Implementation:AUTOMATIC1111 Stable diffusion webui Shared State

Knowledge Sources	AUTOMATIC1111_Stable_diffusion_webui
Domains	Job Management, Progress Tracking
Last Updated	2025-05-15 00:00 GMT

Overview

Implements the State class that tracks the lifecycle of image generation jobs, including progress, interruption, live preview updates, and server command signaling.

Description

The State class is the central mechanism for managing generation job state across the web UI. It tracks whether a job has been skipped, interrupted, or stopped via boolean flags. It maintains job metadata such as job name, number, count, timestamp, and sampling step progress. The class provides begin() and end() methods to bracket job execution, resetting counters and timing. The skip(), interrupt(), and stop_generating() methods allow the UI or API to signal cancellation. The set_current_image() and do_set_current_image() methods manage live preview generation from current latents at configurable intervals. Server lifecycle commands (restart, stop) are handled through a thread-safe signaling mechanism using threading.Event, with server_command as a property that sets a signal when assigned, and wait_for_server_command() blocking until a command is issued. The dict() method serializes the current state for API responses. A backward-compatible need_restart property maps to the server_command system.

Usage

Use this class through shared.state to check or control the progress of image generation, to implement skip/interrupt behavior in samplers, or to manage server restart/stop commands from the UI.

Code Reference

Source Location

• Repository: AUTOMATIC1111_Stable_diffusion_webui
• File: modules/shared_state.py
• Lines: 1-168

Signature

class State:
    skipped: bool
    interrupted: bool
    stopping_generation: bool
    job: str
    job_no: int
    job_count: int
    sampling_step: int
    sampling_steps: int
    current_latent: Tensor
    current_image: Image
    server_command: Optional[str]

    def __init__(self)
    def need_restart(self) -> bool  # property
    def wait_for_server_command(self, timeout: Optional[float] = None) -> Optional[str]
    def request_restart(self) -> None
    def skip(self) -> None
    def interrupt(self) -> None
    def stop_generating(self) -> None
    def nextjob(self) -> None
    def dict(self) -> dict
    def begin(self, job: str = "(unknown)") -> None
    def end(self) -> None
    def set_current_image(self) -> None
    def do_set_current_image(self) -> None
    def assign_current_image(self, image) -> None

Import

from modules.shared_state import State
# Typically accessed via:
from modules import shared
state = shared.state

I/O Contract

Inputs

Name	Type	Required	Description
job	str	No	Job name passed to begin(), defaults to "(unknown)".
timeout	float or None	No	Timeout in seconds for wait_for_server_command().
image	PIL.Image	Yes	Image to assign as the current live preview in assign_current_image().

Outputs

Name	Type	Description
dict()	dict	Serialized state containing skipped, interrupted, stopping_generation, job, job_count, job_timestamp, job_no, sampling_step, sampling_steps.
server_command	str or None	The server command received (e.g., "restart", "stop") from wait_for_server_command().

Usage Examples

from modules import shared

# Start a new generation job
shared.state.begin("txt2img")
shared.state.job_count = 4

for i in range(4):
    if shared.state.interrupted:
        break
    if shared.state.skipped:
        shared.state.skipped = False
        continue

    # ... perform generation ...
    shared.state.nextjob()

shared.state.end()

# Check state from API
status = shared.state.dict()
print(f"Progress: step {status['sampling_step']}/{status['sampling_steps']}")

Related Pages

• Principle:AUTOMATIC1111_Stable_diffusion_webui_Job_Management