Batch character enrichment by prompt size
This commit is contained in:
parent
4e1646936f
commit
68c327d39b
@ -338,6 +338,7 @@ var tests = new (string Name, Action Test)[]
|
||||
("Story Intelligence character review queues enrichment beside plot synthesis", StoryIntelligenceCharacterReviewQueuesEnrichmentBesidePlotSynthesis),
|
||||
("Story Intelligence character enrichment uses persisted SignalR progress", StoryIntelligenceCharacterEnrichmentUsesPersistedSignalRProgress),
|
||||
("Story Intelligence character enrichment preserves author fields", StoryIntelligenceCharacterEnrichmentPreservesAuthorFields),
|
||||
("Story Intelligence character enrichment batches by prompt size", StoryIntelligenceCharacterEnrichmentBatchesByPromptSize),
|
||||
("Character detail can queue single-character enrichment", CharacterDetailCanQueueSingleCharacterEnrichment),
|
||||
("Single-character enrichment is persisted and duplicate safe", SingleCharacterEnrichmentIsPersistedAndDuplicateSafe),
|
||||
("Character detail enrichment status uses scoped SignalR", CharacterDetailEnrichmentStatusUsesScopedSignalR),
|
||||
@ -1256,6 +1257,20 @@ static void StoryIntelligenceCharacterEnrichmentPreservesAuthorFields()
|
||||
Assert(migration.Contains("CREATE TABLE dbo.CharacterEnrichmentEvidence", StringComparison.Ordinal), "Generated facts should retain internal provenance/evidence.");
|
||||
}
|
||||
|
||||
static void StoryIntelligenceCharacterEnrichmentBatchesByPromptSize()
|
||||
{
|
||||
var root = PlotLineRoot();
|
||||
var service = File.ReadAllText(Path.Combine(root, "Services/CharacterEnrichmentService.cs"));
|
||||
|
||||
Assert(service.Contains("BuildPromptBatches", StringComparison.Ordinal), "Character enrichment should batch by prompt size rather than fixed alphabetical chunks only.");
|
||||
Assert(service.Contains("MainCharacterPromptCharacters", StringComparison.Ordinal), "Evidence-heavy characters should be processed singly before context is restrained.");
|
||||
Assert(service.Contains("RestrainSingleCharacterContext", StringComparison.Ordinal), "Single-character prompts should be explicitly restrained only when one character exceeds the safe limit.");
|
||||
Assert(service.Contains("SelectRepresentativeScenes", StringComparison.Ordinal), "Oversized single-character prompts should use deterministic whole-scene selection.");
|
||||
Assert(!service.Contains("character.Scenes.Take(60)", StringComparison.Ordinal), "Prompt construction should not silently take the first 60 scenes for every character.");
|
||||
Assert(service.Contains("OriginalPromptCharacters", StringComparison.Ordinal)
|
||||
&& service.Contains("IncludedScenes", StringComparison.Ordinal), "Restrained character context should be logged with auditable scene counts.");
|
||||
}
|
||||
|
||||
static void CharacterDetailCanQueueSingleCharacterEnrichment()
|
||||
{
|
||||
var root = PlotLineRoot();
|
||||
|
||||
@ -33,6 +33,8 @@ public sealed class CharacterEnrichmentService(
|
||||
private const string PromptVersion = "Character-Enrichment-V1";
|
||||
private const int BatchSize = 8;
|
||||
private const int MaxPromptCharacters = 350_000;
|
||||
private const int MainCharacterPromptCharacters = 175_000;
|
||||
private const int MinimumSingleCharacterScenes = 8;
|
||||
private const decimal SummaryMinimumConfidence = 0.65m;
|
||||
private const decimal FactMinimumConfidence = 0.9m;
|
||||
private static readonly JsonSerializerOptions JsonOptions = new(JsonSerializerDefaults.Web)
|
||||
@ -134,29 +136,22 @@ public sealed class CharacterEnrichmentService(
|
||||
characters.Count);
|
||||
|
||||
