ssml-builder-js 2.13.0 → 2.15.0

package/dist/core.js

@@ -30,11 +30,13 @@ __export(core_exports, {
   areAzureLanguagesEquivalent: () => areAzureLanguagesEquivalent,
   buildPartialSsml: () => buildPartialSsml,
   buildSsml: () => buildSsml,
+  createAzureUrlValidatorRunner: () => createAzureUrlValidatorRunner,
   extractSsmlText: () => extractSsmlText,
   extractSsmlTranslatableText: () => extractSsmlTranslatableText,
   fromPlainTextToSsml: () => fromPlainTextToSsml,
   getAzureVoiceCatalogMetadata: () => getAzureVoiceCatalogMetadata,
   getBuiltInVoiceCatalogMetadata: () => getBuiltInVoiceCatalogMetadata,
+  getSsmlSourceMap: () => getSsmlSourceMap,
   isValidAzureAudioDuration: () => isValidAzureAudioDuration,
   mapSsmlTextNodes: () => mapSsmlTextNodes,
   normalizeAzureLanguage: () => normalizeAzureLanguage,
@@ -973,6 +975,236 @@ function buildPartialSsml(textOrOptions, context) {
   return serializePartialSsml(textOrOptions.text, textOrOptions);
 }
 
+// packages/ssml-core/src/textNodes.ts
+function decodeXmlText(value) {
+  return value.replace(/&(?:amp|apos|gt|lt|quot);|&#(?:x[\da-f]+|\d+);/gi, (entity) => {
+    if (entity === "&") return "&";
+    if (entity === "'") return "'";
+    if (entity === ">") return ">";
+    if (entity === "&lt;") return "<";
+    if (entity === "&quot;") return '"';
+    const hexadecimal = entity.toLowerCase().startsWith("&#x");
+    const digits = entity.slice(hexadecimal ? 3 : 2, -1);
+    return String.fromCodePoint(Number.parseInt(digits, hexadecimal ? 16 : 10));
+  });
+}
+function encodeXmlText(value) {
+  return value.replace(/&/g, "&amp;").replace(/</g, "&lt;").replace(/>/g, "&gt;");
+}
+function decodeXmlAttribute(value) {
+  return decodeXmlText(value);
+}
+function findTagEnd(source, start) {
+  let quote = "";
+  for (let index = start; index < source.length; index += 1) {
+    const character = source[index];
+    if (quote) {
+      if (character === quote) quote = "";
+    } else if (character === '"' || character === "'") {
+      quote = character;
+    } else if (character === ">") {
+      return index;
+    }
+  }
+  return source.length - 1;
+}
+function readTagName(tag) {
+  const match = /^<\s*([A-Za-z_][A-Za-z0-9_.:-]*)/.exec(tag);
+  return match?.[1];
+}
+function readTagAttributes(tag, name) {
+  const attributes = {};
+  const nameStart = tag.indexOf(name);
+  const attributeSource = tag.slice(nameStart + name.length, tag.length - 1).replace(/\/\s*$/, "");
+  const attributePattern = /([A-Za-z_][A-Za-z0-9_.:-]*)\s*=\s*(["'])([\s\S]*?)\2/g;
+  for (const match of attributeSource.matchAll(attributePattern)) {
+    attributes[match[1].toLowerCase()] = decodeXmlAttribute(match[3]);
+  }
+  return attributes;
+}
+function collectTextNodes(source) {
+  const nodes = [];
+  const elements = [];
+  let index = 0;
+  const addText = (start, end, rawText, sourceStart = start, sourceEnd = end) => {
+    if (!rawText) return;
+    const path = elements.map((element) => element.name);
+    const parent = elements[elements.length - 1];
+    nodes.push({
+      context: {
+        ancestorTags: path.slice(0, -1),
+        parentAttributes: { ...parent?.attributes ?? {} },
+        parentTag: parent?.name ?? "",
+        path
+      },
+      decodedText: decodeXmlText(rawText),
+      end,
+      sourceEnd,
+      sourceStart,
+      start
+    });
+  };
+  while (index < source.length) {
+    if (source[index] !== "<") {
+      const nextTag = source.indexOf("<", index);
+      const end2 = nextTag === -1 ? source.length : nextTag;
+      addText(index, end2, source.slice(index, end2));
+      index = end2;
+      continue;
+    }
+    if (source.startsWith("<!--", index)) {
