FirstCharsStatistics,
+    LengthEntry,
+    LengthsStatistics,
+    MaskedFormEntry,
+    NumericBin,
+    NumericHistograms,
+    NumericPreview,
+    NumericStatistics,
+    NumericSummaries,
+    NumericTemporalDisclosureLevel,
+    PatternComposition,
+    PatternEntry,
+    PatternsStatistics,
+    SafeDataPreview,
+    SafeDataPreviewOptions,
+    StringDisclosureLevel,
+    StringPreview,
+    StringStatistics,
+    TemporalBin,
+    TemporalHistograms,
+    TemporalPreview,
+    TemporalStatistics,
+    TemporalSummaries,
+    UnsupportedPreview,
+    UnsupportedStatistics,
+)
 from datamasque.client.models.status import (
     AsyncRulesetGenerationTaskStatus,
     MaskingRunStatus,
@@ -123,6 +155,12 @@ from datamasque.client.models.status import (
     ValidationErrorType,
     ValidationStatus,
 )
+from datamasque.client.models.table_reference import (
+    TableReference,
+    TableReferenceFormat,
+    TableReferenceId,
+    TableReferenceOptions,
+)
 from datamasque.client.models.user import User, UserId, UserRole
 
 __version__ = version("datamasque-python")
@@ -131,6 +169,11 @@ __all__ = [
     "AsyncRulesetGenerationInProgressError",
     "AsyncRulesetGenerationTaskStatus",
     "AzureConnectionConfig",
+    "BooleanPreview",
+    "BooleanStatistics",
+    "ColumnKind",
+    "ColumnPreview",
+    "CommonStatistics",
     "ConnectionConfig",
     "ConnectionId",
     "ConstraintColumns",
@@ -171,6 +214,7 @@ __all__ = [
     "FileId",
     "FileOrContent",
     "FileRulesetGenerationRequest",
+    "FirstCharsStatistics",
     "ForeignKeyRef",
     "GitSnapshot",
     "HashColumnsTableConfig",
@@ -186,16 +230,28 @@ __all__ = [
     "InvalidLibraryError",
     "InvalidRulesetError",
     "JsonPath",
+    "LengthEntry",
+    "LengthsStatistics",
     "LicenseInfo",
     "Locator",
     "MaskType",
+    "MaskedFormEntry",
     "MaskingRunOptions",
     "MaskingRunRequest",
     "MaskingRunStatus",
     "MongoConnectionConfig",
     "MountedShareConnectionConfig",
     "MssqlLinkedServerConnectionConfig",
+    "NumericBin",
+    "NumericHistograms",
+    "NumericPreview",
+    "NumericStatistics",
+    "NumericSummaries",
+    "NumericTemporalDisclosureLevel",
     "OracleWalletFile",
+    "PatternComposition",
+    "PatternEntry",
+    "PatternsStatistics",
     "ReferencingForeignKey",
     "Ruleset",
     "RulesetGenerationRequest",
@@ -213,6 +269,8 @@ __all__ = [
     "RunInfo",
     "RunNotCancellableError",
     "S3ConnectionConfig",
+    "SafeDataPreview",
+    "SafeDataPreviewOptions",
     "SchemaDiscoveryColumn",
     "SchemaDiscoveryFromConfigRequest",
     "SchemaDiscoveryPage",
@@ -228,9 +286,23 @@ __all__ = [
     "SseConfig",
     "SseSelection",
     "SslZipFile",
+    "StringDisclosureLevel",
+    "StringPreview",
+    "StringStatistics",
     "SwitchableLicenseMetadata",
     "TableConstraints",
+    "TableReference",
+    "TableReferenceFormat",
+    "TableReferenceId",
+    "TableReferenceOptions",
+    "TemporalBin",
+    "TemporalHistograms",
+    "TemporalPreview",
+    "TemporalStatistics",
+    "TemporalSummaries",
     "UnfinishedRun",
+    "UnsupportedPreview",
+    "UnsupportedStatistics",
     "User",
     "UserId",
     "UserRole",

{datamasque_python-1.1.7 → datamasque_python-1.1.8}/datamasque/client/dmclient.py

@@ -9,6 +9,7 @@ from datamasque.client.ruleset_libraries import RulesetLibraryClient
 from datamasque.client.rulesets import RulesetClient
 from datamasque.client.runs import RunClient
 from datamasque.client.settings import SettingsClient
+from datamasque.client.table_references import TableReferenceClient
 from datamasque.client.users import UserClient
 