await PublishProgressAsync(run, characters.Count, 0, "Analysing characters", "Analysing characters...", cancellationToken);
|
||||
foreach (var batch in characters.Chunk(BatchSize))
|
||||
foreach (var batchContext in BuildPromptBatches(context, characters, run))
|
||||
{
|
||||
cancellationToken.ThrowIfCancellationRequested();
|
||||
var batchContext = new CharacterEnrichmentContext
|
||||
{
|
||||
ProjectID = context.ProjectID,
|
||||
BookID = context.BookID,
|
||||
BookTitle = context.BookTitle,
|
||||
StoryEra = context.StoryEra,
|
||||
SeriesStartDate = context.SeriesStartDate,
|
||||
Characters = batch
|
||||
};
|
||||
var prompt = BuildPrompt(batchContext);
|
||||
if (prompt.Length > MaxPromptCharacters)
|
||||
{
|
||||
throw new InvalidOperationException("Character enrichment prompt would exceed the configured safe context limit. No manuscript text was silently truncated.");
|
||||
throw new InvalidOperationException($"Character enrichment prompt would exceed the configured safe context limit. PromptCharacters={prompt.Length:N0} Limit={MaxPromptCharacters:N0}. No manuscript text was silently truncated.");
|
||||
}
|
||||
|
||||
logger.LogInformation(
|
||||
"Character enrichment batch starting. RunID={RunID} CharacterID={CharacterID} BatchCharacters={BatchCharacters} Processed={Processed}/{Total}",
|
||||
"Character enrichment batch starting. RunID={RunID} CharacterID={CharacterID} BatchCharacters={BatchCharacters} PromptCharacters={PromptCharacters} SceneCount={SceneCount} Processed={Processed}/{Total}",
|
||||
run.CharacterEnrichmentRunID,
|
||||
run.CharacterID,
|
||||
batch.Length,
|
||||
batchContext.Characters.Count,
|
||||
prompt.Length,
|
||||
batchContext.Characters.Sum(character => character.Scenes.Count),
|
||||
results.Count,
|
||||
characters.Count);
|
||||
await PublishProgressAsync(run, characters.Count, results.Count, "Creating summaries and extracting character details", "Creating summaries and extracting character details...", cancellationToken);
|
||||
@ -172,7 +167,7 @@ public sealed class CharacterEnrichmentService(
|
||||
inputTokens += clientResult.InputTokens ?? 0;
|
||||
outputTokens += clientResult.OutputTokens ?? 0;
|
||||
var parsed = CharacterEnrichmentResult.FromJson(ExtractOutputText(clientResult.RawResponseText));
|
||||
results.AddRange(FilterResults(parsed.Characters, batch.Select(character => character.CharacterID).ToHashSet()));
|
||||
results.AddRange(FilterResults(parsed.Characters, batchContext.Characters.Select(character => character.CharacterID).ToHashSet()));
|
||||
await PublishProgressAsync(run, characters.Count, Math.Min(characters.Count, results.Count), "Saving character analysis", "Saving character analysis...", cancellationToken);
|
||||
}
|
||||
|
||||
@ -213,6 +208,186 @@ public sealed class CharacterEnrichmentService(
|
||||
public StoryIntelligenceResponseContract BuildResponseContractForTest()
|
||||
=> CharacterEnrichmentStructuredOutputSchema.Contract;
|
||||
|
||||
private IReadOnlyList<CharacterEnrichmentContext> BuildPromptBatches(
|
||||
CharacterEnrichmentContext context,
|
||||
IReadOnlyList<CharacterEnrichmentCharacterContext> characters,
|
||||
CharacterEnrichmentRun run)
|
||||
{
|
||||
var batches = new List<CharacterEnrichmentContext>();
|
||||
var current = new List<CharacterEnrichmentCharacterContext>();
|
||||
|
||||
foreach (var character in characters)
|
||||
{
|
||||
var single = BuildContextForCharacters(context, [character]);
|
||||
var singlePromptLength = BuildPrompt(single).Length;
|
||||
if (singlePromptLength > MaxPromptCharacters)
|
||||
{
|
||||