+      const end2 = source.indexOf("-->", index + 4);
+      index = end2 === -1 ? source.length : end2 + 3;
+      continue;
+    }
+    if (source.startsWith("<![CDATA[", index)) {
+      const contentStart = index + 9;
+      const end2 = source.indexOf("]]>", contentStart);
+      const contentEnd = end2 === -1 ? source.length : end2;
+      addText(
+        contentStart,
+        contentEnd,
+        source.slice(contentStart, contentEnd),
+        index,
+        end2 === -1 ? source.length : end2 + 3
+      );
+      index = end2 === -1 ? source.length : end2 + 3;
+      continue;
+    }
+    if (source.startsWith("<?", index)) {
+      const end2 = source.indexOf("?>", index + 2);
+      index = end2 === -1 ? source.length : end2 + 2;
+      continue;
+    }
+    if (source.startsWith("</", index)) {
+      const end2 = findTagEnd(source, index + 2);
+      elements.pop();
+      index = end2 + 1;
+      continue;
+    }
+    const end = findTagEnd(source, index + 1);
+    const tag = source.slice(index, end + 1);
+    const name = readTagName(tag);
+    if (name && !/\/\s*>$/.test(tag)) elements.push({ attributes: readTagAttributes(tag, name), name });
+    index = end + 1;
+  }
+  return nodes;
+}
+function collectSourceMap(source) {
+  const segments = [];
+  const markers = [];
+  const elements = [];
+  let textOffset = 0;
+  let index = 0;
+  const textParts = [];
+  const addText = (value) => {
+    if (!value) return;
+    const parent = elements[elements.length - 1];
+    if (parent) parent.nextChildIndex += 1;
+    const sourceNodePath = parent?.path ?? ["speak"];
+    const start = textOffset;
+    textOffset += value.length;
+    textParts.push(value);
+    segments.push({ text: value, range: { start, end: textOffset }, sourceNodePath: [...sourceNodePath] });
+  };
+  while (index < source.length) {
+    if (source[index] !== "<") {
+      const end2 = source.indexOf("<", index);
+      const textEnd = end2 === -1 ? source.length : end2;
+      addText(decodeXmlText(source.slice(index, textEnd)));
+      index = textEnd;
+      continue;
+    }
+    if (source.startsWith("<!--", index)) {
+      const end2 = source.indexOf("-->", index + 4);
+      index = end2 === -1 ? source.length : end2 + 3;
+      continue;
+    }
+    if (source.startsWith("<![CDATA[", index)) {
+      const contentStart = index + 9;
+      const end2 = source.indexOf("]]>", contentStart);
+      const contentEnd = end2 === -1 ? source.length : end2;
+      addText(source.slice(contentStart, contentEnd));
+      index = end2 === -1 ? source.length : end2 + 3;
+      continue;
+    }
+    if (source.startsWith("<?", index)) {
+      const end2 = source.indexOf("?>", index + 2);
+      index = end2 === -1 ? source.length : end2 + 2;
+      continue;
+    }
+    const end = findTagEnd(source, index + 1);
+    const rawTag = source.slice(index, end + 1);
+    if (rawTag.startsWith("</")) {
+      elements.pop();
+      index = end + 1;
+      continue;
+    }
+    const name = readTagName(rawTag);
+    if (!name) {
+      index = end + 1;
+      continue;
+    }
+    const parent = elements[elements.length - 1];
+    const childIndex = parent?.nextChildIndex ?? 0;
+    if (parent) parent.nextChildIndex += 1;
+    const path = parent ? [...parent.path, `${name}[${childIndex}]`] : [name];
+    const attributes = readTagAttributes(rawTag, name);
+    const normalizedName = name.toLowerCase();
+    if (normalizedName === "mark" || normalizedName === "bookmark") {
+      const markerName = attributes[normalizedName === "mark" ? "name" : "mark"];
+      if (markerName) {
+        markers.push({
+          kind: normalizedName,
+          name: markerName,
