opensteer 0.9.4 → 0.9.5

package/dist/index.cjs

@@ -4445,10 +4445,10 @@ var opensteerDomScrollInputSchema = objectSchema(
     required: ["target", "direction", "amount"]
   }
 );
-var opensteerExtractSchemaSchema = objectSchema(
+var opensteerExtractTemplateSchema = objectSchema(
   {},
   {
-    title: "OpensteerExtractSchema",
+    title: "OpensteerExtractTemplate",
     additionalProperties: true
   }
 );
@@ -4456,13 +4456,13 @@ var opensteerDomExtractInputSchema = defineSchema({
   ...objectSchema(
     {
       persist: stringSchema(),
-      schema: opensteerExtractSchemaSchema
+      template: opensteerExtractTemplateSchema
     },
     {
       title: "OpensteerDomExtractInput"
     }
   ),
-  anyOf: [defineSchema({ required: ["persist"] }), defineSchema({ required: ["schema"] })]
+  anyOf: [defineSchema({ required: ["persist"] }), defineSchema({ required: ["template"] })]
 });
 var jsonValueSchema2 = recordSchema({}, { title: "JsonValueRecord" });
 var opensteerDomExtractOutputSchema = objectSchema(
@@ -6912,6 +6912,7 @@ var defaultNavigationSettleObserver = {
       return false;
     }
     try {
+      const startedAt = Date.now();
       await input.engine.waitForPostLoadQuiet({
         pageRef: input.pageRef,
         timeoutMs: effectiveTimeout,
@@ -6919,9 +6920,13 @@ var defaultNavigationSettleObserver = {
         captureWindowMs: Math.min(NAVIGATION_POST_LOAD_CAPTURE_WINDOW_MS, effectiveTimeout),
         signal: input.signal
       });
+      const visualTimeout = Math.max(0, effectiveTimeout - (Date.now() - startedAt));
+      if (visualTimeout <= 0) {
+        return true;
+      }
       await input.engine.waitForVisualStability({
         pageRef: input.pageRef,
-        timeoutMs: effectiveTimeout,
+        timeoutMs: visualTimeout,
         settleMs: profile.settleMs,
         scope: profile.scope
       });
@@ -11200,9 +11205,9 @@ function isPersistedObjectNode(node) {
 }
 
