Add optional Stream QA pre-processor

main.js

@@ -9,6 +9,7 @@ import { eventSource, event_types, streamingProcessor, saveSettingsDebounced } f
 import { Handlebars, hljs } from "../../../../lib.js";
 import { NAME } from "./common.js";
 import { loadSettings } from "./settings.js";
+import { processStreamBuffer } from "./qa_validator.js";
 
 
 // @ts-ignore: Hack to suppress IDE errors due to SillyTavern's
@@ -122,15 +123,17 @@ function runScript(script) {
 }
 
 function installStreamHook() {
-    if (hookToBeInstalled) {
+    if (hookToBeInstalled || settings.qaProfile !== "disabled") {
         const markerRegex = new RegExp(settings.markerRegex, "g");
         const partialMarkerRegex = new RegExp(settings.partialMarkerRegex);
 
         const processedIndices = new Set();
 
         const originalOnProgressStreaming = streamingProcessor.onProgressStreaming.bind(streamingProcessor);
 
-        function onProgressStreaming(messageId, text, isFinal) {
+        async function onProgressStreaming(messageId, text, isFinal) {
+            text = await processStreamBuffer(messageId, text, isFinal, settings);
+
             for (const match of text.matchAll(markerRegex)) {
                 if (!processedIndices.has(match.index)) {
                     // `match[0]` is the whole match, `match[1]` is the first capturing group, i.e. the script ID.
@@ -218,6 +221,11 @@ $(async () => {
         saveSettingsDebounced();
     });
 
+    $("#sorcery-qa-profile").val(settings.qaProfile).on("change", function () {
+        settings.qaProfile = String($(this).val());
+        saveSettingsDebounced();
+    });
+
     $("#sorcery-flash-icon").prop("checked", settings.flashIcon).change(function () {
         settings.flashIcon = this.checked;
         saveSettingsDebounced();

qa_validator.js

@@ -0,0 +1,144 @@
+// SPDX-License-Identifier: AGPL-3.0-or-later
+// Copyright (C) 2025  Sorcery Stream QA contributors
+
+import { chat, eventSource, event_types, Generate, setSendButtonState, stopGeneration } from "../../../../script.js";
+import { textgenerationwebui_settings } from "../../../textgen-settings.js";
+import { oai_settings } from "../../../openai.js";
+import { kai_settings } from "../../../kai-settings.js";
+import { getContext } from "../../../extensions.js";
+
+let rulesCache = null;
+/** @type {object | null} Active profile waiting for GENERATION_STOPPED trampoline */
+let pendingCorrectionProfile = null;
+/** @type {{ target: ReturnType<typeof getSamplerTarget>, temp: number, top_k: number } | null} */
+let originalParams = null;
+
+function getSamplerTarget() {
+    const api = getContext().mainApi;
+    if (api === "textgenerationwebui") {
+        return { settings: textgenerationwebui_settings, tempKey: "temp", topKKey: "top_k" };
+    }
+    if (api === "openai") {
+        return { settings: oai_settings, tempKey: "temp_openai", topKKey: "top_k_openai" };
+    }
+    if (api === "kobold") {
+        return { settings: kai_settings, tempKey: "temp", topKKey: "top_k" };
+    }
+    return null;
+}
+
+function compileRegexRules(regexRules) {
+    const compiled = {};
+    for (const [level, categories] of Object.entries(regexRules)) {
+        compiled[level] = {};
+        for (const [category, patterns] of Object.entries(categories)) {
+            compiled[level][category] = patterns.map((pattern) => new RegExp(pattern, "gi"));
+        }
+    }
+    return compiled;
+}
+
+async function loadRules() {
+    if (rulesCache) return rulesCache;
+    try {
+        const response = await fetch("/scripts/extensions/third-party/sorcery/rules.json");
+        const data = await response.json();
+        rulesCache = { profiles: data.profiles, compiled: compileRegexRules(data.regex_rules) };
+        return rulesCache;
+    } catch (e) {
+        console.error("[Sorcery QA] Failed to load rules.json", e);
+        return null;
+    }
+}
+
+function scrambleSamplers(profile) {
+    if (profile.temp_min == null || profile.top_k_min == null) return;
+    const target = getSamplerTarget();
+    if (!target) return;
+    const { settings, tempKey, topKKey } = target;
+    if (!originalParams) {
+        originalParams = {
+            target,
+            temp: settings[tempKey],
+            top_k: settings[topKKey],
+        };
+    }
+    const randomTemp = Math.random() * (profile.temp_max - profile.temp_min) + profile.temp_min;
+    const randomTopK = Math.floor(Math.random() * (profile.top_k_max - profile.top_k_min + 1)) + profile.top_k_min;
+    settings[tempKey] = parseFloat(randomTemp.toFixed(2));
+    settings[topKKey] = randomTopK;
+}
+
+function restoreSamplers() {
+    if (!originalParams) return;
+    const { target, temp, top_k } = originalParams;
+    target.settings[target.tempKey] = temp;
+    target.settings[target.topKKey] = top_k;
+    originalParams = null;
+}
+
+function resetQaState() {
+    pendingCorrectionProfile = null;
+    restoreSamplers();
+}
+
+function scheduleContinueAfterStop() {
+    queueMicrotask(() => {
+        setSendButtonState(true);
+        void Generate("continue");
+    });
+}
+
+eventSource.on(event_types.GENERATION_STOPPED, () => {
+    if (pendingCorrectionProfile) {
+        const profile = pendingCorrectionProfile;
+        pendingCorrectionProfile = null;
+        scrambleSamplers(profile);
+        scheduleContinueAfterStop();
+    } else {
+        resetQaState();
+    }
+});
+
+eventSource.on(event_types.GENERATION_ENDED, () => {
+    if (!pendingCorrectionProfile) {
+        resetQaState();
+    }
+});
+
+export async function processStreamBuffer(messageId, text, isFinal, settings) {
+    if (!settings.qaProfile || settings.qaProfile === "disabled") return text;
+
+    const rules = await loadRules();
+    if (!rules) return text;
+
+    const profile = rules.profiles[settings.qaProfile];
+    const strictness = profile && rules.compiled[profile.regex_strictness];
+    if (!profile || !strictness) return text;
+
+    let cleanedText = text;
+    let hitDetected = false;
+
+    for (const regex of strictness.looping ?? []) {
+        const next = cleanedText.replace(regex, "$1");
+        if (next !== cleanedText) {
+            cleanedText = next;
+            hitDetected = true;
+        }
+    }
+    for (const regex of strictness.slop ?? []) {
+        const next = cleanedText.replace(regex, "");
+        if (next !== cleanedText) {
+            cleanedText = next;
+            hitDetected = true;
+        }
+    }
+
+    if (hitDetected && !isFinal && !pendingCorrectionProfile) {
+        pendingCorrectionProfile = profile;
+        if (chat[messageId]) chat[messageId].mes = cleanedText.trim();
+        stopGeneration();
+    }
+
+    return cleanedText;
+}