+          originalTextRange: { start: textOffset, end: textOffset },
+          sourceNodePath: [...path]
+        });
+      }
+    }
+    if (!/\/\s*>$/.test(rawTag)) elements.push({ name, path, nextChildIndex: 0 });
+    index = end + 1;
+  }
+  return { text: textParts.join(""), segments, markers };
+}
+function getSsmlSourceMap(ssml) {
+  parseSsml(ssml);
+  return collectSourceMap(ssml);
+}
+function extractSsmlText(ssml) {
+  parseSsml(ssml);
+  return collectTextNodes(ssml).map((node) => node.decodedText);
+}
+async function mapSsmlTextNodes(ssml, transform, options = {}) {
+  parseSsml(ssml);
+  const nodes = collectTextNodes(ssml);
+  const skipTags = new Set((options.skipTags ?? ["phoneme", "say-as", "sub"]).map((tag) => tag.toLowerCase()));
+  const replacements = await Promise.all(
+    nodes.map(async (node) => {
+      const context = {
+        ancestorTags: [...node.context.ancestorTags],
+        parentAttributes: { ...node.context.parentAttributes },
+        parentTag: node.context.parentTag,
+        path: [...node.context.path]
+      };
+      const shouldTransform = !skipTags.has(context.parentTag.toLowerCase()) && (options.filter?.(context) ?? true);
+      if (!shouldTransform) return ssml.slice(node.sourceStart, node.sourceEnd);
+      const transformed = await transform(node.decodedText, context);
+      if (typeof transformed !== "string") {
+        throw new TypeError("SSML text node transform must return a string");
+      }
+      return transformed === node.decodedText ? ssml.slice(node.sourceStart, node.sourceEnd) : encodeXmlText(transformed);
+    })
+  );
+  let result = "";
+  let cursor = 0;
+  nodes.forEach((node, nodeIndex) => {
+    result += ssml.slice(cursor, node.sourceStart) + replacements[nodeIndex];
+    cursor = node.sourceEnd;
+  });
+  return result + ssml.slice(cursor);
+}
+
 // packages/ssml-core/src/split.ts
 var DEFAULT_MAX_LENGTH = 1e4;
 function cloneElement(element, children) {
@@ -1076,7 +1308,35 @@ function collectInheritedContext(nodes) {
   nodes.forEach(visit);
   return context;
 }
-function createChunk(document, nodes, chunkIndex, textStart, backgroundAudio, replicateBackgroundAudio) {
+function elementName(node) {
+  return node.type === "custom" || node.type === "element" ? node.name : node.type;
+}
+function findSourceNodePath(nodes, targetOffset) {
+  let textOffset = 0;
+  let firstPath;
+  let foundPath;
+  const visit = (node, path) => {
+    if (typeof node === "string" || node.type === "text") {
+      const text = typeof node === "string" ? node : node.value;
+      if (text && firstPath === void 0) firstPath = [...path];
+      if (text && foundPath === void 0 && targetOffset < textOffset + text.length) foundPath = [...path];
+      textOffset += text.length;
+      return;
+    }
+    node.children?.forEach((child, index) => {
+      const childPath = typeof child === "string" || child.type === "text" ? path : [...path, `${elementName(child)}[${index}]`];
+      visit(child, childPath);
+    });
+  };
+  nodes.forEach((node, index) => {
+    if (!foundPath) {
+      if (typeof node === "string" || node.type === "text") visit(node, ["speak"]);
+      else visit(node, ["speak", `${elementName(node)}[${index}]`]);
+    }
+  });
+  return foundPath ?? firstPath;
+}
+function createChunk(document, nodes, chunkIndex, textStart, backgroundAudio, replicateBackgroundAudio, sourceMap, includeEndMarkers) {
   const chunkNodes = backgroundAudio && (replicateBackgroundAudio || chunkIndex === 0) ? [backgroundAudio, ...nodes] : nodes;
   const text = nodes.map(textFromNode).join("");
   const marks = [];
@@ -1091,7 +1351,24 @@ function createChunk(document, nodes, chunkIndex, textStart, backgroundAudio, re
