sarvamai 0.1.13a2__tar.gz → 0.1.14__tar.gz

{sarvamai-0.1.13a2 → sarvamai-0.1.14}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sarvamai
-Version: 0.1.13a2
+Version: 0.1.14
 Summary: 
 Requires-Python: >=3.8,<4.0
 Classifier: Intended Audience :: Developers

{sarvamai-0.1.13a2 → sarvamai-0.1.14}/pyproject.toml

@@ -3,7 +3,7 @@ name = "sarvamai"
 
 [tool.poetry]
 name = "sarvamai"
-version = "0.1.13a2"
+version = "0.1.14"
 description = ""
 readme = "README.md"
 authors = []

{sarvamai-0.1.13a2 → sarvamai-0.1.14}/src/sarvamai/core/client_wrapper.py

@@ -23,10 +23,10 @@ class BaseClientWrapper:
 
     def get_headers(self) -> typing.Dict[str, str]:
         headers: typing.Dict[str, str] = {
-            "User-Agent": "sarvamai/0.1.13a2",
+            "User-Agent": "sarvamai/0.1.14",
             "X-Fern-Language": "Python",
             "X-Fern-SDK-Name": "sarvamai",
-            "X-Fern-SDK-Version": "0.1.13a2",
+            "X-Fern-SDK-Version": "0.1.14",
             **(self.get_custom_headers() or {}),
         }
         headers["api-subscription-key"] = self.api_subscription_key

{sarvamai-0.1.13a2 → sarvamai-0.1.14}/src/sarvamai/speech_to_text_job/client.py

@@ -10,7 +10,10 @@ from ..types.bulk_job_init_response_v_1 import BulkJobInitResponseV1
 from ..types.files_download_response import FilesDownloadResponse
 from ..types.files_upload_response import FilesUploadResponse
 from ..types.job_status_v_1_response import JobStatusV1Response
+from ..types.speech_to_text_model import SpeechToTextModel
+from ..types.speech_to_text_language import SpeechToTextLanguage
 from .raw_client import AsyncRawSpeechToTextJobClient, RawSpeechToTextJobClient
+from .job import AsyncSpeechToTextJob, SpeechToTextJob
 
 # this is used as the default value for optional parameters
 OMIT = typing.cast(typing.Any, ...)
@@ -215,6 +218,76 @@ class SpeechToTextJobClient:
         _response = self._raw_client.get_download_links(job_id=job_id, files=files, request_options=request_options)
         return _response.data
 
+    def create_job(
+        self,
+        model: SpeechToTextModel = "saarika:v2.5",
+        with_diarization: bool = False,
+        with_timestamps: bool = False,
+        language_code: typing.Optional[SpeechToTextLanguage] = None,
+        num_speakers: typing.Optional[int] = None,
+        callback: typing.Optional[BulkJobCallbackParams] = OMIT,
+        request_options: typing.Optional[RequestOptions] = None,
+    ) -> SpeechToTextJob:
+        """
+        Create a new Speech-to-Text bulk job.
+
+        Parameters
+        ----------
+        model : SpeechToTextModel, default="saarika:v2.5"
+            The model to use for transcription.
+
+        with_diarization : typing.Optional[bool], default=False
+            Whether to enable speaker diarization (distinguishing who said what).
+
+        with_timestamps : typing.Optional[bool], default=False
+            Whether to include word-level timestamps in the transcription output.
+
+        language_code : typing.Optional[SpeechToTextLanguage], default=None
+            The language code of the input audio (e.g., "hi-IN", "bn-IN").
+            
+        num_speakers : typing.Optional[int], default=None
+            The number of distinct speakers in the audio, if known.
+
+        callback : typing.Optional[BulkJobCallbackParams], default=OMIT
+            Optional callback configuration to receive job completion events.
+
+        request_options : typing.Optional[RequestOptions], default=None
+            Request-specific configuration.
+
+        Returns
+        -------
+        SpeechToTextJob
+            A handle to the newly created Speech-to-Text job.
+        """
+        response = self.initialise(
+            job_parameters=SpeechToTextJobParametersParams(
+                language_code=language_code,
+                model=model,
+                num_speakers=num_speakers,  # type: ignore[typeddict-item]
+                with_diarization=with_diarization,
+                with_timestamps=with_timestamps,
+            ),
+            callback=callback,
+            request_options=request_options,
+        )
+        return SpeechToTextJob(job_id=response.job_id, client=self)
+
+    def get_job(self, job_id: str) -> SpeechToTextJob:
+        """
+        Get an existing Speech-to-Text job handle by job ID.
+
+        Parameters
+        ----------
+        job_id : str
+            The job ID of the previously created Speech-to-Text job.
+
+        Returns
+        -------
+        SpeechToTextJob
+            A job handle which can be used to check status or retrieve results.
+        """
+        return SpeechToTextJob(job_id=job_id, client=self)
+
 
