@pie-players/pie-assessment-toolkit 0.3.44 → 0.3.45

package/dist/services/tts/highlight-pipeline/readable-region.js

@@ -0,0 +1,19 @@
+const READABLE_REGION_SELECTOR = [
+    "[data-pie-tts-region]",
+    "[data-catalog-region]",
+    "li",
+    "p",
+    "label",
+    "section",
+    "[role='group']",
+    "[role='region']",
+    "div",
+].join(",");
+export const resolveReadableRegion = (sourceElement, contentRoot) => {
+    const region = sourceElement.closest(READABLE_REGION_SELECTOR);
+    if (region && contentRoot.contains(region)) {
+        return region;
+    }
+    return sourceElement;
+};
+//# sourceMappingURL=readable-region.js.map

package/dist/services/tts/highlight-pipeline/readable-region.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"readable-region.js","sourceRoot":"","sources":["../../../../src/services/tts/highlight-pipeline/readable-region.ts"],"names":[],"mappings":"AAAA,MAAM,wBAAwB,GAAG;IAChC,uBAAuB;IACvB,uBAAuB;IACvB,IAAI;IACJ,GAAG;IACH,OAAO;IACP,SAAS;IACT,gBAAgB;IAChB,iBAAiB;IACjB,KAAK;CACL,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AAEZ,MAAM,CAAC,MAAM,qBAAqB,GAAG,CACpC,aAAsB,EACtB,WAAoB,EACV,EAAE;IACZ,MAAM,MAAM,GAAG,aAAa,CAAC,OAAO,CAAC,wBAAwB,CAAC,CAAC;IAC/D,IAAI,MAAM,IAAI,WAAW,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QAC5C,OAAO,MAAM,CAAC;IACf,CAAC;IACD,OAAO,aAAa,CAAC;AACtB,CAAC,CAAC","sourcesContent":["const READABLE_REGION_SELECTOR = [\n\t\"[data-pie-tts-region]\",\n\t\"[data-catalog-region]\",\n\t\"li\",\n\t\"p\",\n\t\"label\",\n\t\"section\",\n\t\"[role='group']\",\n\t\"[role='region']\",\n\t\"div\",\n].join(\",\");\n\nexport const resolveReadableRegion = (\n\tsourceElement: Element,\n\tcontentRoot: Element,\n): Element => {\n\tconst region = sourceElement.closest(READABLE_REGION_SELECTOR);\n\tif (region && contentRoot.contains(region)) {\n\t\treturn region;\n\t}\n\treturn sourceElement;\n};\n"]}

package/dist/services/tts/highlight-pipeline/rendered-math-target-resolver.d.ts

@@ -0,0 +1,6 @@
+import type { SemanticHighlightTarget, RenderableHighlightTarget } from "./types.js";
+export type MathHighlightCapability = "token" | "expression";
+export declare const classifyMathHighlightCapability: (mathElement: Element) => MathHighlightCapability;
+export type RenderedMathTargetResolver = (target: SemanticHighlightTarget) => RenderableHighlightTarget | null;
+export declare const resolveRenderedMathTarget: RenderedMathTargetResolver;
+export declare const createRenderedMathTargetResolver: () => RenderedMathTargetResolver;

package/dist/services/tts/highlight-pipeline/rendered-math-target-resolver.js

@@ -0,0 +1,198 @@
+import { isNodeHiddenForTTS } from "../text-processing.js";
+const MATHML_LEAF_TO_MATHJAX = new Map([
+    ["mi", "mjx-mi"],
+    ["mn", "mjx-mn"],
+    ["mo", "mjx-mo"],
+    ["mtext", "mjx-mtext"],
+]);
+const MATHML_LEAF_SELECTOR = "mi,mn,mo,mtext";
+const MATHJAX_LEAF_SELECTOR = "mjx-mi,mjx-mn,mjx-mo,mjx-mtext";
+// 2D layouts MathJax renders with extra, glyph-only nodes (radical surds,
+// fraction bars drawn as borders, table scaffolding) that have no source-MathML
+// leaf counterpart. They break the 1:1 token correspondence we rely on below,
+// so we refuse positional mapping inside them and fall back to the whole
+// expression instead of guessing.
+const UNSUPPORTED_MATHJAX_LAYOUT_SELECTOR = [
+    "mjx-mfrac",
+    "mjx-msqrt",
+    "mjx-mroot",
+    "mjx-mtable",
+    "mjx-munder",
+    "mjx-mover",
+    "mjx-munderover",
+].join(",");
+const nearestElement = (target) => {
+    if (target.type === "text-range")
+        return target.node.parentElement;
+    return target.element;
+};
+const owningMathJaxContainer = (element) => {
+    let current = element;
+    while (current) {
+        if (current.localName?.toLowerCase() === "mjx-container")
+            return current;
+        current = current.parentElement;
+    }
+    return null;
+};
+const isVisibleNativeTarget = (element) => {
+    if (!element)
+        return false;
+    if (owningMathJaxContainer(element))
+        return false;
+    return !isNodeHiddenForTTS(element);
+};
+const hasVisibleGeometry = (element) => {
+    const rect = element.getBoundingClientRect?.();
+    return Boolean(rect && rect.width > 0 && rect.height > 0);
+};
+const closestMathElement = (element) => {
+    let current = element;
+    while (current) {
+        if (current.localName?.toLowerCase() === "math")
+            return current;
+        current = current.parentElement;
+    }
+    return null;
+};
+// MathJax CHTML renders each source-MathML token (mi/mn/mo/mtext) as exactly
+// one mjx-* token in the same document order — including invisible operators,
+// which it keeps as zero-size nodes. The original <math> survives, clipped, in
+// mjx-assistive-mml, so its glyph text nodes are not visible and the visible
+// CHTML tokens carry no readable textContent (glyphs come from web fonts).
+// We therefore map a source token to its rendered glyph by *structural
+// position* rather than text, but only after proving the two leaf lists are
+// isomorphic (equal length, matching token type at every index). When the
+// structure diverges we refuse to guess and let the caller fall back to the
+// whole expression. "False positives are worse than coarse highlighting."
+const tokensAreIsomorphic = (mathLeaves, chtmlLeaves) => {
+    if (mathLeaves.length === 0)
+        return false;
+    if (mathLeaves.length !== chtmlLeaves.length)
+        return false;
+    for (let index = 0; index < mathLeaves.length; index += 1) {
