svo-client 2.3.2

Async client for SVO semantic chunker microservice.

svo-client

Async Python client for SVO Semantic Chunker microservice.

Installation

pip install svo-client

Quick Start

Text Chunking

from svo_client import ChunkerClient
import asyncio

async def main():
    async with ChunkerClient(
        host="localhost", port=8009,
        cert="client.crt", key="client.key", ca="ca.crt",
    ) as client:
        chunks = await client.chunk(["Your text here."])
        for text_chunks in chunks:
            for chunk in text_chunks:
                print(chunk.text)

asyncio.run(main())

File Chunking

async with ChunkerClient(
    host="localhost", port=8009,
    cert=cert, key=key, ca=ca,
) as client:
    # Any file, any format — server handles processing.
    # Default: no timeout limit (server can work up to 1 hour).
    chunks = await client.chunk_file(
        filepath="/path/to/document.pdf",
        filter_name="plain_text",
    )

    # Explicit timeout limit (optional):
    chunks = await client.chunk_file(
        filepath="/path/to/document.pdf",
        timeout=1800,  # 30 min max
    )

API Reference

ChunkerClient

ChunkerClient(
    *,
    config: Optional[Dict[str, Any]] = None,
    host: str = "localhost",
    port: int = 8009,
    cert: Optional[str] = None,
    key: Optional[str] = None,
    ca: Optional[str] = None,
    token: Optional[str] = None,
    token_header: str = "X-API-Key",
    check_hostname: bool = False,
    timeout: Optional[float] = None,
)

Parameter	Type	Default	Description
config	Optional[Dict]	None	Pre-built config; if set, other args ignored
host	str	"localhost"	Server host
port	int	8009	Server port
cert	Optional[str]	None	Client certificate path (mTLS)
key	Optional[str]	None	Client key path (mTLS)
ca	Optional[str]	None	CA certificate path (mTLS)
token	Optional[str]	None	API key for authentication
token_header	str	"X-API-Key"	HTTP header for API key
check_hostname	bool	False	Verify SSL hostname
timeout	Optional[float]	None	Default request timeout (seconds); None = no timeout

Methods

chunk(texts, use_sv=False, timeout=0.0, verify_integrity=False, **params)

Chunk a list of texts via WebSocket. Returns List[List[SemanticChunk]] — one list of chunks per input text.

Parameter	Type	Default	Description
texts	List[str]	required	Texts to chunk
use_sv	bool	False	Use semantic verification
timeout	float	0.0	Timeout in seconds; 0 = no limit
verify_integrity	bool	False	Check text integrity after chunking
**params	Any	—	Additional chunk parameters

file(*, filepath=None, filename=None, file_content=None, filter_name="plain_text", timeout=None)

Send a file to the server for text extraction. Returns FileResponse.

Two input modes:

• CLI channel: provide filepath (server reads the file).
• API channel: provide filename + file_content (bytes; base64-encoded internally).

Parameter	Type	Default	Description
filepath	Optional[str]	None	Local file path (CLI channel)
filename	Optional[str]	None	Filename (API channel)
file_content	Optional[bytes]	None	Raw file bytes (API channel)
filter_name	str	"plain_text"	Server-side filter name
timeout	Optional[float]	None	Per-call timeout in seconds

chunk_file(*, filepath=None, filename=None, file_content=None, filter_name="plain_text", use_sv=False, timeout=0, verify_integrity=False, **chunk_params)

Convenience method: file() + chunk() in one call. Returns List[List[SemanticChunk]].

Any file format supported (PDF, DOCX, images, text, markdown, etc.) — server handles all processing.

Parameter	Type	Default	Description
filepath	Optional[str]	None	Local file path
filename	Optional[str]	None	Filename for API channel
file_content	Optional[bytes]	None	Raw file bytes
filter_name	str	"plain_text"	Server-side filter name
use_sv	bool	False	Use semantic verification
timeout	Optional[float]	0	Timeout in seconds; 0 = no limit
verify_integrity	bool	False	Check text integrity
**chunk_params	Any	—	Additional chunk parameters

config(timeout=None)

Retrieve server configuration. Returns Dict[str, Any].

help_cmd(command=None, timeout=None)

Retrieve server help information. Returns Dict[str, Any]. If command is given, returns help for that specific command.

health()

Health check — verifies the server is up. Returns Dict[str, Any].

open_ws_channel(receive_timeout=60.0, heartbeat=30.0)

Open a bidirectional WebSocket channel for multiple requests. Returns BidirectionalWsChannel.

Parameter	Type	Default	Description
receive_timeout	float	60.0	Per-message receive timeout (seconds)
heartbeat	float	30.0	WebSocket keepalive interval (seconds)

close()

Close the underlying client connection. Also available via async context manager (async with).

Long-Running File Operations

File processing is entirely server-side. The client sends the file as-is (any format: PDF, DOCX, images, text, markdown, etc.).

• chunk_file() defaults to no timeout limit — the client waits as long as the server needs (up to 1 hour for large files).
• Results arrive via WebSocket; the adapter heartbeat (30s) keeps the connection alive.
• To set an explicit limit, pass timeout=N (seconds):

# Wait up to 30 minutes:
chunks = await client.chunk_file(
    filepath="/path/to/large.pdf",
    timeout=1800,
)

CLI Reference

# Chunk text
svo-chunker chunk --text "Your text here" [--use-sv] [--type Draft]

# Chunk multiple texts (batch)
svo-chunker chunk-batch --text "Text one" --text "Text two"

# Process file (extraction + optional chunking)
svo-chunker file --filepath /path/to/file [--filter plain_text] [--chunk]

# Server configuration
svo-chunker config

# Health check
svo-chunker health

Error Handling

All exceptions are importable from svo_client:

from svo_client import SVOServerError, SVOTimeoutError, SVOFileError

Exception	When Raised
SVOServerError	Server application-level error
SVOChunkingIntegrityError	Text integrity check fails (subclass of SVOServerError)
SVOJSONRPCError	JSON-RPC error response
SVOHTTPError	HTTP error or invalid response
SVOWebSocketRequiredError	WebSocket required but unavailable
SVOConnectionError	Network/connection issues
SVOTimeoutError	Request timeout exceeded
SVOEmbeddingError	Embedding service error
SVOFileError	Base file-command error
SVOFilePayloadError	Invalid file payload (subclass of SVOFileError)
SVOFileTypeError	Unknown filter_name (subclass of SVOFileError)
SVOFileNotFoundError	File not found (subclass of SVOFileError)
SVOFilePermissionError	Permission denied (subclass of SVOFileError)
SVOFileReadError	OS-level file read error (subclass of SVOFileError)

Filter Names

Supported server-side filter names for file processing:

• plain_text — extract plain text content (default)
• markdown — extract and process markdown
• txt — raw text extraction