 __all__ = ["DataMasqueClient", "FileOrContent", "UploadFile"]
@@ -24,6 +25,7 @@ class DataMasqueClient(
     DiscoveryClient,
     DiscoveryConfigClient,
     DiscoveryConfigLibraryClient,
+    TableReferenceClient,
     UserClient,
     SettingsClient,
 ):

{datamasque_python-1.1.7 → datamasque_python-1.1.8}/datamasque/client/models/discovery.py

@@ -10,6 +10,10 @@ from datamasque.client.models.data_selection import HashColumnsTableConfig, Loca
 from datamasque.client.models.discovery_config import DiscoveryConfig, DiscoveryConfigId, unwrap_discovery_config_id
 from datamasque.client.models.pagination import Page
 from datamasque.client.models.runs import RunConnectionRef
+from datamasque.client.models.safe_data_preview import (
+    SafeDataPreview,
+    SafeDataPreviewOptions,
+)
 
 
 class InDataDiscoveryRule(BaseModel):
@@ -32,6 +36,7 @@ class InDataDiscoveryConfig(BaseModel):
     non_sensitive_rules: Optional[list[InDataDiscoveryRule]] = None
     ignore_rules: Optional[list[InDataDiscoveryRule]] = None
     force: Optional[bool] = None
+    safe_data_preview: Optional[SafeDataPreviewOptions] = None
 
 
 class SchemaDiscoveryRequest(BaseModel):
@@ -294,6 +299,7 @@ class SchemaDiscoveryColumn(BaseModel):
     unique_index_names: list[str]
     referencing_foreign_keys: list[ReferencingForeignKey]
     constraint: str  # Primary or Unique, or empty string if column does not participate in a PK/UK
+    safe_data_preview: Optional[SafeDataPreview] = None
 
 
 class SchemaDiscoveryResult(BaseModel):
@@ -324,6 +330,7 @@ class TableConstraints(BaseModel):
     primary_keys: Optional[list[ConstraintColumns]] = None
     unique_keys: Optional[list[ConstraintColumns]] = None
     foreign_keys: Optional[list[ConstraintColumns]] = None
+    row_count: Optional[int] = None
 
 
 class SchemaDiscoveryPage(Page[SchemaDiscoveryResult]):
@@ -356,6 +363,7 @@ class FileDiscoveryLocatorResult(BaseModel):
     locator: Locator
     matches: list[FileDiscoveryMatch]
     data_types: list[str]
+    safe_data_preview: Optional[SafeDataPreview] = None
 
 
 class FileDiscoveryFile(BaseModel):