+        const expected = MATHML_LEAF_TO_MATHJAX.get(mathLeaves[index].localName.toLowerCase());
+        if (!expected)
+            return false;
+        if (chtmlLeaves[index].localName.toLowerCase() !== expected)
+            return false;
+    }
+    return true;
+};
+const resolveMathJaxTokenByPosition = (semanticElement, container) => {
+    if (!semanticElement)
+        return null;
+    if (!MATHML_LEAF_TO_MATHJAX.has(semanticElement.localName.toLowerCase()))
+        return null;
+    if (container.querySelector(UNSUPPORTED_MATHJAX_LAYOUT_SELECTOR))
+        return null;
+    const sourceMath = closestMathElement(semanticElement);
+    const renderedMath = container.querySelector("mjx-math");
+    if (!sourceMath || !renderedMath)
+        return null;
+    const mathLeaves = Array.from(sourceMath.querySelectorAll(MATHML_LEAF_SELECTOR));
+    const index = mathLeaves.indexOf(semanticElement);
+    if (index < 0)
+        return null;
+    const chtmlLeaves = Array.from(renderedMath.querySelectorAll(MATHJAX_LEAF_SELECTOR));
+    if (!tokensAreIsomorphic(mathLeaves, chtmlLeaves))
+        return null;
+    const token = chtmlLeaves[index];
+    // The matched token must be a real, visible glyph. We deliberately do NOT
+    // run isNodeHiddenForTTS here: MathJax marks its visual CHTML
+    // (mjx-math) with aria-hidden="true" because the accessible tree lives in
+    // the clipped mjx-assistive-mml. aria-hidden is correct for screen readers
+    // but irrelevant for a visual highlight — the glyph is painted on screen.
+    // Visible geometry is the right test: it rejects display:none nodes and
+    // zero-size invisible operators (e.g. U+2062 invisible-times, which is
+    // never spoken anyway) without discarding the real, on-screen tokens.
+    if (!token || !hasVisibleGeometry(token))
+        return null;
+    return token;
+};
+export const classifyMathHighlightCapability = (mathElement) => {
+    const container = owningMathJaxContainer(mathElement);
+    if (container) {
+        // 2D MathJax layouts add glyph-only nodes with no source-MathML
+        // counterpart, so positional mapping is unsafe — commit to expression.
+        if (container.querySelector(UNSUPPORTED_MATHJAX_LAYOUT_SELECTOR)) {
+            return "expression";
+        }
+        const sourceMath = closestMathElement(mathElement) ?? mathElement;
+        const renderedMath = container.querySelector("mjx-math");
+        if (!renderedMath)
+            return "expression";
+        const mathLeaves = Array.from(sourceMath.querySelectorAll(MATHML_LEAF_SELECTOR));
+        const chtmlLeaves = Array.from(renderedMath.querySelectorAll(MATHJAX_LEAF_SELECTOR));
+        if (!tokensAreIsomorphic(mathLeaves, chtmlLeaves))
+            return "expression";
+        // At least one rendered token must paint a real glyph; otherwise a
+        // per-token highlight would have nothing visible to land on.
+        return chtmlLeaves.some((leaf) => hasVisibleGeometry(leaf))
+            ? "token"
+            : "expression";
+    }
+    // Native MathML: every leaf is a real, browser-positioned element, so
+    // per-token highlighting is reliable even inside 2D layouts (fractions,
+    // radicals). It is token-capable as long as a visible leaf exists.
+    const sourceMath = closestMathElement(mathElement) ?? mathElement;
+    const leaves = Array.from(sourceMath.querySelectorAll(MATHML_LEAF_SELECTOR));
+    return leaves.some((leaf) => !isNodeHiddenForTTS(leaf))
+        ? "token"
+        : "expression";
+};
+const resolveRenderedMathTargetUncached = (target) => {
+    const element = nearestElement(target);
+    const mathJaxContainer = owningMathJaxContainer(element);
+    if (mathJaxContainer) {
+        const visibleToken = resolveMathJaxTokenByPosition(element, mathJaxContainer);
+        if (visibleToken) {
+            return {
+                type: "element",
+                quality: "semantic-token",
+                element: visibleToken,
+            };
+        }
+        return {
+            type: "element",
+            quality: "expression",
+            element: mathJaxContainer,
+        };
+    }
+    if (target.type === "text-range") {
+        if (!isVisibleNativeTarget(target.node.parentElement))
+            return null;
+        return {
+            type: "text-range",
+            quality: "semantic-token",
+            node: target.node,
+            startOffset: target.startOffset,
+            endOffset: target.endOffset,
+        };
+    }
+    if (!isVisibleNativeTarget(target.element))
+        return null;
+    return {
+        type: "element",
+        quality: target.quality === "region-fallback" ? "expression" : "semantic-token",
+        element: target.element,
+    };
+};
+export const resolveRenderedMathTarget = resolveRenderedMathTargetUncached;
+export const createRenderedMathTargetResolver = () => {
+    const mathJaxTargetCache = new WeakMap();
+    return (target) => {
+        const element = nearestElement(target);
+        const mathJaxContainer = owningMathJaxContainer(element);
+        if (!element || !mathJaxContainer) {
+            return resolveRenderedMathTargetUncached(target);
+        }
+        if (mathJaxTargetCache.has(element)) {
+            return mathJaxTargetCache.get(element) ?? null;
+        }
+        const rendered = resolveRenderedMathTargetUncached(target);
+        mathJaxTargetCache.set(element, rendered);
+        return rendered;
+    };
+};
+//# sourceMappingURL=rendered-math-target-resolver.js.map