var restrained = RestrainSingleCharacterContext(context, character, run, singlePromptLength);
|
||||
AddCurrentBatch();
|
||||
batches.Add(restrained);
|
||||
continue;
|
||||
}
|
||||
|
||||
if (singlePromptLength >= MainCharacterPromptCharacters)
|
||||
{
|
||||
AddCurrentBatch();
|
||||
batches.Add(single);
|
||||
continue;
|
||||
}
|
||||
|
||||
var candidate = BuildContextForCharacters(context, current.Concat([character]).ToList());
|
||||
if (current.Count > 0 && (current.Count >= BatchSize || BuildPrompt(candidate).Length > MaxPromptCharacters))
|
||||
{
|
||||
AddCurrentBatch();
|
||||
}
|
||||
|
||||
current.Add(character);
|
||||
}
|
||||
|
||||
AddCurrentBatch();
|
||||
return batches;
|
||||
|
||||
void AddCurrentBatch()
|
||||
{
|
||||
if (current.Count == 0)
|
||||
{
|
||||
return;
|
||||
}
|
||||
|
||||
batches.Add(BuildContextForCharacters(context, current.ToList()));
|
||||
current.Clear();
|
||||
}
|
||||
}
|
||||
|
||||
private CharacterEnrichmentContext RestrainSingleCharacterContext(
|
||||
CharacterEnrichmentContext context,
|
||||
CharacterEnrichmentCharacterContext character,
|
||||
CharacterEnrichmentRun run,
|
||||
int originalPromptLength)
|
||||
{
|
||||
var scenes = SelectRepresentativeScenes(character.Scenes, character.Scenes.Count).ToList();
|
||||
while (scenes.Count > MinimumSingleCharacterScenes)
|
||||
{
|
||||
var candidate = BuildContextForCharacters(context, [CopyCharacter(character, scenes)]);
|
||||
if (BuildPrompt(candidate).Length <= MaxPromptCharacters)
|
||||
{
|
||||
logger.LogWarning(
|
||||
"Character enrichment constrained an evidence-heavy character to fit the safe context limit. RunID={RunID} ProjectID={ProjectID} BookID={BookID} CharacterID={CharacterID} CharacterName={CharacterName} OriginalPromptCharacters={OriginalPromptCharacters} FinalPromptCharacters={FinalPromptCharacters} OriginalScenes={OriginalScenes} IncludedScenes={IncludedScenes} Limit={Limit}",
|
||||
run.CharacterEnrichmentRunID,
|
||||
run.ProjectID,
|
||||
run.BookID,
|
||||
character.CharacterID,
|
||||
character.CharacterName,
|
||||
originalPromptLength,
|
||||
BuildPrompt(candidate).Length,
|
||||
character.Scenes.Count,
|
||||
scenes.Count,
|
||||
MaxPromptCharacters);
|
||||
return candidate;
|
||||
}
|
||||
|
||||
scenes = SelectRepresentativeScenes(scenes, Math.Max(MinimumSingleCharacterScenes, scenes.Count - 4)).ToList();
|
||||
}
|
||||
|
||||
var minimum = BuildContextForCharacters(context, [CopyCharacter(character, scenes)]);
|
||||
var minimumPromptLength = BuildPrompt(minimum).Length;
|
||||
if (minimumPromptLength <= MaxPromptCharacters)
|
||||
{
|
||||
logger.LogWarning(
|
||||
"Character enrichment constrained an evidence-heavy character to the minimum scene set. RunID={RunID} ProjectID={ProjectID} BookID={BookID} CharacterID={CharacterID} CharacterName={CharacterName} OriginalPromptCharacters={OriginalPromptCharacters} FinalPromptCharacters={FinalPromptCharacters} OriginalScenes={OriginalScenes} IncludedScenes={IncludedScenes} Limit={Limit}",
|
||||
run.CharacterEnrichmentRunID,
|
||||
run.ProjectID,
|
||||
run.BookID,
|
||||
character.CharacterID,
|
||||
character.CharacterName,
|
||||
originalPromptLength,
|
||||
minimumPromptLength,
|
||||
character.Scenes.Count,
|
||||
scenes.Count,
|
||||
MaxPromptCharacters);
|
||||
return minimum;
|
||||
}
|
||||
|
||||
throw new InvalidOperationException($"Character enrichment prompt would exceed the configured safe context limit for character {character.CharacterID} even when limited to {scenes.Count:N0} whole scenes. PromptCharacters={minimumPromptLength:N0} Limit={MaxPromptCharacters:N0}. No manuscript text was silently truncated.");
