hirundo 0.1.18__py3-none-any.whl → 0.2.3.post1__py3-none-any.whl

hirundo/{dataset_optimization_results.py → dataset_qa_results.py}

@@ -11,17 +11,17 @@ DataFrameType = TypeAliasType("DataFrameType", None)
 if has_pandas:
     from hirundo._dataframe import pd
 
-    DataFrameType = TypeAliasType("DataFrameType", typing.Union[pd.DataFrame, None])
+    DataFrameType = TypeAliasType("DataFrameType", pd.DataFrame | None)
 if has_polars:
     from hirundo._dataframe import pl
 
-    DataFrameType = TypeAliasType("DataFrameType", typing.Union[pl.DataFrame, None])
+    DataFrameType = TypeAliasType("DataFrameType", pl.DataFrame | None)
 
 
 T = typing.TypeVar("T")
 
 
-class DatasetOptimizationResults(BaseModel, typing.Generic[T]):
+class DatasetQAResults(BaseModel, typing.Generic[T]):
     model_config = {"arbitrary_types_allowed": True}
 
     cached_zip_path: Path
@@ -30,13 +30,13 @@ class DatasetOptimizationResults(BaseModel, typing.Generic[T]):
     """
     suspects: T
     """
-    A polars/pandas DataFrame containing the results of the optimization run
+    A polars/pandas DataFrame containing the results of the data QA run
     """
-    object_suspects: typing.Optional[T]
+    object_suspects: T | None
     """
-    A polars/pandas DataFrame containing the object-level results of the optimization run
+    A polars/pandas DataFrame containing the object-level results of the data QA run
     """
     warnings_and_errors: T
     """
-    A polars/pandas DataFrame containing the warnings and errors of the optimization run
+    A polars/pandas DataFrame containing the warnings and errors of the data QA run
     """

hirundo/git.py

@@ -1,15 +1,13 @@
 import datetime
 import re
-import typing
 
 import pydantic
-import requests
 from pydantic import BaseModel, field_validator
 from pydantic_core import Url
 
 from hirundo._env import API_HOST
 from hirundo._headers import get_headers
-from hirundo._http import raise_for_status_with_reason
+from hirundo._http import raise_for_status_with_reason, requests
 from hirundo._timeouts import MODIFY_TIMEOUT, READ_TIMEOUT
 from hirundo._urls import RepoUrl
 from hirundo.logger import get_logger
@@ -33,14 +31,14 @@ class GitSSHAuth(BaseModel):
     """
     The SSH key for the Git repository
     """
-    ssh_password: typing.Optional[str]
+    ssh_password: str | None
     """
     The password for the SSH key for the Git repository.
     """
 
 
 class GitRepo(BaseModel):
-    id: typing.Optional[int] = None
+    id: int | None = None
     """
     The ID of the Git repository.
     """
@@ -49,25 +47,25 @@ class GitRepo(BaseModel):
     """
     A name to identify the Git repository in the Hirundo system.
     """
-    repository_url: typing.Union[str, RepoUrl]
+    repository_url: str | RepoUrl
     """
     The URL of the Git repository, it should start with `ssh://` or `https://` or be in the form `user@host:path`.
     If it is in the form `user@host:path`, it will be rewritten to `ssh://user@host/path`.
     """
-    organization_id: typing.Optional[int] = None
+    organization_id: int | None = None
     """
     The ID of the organization that the Git repository belongs to.
     If not provided, it will be assigned to your default organization.
     """
 
-    plain_auth: typing.Optional[GitPlainAuth] = pydantic.Field(
+    plain_auth: GitPlainAuth | None = pydantic.Field(
         default=None, examples=[None, {"username": "ben", "password": "password"}]
     )
     """
     The plain authentication details for the Git repository.
     Use this if using a special user with a username and password for authentication.
     """
-    ssh_auth: typing.Optional[GitSSHAuth] = pydantic.Field(
+    ssh_auth: GitSSHAuth | None = pydantic.Field(
         default=None,
         examples=[
             {
@@ -85,7 +83,7 @@ class GitRepo(BaseModel):
 
     @field_validator("repository_url", mode="before", check_fields=True)
     @classmethod
-    def check_valid_repository_url(cls, repository_url: typing.Union[str, RepoUrl]):
+    def check_valid_repository_url(cls, repository_url: str | RepoUrl):
         # Check if the URL has the `@` and `:` pattern with a non-numeric section before the next slash
         match = re.match("([^@]+@[^:]+):([^0-9/][^/]*)/(.+)", str(repository_url))
         if match:

hirundo/labeling.py

@@ -3,11 +3,9 @@ from abc import ABC
 
 from pydantic import BaseModel, Field
 
+from hirundo._urls import HirundoUrl
 from hirundo.dataset_enum import DatasetMetadataType
 