package/dist/services/tts/highlight-pipeline/rendered-math-target-resolver.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"rendered-math-target-resolver.js","sourceRoot":"","sources":["../../../../src/services/tts/highlight-pipeline/rendered-math-target-resolver.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,kBAAkB,EAAE,MAAM,uBAAuB,CAAC;AAM3D,MAAM,sBAAsB,GAAG,IAAI,GAAG,CAAC;IACtC,CAAC,IAAI,EAAE,QAAQ,CAAC;IAChB,CAAC,IAAI,EAAE,QAAQ,CAAC;IAChB,CAAC,IAAI,EAAE,QAAQ,CAAC;IAChB,CAAC,OAAO,EAAE,WAAW,CAAC;CACtB,CAAC,CAAC;AAEH,MAAM,oBAAoB,GAAG,gBAAgB,CAAC;AAC9C,MAAM,qBAAqB,GAAG,gCAAgC,CAAC;AAE/D,0EAA0E;AAC1E,gFAAgF;AAChF,8EAA8E;AAC9E,yEAAyE;AACzE,kCAAkC;AAClC,MAAM,mCAAmC,GAAG;IAC3C,WAAW;IACX,WAAW;IACX,WAAW;IACX,YAAY;IACZ,YAAY;IACZ,WAAW;IACX,gBAAgB;CAChB,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AAEZ,MAAM,cAAc,GAAG,CAAC,MAA+B,EAAkB,EAAE;IAC1E,IAAI,MAAM,CAAC,IAAI,KAAK,YAAY;QAAE,OAAO,MAAM,CAAC,IAAI,CAAC,aAAa,CAAC;IACnE,OAAO,MAAM,CAAC,OAAO,CAAC;AACvB,CAAC,CAAC;AAEF,MAAM,sBAAsB,GAAG,CAAC,OAAuB,EAAkB,EAAE;IAC1E,IAAI,OAAO,GAAmB,OAAO,CAAC;IACtC,OAAO,OAAO,EAAE,CAAC;QAChB,IAAI,OAAO,CAAC,SAAS,EAAE,WAAW,EAAE,KAAK,eAAe;YAAE,OAAO,OAAO,CAAC;QACzE,OAAO,GAAG,OAAO,CAAC,aAAa,CAAC;IACjC,CAAC;IACD,OAAO,IAAI,CAAC;AACb,CAAC,CAAC;AAEF,MAAM,qBAAqB,GAAG,CAAC,OAAuB,EAAW,EAAE;IAClE,IAAI,CAAC,OAAO;QAAE,OAAO,KAAK,CAAC;IAC3B,IAAI,sBAAsB,CAAC,OAAO,CAAC;QAAE,OAAO,KAAK,CAAC;IAClD,OAAO,CAAC,kBAAkB,CAAC,OAAO,CAAC,CAAC;AACrC,CAAC,CAAC;AAEF,MAAM,kBAAkB,GAAG,CAAC,OAAgB,EAAW,EAAE;IACxD,MAAM,IAAI,GAAI,OAAuB,CAAC,qBAAqB,EAAE,EAAE,CAAC;IAChE,OAAO,OAAO,CAAC,IAAI,IAAI,IAAI,CAAC,KAAK,GAAG,CAAC,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;AAC3D,CAAC,CAAC;AAEF,MAAM,kBAAkB,GAAG,CAAC,OAAuB,EAAkB,EAAE;IACtE,IAAI,OAAO,GAAmB,OAAO,CAAC;IACtC,OAAO,OAAO,EAAE,CAAC;QAChB,IAAI,OAAO,CAAC,SAAS,EAAE,WAAW,EAAE,KAAK,MAAM;YAAE,OAAO,OAAO,CAAC;QAChE,OAAO,GAAG,OAAO,CAAC,aAAa,CAAC;IACjC,CAAC;IACD,OAAO,IAAI,CAAC;AACb,CAAC,CAAC;AAEF,6EAA6E;AAC7E,8EAA8E;AAC9E,+EAA+E;AAC/E,6EAA6E;AAC7E,2EAA2E;AAC3E,uEAAuE;AACvE,4EAA4E;AAC5E,0EAA0E;AAC1E,4EAA4E;AAC5E,0EAA0E;AAC1E,MAAM,mBAAmB,GAAG,CAC3B,UAAqB,EACrB,WAAsB,EACZ,EAAE;IACZ,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,KAAK,CAAC;IAC1C,IAAI,UAAU,CAAC,MAAM,KAAK,WAAW,CAAC,MAAM;QAAE,OAAO,KAAK,CAAC;IAC3D,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,UAAU,CAAC,MAAM,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;QAC3D,MAAM,QAAQ,GAAG,sBAAsB,CAAC,GAAG,CAC1C,UAAU,CAAC,KAAK,CAAC,CAAC,SAAS,CAAC,WAAW,EAAE,CACzC,CAAC;QACF,IAAI,CAAC,QAAQ;YAAE,OAAO,KAAK,CAAC;QAC5B,IAAI,WAAW,CAAC,KAAK,CAAC,CAAC,SAAS,CAAC,WAAW,EAAE,KAAK,QAAQ;YAAE,OAAO,KAAK,CAAC;IAC3E,CAAC;IACD,OAAO,IAAI,CAAC;AACb,CAAC,CAAC;AAEF,MAAM,6BAA6B,GAAG,CACrC,eAA+B,EAC/B,SAAkB,EACD,EAAE;IACnB,IAAI,CAAC,eAAe;QAAE,OAAO,IAAI,CAAC;IAClC,IAAI,CAAC,sBAAsB,CAAC,GAAG,CAAC,eAAe,CAAC,SAAS,CAAC,WAAW,EAAE,CAAC;QACvE,OAAO,IAAI,CAAC;IACb,IAAI,SAAS,CAAC,aAAa,CAAC,mCAAmC,CAAC;QAAE,OAAO,IAAI,CAAC;IAC9E,MAAM,UAAU,GAAG,kBAAkB,CAAC,eAAe,CAAC,CAAC;IACvD,MAAM,YAAY,GAAG,SAAS,CAAC,aAAa,CAAC,UAAU,CAAC,CAAC;IACzD,IAAI,CAAC,UAAU,IAAI,CAAC,YAAY;QAAE,OAAO,IAAI,CAAC;IAC9C,MAAM,UAAU,GAAG,KAAK,CAAC,IAAI,CAC5B,UAAU,CAAC,gBAAgB,CAAC,oBAAoB,CAAC,CACjD,CAAC;IACF,MAAM,KAAK,GAAG,UAAU,CAAC,OAAO,CAAC,eAAe,CAAC,CAAC;IAClD,IAAI,KAAK,GAAG,CAAC;QAAE,OAAO,IAAI,CAAC;IAC3B,MAAM,WAAW,GAAG,KAAK,CAAC,IAAI,CAC7B,YAAY,CAAC,gBAAgB,CAAC,qBAAqB,CAAC,CACpD,CAAC;IACF,IAAI,CAAC,mBAAmB,CAAC,UAAU,EAAE,WAAW,CAAC;QAAE,OAAO,IAAI,CAAC;IAC/D,MAAM,KAAK,GAAG,WAAW,CAAC,KAAK,CAAC,CAAC;IACjC,0EAA0E;IAC1E,8DAA8D;IAC9D,0EAA0E;IAC1E,2EAA2E;IAC3E,0EAA0E;IAC1E,wEAAwE;IACxE,uEAAuE;IACvE,sEAAsE;IACtE,IAAI,CAAC,KAAK,IAAI,CAAC,kBAAkB,CAAC,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IACtD,OAAO,KAAK,CAAC;AACd,CAAC,CAAC;AAYF,MAAM,CAAC,MAAM,+BAA+B,GAAG,CAC9C,WAAoB,EACM,EAAE;IAC5B,MAAM,SAAS,GAAG,sBAAsB,CAAC,WAAW,CAAC,CAAC;IACtD,IAAI,SAAS,EAAE,CAAC;QACf,gEAAgE;QAChE,uEAAuE;QACvE,IAAI,SAAS,CAAC,aAAa,CAAC,mCAAmC,CAAC,EAAE,CAAC;YAClE,OAAO,YAAY,CAAC;QACrB,CAAC;QACD,MAAM,UAAU,GAAG,kBAAkB,CAAC,WAAW,CAAC,IAAI,WAAW,CAAC;QAClE,MAAM,YAAY,GAAG,SAAS,CAAC,aAAa,CAAC,UAAU,CAAC,CAAC;QACzD,IAAI,CAAC,YAAY;YAAE,OAAO,YAAY,CAAC;QACvC,MAAM,UAAU,GAAG,KAAK,CAAC,IAAI,CAC5B,UAAU,CAAC,gBAAgB,CAAC,oBAAoB,CAAC,CACjD,CAAC;QACF,MAAM,WAAW,GAAG,KAAK,CAAC,IAAI,CAC7B,YAAY,CAAC,gBAAgB,CAAC,qBAAqB,CAAC,CACpD,CAAC;QACF,IAAI,CAAC,mBAAmB,CAAC,UAAU,EAAE,WAAW,CAAC;YAAE,OAAO,YAAY,CAAC;QACvE,mEAAmE;QACnE,6DAA6D;QAC7D,OAAO,WAAW,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,kBAAkB,CAAC,IAAI,CAAC,CAAC;YAC1D,CAAC,CAAC,OAAO;YACT,CAAC,CAAC,YAAY,CAAC;IACjB,CAAC;IACD,sEAAsE;IACtE,wEAAwE;IACxE,mEAAmE;IACnE,MAAM,UAAU,GAAG,kBAAkB,CAAC,WAAW,CAAC,IAAI,WAAW,CAAC;IAClE,MAAM,MAAM,GAAG,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,gBAAgB,CAAC,oBAAoB,CAAC,CAAC,CAAC;IAC7E,OAAO,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,kBAAkB,CAAC,IAAI,CAAC,CAAC;QACtD,CAAC,CAAC,OAAO;QACT,CAAC,CAAC,YAAY,CAAC;AACjB,CAAC,CAAC;AAMF,MAAM,iCAAiC,GAAG,CACzC,MAA+B,EACI,EAAE;IACrC,MAAM,OAAO,GAAG,cAAc,CAAC,MAAM,CAAC,CAAC;IACvC,MAAM,gBAAgB,GAAG,sBAAsB,CAAC,OAAO,CAAC,CAAC;IACzD,IAAI,gBAAgB,EAAE,CAAC;QACtB,MAAM,YAAY,GAAG,6BAA6B,CACjD,OAAO,EACP,gBAAgB,CAChB,CAAC;QACF,IAAI,YAAY,EAAE,CAAC;YAClB,OAAO;gBACN,IAAI,EAAE,SAAS;gBACf,OAAO,EAAE,gBAAgB;gBACzB,OAAO,EAAE,YAAY;aACrB,CAAC;QACH,CAAC;QACD,OAAO;YACN,IAAI,EAAE,SAAS;YACf,OAAO,EAAE,YAAY;YACrB,OAAO,EAAE,gBAAgB;SACzB,CAAC;IACH,CAAC;IACD,IAAI,MAAM,CAAC,IAAI,KAAK,YAAY,EAAE,CAAC;QAClC,IAAI