|
||||
}
|
||||
|
||||
private static CharacterEnrichmentContext BuildContextForCharacters(
|
||||
CharacterEnrichmentContext context,
|
||||
IReadOnlyList<CharacterEnrichmentCharacterContext> characters)
|
||||
=> new()
|
||||
{
|
||||
ProjectID = context.ProjectID,
|
||||
BookID = context.BookID,
|
||||
BookTitle = context.BookTitle,
|
||||
StoryEra = context.StoryEra,
|
||||
SeriesStartDate = context.SeriesStartDate,
|
||||
Characters = characters
|
||||
};
|
||||
|
||||
private static CharacterEnrichmentCharacterContext CopyCharacter(
|
||||
CharacterEnrichmentCharacterContext character,
|
||||
IReadOnlyList<CharacterEnrichmentSceneContext> scenes)
|
||||
=> new()
|
||||
{
|
||||
CharacterID = character.CharacterID,
|
||||
CharacterName = character.CharacterName,
|
||||
BirthDate = character.BirthDate,
|
||||
AgeAtSeriesStart = character.AgeAtSeriesStart,
|
||||
Height = character.Height,
|
||||
EyeColour = character.EyeColour,
|
||||
DefaultDescription = character.DefaultDescription,
|
||||
Aliases = character.Aliases,
|
||||
Scenes = scenes
|
||||
};
|
||||
|
||||
private static IReadOnlyList<CharacterEnrichmentSceneContext> SelectRepresentativeScenes(
|
||||
IReadOnlyList<CharacterEnrichmentSceneContext> scenes,
|
||||
int maxScenes)
|
||||
{
|
||||
if (scenes.Count <= maxScenes)
|
||||
{
|
||||
return scenes;
|
||||
}
|
||||
|
||||
maxScenes = Math.Clamp(maxScenes, 1, scenes.Count);
|
||||
var selected = new SortedDictionary<int, CharacterEnrichmentSceneContext>();
|
||||
var firstCount = Math.Min(scenes.Count, Math.Min(12, Math.Max(1, maxScenes / 3)));
|
||||
var lastCount = Math.Min(scenes.Count - firstCount, Math.Min(8, Math.Max(0, maxScenes / 4)));
|
||||
|
||||
for (var index = 0; index < firstCount; index++)
|
||||
{
|
||||
selected[index] = scenes[index];
|
||||
}
|
||||
|
||||
for (var index = scenes.Count - lastCount; index < scenes.Count; index++)
|
||||
{
|
||||
if (index >= 0)
|
||||
{
|
||||
selected[index] = scenes[index];
|
||||
}
|
||||
}
|
||||
|
||||
var remaining = maxScenes - selected.Count;
|
||||
if (remaining > 0)
|
||||
{
|
||||
var start = firstCount;
|
||||
var endExclusive = scenes.Count - lastCount;
|
||||
var span = Math.Max(0, endExclusive - start);
|
||||
for (var slot = 1; slot <= remaining && span > 0; slot++)
|
||||
{
|
||||
var offset = (int)Math.Round(slot * (span - 1) / (double)(remaining + 1), MidpointRounding.AwayFromZero);
|
||||
selected.TryAdd(start + offset, scenes[start + offset]);
|
||||
}
|
||||
}
|
||||
|
||||
for (var index = 0; selected.Count < maxScenes && index < scenes.Count; index++)
|
||||
{
|
||||
selected.TryAdd(index, scenes[index]);
|
||||
}
|
||||
|
||||
return selected.Values.ToList();
|
||||
}
|
||||
|
||||
private async Task PublishProgressAsync(CharacterEnrichmentRun run, int total, int processed, string stage, string message, CancellationToken cancellationToken)
|
||||
{
|
||||
await repository.UpdateProgressAsync(new CharacterEnrichmentProgressUpdate
|
||||
@ -288,7 +463,7 @@ public sealed class CharacterEnrichmentService(
|
||||
builder.AppendLine($"Existing eye colour: {character.EyeColour}");
|
||||
builder.AppendLine($"Existing description: {character.DefaultDescription}");
|
||||
builder.AppendLine("[SCENES]");
|
||||
foreach (var scene in character.Scenes.Take(60))
|
||||
foreach (var scene in character.Scenes)
|
||||
{
|
||||
builder.AppendLine($"SceneID: {scene.SceneID}; Chapter {scene.ChapterNumber:0.##} {scene.ChapterTitle}; Scene {scene.SceneNumber:0.##} {scene.SceneTitle}");
|
||||
builder.AppendLine("<text>");
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user