     containedMarks: marks,
     hasBackgroundAudio: chunkNodes.some(
       (node) => typeof node !== "string" && node.type !== "text" && node.type === "mstts:backgroundaudio"
-    )
+    ),
+    sourceNodePath: findSourceNodePath(document.children ?? [], textStart),
+    sourceTextSegments: sourceMap.segments.filter(({ range }) => range.end > textStart && range.start < textStart + text.length).map((segment) => {
+      const start = Math.max(segment.range.start, textStart);
+      const end = Math.min(segment.range.end, textStart + text.length);
+      return {
+        text: segment.text.slice(start - segment.range.start, end - segment.range.start),
+        range: { start, end },
+        sourceNodePath: [...segment.sourceNodePath]
+      };
+    }),
+    sourceMarkers: sourceMap.markers.filter(
+      ({ originalTextRange }) => originalTextRange.start >= textStart && (originalTextRange.start < textStart + text.length || includeEndMarkers && originalTextRange.start === textStart + text.length)
+    ).map((marker) => ({
+      ...marker,
+      originalTextRange: { ...marker.originalTextRange },
+      sourceNodePath: [...marker.sourceNodePath]
+    }))
   };
 }
 function splitSsmlDocument(ssml, maxLength = DEFAULT_MAX_LENGTH, options = {}) {
@@ -1101,11 +1378,12 @@ function splitSsmlDocument(ssml, maxLength = DEFAULT_MAX_LENGTH, options = {}) {
     throw new RangeError("maxLength must be a positive integer");
   }
   const document = parseSsml(ssml);
+  const sourceMap = getSsmlSourceMap(ssml);
   const backgroundAudio = (document.children ?? []).find(
     (node) => typeof node !== "string" && node.type !== "text" && node.type === "mstts:backgroundaudio"
   );
   if (ssml.length <= resolvedMaxLength) {
-    return [createChunk(document, document.children ?? [], 0, 0, backgroundAudio, true)];
+    return [createChunk(document, document.children ?? [], 0, 0, backgroundAudio, true, sourceMap, true)];
   }
   const contentChildren = (document.children ?? []).filter((node) => node !== backgroundAudio);
   const plainDocumentLength = documentWithChildren(document, []).length;
@@ -1129,7 +1407,16 @@ function splitSsmlDocument(ssml, maxLength = DEFAULT_MAX_LENGTH, options = {}) {
   }
   if (group.length > 0) chunks.push(group);
   if (chunks.length === 0) {
-    const result = createChunk(document, [], 0, 0, backgroundAudio, resolvedOptions.replicateBackgroundAudio ?? false);
+    const result = createChunk(
+      document,
+      [],
+      0,
+      0,
+      backgroundAudio,
+      resolvedOptions.replicateBackgroundAudio ?? false,
+      sourceMap,
+      true
+    );
     if (result.ssml.length > resolvedMaxLength) {
       throw new RangeError("maxLength is too small to contain the SSML document wrapper");
     }
@@ -1143,7 +1430,9 @@ function splitSsmlDocument(ssml, maxLength = DEFAULT_MAX_LENGTH, options = {}) {
       chunkIndex,
       textStart,
       backgroundAudio,
-      resolvedOptions.replicateBackgroundAudio ?? false
+      resolvedOptions.replicateBackgroundAudio ?? false,
+      sourceMap,
+      chunkIndex === chunks.length - 1
     );
     textStart = result.originalTextRange.end;
     return result;
@@ -1166,158 +1455,9 @@ function validateSsml(xmlString) {
   }
 }
 
-// packages/ssml-core/src/textNodes.ts
-function decodeXmlText(value) {
-  return value.replace(/&(?:amp|apos|gt|lt|quot);|&#(?:x[\da-f]+|\d+);/gi, (entity) => {
-    if (entity === "&amp;") return "&";
-    if (entity === "&apos;") return "'";
-    if (entity === "&gt;") return ">";
-    if (entity === "&lt;") return "<";
-    if (entity === "&quot;") return '"';