,CAAC,qBAAqB,CAAC,MAAM,CAAC,IAAI,CAAC,aAAa,CAAC;YAAE,OAAO,IAAI,CAAC;QACnE,OAAO;YACN,IAAI,EAAE,YAAY;YAClB,OAAO,EAAE,gBAAgB;YACzB,IAAI,EAAE,MAAM,CAAC,IAAI;YACjB,WAAW,EAAE,MAAM,CAAC,WAAW;YAC/B,SAAS,EAAE,MAAM,CAAC,SAAS;SAC3B,CAAC;IACH,CAAC;IACD,IAAI,CAAC,qBAAqB,CAAC,MAAM,CAAC,OAAO,CAAC;QAAE,OAAO,IAAI,CAAC;IACxD,OAAO;QACN,IAAI,EAAE,SAAS;QACf,OAAO,EACN,MAAM,CAAC,OAAO,KAAK,iBAAiB,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,gBAAgB;QACvE,OAAO,EAAE,MAAM,CAAC,OAAO;KACvB,CAAC;AACH,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,yBAAyB,GACrC,iCAAiC,CAAC;AAEnC,MAAM,CAAC,MAAM,gCAAgC,GAC5C,GAA+B,EAAE;IAChC,MAAM,kBAAkB,GAAG,IAAI,OAAO,EAGnC,CAAC;IACJ,OAAO,CAAC,MAAM,EAAE,EAAE;QACjB,MAAM,OAAO,GAAG,cAAc,CAAC,MAAM,CAAC,CAAC;QACvC,MAAM,gBAAgB,GAAG,sBAAsB,CAAC,OAAO,CAAC,CAAC;QACzD,IAAI,CAAC,OAAO,IAAI,CAAC,gBAAgB,EAAE,CAAC;YACnC,OAAO,iCAAiC,CAAC,MAAM,CAAC,CAAC;QAClD,CAAC;QACD,IAAI,kBAAkB,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC;YACrC,OAAO,kBAAkB,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC;QAChD,CAAC;QACD,MAAM,QAAQ,GAAG,iCAAiC,CAAC,MAAM,CAAC,CAAC;QAC3D,kBAAkB,CAAC,GAAG,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC;QAC1C,OAAO,QAAQ,CAAC;IACjB,CAAC,CAAC;AACH,CAAC,CAAC","sourcesContent":["import { isNodeHiddenForTTS } from \"../text-processing.js\";\nimport type {\n\tSemanticHighlightTarget,\n\tRenderableHighlightTarget,\n} from \"./types.js\";\n\nconst MATHML_LEAF_TO_MATHJAX = new Map([\n\t[\"mi\", \"mjx-mi\"],\n\t[\"mn\", \"mjx-mn\"],\n\t[\"mo\", \"mjx-mo\"],\n\t[\"mtext\", \"mjx-mtext\"],\n]);\n\nconst MATHML_LEAF_SELECTOR = \"mi,mn,mo,mtext\";\nconst MATHJAX_LEAF_SELECTOR = \"mjx-mi,mjx-mn,mjx-mo,mjx-mtext\";\n\n// 2D layouts MathJax renders with extra, glyph-only nodes (radical surds,\n// fraction bars drawn as borders, table scaffolding) that have no source-MathML\n// leaf counterpart. They break the 1:1 token correspondence we rely on below,\n// so we refuse positional mapping inside them and fall back to the whole\n// expression instead of guessing.\nconst UNSUPPORTED_MATHJAX_LAYOUT_SELECTOR = [\n\t\"mjx-mfrac\",\n\t\"mjx-msqrt\",\n\t\"mjx-mroot\",\n\t\"mjx-mtable\",\n\t\"mjx-munder\",\n\t\"mjx-mover\",\n\t\"mjx-munderover\",\n].join(\",\");\n\nconst nearestElement = (target: SemanticHighlightTarget): Element | null => {\n\tif (target.type === \"text-range\") return target.node.parentElement;\n\treturn target.element;\n};\n\nconst owningMathJaxContainer = (element: Element | null): Element | null => {\n\tlet current: Element | null = element;\n\twhile (current) {\n\t\tif (current.localName?.toLowerCase() === \"mjx-container\") return current;\n\t\tcurrent = current.parentElement;\n\t}\n\treturn null;\n};\n\nconst isVisibleNativeTarget = (element: Element | null): boolean => {\n\tif (!element) return false;\n\tif (owningMathJaxContainer(element)) return false;\n\treturn !isNodeHiddenForTTS(element);\n};\n\nconst hasVisibleGeometry = (element: Element): boolean => {\n\tconst rect = (element as HTMLElement).getBoundingClientRect?.();\n\treturn Boolean(rect && rect.width > 0 && rect.height > 0);\n};\n\nconst closestMathElement = (element: Element | null): Element | null => {\n\tlet current: Element | null = element;\n\twhile (current) {\n\t\tif (current.localName?.toLowerCase() === \"math\") return current;\n\t\tcurrent = current.parentElement;\n\t}\n\treturn null;\n};\n\n// MathJax CHTML renders each source-MathML token (mi/mn/mo/mtext) as exactly\n// one mjx-* token in the same document order — including invisible operators,\n// which it keeps as zero-size nodes. The original <math> survives, clipped, in\n// mjx-assistive-mml, so its glyph text nodes are not visible and the visible\n// CHTML tokens carry no readable textContent (glyphs come from web fonts).\n// We therefore map a source token to its rendered glyph by *structural\n// position* rather than text, but only after proving the two leaf lists are\n// isomorphic (equal length, matching token type at every index). When the\n// structure diverges we refuse to guess and let the caller fall back to the\n// whole expression. \"False positives are worse than coarse highlighting.\"\nconst tokensAreIsomorphic = (\n\tmathLeaves: Element[],\n\tchtmlLeaves: Element[],\n): boolean => {\n\tif (mathLeaves.length === 0) return false;\n\tif (mathLeaves.length !== chtmlLeaves.length) return false;\n\tfor (let index = 0; index < mathLeaves.length; index += 1) {\n\t\tconst expected = MATHML_LEAF_TO_MATHJAX.get(\n\t\t\tmathLeaves[index].localName.toLowerCase(),\n\t\t);\n\t\tif (!expected) return false;\n\t\tif (chtmlLeaves[index].localName.toLowerCase() !== expected) return false;\n\t}\n\treturn true;\n};\n\nconst resolveMathJaxTokenByPosition = (\n\tsemanticElement: Element | null,\n\tcontainer: Element,\n): Element | null => {\n\tif (!semanticElement) return null;\n\tif (!MATHML_LEAF_TO_MATHJAX.has(semanticElement.localName.toLowerCase()))\n\t\treturn null;\n\tif (container.querySelector(UNSUPPORTED_MATHJAX_LAYOUT_SELECTOR)) return null;\n\tconst sourceMath = closestMathElement(semanticElement);\n\tconst renderedMath = container.querySelector(\"mjx-math\");\n\tif (!sourceMath || !renderedMath) return null;\n\tconst mathLeaves = Array.from(\n\t\tsourceMath.querySelectorAll(MATHML_LEAF_SELECTOR),\n\t);\n\tconst index = mathLeaves.indexOf(semanticElement);\n\tif (index < 0) return null;\n\tconst chtmlLeaves = Array.from(\n\t\trenderedMath.querySelectorAll(MATHJAX_LEAF_SELECTOR),\n\t);\n\tif (!tokensAreIsomorphic(mathLeaves, chtmlLeaves)) return null;\n\tconst token = chtmlLeaves[index];\n\t// The matched token must be a real, visible glyph. We deliberately do NOT\n\t// run isNodeHiddenForTTS here: MathJax marks its visual CHTML\n\t// (mjx-math) with aria-hidden=\"true\" because the accessible tree lives in\n\t// the clipped mjx-assistive-mml. aria-hidden is correct for screen readers\n\t// but irrelevant for a visual highlight — the glyph is painted on screen.