datamasque_python-1.1.8/datamasque/client/models/safe_data_preview.py

@@ -0,0 +1,282 @@
+"""Typed request and response shapes for Safe Data Preview, part of In-Data Discovery."""
+
+from enum import Enum
+from typing import Annotated, Literal, Optional, Union
+
+from pydantic import (
+    BaseModel,
+    ConfigDict,
+    Field,
+)
+
+
+class StringDisclosureLevel(Enum):
+    """How much detail a string Safe Data Preview reveals."""
+
+    lengths = "lengths"
+    patterns = "patterns"
+    first_chars = "first_chars"
+
+
+class NumericTemporalDisclosureLevel(Enum):
+    """How much detail a numeric or temporal Safe Data Preview reveals."""
+
+    summaries = "summaries"
+    histograms = "histograms"
+
+
+class SafeDataPreviewOptions(BaseModel):
+    """Configures whether Safe Data Preview runs and how much detail each column reveals."""
+
+    model_config = ConfigDict(extra="forbid")
+
+    enabled: Optional[bool] = None
+    min_group_size: Optional[int] = None
+    string_level: Optional[StringDisclosureLevel] = None
+    numeric_level: Optional[NumericTemporalDisclosureLevel] = None
+    temporal_level: Optional[NumericTemporalDisclosureLevel] = None
+
+
+class ColumnKind(str, Enum):
+    """Which shape a column's Safe Data Preview takes."""
+
+    boolean = "boolean"
+    numeric = "numeric"
+    string = "string"
+    temporal = "temporal"
+    unsupported = "unsupported"
+
+
+class CommonStatistics(BaseModel):
+    """Row counts shared by every preview kind."""
+
+    model_config = ConfigDict(extra="allow")
+
+    count_row: int
+    count_null: int
+    count_distinct: Optional[int] = None
+
+
+class LengthEntry(BaseModel):
+    """A value length and how many sampled rows had it."""
+
+    model_config = ConfigDict(extra="allow")
+
+    length: int
+    count: int
+
+
+class LengthsStatistics(BaseModel):
+    """Distribution of value lengths."""
+
+    model_config = ConfigDict(extra="allow")
+
+    min: int
+    max: int
+    mean: float
+    median: float
+    most_common: list[LengthEntry]
+
+
+class PatternEntry(BaseModel):
+    """A masked character pattern and how many sampled rows matched it."""
+
+    model_config = ConfigDict(extra="allow")
+
+    pattern: str
+    count: int
+
+
+class PatternComposition(BaseModel):
+    """The proportion of letter, digit, and other characters across sampled values."""
+
+    model_config = ConfigDict(extra="allow")
+
+    letter: float
+    digit: float
+    other: float
+
+
+class PatternsStatistics(BaseModel):
+    """The most common character patterns and the overall character composition."""
+
+    model_config = ConfigDict(extra="allow")
+
+    top: list[PatternEntry]
+    composition: PatternComposition
+
+
+class MaskedFormEntry(BaseModel):
+    """A masked leading form of a value and how many sampled rows had it."""
+
+    model_config = ConfigDict(extra="allow")
+
+    masked: str
+    count: int
+
+
+class FirstCharsStatistics(BaseModel):
+    """The most common masked leading forms."""
+
+    model_config = ConfigDict(extra="allow")
+
+    top: list[MaskedFormEntry]
+
+
+class StringStatistics(BaseModel):
+    """String preview statistics; `patterns`/`first_chars` appear at or above certain disclosure levels."""
+
+    model_config = ConfigDict(extra="allow")
+
+    lengths: LengthsStatistics
+    patterns: Optional[PatternsStatistics] = None
+    first_chars: Optional[FirstCharsStatistics] = None
+
+
+class NumericSummaries(BaseModel):
+    """Numeric mean and percentile summary."""
+
+    model_config = ConfigDict(extra="allow")
+
+    mean: float
+    q1: float
+    q2: float
+    q3: float
+    p5: float
+    p95: float
+
+
+class NumericBin(BaseModel):
+    """One half-open interval `[lower_bound, upper_bound)` of a numeric histogram and its row count."""
+
+    model_config = ConfigDict(extra="allow")
+
+    lower_bound: float
+    upper_bound: float
+    count: int
+
+
+class NumericHistograms(BaseModel):
+    """The bins of a numeric histogram."""
+
+    model_config = ConfigDict(extra="allow")
+
+    bins: list[NumericBin]
+
+
+class NumericStatistics(BaseModel):
+    """Numeric preview statistics; `histograms` is present only at the histogram disclosure level."""
+
+    model_config = ConfigDict(extra="allow")
+
+    summaries: NumericSummaries
+    histograms: Optional[NumericHistograms] = None
+
+
+class TemporalSummaries(BaseModel):
+    """Temporal mean and percentile summary, each an ISO-formatted string."""
+
+    model_config = ConfigDict(extra="allow")
+
+    mean: str
+    q1: str
+    q2: str
+    q3: str
+    p5: str
+    p95: str
+
+
+class TemporalBin(BaseModel):
+    """One interval of a temporal histogram, with ISO-formatted bounds and a human-readable label."""
+
+    model_config = ConfigDict(extra="allow")
+
+    lower_bound: str
+    upper_bound: str
+    label: str
+    count: int
+
+
+class TemporalHistograms(BaseModel):
+    """The bins of a temporal histogram."""
+
+    model_config = ConfigDict(extra="allow")
+
+    bins: list[TemporalBin]
+
+
+class TemporalStatistics(BaseModel):
+    """Temporal preview statistics; `histograms` is present only at the histogram disclosure level."""
+
+    model_config = ConfigDict(extra="allow")
+
+    summaries: TemporalSummaries
+    histograms: Optional[TemporalHistograms] = None
+
+
+class BooleanStatistics(BaseModel):
+    """Counts of true and false values."""
+
+    model_config = ConfigDict(extra="allow")
+
+    count_true: int
+    count_false: int
+
+
+class UnsupportedStatistics(BaseModel):
+    """Why a column could not be previewed."""
+
+    model_config = ConfigDict(extra="allow")
+
+    reason: str
+
+
+class ColumnPreview(BaseModel):
+    """Fields present on every Safe Data Preview."""
+
+    model_config = ConfigDict(extra="allow")
+
+    # For grouped file discovery, the file whose sample produced this preview; None for database columns.
+    sampled_from: Optional[str] = None
+    statistics_common: CommonStatistics
+
+
+class StringPreview(ColumnPreview):
+    """A preview of a string column."""
+
+    kind: Literal[ColumnKind.string] = ColumnKind.string
+    statistics_kind: StringStatistics
+
+
+class NumericPreview(ColumnPreview):
+    """A preview of a numeric column."""
+
+    kind: Literal[ColumnKind.numeric] = ColumnKind.numeric
+    statistics_kind: NumericStatistics
+
+
+class TemporalPreview(ColumnPreview):
+    """A preview of a date/time column."""
+
+    kind: Literal[ColumnKind.temporal] = ColumnKind.temporal
+    statistics_kind: TemporalStatistics
+
+
+class BooleanPreview(ColumnPreview):
+    """A preview of a boolean column."""
+
+    kind: Literal[ColumnKind.boolean] = ColumnKind.boolean
+    statistics_kind: BooleanStatistics
+
+
+class UnsupportedPreview(ColumnPreview):
+    """A column the preview ran against but declined; `statistics_kind.reason` says why."""
+
+    kind: Literal[ColumnKind.unsupported] = ColumnKind.unsupported
+    statistics_kind: UnsupportedStatistics
+
+
+SafeDataPreview = Annotated[
+    Union[StringPreview, NumericPreview, TemporalPreview, BooleanPreview, UnsupportedPreview],
+    Field(discriminator="kind"),
+]