-if typing.TYPE_CHECKING:
-    from hirundo._urls import HirundoUrl
-
 
 class Metadata(BaseModel, ABC, frozen=True):
     type: DatasetMetadataType
@@ -21,7 +19,7 @@ class HirundoCSV(Metadata, frozen=True):
     type: typing.Literal[DatasetMetadataType.HIRUNDO_CSV] = (
         DatasetMetadataType.HIRUNDO_CSV
     )
-    csv_url: "HirundoUrl"
+    csv_url: HirundoUrl
     """
     The URL to access the dataset metadata CSV file.
     e.g. `s3://my-bucket-name/my-folder/my-metadata.csv`, `gs://my-bucket-name/my-folder/my-metadata.csv`,
@@ -36,7 +34,7 @@ class COCO(Metadata, frozen=True):
     """
 
     type: typing.Literal[DatasetMetadataType.COCO] = DatasetMetadataType.COCO
-    json_url: "HirundoUrl"
+    json_url: HirundoUrl
     """
     The URL to access the dataset metadata JSON file.
     e.g. `s3://my-bucket-name/my-folder/my-metadata.json`, `gs://my-bucket-name/my-folder/my-metadata.json`,
@@ -47,8 +45,18 @@ class COCO(Metadata, frozen=True):
 
 class YOLO(Metadata, frozen=True):
     type: typing.Literal[DatasetMetadataType.YOLO] = DatasetMetadataType.YOLO
-    data_yaml_url: "typing.Optional[HirundoUrl]" = None
-    labels_dir_url: "HirundoUrl"
+    data_yaml_url: HirundoUrl | None = None
+    labels_dir_url: HirundoUrl
+
+
+class HuggingFaceAudio(Metadata, frozen=True):
+    type: typing.Literal[DatasetMetadataType.HuggingFaceAudio] = (
+        DatasetMetadataType.HuggingFaceAudio
+    )
+    audio_column: str
+    text_column: str
+    subset: str | None = None
+    split: str | None = None
 
 
 class KeylabsAuth(BaseModel):
@@ -63,7 +71,7 @@ class Keylabs(Metadata, frozen=True):
     Keylabs project ID.
     """
 
-    labels_dir_url: "HirundoUrl"
+    labels_dir_url: HirundoUrl
     """
     URL to the directory containing the Keylabs labels.
     """
@@ -73,11 +81,11 @@ class Keylabs(Metadata, frozen=True):
     Whether to include attributes in the class name.
     """
 
-    project_name: typing.Optional[str] = None
+    project_name: str | None = None
     """
     Keylabs project name (optional; added to output CSV if provided).
     """
-    keylabs_auth: typing.Optional[KeylabsAuth] = None
+    keylabs_auth: KeylabsAuth | None = None
     """
     Keylabs authentication credentials (optional; if provided, used to provide links to each sample).
     """
@@ -107,9 +115,9 @@ class KeylabsObjSegVideo(Keylabs, frozen=True):
     )
 
 
-KeylabsInfo = typing.Union[
-    KeylabsObjDetImages, KeylabsObjDetVideo, KeylabsObjSegImages, KeylabsObjSegVideo
-]
+KeylabsInfo = (
+    KeylabsObjDetImages | KeylabsObjDetVideo | KeylabsObjSegImages | KeylabsObjSegVideo
+)
 """
 The dataset labeling info for Keylabs. The dataset labeling info can be one of the following:
 - `DatasetMetadataType.KeylabsObjDetImages`: Indicates that the dataset metadata file is in the Keylabs object detection image format
@@ -118,12 +126,7 @@ The dataset labeling info for Keylabs. The dataset labeling info can be one of t
 - `DatasetMetadataType.KeylabsObjSegVideo`: Indicates that the dataset metadata file is in the Keylabs object segmentation video format
 """
 LabelingInfo = typing.Annotated[
-    typing.Union[
-        HirundoCSV,
-        COCO,
-        YOLO,
-        KeylabsInfo,
-    ],
+    HirundoCSV | COCO | YOLO | KeylabsInfo | HuggingFaceAudio,
     Field(discriminator="type"),
 ]
 """

hirundo/storage.py

@@ -1,14 +1,13 @@
-import typing
+import datetime
 from pathlib import Path
 
 import pydantic
-import requests
 from pydantic import BaseModel, model_validator
 from pydantic_core import Url
 
 from hirundo._env import API_HOST
 from hirundo._headers import get_headers
-from hirundo._http import raise_for_status_with_reason
+from hirundo._http import raise_for_status_with_reason, requests
 from hirundo._timeouts import MODIFY_TIMEOUT, READ_TIMEOUT
 from hirundo._urls import S3BucketUrl, StorageConfigName
 from hirundo.dataset_enum import StorageTypes
@@ -21,13 +20,13 @@ S3_PREFIX = "s3://"
 
 
 class StorageS3Base(BaseModel):
-    endpoint_url: typing.Optional[Url] = None
+    endpoint_url: Url | None = None
     bucket_url: S3BucketUrl
     region_name: str
     # ⬆️ We could restrict this, but if we're allowing custom endpoints then the validation may be wrong
-    access_key_id: typing.Optional[str] = None
+    access_key_id: str | None = None
 
-    def get_url(self, path: typing.Union[str, Path]) -> Url:
+    def get_url(self, path: str | Path) -> Url:
         """
         Get the full URL for a file in the S3 bucket
 