\n\t// Visible geometry is the right test: it rejects display:none nodes and\n\t// zero-size invisible operators (e.g. U+2062 invisible-times, which is\n\t// never spoken anyway) without discarding the real, on-screen tokens.\n\tif (!token || !hasVisibleGeometry(token)) return null;\n\treturn token;\n};\n\n// Per-equation highlight capability, decided once before playback. \"token\"\n// means we are confident every spoken token can be mapped to a visible glyph,\n// so the equation tracks word-by-word and the whole expression is never\n// painted. \"expression\" means we cannot reliably map tokens (a 2D MathJax\n// layout, a non-isomorphic render, or nothing visible), so the equation is\n// highlighted as one stable block for its entire spoken duration. Deciding\n// once and committing is what stops the highlight from flickering between a\n// token and the whole expression during breaks/pauses.\nexport type MathHighlightCapability = \"token\" | \"expression\";\n\nexport const classifyMathHighlightCapability = (\n\tmathElement: Element,\n): MathHighlightCapability => {\n\tconst container = owningMathJaxContainer(mathElement);\n\tif (container) {\n\t\t// 2D MathJax layouts add glyph-only nodes with no source-MathML\n\t\t// counterpart, so positional mapping is unsafe — commit to expression.\n\t\tif (container.querySelector(UNSUPPORTED_MATHJAX_LAYOUT_SELECTOR)) {\n\t\t\treturn \"expression\";\n\t\t}\n\t\tconst sourceMath = closestMathElement(mathElement) ?? mathElement;\n\t\tconst renderedMath = container.querySelector(\"mjx-math\");\n\t\tif (!renderedMath) return \"expression\";\n\t\tconst mathLeaves = Array.from(\n\t\t\tsourceMath.querySelectorAll(MATHML_LEAF_SELECTOR),\n\t\t);\n\t\tconst chtmlLeaves = Array.from(\n\t\t\trenderedMath.querySelectorAll(MATHJAX_LEAF_SELECTOR),\n\t\t);\n\t\tif (!tokensAreIsomorphic(mathLeaves, chtmlLeaves)) return \"expression\";\n\t\t// At least one rendered token must paint a real glyph; otherwise a\n\t\t// per-token highlight would have nothing visible to land on.\n\t\treturn chtmlLeaves.some((leaf) => hasVisibleGeometry(leaf))\n\t\t\t? \"token\"\n\t\t\t: \"expression\";\n\t}\n\t// Native MathML: every leaf is a real, browser-positioned element, so\n\t// per-token highlighting is reliable even inside 2D layouts (fractions,\n\t// radicals). It is token-capable as long as a visible leaf exists.\n\tconst sourceMath = closestMathElement(mathElement) ?? mathElement;\n\tconst leaves = Array.from(sourceMath.querySelectorAll(MATHML_LEAF_SELECTOR));\n\treturn leaves.some((leaf) => !isNodeHiddenForTTS(leaf))\n\t\t? \"token\"\n\t\t: \"expression\";\n};\n\nexport type RenderedMathTargetResolver = (\n\ttarget: SemanticHighlightTarget,\n) => RenderableHighlightTarget | null;\n\nconst resolveRenderedMathTargetUncached = (\n\ttarget: SemanticHighlightTarget,\n): RenderableHighlightTarget | null => {\n\tconst element = nearestElement(target);\n\tconst mathJaxContainer = owningMathJaxContainer(element);\n\tif (mathJaxContainer) {\n\t\tconst visibleToken = resolveMathJaxTokenByPosition(\n\t\t\telement,\n\t\t\tmathJaxContainer,\n\t\t);\n\t\tif (visibleToken) {\n\t\t\treturn {\n\t\t\t\ttype: \"element\",\n\t\t\t\tquality: \"semantic-token\",\n\t\t\t\telement: visibleToken,\n\t\t\t};\n\t\t}\n\t\treturn {\n\t\t\ttype: \"element\",\n\t\t\tquality: \"expression\",\n\t\t\telement: mathJaxContainer,\n\t\t};\n\t}\n\tif (target.type === \"text-range\") {\n\t\tif (!isVisibleNativeTarget(target.node.parentElement)) return null;\n\t\treturn {\n\t\t\ttype: \"text-range\",\n\t\t\tquality: \"semantic-token\",\n\t\t\tnode: target.node,\n\t\t\tstartOffset: target.startOffset,\n\t\t\tendOffset: target.endOffset,\n\t\t};\n\t}\n\tif (!isVisibleNativeTarget(target.element)) return null;\n\treturn {\n\t\ttype: \"element\",\n\t\tquality:\n\t\t\ttarget.quality === \"region-fallback\" ? \"expression\" : \"semantic-token\",\n\t\telement: target.element,\n\t};\n};\n\nexport const resolveRenderedMathTarget: RenderedMathTargetResolver =\n\tresolveRenderedMathTargetUncached;\n\nexport const createRenderedMathTargetResolver =\n\t(): RenderedMathTargetResolver => {\n\t\tconst mathJaxTargetCache = new WeakMap<\n\t\t\tElement,\n\t\t\tRenderableHighlightTarget | null\n\t\t>();\n\t\treturn (target) => {\n\t\t\tconst element = nearestElement(target);\n\t\t\tconst mathJaxContainer = owningMathJaxContainer(element);\n\t\t\tif (!element || !mathJaxContainer) {\n\t\t\t\treturn resolveRenderedMathTargetUncached(target);\n\t\t\t}\n\t\t\tif (mathJaxTargetCache.has(element)) {\n\t\t\t\treturn mathJaxTargetCache.get(element) ?? null;\n\t\t\t}\n\t\t\tconst rendered = resolveRenderedMathTargetUncached(target);\n\t\t\tmathJaxTargetCache.set(element, rendered);\n\t\t\treturn rendered;\n\t\t};\n\t};\n"]}