 class AsyncSpeechToTextJobClient:
     def __init__(self, *, client_wrapper: AsyncClientWrapper):
@@ -456,3 +529,73 @@ class AsyncSpeechToTextJobClient:
             job_id=job_id, files=files, request_options=request_options
         )
         return _response.data
+
+    async def create_job(
+        self,
+        model: SpeechToTextModel = "saarika:v2.5",
+        with_diarization: bool = False,
+        with_timestamps: bool = False,
+        language_code: typing.Optional[SpeechToTextLanguage] = None,
+        num_speakers: typing.Optional[int] = None,
+        callback: typing.Optional[BulkJobCallbackParams] = OMIT,
+        request_options: typing.Optional[RequestOptions] = None,
+    ) -> "AsyncSpeechToTextJob":
+        """
+        Create a new Speech-to-Text bulk job.
+
+        Parameters
+        ----------
+        model : SpeechToTextModel, default="saarika:v2.5"
+            The model to use for transcription.
+
+        with_diarization : typing.Optional[bool], default=False
+            Whether to enable speaker diarization (distinguishing who said what).
+
+        with_timestamps : typing.Optional[bool], default=False
+            Whether to include word-level timestamps in the transcription output.
+
+        language_code : typing.Optional[SpeechToTextLanguage], default=None
+            The language code of the input audio (e.g., "hi-IN", "bn-IN").
+            
+        num_speakers : typing.Optional[int], default=None
+            The number of distinct speakers in the audio, if known.
+
+        callback : typing.Optional[BulkJobCallbackParams], default=OMIT
+            Optional callback configuration to receive job completion events.
+
+        request_options : typing.Optional[RequestOptions], default=None
+            Request-specific configuration.
+
+        Returns
+        -------
+        AsyncSpeechToTextJob
+            A handle to the newly created job.
+        """
+        response = await self.initialise(
+            job_parameters=SpeechToTextJobParametersParams(
+                language_code=language_code,
+                model=model,
+                with_diarization=with_diarization,
+                with_timestamps=with_timestamps,
+                num_speakers=num_speakers,  # type: ignore[typeddict-item]
+            ),
+            callback=callback,
+            request_options=request_options,
+        )
+        return AsyncSpeechToTextJob(job_id=response.job_id, client=self)
+
+    async def get_job(self, job_id: str) -> "AsyncSpeechToTextJob":
+        """
+        Get an existing Speech-to-Text job handle by job ID.
+
+        Parameters
+        ----------
+        job_id : str
+            The job ID of the previously created speech-to-text job.
+
+        Returns
+        -------
+        AsyncSpeechToTextJob
+            A job handle which can be used to check status or retrieve results.
+        """
+        return AsyncSpeechToTextJob(job_id=job_id, client=self)