datamasque_python-1.1.8/datamasque/client/models/table_reference.py

@@ -0,0 +1,101 @@
+"""Table reference models for the DataMasque API."""
+
+import enum
+from datetime import datetime
+from typing import NewType, Optional, Union
+
+from pydantic import BaseModel, ConfigDict, Field, field_validator
+
+from datamasque.client.models.connection import ConnectionConfig, ConnectionId, unwrap_connection_id
+
+TableReferenceId = NewType("TableReferenceId", str)
+
+
+class TableReferenceFormat(enum.Enum):
+    """
+    File format of the data a table reference points at.
+
+    Only meaningful for a table reference on a file connection,
+    and an explicit choice: the source path's suffix carries no meaning.
+    """
+
+    csv = "csv"
+    parquet = "parquet"
+
+
+class TableReferenceOptions(BaseModel):
+    """
+    Format and CSV parsing options for a table reference.
+
+    These apply to file connections only —
+    they are ignored for a table reference on a database connection,
+    and the CSV options are ignored when the `format` is `parquet`.
+
+    There is deliberately no header toggle:
+    a CSV identity map's columns must be addressable by name,
+    so the first row is always read as the header.
+
+    Every option has a default, and the server fills in any option omitted from a request,
+    so the values here are the same defaults the server applies.
+    """
+
+    model_config = ConfigDict(extra="allow", populate_by_name=True)
+
+    format: TableReferenceFormat = TableReferenceFormat.csv
+    delimiter: str = ","
+    encoding: str = "utf-8"
+    quotechar: str = '"'
+    null_string: Optional[str] = None
+    """The string in the source data to read as a null value, or `None` to read no value as null."""
+
+
+class TableReference(BaseModel):
+    """
+    Represents a named, persisted reference to a table of identity data held outside DataMasque.
+
+    The data lives in the referenced connection —
+    a CSV or Parquet file in a file connection, or a table in a database connection —
+    so a ruleset can reference the identity map once, by name,
+    instead of re-declaring the source each time.
+
+    A table reference stores no credentials of its own; it borrows the referenced connection's.
+
+    `source` is interpreted according to that connection's own type:
+
+    * file connection — a path to the file within the connection's fileset.
+      The format comes from `options.format`, never from the path's suffix.
+    * database connection — a dotted, schema-qualified `schema.table` reference.
+
+    Names are unique across live table references (a deleted reference frees its name for reuse),
+    and the referenced connection must not be archived.
+    """
+
+    model_config = ConfigDict(extra="allow", populate_by_name=True)
+
+    name: str
+    connection: Union[ConnectionId, ConnectionConfig]
+    """The connection the data lives in. Accepts a `ConnectionId` or a `ConnectionConfig` (its `id` is extracted)."""
+    source: str
+    options: Optional[TableReferenceOptions] = None
+    """Format and CSV parsing options; when `None`, the server applies its defaults on create."""
+
+    # Server-populated read-only fields, excluded from request bodies.
+    id: Optional[TableReferenceId] = Field(default=None, exclude=True)
+    created: Optional[datetime] = Field(default=None, exclude=True)
+    modified: Optional[datetime] = Field(default=None, exclude=True)
+
+    @field_validator("connection", mode="before")
+    @classmethod
+    def _unwrap_connection(cls, value: object) -> object:
+        return unwrap_connection_id(value)
+
+    @property
+    def connection_id(self) -> ConnectionId:
+        """
+        The ID of the referenced connection.
+
+        `connection` accepts a whole `ConnectionConfig` for convenience but always holds an ID
+        once validated, so this narrows the declared type back down for callers reading it.
+        """
+
+        return ConnectionId(unwrap_connection_id(self.connection))