package/dist/services/tts/highlight-pipeline/types.d.ts

@@ -0,0 +1,72 @@
+import type { CatalogSpanAlignment } from "../catalog-span-alignment.js";
+import type { MathAwareAlignment, HighlightTarget } from "../math-alignment/index.js";
+import type { NormalizedTextMap } from "../text-processing.js";
+export type ProviderOffsetSpace = "plain-spoken-text" | "raw-ssml" | "provider-speech-marks" | "unknown";
+export type ChunkOffsetSpace = Exclude<ProviderOffsetSpace, "unknown"> | "unsupported";
+export interface TTSHighlightChunk {
+    id: string;
+    speechText: string;
+    visibleText: string;
+    sourceElement: Element | null;
+    contentRoot: Element;
+    regionElement: Element | null;
+    regionRange?: Range;
+    visibleMap?: NormalizedTextMap;
+    catalogAlignment?: CatalogSpanAlignment;
+    mathAlignments: Array<{
+        element: Element;
+        alignment: MathAwareAlignment;
+    }>;
+    offsetSpace: ChunkOffsetSpace;
+}
+export interface TTSBoundaryEvent {
+    chunkId: string;
+    word: string;
+    position: number;
+    length?: number;
+    providerOffsetSpace: ProviderOffsetSpace;
+}
+export interface NormalizedBoundaryEvent {
+    chunkId: string;
+    normalizedWord: string | null;
+    chunkSpokenStart: number | null;
+    chunkSpokenEnd: number | null;
+    confidence: number;
+    reason: string;
+}
+export type RenderableHighlightTarget = {
+    type: "text-range";
+    quality: "exact-word" | "semantic-token";
+    node: Text;
+    startOffset: number;
+    endOffset: number;
+} | {
+    type: "range";
+    quality: "exact-word" | "semantic-token" | "expression" | "region";
+    range: Range;
+} | {
+    type: "element";
+    quality: "semantic-token" | "expression" | "region";
+    element: Element;
+};
+export type HighlightDecisionQuality = "exact-word" | "semantic-token" | "expression" | "region";
+export interface HighlightDecision {
+    activeTarget: RenderableHighlightTarget | null;
+    regionTarget: RenderableHighlightTarget | null;
+    quality: HighlightDecisionQuality;
+    confidence: number;
+    reason: string;
+}
+export interface HighlightDecisionInput {
+    semanticTarget: RenderableHighlightTarget | null;
+    expressionTarget: RenderableHighlightTarget | null;
+    regionTarget: RenderableHighlightTarget | null;
+    confidence: number;
+    reason: string;
+}
+export interface TTSHighlightPlan {
+    chunks: TTSHighlightChunk[];
+    resolveInitial(chunkId: string): HighlightDecision;
+    resolveBoundary(event: TTSBoundaryEvent): HighlightDecision;
+}
+export type SemanticHighlightTarget = HighlightTarget;

package/dist/services/tts/highlight-pipeline/types.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=types.js.map

package/dist/services/tts/highlight-pipeline/types.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.js","sourceRoot":"","sources":["../../../../src/services/tts/highlight-pipeline/types.ts"],"names":[],"mappings":"","sourcesContent":["import type { CatalogSpanAlignment } from \"../catalog-span-alignment.js\";\nimport type {\n\tMathAwareAlignment,\n\tHighlightTarget,\n} from \"../math-alignment/index.js\";\nimport type { NormalizedTextMap } from \"../text-processing.js\";\n\nexport type ProviderOffsetSpace =\n\t| \"plain-spoken-text\"\n\t| \"raw-ssml\"\n\t| \"provider-speech-marks\"\n\t| \"unknown\";\n\nexport type ChunkOffsetSpace =\n\t| Exclude<ProviderOffsetSpace, \"unknown\">\n\t| \"unsupported\";\n\nexport interface TTSHighlightChunk {\n\tid: string;\n\tspeechText: string;\n\tvisibleText: string;\n\tsourceElement: Element | null;\n\tcontentRoot: Element;\n\tregionElement: Element | null;\n\tregionRange?: Range;\n\tvisibleMap?: NormalizedTextMap;\n\tcatalogAlignment?: CatalogSpanAlignment;\n\tmathAlignments: Array<{ element: Element; alignment: MathAwareAlignment }>;\n\toffsetSpace: ChunkOffsetSpace;\n}\n\nexport interface TTSBoundaryEvent {\n\tchunkId: string;\n\tword: string;\n\tposition: number;\n\tlength?: number;\n\tproviderOffsetSpace: ProviderOffsetSpace;\n}\n\nexport interface NormalizedBoundaryEvent {\n\tchunkId: string;\n\tnormalizedWord: string | null;\n\tchunkSpokenStart: number | null;\n\tchunkSpokenEnd: number | null;\n\tconfidence: number;\n\treason: string;\n}\n\nexport type RenderableHighlightTarget =\n\t| {\n\t\t\ttype: \"text-range\";\n\t\t\tquality: \"exact-word\" | \"semantic-token\";\n\t\t\tnode: Text;\n\t\t\tstartOffset: number;\n\t\t\tendOffset: number;\n\t  }\n\t| {\n\t\t\ttype: \"range\";\n\t\t\tquality: \"exact-word\" | \"semantic-token\" | \"expression\" | \"region\";\n\t\t\trange: Range;\n\t  }\n\t| {\n\t\t\ttype: \"element\";\n\t\t\tquality: \"semantic-token\" | \"expression\" | \"region\";\n\t\t\telement: Element;\n\t  };\n\nexport type HighlightDecisionQuality =\n\t| \"exact-word\"\n\t| \"semantic-token\"\n\t| \"expression\"\n\t| \"region\";\n\nexport interface HighlightDecision {\n\tactiveTarget: RenderableHighlightTarget | null;\n\tregionTarget: RenderableHighlightTarget | null;\n\tquality: HighlightDecisionQuality;\n\tconfidence: number;\n\treason: string;\n}\n\nexport interface HighlightDecisionInput {\n\tsemanticTarget: RenderableHighlightTarget | null;\n\texpressionTarget: RenderableHighlightTarget | null;\n\tregionTarget: RenderableHighlightTarget | null;\n\tconfidence: number;\n\treason: string;\n}\n\nexport interface TTSHighlightPlan {\n\tchunks: TTSHighlightChunk[];\n\tresolveInitial(chunkId: string): HighlightDecision;\n\tresolveBoundary(event: TTSBoundaryEvent): HighlightDecision;\n}\n\nexport type SemanticHighlightTarget = HighlightTarget;\n"]}

package/dist/services/tts/highlight-pipeline/visible-map-range.d.ts

@@ -0,0 +1,2 @@
+import type { NormalizedTextMap } from "../text-processing.js";
+export declare const createRangeFromVisibleMap: (visibleMap: NormalizedTextMap | undefined, start: number, end: number) => Range | null;

package/dist/services/tts/highlight-pipeline/visible-map-range.js

@@ -0,0 +1,22 @@
+export const createRangeFromVisibleMap = (visibleMap, start, end) => {
+    if (!visibleMap || typeof document === "undefined")
+        return null;
+    if (end <= start)
+        return null;
+    const startMapping = visibleMap.get(start);
+    let endMapping = visibleMap.get(Math.max(start, end - 1));
+    if (!startMapping)
+        return null;
+    if (!endMapping)
+        endMapping = startMapping;
+    try {
+        const range = document.createRange();
+        range.setStart(startMapping.node, startMapping.offset);
+        range.setEnd(endMapping.node, endMapping.offset + 1);
+        return range;
+    }
+    catch {
+        return null;
+    }
+};
+//# sourceMappingURL=visible-map-range.js.map