-    const hexadecimal = entity.toLowerCase().startsWith("&#x");
-    const digits = entity.slice(hexadecimal ? 3 : 2, -1);
-    return String.fromCodePoint(Number.parseInt(digits, hexadecimal ? 16 : 10));
-  });
-}
-function encodeXmlText(value) {
-  return value.replace(/&/g, "&amp;").replace(/</g, "&lt;").replace(/>/g, "&gt;");
-}
-function decodeXmlAttribute(value) {
-  return decodeXmlText(value);
-}
-function findTagEnd(source, start) {
-  let quote = "";
-  for (let index = start; index < source.length; index += 1) {
-    const character = source[index];
-    if (quote) {
-      if (character === quote) quote = "";
-    } else if (character === '"' || character === "'") {
-      quote = character;
-    } else if (character === ">") {
-      return index;
-    }
-  }
-  return source.length - 1;
-}
-function readTagName(tag) {
-  const match = /^<\s*([A-Za-z_][A-Za-z0-9_.:-]*)/.exec(tag);
-  return match?.[1];
-}
-function readTagAttributes(tag, name) {
-  const attributes = {};
-  const nameStart = tag.indexOf(name);
-  const attributeSource = tag.slice(nameStart + name.length, tag.length - 1).replace(/\/\s*$/, "");
-  const attributePattern = /([A-Za-z_][A-Za-z0-9_.:-]*)\s*=\s*(["'])([\s\S]*?)\2/g;
-  for (const match of attributeSource.matchAll(attributePattern)) {
-    attributes[match[1].toLowerCase()] = decodeXmlAttribute(match[3]);
-  }
-  return attributes;
-}
-function collectTextNodes(source) {
-  const nodes = [];
-  const elements = [];
-  let index = 0;
-  const addText = (start, end, rawText, sourceStart = start, sourceEnd = end) => {
-    if (!rawText) return;
-    const path = elements.map((element) => element.name);
-    const parent = elements[elements.length - 1];
-    nodes.push({
-      context: {
-        ancestorTags: path.slice(0, -1),
-        parentAttributes: { ...parent?.attributes ?? {} },
-        parentTag: parent?.name ?? "",
-        path
-      },
-      decodedText: decodeXmlText(rawText),
-      end,
-      sourceEnd,
-      sourceStart,
-      start
-    });
-  };
-  while (index < source.length) {
-    if (source[index] !== "<") {
-      const nextTag = source.indexOf("<", index);
-      const end2 = nextTag === -1 ? source.length : nextTag;
-      addText(index, end2, source.slice(index, end2));
-      index = end2;
-      continue;
-    }
-    if (source.startsWith("<!--", index)) {
-      const end2 = source.indexOf("-->", index + 4);
-      index = end2 === -1 ? source.length : end2 + 3;
-      continue;
-    }
-    if (source.startsWith("<![CDATA[", index)) {
-      const contentStart = index + 9;
-      const end2 = source.indexOf("]]>", contentStart);
-      const contentEnd = end2 === -1 ? source.length : end2;
-      addText(
-        contentStart,
-        contentEnd,
-        source.slice(contentStart, contentEnd),
-        index,
-        end2 === -1 ? source.length : end2 + 3
-      );
-      index = end2 === -1 ? source.length : end2 + 3;
-      continue;
-    }
-    if (source.startsWith("<?", index)) {
-      const end2 = source.indexOf("?>", index + 2);
-      index = end2 === -1 ? source.length : end2 + 2;
-      continue;
-    }
-    if (source.startsWith("</", index)) {
-      const end2 = findTagEnd(source, index + 2);
-      elements.pop();
-      index = end2 + 1;
-      continue;
-    }
-    const end = findTagEnd(source, index + 1);
-    const tag = source.slice(index, end + 1);
-    const name = readTagName(tag);
-    if (name && !/\/\s*>$/.test(tag)) elements.push({ attributes: readTagAttributes(tag, name), name });
-    index = end + 1;
-  }
-  return nodes;
-}
-function extractSsmlText(ssml) {
-  parseSsml(ssml);
-  return collectTextNodes(ssml).map((node) => node.decodedText);
-}