sarvamai-0.1.14/src/sarvamai/speech_to_text_job/job.py

@@ -0,0 +1,484 @@
+import asyncio
+import mimetypes
+import os
+import time
+import typing
+import httpx
+
+from ..types import JobStatusV1Response
+
+if typing.TYPE_CHECKING:
+    from .client import AsyncSpeechToTextJobClient, SpeechToTextJobClient
+
+
+class AsyncSpeechToTextJob:
+    def __init__(self, job_id: str, client: "AsyncSpeechToTextJobClient"):
+        """
+        Initialize the asynchronous speech-to-text job.
+
+        Parameters
+        ----------
+        job_id : str
+            The unique job identifier returned from a previous job initialization.
+
+        client : AsyncSpeechToTextJobClient
+            The async client instance used to create the job.
+
+            !!! important
+                This must be the **same client instance** that was used to initialize
+                the job originally, as it contains the subscription key and configuration
+                required to authenticate and manage the job.
+
+        """
+        self._job_id = job_id
+        self._client = client
+
+    @property
+    def job_id(self) -> str:
+        """
+        Returns the job ID associated with this job instance.
+
+        Returns
+        -------
+        str
+        """
+        return self._job_id
+
+    async def upload_files(
+        self, file_paths: typing.Sequence[str], timeout: float = 60.0
+    ) -> bool:
+        """
+        Upload input audio files for the speech-to-text job.
+
+        Parameters
+        ----------
+        file_paths : Sequence[str]
+            List of full paths to local audio files.
+
+        timeout : float, optional
+            The maximum time to wait for the upload to complete (in seconds),
+            by default 60.0
+        Returns
+        -------
+        bool
+            True if all files are uploaded successfully.
+        """
+        upload_links = await self._client.get_upload_links(
+            job_id=self._job_id,
+            files=[os.path.basename(p) for p in file_paths],
+        )
+        client_timeout = httpx.Timeout(timeout=timeout)
+        async with httpx.AsyncClient(timeout=client_timeout) as session:
+            for path in file_paths:
+                file_name = os.path.basename(path)
+                url = upload_links.upload_urls[file_name].file_url
+                with open(path, "rb") as f:
+                    content_type, _ = mimetypes.guess_type(path)
+                    if content_type is None:
+                        content_type = "audio/wav"
+                    response = await session.put(
+                        url,
+                        content=f.read(),
+                        headers={
+                            "x-ms-blob-type": "BlockBlob",
+                            "Content-Type": content_type,
+                        },
+                    )
+                if response.status_code != 201:
+                    raise RuntimeError(
+                        f"Upload failed for {file_name}: {response.status_code}"
+                    )
+        return True
+
+    async def wait_until_complete(
+        self, poll_interval: int = 5, timeout: int = 600
+    ) -> JobStatusV1Response:
+        """
+        Polls job status until it completes or fails.
+
+        Parameters
+        ----------
+        poll_interval : int, optional
+            Time in seconds between polling attempts (default is 5).
+
+        timeout : int, optional
+            Maximum time to wait for completion in seconds (default is 600).
+
+        Returns
+        -------
+        JobStatusV1Response
+            Final job status.
+
+        Raises
+        ------
+        TimeoutError
+            If the job does not complete within the given timeout.
+        """
+        start = asyncio.get_event_loop().time()
+        while True:
+            status = await self.get_status()
+            state = status.job_state.lower()
+            if state in {"completed", "failed"}:
+                return status
+            if asyncio.get_event_loop().time() - start > timeout:
+                raise TimeoutError(
+                    f"Job {self._job_id} did not complete within {timeout} seconds."
+                )
+            await asyncio.sleep(poll_interval)
+
+    async def get_output_mappings(self) -> typing.List[typing.Dict[str, str]]:
+        """
+        Get the mapping of input files to their corresponding output files.
+
+        Returns
+        -------
+        List[Dict[str, str]]
+            List of mappings with keys 'input_file' and 'output_file'.
+        """
+        job_status = await self.get_status()
+        return [
+            {
+                "input_file": detail.inputs[0].file_name,
+                "output_file": detail.outputs[0].file_name,
+            }
+            for detail in (job_status.job_details or [])
+            if detail.inputs and detail.outputs
+        ]
+
+    async def download_outputs(self, output_dir: str) -> bool:
+        """