package/dist/services/tts/highlight-pipeline/visible-map-range.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"visible-map-range.js","sourceRoot":"","sources":["../../../../src/services/tts/highlight-pipeline/visible-map-range.ts"],"names":[],"mappings":"AAEA,MAAM,CAAC,MAAM,yBAAyB,GAAG,CACxC,UAAyC,EACzC,KAAa,EACb,GAAW,EACI,EAAE;IACjB,IAAI,CAAC,UAAU,IAAI,OAAO,QAAQ,KAAK,WAAW;QAAE,OAAO,IAAI,CAAC;IAChE,IAAI,GAAG,IAAI,KAAK;QAAE,OAAO,IAAI,CAAC;IAC9B,MAAM,YAAY,GAAG,UAAU,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;IAC3C,IAAI,UAAU,GAAG,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC;IAC1D,IAAI,CAAC,YAAY;QAAE,OAAO,IAAI,CAAC;IAC/B,IAAI,CAAC,UAAU;QAAE,UAAU,GAAG,YAAY,CAAC;IAC3C,IAAI,CAAC;QACJ,MAAM,KAAK,GAAG,QAAQ,CAAC,WAAW,EAAE,CAAC;QACrC,KAAK,CAAC,QAAQ,CAAC,YAAY,CAAC,IAAI,EAAE,YAAY,CAAC,MAAM,CAAC,CAAC;QACvD,KAAK,CAAC,MAAM,CAAC,UAAU,CAAC,IAAI,EAAE,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACrD,OAAO,KAAK,CAAC;IACd,CAAC;IAAC,MAAM,CAAC;QACR,OAAO,IAAI,CAAC;IACb,CAAC;AACF,CAAC,CAAC","sourcesContent":["import type { NormalizedTextMap } from \"../text-processing.js\";\n\nexport const createRangeFromVisibleMap = (\n\tvisibleMap: NormalizedTextMap | undefined,\n\tstart: number,\n\tend: number,\n): Range | null => {\n\tif (!visibleMap || typeof document === \"undefined\") return null;\n\tif (end <= start) return null;\n\tconst startMapping = visibleMap.get(start);\n\tlet endMapping = visibleMap.get(Math.max(start, end - 1));\n\tif (!startMapping) return null;\n\tif (!endMapping) endMapping = startMapping;\n\ttry {\n\t\tconst range = document.createRange();\n\t\trange.setStart(startMapping.node, startMapping.offset);\n\t\trange.setEnd(endMapping.node, endMapping.offset + 1);\n\t\treturn range;\n\t} catch {\n\t\treturn null;\n\t}\n};\n"]}

package/dist/services/tts/math-alignment/index.d.ts

@@ -0,0 +1,30 @@
+import { type SpeechSourceTokenization } from "./speech-tokenizer.js";
+import type { AlignmentResult, HighlightTarget, MathMLTokenizationResult } from "./types.js";
+export interface MathAwareAlignment {
+    speech: SpeechSourceTokenization;
+    math: MathMLTokenizationResult;
+    result: AlignmentResult;
+}
+export declare const createMathAwareAlignment: (args: {
+    mathElement: Element;
+    speechText: string;
+}) => MathAwareAlignment;
+export declare const resolveHighlightTargetForBoundary: (alignment: MathAwareAlignment, args: {
+    position: number;
+    length?: number;
+    boundaryWord?: string;
+}) => HighlightTarget;
+export declare const resolveMatchedHighlightTargetForBoundary: (alignment: MathAwareAlignment, args: {
+    position: number;
+    length?: number;
+    boundaryWord?: string;
+}) => HighlightTarget | null;
+export declare const resolveMatchedHighlightTargetForSpokenBoundary: (alignment: MathAwareAlignment, args: {
+    position: number;
+    length?: number;
+    boundaryWord?: string;
+}) => HighlightTarget | null;
+export declare const resolveUniqueMathTargetForBoundaryWord: (alignment: MathAwareAlignment, args: {
+    boundaryWord?: string;
+}) => HighlightTarget | null;
+export type * from "./types.js";

package/dist/services/tts/math-alignment/index.js

@@ -0,0 +1,66 @@
+import { tokenizeMathML } from "./mathml-tokenizer.js";
+import { resolveMatchedHighlightTargetForBoundary as resolveMatchedTargetForBoundary, resolveHighlightTargetForBoundary as resolveTargetForBoundary, } from "./range-resolver.js";
+import { alignSpeechToMath } from "./sequence-aligner.js";
+import { resolveBoundaryToSpeechToken, resolveSpokenOffsetToSpeechToken, tokenizeSpeechSource, } from "./speech-tokenizer.js";
+export const createMathAwareAlignment = (args) => {
+    const speech = tokenizeSpeechSource({ speechText: args.speechText });
+    const math = tokenizeMathML(args.mathElement);
+    return {
+        speech,
+        math,
+        result: alignSpeechToMath({ speech, math }),
+    };
+};
+export const resolveHighlightTargetForBoundary = (alignment, args) => {
+    const boundary = resolveBoundaryToSpeechToken({
+        tokenization: alignment.speech,
+        position: args.position,
+        length: args.length,
+        boundaryWord: args.boundaryWord,
+    });
+    return resolveTargetForBoundary({
+        alignment: alignment.result,
+        boundary,
+    });
+};
+export const resolveMatchedHighlightTargetForBoundary = (alignment, args) => {
+    const boundary = resolveBoundaryToSpeechToken({
+        tokenization: alignment.speech,
+        position: args.position,
+        length: args.length,
+        boundaryWord: args.boundaryWord,
+    });
+    return resolveMatchedTargetForBoundary({
+        alignment: alignment.result,
+        boundary,
+    });
+};
+export const resolveMatchedHighlightTargetForSpokenBoundary = (alignment, args) => {
+    const boundary = resolveSpokenOffsetToSpeechToken({
+        tokenization: alignment.speech,
+        position: args.position,
+        length: args.length,
+        boundaryWord: args.boundaryWord,
+    });
+    return resolveMatchedTargetForBoundary({
+        alignment: alignment.result,
+        boundary,
+    });
+};
+export const resolveUniqueMathTargetForBoundaryWord = (alignment, args) => {
+    if (!args.boundaryWord || /^<[^>]+>$/.test(args.boundaryWord.trim())) {
+        return null;
+    }
+    const boundaryToken = tokenizeSpeechSource({
+        speechText: args.boundaryWord,
+    }).tokens[0];
+    if (!boundaryToken)
+        return null;
+    const matches = alignment.math.tokens.filter((token) => {
+        const normalized = boundaryToken.normalized.toLowerCase();
+        return (token.normalized.toLowerCase() === normalized ||
+            token.spokenAliases.some((alias) => alias.toLowerCase() === normalized));
+    });
+    return matches.length === 1 ? matches[0].target : null;
+};
+//# sourceMappingURL=index.js.map