-async function mapSsmlTextNodes(ssml, transform, options = {}) {
-  parseSsml(ssml);
-  const nodes = collectTextNodes(ssml);
-  const skipTags = new Set((options.skipTags ?? ["phoneme", "say-as", "sub"]).map((tag) => tag.toLowerCase()));
-  const replacements = await Promise.all(
-    nodes.map(async (node) => {
-      const context = {
-        ancestorTags: [...node.context.ancestorTags],
-        parentAttributes: { ...node.context.parentAttributes },
-        parentTag: node.context.parentTag,
-        path: [...node.context.path]
-      };
-      const shouldTransform = !skipTags.has(context.parentTag.toLowerCase()) && (options.filter?.(context) ?? true);
-      if (!shouldTransform) return ssml.slice(node.sourceStart, node.sourceEnd);
-      const transformed = await transform(node.decodedText, context);
-      if (typeof transformed !== "string") {
-        throw new TypeError("SSML text node transform must return a string");
-      }
-      return transformed === node.decodedText ? ssml.slice(node.sourceStart, node.sourceEnd) : encodeXmlText(transformed);
-    })
-  );
-  let result = "";
-  let cursor = 0;
-  nodes.forEach((node, nodeIndex) => {
-    result += ssml.slice(cursor, node.sourceStart) + replacements[nodeIndex];
-    cursor = node.sourceEnd;
-  });
-  return result + ssml.slice(cursor);
-}
-
 // packages/ssml-core/src/migration.ts
 var DEFAULT_TRANSLATION_SKIP_TAGS = ["phoneme", "say-as", "sayAs", "sub"];
-function elementName(element) {
+function elementName2(element) {
   switch (element.type) {
     case "custom":
     case "element":
@@ -1470,7 +1610,7 @@ function extractSsmlTranslatableText(ssml, options = {}) {
         }
         return;
       }
-      const tag = elementName(node);
+      const tag = elementName2(node);
       if (skipTags.has(tag.toLowerCase())) return;
       visit(childrenOf(node), [...ancestors, tag], [...path, String(index)]);
     });
@@ -1516,7 +1656,7 @@ function serializeDocument2(document) {
   const serialize = (node) => {
     if (typeof node === "string") return node.replace(/&/g, "&amp;").replace(/</g, "&lt;").replace(/>/g, "&gt;");
     if (node.type === "text") return serialize(node.value);
-    const tag = elementName(node);
+    const tag = elementName2(node);
     const nodeAttributes = elementAttributes(node);
     const serializedAttributes = Object.entries(nodeAttributes).map(([name, value]) => ` ${name}="${serialize(value).replace(/"/g, "&quot;")}"`).join("");
     const children = childrenOf(node).map(serialize).join("");
@@ -1530,7 +1670,7 @@ function flatten(document) {
     nodes.forEach((node, index) => {
       if (typeof node === "string" || node.type === "text") return;
       const currentPath = `${path}/${index}`;
-      result.push({ name: elementName(node), attributes: elementAttributes(node), path: currentPath });
+      result.push({ name: elementName2(node), attributes: elementAttributes(node), path: currentPath });
       visit(childrenOf(node), currentPath);
     });
   };
@@ -1728,6 +1868,86 @@ var AZURE_VOICE_DEFINITIONS = [
 ];
 
 // packages/ssml-core/src/azureValidation.ts
+function createAzureUrlValidatorRunner(validator, options = {}) {
+  if (typeof validator !== "function") throw new TypeError("A URL validator function is required.");
+  const concurrency = options.concurrency === void 0 ? Infinity : Number.isFinite(options.concurrency) ? Math.max(1, Math.floor(options.concurrency)) : Infinity;
+  const cache = options.cache ?? /* @__PURE__ */ new Map();
+  const inFlight = /* @__PURE__ */ new Map();
+  const waiters = [];
+  let active = 0;
+  const configuredSignal = options.signal ?? new AbortController().signal;
+  const acquire = async (signal) => {