+        Download output files to the specified directory.
+
+        Parameters
+        ----------
+        output_dir : str
+            Local directory where outputs will be saved.
+
+        Returns
+        -------
+        bool
+            True if all files downloaded successfully.
+
+        Raises
+        ------
+        RuntimeError
+            If a file fails to download.
+        """
+        mappings = await self.get_output_mappings()
+        file_names = [m["output_file"] for m in mappings]
+        download_links = await self._client.get_download_links(
+            job_id=self._job_id, files=file_names
+        )
+
+        os.makedirs(output_dir, exist_ok=True)
+        async with httpx.AsyncClient() as session:
+            for m in mappings:
+                url = download_links.download_urls[m["output_file"]].file_url
+                response = await session.get(url)
+                if response.status_code != 200:
+                    raise RuntimeError(
+                        f"Download failed for {m['output_file']}: {response.status_code}"
+                    )
+                output_path = os.path.join(output_dir, f"{m['input_file']}.json")
+                with open(output_path, "wb") as f:
+                    f.write(response.content)
+        return True
+
+    async def get_status(self) -> JobStatusV1Response:
+        """
+        Retrieve the current status of the job.
+
+        Returns
+        -------
+        JobStatusV1Response
+        """
+        return await self._client.get_status(self._job_id)
+
+    async def start(self) -> JobStatusV1Response:
+        """
+        Start the speech-to-text job processing.
+
+        Returns
+        -------
+        JobStatusV1Response
+        """
+        return await self._client.start(job_id=self._job_id)
+
+    async def exists(self) -> bool:
+        """
+        Check if the job exists in the system.
+
+        Returns
+        -------
+        bool
+        """
+        try:
+            await self.get_status()
+            return True
+        except httpx.HTTPStatusError:
+            return False
+
+    async def is_complete(self) -> bool:
+        """
+        Check if the job is either completed or failed.
+
+        Returns
+        -------
+        bool
+        """
+        state = (await self.get_status()).job_state.lower()
+        return state in {"completed", "failed"}
+
+    async def is_successful(self) -> bool:
+        """
+        Check if the job completed successfully.
+
+        Returns
+        -------
+        bool
+        """
+        return (await self.get_status()).job_state.lower() == "completed"
+
+    async def is_failed(self) -> bool:
+        """
+        Check if the job has failed.
+
+        Returns
+        -------
+        bool
+        """
+        return (await self.get_status()).job_state.lower() == "failed"
+
+
+class SpeechToTextJob:
+    def __init__(self, job_id: str, client: "SpeechToTextJobClient"):
+        """
+        Initialize the synchronous speech-to-text job.
+
+        Parameters
+        ----------
+        job_id : str
+            The unique job identifier returned from a previous job initialization.
+
+        client : SpeechToTextJobClient
+            The client instance used to create the job.
+
+            !!! important
+                This must be the **same client instance** that was used to initialize
+                the job originally, as it contains the subscription key and configuration
+                required to authenticate and manage the job.
+
+        """
+        self._job_id = job_id
+        self._client = client
+
+    @property
+    def job_id(self) -> str:
+        """
+        Returns the job ID associated with this job instance.
+
+        Returns
+        -------
+        str
+        """
+        return self._job_id
+
+    def upload_files(
+        self, file_paths: typing.Sequence[str], timeout: float = 60.0
+    ) -> bool:
+        """
+        Upload input audio files for the speech-to-text job.
+
+        Parameters
+        ----------
+        file_paths : Sequence[str]
+            List of full paths to local audio files.
+
+        timeout : float, optional
+            The maximum time to wait for the upload to complete (in seconds),
+            by default 60.0
+        Returns
+        -------
+        bool
+            True if all files are uploaded successfully.
+        """
+        upload_links = self._client.get_upload_links(
+            job_id=self._job_id, files=[os.path.basename(p) for p in file_paths]
+        )
+        client_timeout = httpx.Timeout(timeout=timeout)
+        with httpx.Client(timeout=client_timeout) as client:
+            for path in file_paths:
+                file_name = os.path.basename(path)
+                url = upload_links.upload_urls[file_name].file_url
+                with open(path, "rb") as f:
+                    response = client.put(
+                        url,
+                        content=f,