package/dist/services/tts/math-alignment/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../src/services/tts/math-alignment/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAC;AACvD,OAAO,EACN,wCAAwC,IAAI,+BAA+B,EAC3E,iCAAiC,IAAI,wBAAwB,GAC7D,MAAM,qBAAqB,CAAC;AAC7B,OAAO,EAAE,iBAAiB,EAAE,MAAM,uBAAuB,CAAC;AAC1D,OAAO,EACN,4BAA4B,EAC5B,gCAAgC,EAChC,oBAAoB,GAEpB,MAAM,uBAAuB,CAAC;AAa/B,MAAM,CAAC,MAAM,wBAAwB,GAAG,CAAC,IAGxC,EAAsB,EAAE;IACxB,MAAM,MAAM,GAAG,oBAAoB,CAAC,EAAE,UAAU,EAAE,IAAI,CAAC,UAAU,EAAE,CAAC,CAAC;IACrE,MAAM,IAAI,GAAG,cAAc,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;IAC9C,OAAO;QACN,MAAM;QACN,IAAI;QACJ,MAAM,EAAE,iBAAiB,CAAC,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;KAC3C,CAAC;AACH,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,iCAAiC,GAAG,CAChD,SAA6B,EAC7B,IAIC,EACiB,EAAE;IACpB,MAAM,QAAQ,GAAG,4BAA4B,CAAC;QAC7C,YAAY,EAAE,SAAS,CAAC,MAAM;QAC9B,QAAQ,EAAE,IAAI,CAAC,QAAQ;QACvB,MAAM,EAAE,IAAI,CAAC,MAAM;QACnB,YAAY,EAAE,IAAI,CAAC,YAAY;KAC/B,CAAC,CAAC;IACH,OAAO,wBAAwB,CAAC;QAC/B,SAAS,EAAE,SAAS,CAAC,MAAM;QAC3B,QAAQ;KACR,CAAC,CAAC;AACJ,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,wCAAwC,GAAG,CACvD,SAA6B,EAC7B,IAIC,EACwB,EAAE;IAC3B,MAAM,QAAQ,GAAG,4BAA4B,CAAC;QAC7C,YAAY,EAAE,SAAS,CAAC,MAAM;QAC9B,QAAQ,EAAE,IAAI,CAAC,QAAQ;QACvB,MAAM,EAAE,IAAI,CAAC,MAAM;QACnB,YAAY,EAAE,IAAI,CAAC,YAAY;KAC/B,CAAC,CAAC;IACH,OAAO,+BAA+B,CAAC;QACtC,SAAS,EAAE,SAAS,CAAC,MAAM;QAC3B,QAAQ;KACR,CAAC,CAAC;AACJ,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,8CAA8C,GAAG,CAC7D,SAA6B,EAC7B,IAIC,EACwB,EAAE;IAC3B,MAAM,QAAQ,GAAG,gCAAgC,CAAC;QACjD,YAAY,EAAE,SAAS,CAAC,MAAM;QAC9B,QAAQ,EAAE,IAAI,CAAC,QAAQ;QACvB,MAAM,EAAE,IAAI,CAAC,MAAM;QACnB,YAAY,EAAE,IAAI,CAAC,YAAY;KAC/B,CAAC,CAAC;IACH,OAAO,+BAA+B,CAAC;QACtC,SAAS,EAAE,SAAS,CAAC,MAAM;QAC3B,QAAQ;KACR,CAAC,CAAC;AACJ,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,sCAAsC,GAAG,CACrD,SAA6B,EAC7B,IAEC,EACwB,EAAE;IAC3B,IAAI,CAAC,IAAI,CAAC,YAAY,IAAI,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;QACtE,OAAO,IAAI,CAAC;IACb,CAAC;IACD,MAAM,aAAa,GAAG,oBAAoB,CAAC;QAC1C,UAAU,EAAE,IAAI,CAAC,YAAY;KAC7B,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;IACb,IAAI,CAAC,aAAa;QAAE,OAAO,IAAI,CAAC;IAChC,MAAM,OAAO,GAAG,SAAS,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE;QACtD,MAAM,UAAU,GAAG,aAAa,CAAC,UAAU,CAAC,WAAW,EAAE,CAAC;QAC1D,OAAO,CACN,KAAK,CAAC,UAAU,CAAC,WAAW,EAAE,KAAK,UAAU;YAC7C,KAAK,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,WAAW,EAAE,KAAK,UAAU,CAAC,CACvE,CAAC;IACH,CAAC,CAAC,CAAC;IACH,OAAO,OAAO,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC;AACxD,CAAC,CAAC","sourcesContent":["import { tokenizeMathML } from \"./mathml-tokenizer.js\";\nimport {\n\tresolveMatchedHighlightTargetForBoundary as resolveMatchedTargetForBoundary,\n\tresolveHighlightTargetForBoundary as resolveTargetForBoundary,\n} from \"./range-resolver.js\";\nimport { alignSpeechToMath } from \"./sequence-aligner.js\";\nimport {\n\tresolveBoundaryToSpeechToken,\n\tresolveSpokenOffsetToSpeechToken,\n\ttokenizeSpeechSource,\n\ttype SpeechSourceTokenization,\n} from \"./speech-tokenizer.js\";\nimport type {\n\tAlignmentResult,\n\tHighlightTarget,\n\tMathMLTokenizationResult,\n} from \"./types.js\";\n\nexport interface MathAwareAlignment {\n\tspeech: SpeechSourceTokenization;\n\tmath: MathMLTokenizationResult;\n\tresult: AlignmentResult;\n}\n\nexport const createMathAwareAlignment = (args: {\n\tmathElement: Element;\n\tspeechText: string;\n}): MathAwareAlignment => {\n\tconst speech = tokenizeSpeechSource({ speechText: args.speechText });\n\tconst math = tokenizeMathML(args.mathElement);\n\treturn {\n\t\tspeech,\n\t\tmath,\n\t\tresult: alignSpeechToMath({ speech, math }),\n\t};\n};\n\nexport const resolveHighlightTargetForBoundary = (\n\talignment: MathAwareAlignment,\n\targs: {\n\t\tposition: number;\n\t\tlength?: number;\n\t\tboundaryWord?: string;\n\t},\n): HighlightTarget => {\n\tconst boundary = resolveBoundaryToSpeechToken({\n\t\ttokenization: alignment.speech,\n\t\tposition: args.position,\n\t\tlength: args.length,\n\t\tboundaryWord: args.boundaryWord,\n\t});\n\treturn resolveTargetForBoundary({\n\t\talignment: alignment.result,\n\t\tboundary,\n\t});\n};\n\nexport const resolveMatchedHighlightTargetForBoundary = (\n\talignment: MathAwareAlignment,\n\targs: {\n\t\tposition: number;\n\t\tlength?: number;\n\t\tboundaryWord?: string;\n\t},\n): HighlightTarget | null => {\n\tconst boundary = resolveBoundaryToSpeechToken({\n\t\ttokenization: alignment.speech,\n\t\tposition: args.position,\n\t\tlength: args.length,\n\t\tboundaryWord: args.boundaryWord,\n\t});\n\treturn resolveMatchedTargetForBoundary({\n\t\talignment: alignment.result,\n\t\tboundary,\n\t});\n};\n\nexport const resolveMatchedHighlightTargetForSpokenBoundary = (\n\talignment: MathAwareAlignment,\n\targs: {\n\t\tposition: number;\n\t\tlength?: number;\n\t\tboundaryWord?: string;\n\t},\n): HighlightTarget | null => {\n\tconst boundary = resolveSpokenOffsetToSpeechToken({\n\t\ttokenization: alignment.speech,\n\t\tposition: args.position,\n\t\tlength: args.length,\n\t\tboundaryWord: args.boundaryWord,\n\t});\n\treturn resolveMatchedTargetForBoundary({\n\t\talignment: alignment.result,\n\t\tboundary,\n\t});\n};\n\nexport const resolveUniqueMathTargetForBoundaryWord = (\n\talignment: MathAwareAlignment,\n\targs: {\n\t\tboundaryWord?: string;\n\t},\n): HighlightTarget | null => {\n\tif (!args.boundaryWord || /^<[^>]+>$/.test(args.boundaryWord.trim())) {\n\t\treturn null;\n\t}\n\tconst boundaryToken = tokenizeSpeechSource({\n\t\tspeechText: args.boundaryWord,\n\t}).tokens[0];\n\tif (!boundaryToken) return null;\n\tconst matches = alignment.math.tokens.filter((token) => {\n\t\tconst normalized = boundaryToken.normalized.toLowerCase();\n\t\treturn (\n\t\t\ttoken.normalized.toLowerCase() === normalized ||\n\t\t\ttoken.spokenAliases.some((alias) => alias.toLowerCase() === normalized)\n\t\t);\n\t});\n\treturn matches.length === 1 ? matches[0].target : null;\n};\n\nexport type * from \"./types.js\";\n"]}

package/dist/services/tts/math-alignment/mathml-tokenizer.d.ts

@@ -0,0 +1,2 @@
+import type { MathMLTokenizationResult } from "./types.js";
+export declare const tokenizeMathML: (mathElement: Element) => MathMLTokenizationResult;