@@ -46,7 +45,7 @@ class StorageS3Base(BaseModel):
 
 
 class StorageS3(StorageS3Base):
-    secret_access_key: typing.Optional[str] = None
+    secret_access_key: str | None = None
 
 
 class StorageS3Out(StorageS3Base):
@@ -57,7 +56,7 @@ class StorageGCPBase(BaseModel):
     bucket_name: str
     project: str
 
-    def get_url(self, path: typing.Union[str, Path]) -> Url:
+    def get_url(self, path: str | Path) -> Url:
         """
         Get the full URL for a file in the GCP bucket
 
@@ -74,7 +73,7 @@ class StorageGCPBase(BaseModel):
 
 
 class StorageGCP(StorageGCPBase):
-    credentials_json: typing.Optional[dict] = None
+    credentials_json: dict | None = None
 
 
 class StorageGCPOut(StorageGCPBase):
@@ -105,9 +104,7 @@ class StorageGCPOut(StorageGCPBase):
 #     account_url: str
 
 
-def get_git_repo_url(
-    repo_url: typing.Union[str, Url], path: typing.Union[str, Path]
-) -> Url:
+def get_git_repo_url(repo_url: str | Url, path: str | Path) -> Url:
     """
     Get the full URL for a file in the git repository
 
@@ -128,12 +125,12 @@ def get_git_repo_url(
 
 
 class StorageGit(BaseModel):
-    repo_id: typing.Optional[int] = None
+    repo_id: int | None = None
     """
     The ID of the Git repository in the Hirundo system.
     Either :code:`repo_id` or :code:`repo` must be provided.
     """
-    repo: typing.Optional[GitRepo] = None
+    repo: GitRepo | None = None
     """
     The Git repository to link to.
     Either :code:`repo_id` or :code:`repo` must be provided.
@@ -149,7 +146,7 @@ class StorageGit(BaseModel):
             raise ValueError("Either repo_id or repo must be provided")
         return self
 
-    def get_url(self, path: typing.Union[str, Path]) -> Url:
+    def get_url(self, path: str | Path) -> Url:
         """
         Get the full URL for a file in the git repository
 
@@ -172,7 +169,7 @@ class StorageGitOut(BaseModel):
     repo: GitRepoOut
     branch: str
 
-    def get_url(self, path: typing.Union[str, Path]) -> Url:
+    def get_url(self, path: str | Path) -> Url:
         """
         Get the full URL for a file in the git repository
 
@@ -190,12 +187,12 @@ class StorageGitOut(BaseModel):
 
 
 class StorageConfig(BaseModel):
-    id: typing.Optional[int] = None
+    id: int | None = None
     """
     The ID of the :code:`StorageConfig` in the Hirundo system.
     """
 
-    organization_id: typing.Optional[int] = None
+    organization_id: int | None = None
     """
     The ID of the organization that the :code:`StorageConfig` belongs to.
     If not provided, it will be assigned to your default organization.
@@ -205,7 +202,7 @@ class StorageConfig(BaseModel):
     """
     A name to identify the :code:`StorageConfig` in the Hirundo system.
     """
-    type: typing.Optional[StorageTypes] = pydantic.Field(
+    type: StorageTypes | None = pydantic.Field(
         examples=[
             StorageTypes.S3,
             StorageTypes.GCP,
@@ -221,7 +218,7 @@ class StorageConfig(BaseModel):
     - :code:`Azure` (coming soon)
     - :code:`Git`
     """
-    s3: typing.Optional[StorageS3] = pydantic.Field(
+    s3: StorageS3 | None = pydantic.Field(
         default=None,
         examples=[
             {
@@ -239,7 +236,7 @@ class StorageConfig(BaseModel):
     The Amazon Web Services (AWS) S3 storage config details.
     Use this if you want to link to an S3 bucket.
     """
-    gcp: typing.Optional[StorageGCP] = pydantic.Field(
+    gcp: StorageGCP | None = pydantic.Field(
         default=None,
         examples=[
             None,
@@ -282,7 +279,7 @@ class StorageConfig(BaseModel):
     #         None,
     #     ],
     # )  TODO: Azure storage config is coming soon
-    git: typing.Optional[StorageGit] = pydantic.Field(
+    git: StorageGit | None = pydantic.Field(
         default=None,
         examples=[
             None,
@@ -341,7 +338,7 @@ class StorageConfig(BaseModel):
 
     @staticmethod
     def list(
-        organization_id: typing.Optional[int] = None,
+        organization_id: int | None = None,
     ) -> list["ResponseStorageConfig"]:
         """
         Lists all the :code:`StorageConfig`'s created by user's default organization
@@ -441,7 +438,10 @@ class ResponseStorageConfig(BaseModel):
     type: StorageTypes
     organization_name: str
     creator_name: str
-    s3: typing.Optional[StorageS3Out]
-    gcp: typing.Optional[StorageGCPOut]
+    s3: StorageS3Out | None
+    gcp: StorageGCPOut | None
     # azure: typing.Optional[StorageAzureOut]
-    git: typing.Optional[StorageGitOut]
+    git: StorageGitOut | None
+
+    created_at: datetime.datetime
+    updated_at: datetime.datetime