+    if (signal.aborted) throw new Error("URL validation was aborted.");
+    if (active < concurrency) {
+      active += 1;
+      return;
+    }
+    await new Promise((resolve, reject) => {
+      let waiter;
+      const abortHandler = () => {
+        const index = waiters.indexOf(waiter);
+        if (index >= 0) waiters.splice(index, 1);
+        signal.removeEventListener("abort", abortHandler);
+        reject(new Error("URL validation was aborted."));
+      };
+      signal.addEventListener("abort", abortHandler, { once: true });
+      waiter = () => {
+        signal.removeEventListener("abort", abortHandler);
+        resolve();
+      };
+      waiters.push(waiter);
+    });
+    active += 1;
+  };
+  const release = () => {
+    active -= 1;
+    waiters.shift()?.();
+  };
+  const check = async (url, context, signal = configuredSignal) => {
+    if (signal.aborted) throw new Error("URL validation was aborted.");
+    const key = `${context.tag}:${context.attribute}:${url}`;
+    const cached = cache.get(key);
+    if (cached !== void 0) return cached;
+    const existing = inFlight.get(key);
+    if (existing) return existing;
+    const promise = (async () => {
+      await acquire(signal);
+      try {
+        if (signal.aborted) throw new Error("URL validation was aborted.");
+        const validation = Promise.resolve(validator(url, context, signal));
+        let timer;
+        let abortHandler;
+        const cancellation = new Promise((_resolve, reject) => {
+          abortHandler = () => reject(new Error("URL validation was aborted."));
+          signal.addEventListener("abort", abortHandler, { once: true });
+          if (options.timeoutMs !== void 0 && options.timeoutMs > 0) {
+            timer = setTimeout(
+              () => reject(new Error(`URL validation timed out after ${options.timeoutMs} ms.`)),
+              options.timeoutMs
+            );
+          }
+        });
+        try {
+          const result = await (timer || abortHandler ? Promise.race([validation, cancellation]) : validation);
+          cache.set(key, result);
+          return result;
+        } finally {
+          if (timer) clearTimeout(timer);
+          if (abortHandler) signal.removeEventListener("abort", abortHandler);
+        }
+      } finally {
+        release();
+      }
+    })();
+    inFlight.set(key, promise);
+    try {
+      return await promise;
+    } finally {
+      inFlight.delete(key);
+    }
+  };
+  return (url, context, signal) => check(url, context, signal ?? configuredSignal);
+}
 var ALLOWED_BREAK_STRENGTHS = /* @__PURE__ */ new Set(["none", "x-weak", "weak", "medium", "strong", "x-strong"]);
 var ALLOWED_SAY_AS = /* @__PURE__ */ new Set([
   "characters",
@@ -2022,23 +2242,23 @@ function validateVoiceFeatureMatrix(token, source, diagnostics, voiceName, defin
     );
   }
 }
-function validateAudioSource(token, source, diagnostics, options, elementName2) {
+function validateAudioSource(token, source, diagnostics, options, elementName3) {
   const src = attr(token, "src");
   if (!src) {
-    addDiagnostic(diagnostics, source, token.start, `<${elementName2}> requires a "src" attribute.`);
+    addDiagnostic(diagnostics, source, token.start, `<${elementName3}> requires a "src" attribute.`);
     return;
   }
   let parsed;
   try {
     parsed = new URL(src);
   } catch {
-    addDiagnostic(diagnostics, source, token.start, `<${elementName2} src> must be an absolute HTTP(S) URL.`);
+    addDiagnostic(diagnostics, source, token.start, `<${elementName3} src> must be an absolute HTTP(S) URL.`);
     return;
   }
   if (parsed.username || parsed.password)