+                        headers={
+                            "x-ms-blob-type": "BlockBlob",
+                            "Content-Type": "audio/wav",
+                        },
+                    )
+                if response.status_code != 201:
+                    raise RuntimeError(
+                        f"Upload failed for {file_name}: {response.status_code}"
+                    )
+        return True
+
+    def wait_until_complete(
+        self, poll_interval: int = 5, timeout: int = 600
+    ) -> JobStatusV1Response:
+        """
+        Polls job status until it completes or fails.
+
+        Parameters
+        ----------
+        poll_interval : int, optional
+            Time in seconds between polling attempts (default is 5).
+
+        timeout : int, optional
+            Maximum time to wait for completion in seconds (default is 600).
+
+        Returns
+        -------
+        JobStatusV1Response
+            Final job status.
+
+        Raises
+        ------
+        TimeoutError
+            If the job does not complete within the given timeout.
+        """
+        start = time.monotonic()
+        while True:
+            status = self.get_status()
+            state = status.job_state.lower()
+            if state in {"completed", "failed"}:
+                return status
+            if time.monotonic() - start > timeout:
+                raise TimeoutError(
+                    f"Job {self._job_id} did not complete within {timeout} seconds."
+                )
+            time.sleep(poll_interval)
+
+    def get_output_mappings(self) -> typing.List[typing.Dict[str, str]]:
+        """
+        Get the mapping of input files to their corresponding output files.
+
+        Returns
+        -------
+        List[Dict[str, str]]
+            List of mappings with keys 'input_file' and 'output_file'.
+        """
+        job_status = self.get_status()
+        return [
+            {
+                "input_file": detail.inputs[0].file_name,
+                "output_file": detail.outputs[0].file_name,
+            }
+            for detail in (job_status.job_details or [])
+            if detail.inputs and detail.outputs
+        ]
+
+    def download_outputs(self, output_dir: str) -> bool:
+        """
+        Download output files to the specified directory.
+
+        Parameters
+        ----------
+        output_dir : str
+            Local directory where outputs will be saved.
+
+        Returns
+        -------
+        bool
+            True if all files downloaded successfully.
+
+        Raises
+        ------
+        RuntimeError
+            If a file fails to download.
+        """
+        mappings = self.get_output_mappings()
+        file_names = [m["output_file"] for m in mappings]
+        download_links = self._client.get_download_links(
+            job_id=self._job_id, files=file_names
+        )
+
+        os.makedirs(output_dir, exist_ok=True)
+        with httpx.Client() as client:
+            for m in mappings:
+                url = download_links.download_urls[m["output_file"]].file_url
+                response = client.get(url)
+                if response.status_code != 200:
+                    raise RuntimeError(
+                        f"Download failed for {m['output_file']}: {response.status_code}"
+                    )
+                output_path = os.path.join(output_dir, f"{m['input_file']}.json")
+                with open(output_path, "wb") as f:
+                    f.write(response.content)
+        return True
+
+    def get_status(self) -> JobStatusV1Response:
+        """
+        Retrieve the current status of the job.
+
+        Returns
+        -------
+        JobStatusV1Response
+        """
+        return self._client.get_status(self._job_id)
+
+    def start(self) -> JobStatusV1Response:
+        """
+        Start the speech-to-text job processing.
+
+        Returns
+        -------
+        JobStatusV1Response
+        """
+        return self._client.start(job_id=self._job_id)
+
+    def exists(self) -> bool:
+        """
+        Check if the job exists in the system.
+
+        Returns
+        -------
+        bool
+        """
+        try:
+            self.get_status()
+            return True
+        except httpx.HTTPStatusError:
+            return False
+
+    def is_complete(self) -> bool:
+        """
+        Check if the job is either completed or failed.
+
+        Returns
+        -------
+        bool
+        """
+        return self.get_status().job_state.lower() in {"completed", "failed"}
+
+    def is_successful(self) -> bool:
+        """
+        Check if the job completed successfully.
+
+        Returns
+        -------
+        bool
+        """
+        return self.get_status().job_state.lower() == "completed"
+
+    def is_failed(self) -> bool:
+        """
+        Check if the job has failed.
+
+        Returns
+        -------
+        bool
+        """
+        return self.get_status().job_state.lower() == "failed"