rules.json

@@ -0,0 +1,24 @@
+{
+  "profiles": {
+    "me-me": { "temp_min": 0.70, "temp_max": 1.35, "top_k_min": 20, "top_k_max": 120, "regex_strictness": "high" },
+    "mild": { "temp_min": 0.10, "temp_max": 0.85, "top_k_min": 30, "top_k_max": 60, "regex_strictness": "low" },
+    "med": { "temp_min": 0.50, "temp_max": 1.10, "top_k_min": 40, "top_k_max": 90, "regex_strictness": "medium" },
+    "spicy-mild": { "temp_min": 0.85, "temp_max": 1.40, "top_k_min": 15, "top_k_max": 150, "regex_strictness": "medium" },
+    "spicy-spicy": { "temp_min": 1.20, "temp_max": 3.50, "top_k_min": 5, "top_k_max": 300, "regex_strictness": "high" },
+    "static": { "temp_min": null, "temp_max": null, "top_k_min": null, "top_k_max": null, "regex_strictness": "medium" }
+  },
+  "regex_rules": {
+    "high": {
+      "slop": ["(?:shivers down|testament to|tapestry of|dance of|smirked|it's a cold morning|not a luxury but|beacon of|delve into).+?", "\\b(delve|tapestry|beacon|testament|nestled|commendable|furthermore)\\b"],
+      "looping": ["(.{15,})\\1+"]
+    },
+    "medium": {
+      "slop": ["(?:testament to|tapestry of|delve into).+?", "\\b(delve|tapestry|testament)\\b"],
+      "looping": ["(.{30,})\\1+"]
+    },
+    "low": {
+      "slop": ["(?:testament to|tapestry of).+?"],
+      "looping": ["(.{50,})\\1+"]
+    }
+  }
+}

settings.html

@@ -38,6 +38,20 @@ <h3 class="margin0 flex1">
             </label>
         </div>
 
+        <div class="sorcery-qa-container" style="margin-top: 10px;">
+            <label for="sorcery-qa-profile" title="Enables real-time text cleaning and automatic sampler scrambling on slop or generation loops.">Stream QA Profile:</label>
+            <select id="sorcery-qa-profile" class="select_slate">
+                <option value="disabled">Disabled (Stock Sorcery)</option>
+                <option value="me-me">Me-Me</option>
+                <option value="mild">Mild</option>
+                <option value="med">Medium</option>
+                <option value="spicy-mild">Spicy-Mild</option>
+                <option value="spicy-spicy">Spicy-Spicy</option>
+                <option value="static">Static (filter only, no scramble)</option>
+            </select>
+            <small style="display: block; color: var(--grey); margin-top: 2px;">Stream quality auto-correction. Does not affect marker scripts.</small>
+        </div>
+
         <div id="sorcery-scripts"></div>
     </div>
 </div>

settings.js

@@ -36,6 +36,7 @@ Character: "Great idea! Let's do that." *Her eyes sparkle with excitement.*
 
 const DEFAULT_SETTINGS = {
     enabled: true,
+    qaProfile: "disabled",
     flashIcon: true,
     instructions: INSTRUCTIONS,
     markerRegex: "%\\[(\\d+)\\]",