Module de.seggebaeing.sqlanalyzer

Package de.seggebaeing.sqlanalyzer.logic.util.thread

Class GenerationThread

java.lang.Object

java.lang.Thread

de.seggebaeing.sqlanalyzer.logic.util.thread.WorkerThread

de.seggebaeing.sqlanalyzer.logic.util.thread.GenerationThread

All Implemented Interfaces:

Runnable

public class GenerationThread
extends WorkerThread

Worker that generates SQL queries by executing Prompt × LLM combinations, optionally repeated, using a fixed-size pool of subworkers.

Responsibilities:

• Schedules jobs for each Prompt × LLM × repetition and computes a temperature value interpolated between each LLM’s min/max range.
• Respects provider rate limits via PromptAuthorizer: waits before prompting and, on RateLimitException, reports and registers the retry Instant.
• Emits per-LLM progress callbacks (startedProgress/finishedProgress).
• Collects results in a thread-safe set of GeneratedQuery and strips Markdown fences from returned SQL.
• Honors interruption (cancels subworkers) and, on full success, invokes the supplied signalDone callback.

Threading: uses a poolSize-bounded ExecutorService. The result set is synchronized; callers should only read it after the thread finished.

Since:

1.0

Author:

Felix Seggebäing

Nested Class Summary

Nested classes/interfaces inherited from class java.lang.Thread

Thread.Builder, Thread.State, Thread.UncaughtExceptionHandler

Field Summary

Fields inherited from class de.seggebaeing.sqlanalyzer.logic.util.thread.WorkerThread

poolSize, signalDone

Fields inherited from class java.lang.Thread

MAX_PRIORITY, MIN_PRIORITY, NORM_PRIORITY

Constructor Summary

Constructors

Constructor	Description
GenerationThread(int poolSize, int repetitionCount, Collection<LLM> llms, Collection<Prompt> prompts, Runnable signalDone, Consumer<LLM> startedProgress, Consumer<LLM> finishedProgress, BiConsumer<LLM,Instant> rateLimitReporter)	Constructs a generation worker thread.

Method Summary

Modifier and Type	Method	Description
Set<GeneratedQuery>	getResult()	Returns the set of generated queries produced by this worker.
void	run()	Executes generation by dispatching Prompt × LLM × repetition jobs to a fixed thread pool.

Methods inherited from class java.lang.Thread

activeCount, checkAccess, clone, countStackFrames, currentThread, dumpStack, enumerate, getAllStackTraces, getContextClassLoader, getDefaultUncaughtExceptionHandler, getId, getName, getPriority, getStackTrace, getState, getThreadGroup, getUncaughtExceptionHandler, holdsLock, interrupt, interrupted, isAlive, isDaemon, isInterrupted, isVirtual, join, join, join, join, ofPlatform, ofVirtual, onSpinWait, resume, setContextClassLoader, setDaemon, setDefaultUncaughtExceptionHandler, setName, setPriority, setUncaughtExceptionHandler, sleep, sleep, sleep, start, startVirtualThread, stop, suspend, threadId, toString, yield

Methods inherited from class java.lang.Object

equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Details

GenerationThread

public GenerationThread(int poolSize,
 int repetitionCount,
 Collection<LLM> llms,
 Collection<Prompt> prompts,
 Runnable signalDone,
 Consumer<LLM> startedProgress,
 Consumer<LLM> finishedProgress,
 BiConsumer<LLM,Instant> rateLimitReporter)

Constructs a generation worker thread.

Parameters:

poolSize - number of parallel subworkers to use

repetitionCount - number of repetitions per Prompt × LLM

llms - LLMs to use for generation

prompts - prompts to be combined with each LLM

signalDone - callback invoked on successful completion (may be null)

startedProgress - callback invoked when a subworker starts work for an LLM

finishedProgress - callback invoked when a subworker finishes for an LLM

rateLimitReporter - callback to report rate-limit retry instants per LLM

Implementation Note:

The thread name is assigned as "Generation-Worker-<n>" using an atomic counter.

Method Details

run

public void run()

Executes generation by dispatching Prompt × LLM × repetition jobs to a fixed thread pool.

Creates a ExecutorService with poolSize, submits one task per combination, then shuts down the pool and waits for completion. On full success (no timeout/interruption), invokes signalDone. Honors interruption by cancelling subworkers.

Specified by:

run in interface Runnable

Specified by:

run in class WorkerThread

Implementation Note:

Results are accumulated in a synchronized set; read them only after the thread finishes.

getResult

public Set<GeneratedQuery> getResult()

Returns the set of generated queries produced by this worker.

Behavior is undefined if called before the thread has completed successfully.

Specified by:

getResult in class WorkerThread

Returns:

the (synchronized) result set of GeneratedQuery

Implementation Note:

The returned set is a Collections.synchronizedSet; synchronize on it when iterating.