-    addDiagnostic(diagnostics, source, token.start, `<${elementName2} src> must not contain URL credentials.`);
+    addDiagnostic(diagnostics, source, token.start, `<${elementName3} src> must not contain URL credentials.`);
   if (parsed.protocol !== "https:" && !(options.allowHttpAudio && parsed.protocol === "http:"))
-    addDiagnostic(diagnostics, source, token.start, `<${elementName2} src> must use HTTPS.`);
+    addDiagnostic(diagnostics, source, token.start, `<${elementName3} src> must use HTTPS.`);
   const isAllowedOrigin = options.allowedAudioOrigins?.some((allowedOrigin) => {
     try {
       const configured = new URL(allowedOrigin);
@@ -2050,13 +2270,13 @@ function validateAudioSource(token, source, diagnostics, options, elementName2)
     }
   }) ?? false;
   if (options.allowedAudioOrigins && !isAllowedOrigin)
-    addDiagnostic(diagnostics, source, token.start, `<${elementName2} src> origin "${parsed.origin}" is not allowed.`);
+    addDiagnostic(diagnostics, source, token.start, `<${elementName3} src> origin "${parsed.origin}" is not allowed.`);
   else if (!isAllowedOrigin && !options.allowExternalAudio)
     addDiagnostic(
       diagnostics,
       source,
       token.start,
-      `<${elementName2} src> external origin "${parsed.origin}" is blocked by default; set allowExternalAudio to true or provide allowedAudioOrigins.`
+      `<${elementName3} src> external origin "${parsed.origin}" is blocked by default; set allowExternalAudio to true or provide allowedAudioOrigins.`
     );
 }
 function validateElement(token, source, diagnostics, voiceName, options, voiceCatalog) {
@@ -2374,6 +2594,15 @@ function validateAzureSsml(ssml, options = {}) {
   const diagnostics = validateAzureSsmlStatic(ssml, options);
   const validator = options.urlValidator ?? options.customUrlValidator;
   if (!validator || typeof ssml !== "string") return diagnostics;
+  const runnerOptions = options.urlValidation ?? {};
+  const boundedValidator = createAzureUrlValidatorRunner(validator, {
+    ...runnerOptions,
+    ...options.urlValidatorConcurrency !== void 0 ? { concurrency: options.urlValidatorConcurrency } : {},
+    ...options.urlValidatorTimeoutMs !== void 0 ? { timeoutMs: options.urlValidatorTimeoutMs } : {},
+    ...options.urlValidatorSignal ? { signal: options.urlValidatorSignal } : {},
+    ...options.urlValidatorCache ? { cache: options.urlValidatorCache } : {}
+  });
+  const validationSignal = options.urlValidatorSignal ?? options.urlValidation?.signal ?? new AbortController().signal;
   let tokens;
   try {
     tokens = tokenizeElements(ssml);
@@ -2383,7 +2612,11 @@ function validateAzureSsml(ssml, options = {}) {
   const checks = tokens.flatMap(
     (token) => urlAttributes(token).map(async ({ attribute, value }) => {
       try {
-        const result = await validator(value, { tag: token.name, attribute });
+        const result = await boundedValidator(
+          value,
+          { tag: token.name, attribute, ...options.sourceNodePath ? { sourceNodePath: options.sourceNodePath } : {} },
+          validationSignal
+        );
         const valid = typeof result === "boolean" ? result : result.valid;
         if (!valid) {
           const reason = typeof result === "boolean" ? void 0 : result.reason;
@@ -2429,11 +2662,13 @@ var getBuiltInVoiceCatalogMetadata = getAzureVoiceCatalogMetadata;
   areAzureLanguagesEquivalent,
   buildPartialSsml,
   buildSsml,
+  createAzureUrlValidatorRunner,
   extractSsmlText,
   extractSsmlTranslatableText,
   fromPlainTextToSsml,
   getAzureVoiceCatalogMetadata,
   getBuiltInVoiceCatalogMetadata,
+  getSsmlSourceMap,
   isValidAzureAudioDuration,
   mapSsmlTextNodes,
   normalizeAzureLanguage,