 // ../runtime-core/src/sdk/extraction.ts
-function assertValidOpensteerExtractionSchemaRoot(schema) {
-  if (!schema || typeof schema !== "object" || Array.isArray(schema)) {
-    throw new Error("Invalid extraction schema: expected a JSON object at the top level.");
+function assertValidOpensteerExtractionTemplateRoot(template) {
+  if (!template || typeof template !== "object" || Array.isArray(template)) {
+    throw new Error("Invalid extraction template: expected a JSON object at the top level.");
   }
 }
 function isPersistedOpensteerExtractionValueNode2(value) {
@@ -11224,12 +11229,12 @@ function isPersistedOpensteerExtractionArrayNode2(value) {
   return "$array" in value;
 }
 async function compileOpensteerExtractionFieldTargets(options) {
-  assertValidOpensteerExtractionSchemaRoot(options.schema);
+  assertValidOpensteerExtractionTemplateRoot(options.template);
   const fields = [];
-  await collectFieldTargetsFromSchemaObject({
+  await collectFieldTargetsFromTemplateObject({
     dom: options.dom,
     pageRef: options.pageRef,
-    value: options.schema,
+    value: options.template,
     path: "",
     fields,
     insideArray: false
@@ -11281,13 +11286,13 @@ function createOpensteerExtractionDescriptorStore(options) {
   }
   return new MemoryOpensteerExtractionDescriptorStore(namespace);
 }
-async function collectFieldTargetsFromSchemaObject(options) {
+async function collectFieldTargetsFromTemplateObject(options) {
   for (const [key, childValue] of Object.entries(options.value)) {
     const normalizedKey = normalizeKey(key);
     if (!normalizedKey) {
       continue;
     }
-    await collectFieldTargetsFromSchemaValue({
+    await collectFieldTargetsFromTemplateValue({
       dom: options.dom,
       pageRef: options.pageRef,
       value: childValue,
@@ -11297,8 +11302,8 @@ async function collectFieldTargetsFromSchemaObject(options) {
     });
   }
 }
-async function collectFieldTargetsFromSchemaValue(options) {
-  const normalizedField = normalizeSchemaField(options.value);
+async function collectFieldTargetsFromTemplateValue(options) {
+  const normalizedField = normalizeTemplateField(options.value);
   if (normalizedField !== null) {
     options.fields.push(
       await compileFieldTarget({
@@ -11313,12 +11318,12 @@ async function collectFieldTargetsFromSchemaValue(options) {
   if (Array.isArray(options.value)) {
     if (options.insideArray) {
       throw new Error(
-        `Nested arrays are not supported in extraction schema at "${labelForPath(options.path)}".`
+        `Nested arrays are not supported in extraction template at "${labelForPath(options.path)}".`
       );
     }
     if (options.value.length === 0) {
       throw new Error(
-        `Extraction array "${labelForPath(options.path)}" must include at least one representative item.`
+        `Extraction array "${labelForPath(options.path)}" must include at least one representative template item.`
       );
     }
     for (let index = 0; index < options.value.length; index += 1) {
@@ -11329,7 +11334,7 @@ async function collectFieldTargetsFromSchemaValue(options) {
         );
       }
       const fieldCountBeforeItem = options.fields.length;
-      await collectFieldTargetsFromSchemaObject({
+      await collectFieldTargetsFromTemplateObject({
         dom: options.dom,
         pageRef: options.pageRef,
         value: itemValue,
@@ -11340,7 +11345,7 @@ async function collectFieldTargetsFromSchemaValue(options) {
       const itemFields = options.fields.slice(fieldCountBeforeItem);
       if (!itemFields.some((field) => !("source" in field))) {
         throw new Error(
-          `Extraction array "${labelForPath(options.path)}" item ${String(index)} must include at least one element- or selector-backed field.`
+          `Extraction array "${labelForPath(options.path)}" item ${String(index)} must include at least one element number or selector field.`
         );
       }
     }
@@ -11348,10 +11353,10 @@ async function collectFieldTargetsFromSchemaValue(options) {
   }
   if (!options.value || typeof options.value !== "object") {
     throw new Error(
-      `Invalid extraction schema value at "${labelForPath(options.path)}": expected an object, array, or field descriptor.`
+      `Invalid extraction template value at "${labelForPath(options.path)}": expected an object, array, or field descriptor.`
     );
   }
-  await collectFieldTargetsFromSchemaObject({
+  await collectFieldTargetsFromTemplateObject({
     dom: options.dom,
     pageRef: options.pageRef,
     value: options.value,
@@ -11383,7 +11388,7 @@ async function compileFieldTarget(options) {
     path: await resolveSelectorFieldPath({
       dom: options.dom,
       pageRef: options.pageRef,
-      selector: `[c="${String(options.field.element)}"]`
+      selector: `[c="${String(options.field.c)}"]`
     }),
     ...options.field.attribute === void 0 ? {} : { attribute: options.field.attribute }
   };
@@ -11684,24 +11689,29 @@ function countNonNullLeaves(value) {
   }
   return Object.values(value).reduce((sum, item) => sum + countNonNullLeaves(item), 0);
 }
-function normalizeSchemaField(value) {
+function normalizeTemplateField(value) {
+  if (typeof value === "number") {
+    return {
+      c: normalizeExtractionCounter(value)
+    };
+  }
   if (!value || typeof value !== "object" || Array.isArray(value)) {
     return null;
   }
   const raw = value;
-  const hasElement = raw.element !== void 0;
+  const hasCounter = raw.c !== void 0 || raw.element !== void 0;
   const hasSelector = raw.selector !== void 0;
   const hasSource = raw.source !== void 0;
-  const targetCount = Number(hasElement) + Number(hasSelector) + Number(hasSource);
+  const targetCount = Number(hasCounter) + Number(hasSelector) + Number(hasSource);
   if (targetCount === 0) {
     return null;
   }
   if (targetCount !== 1) {
     throw new Error(
-      "Extraction field descriptors must specify exactly one of element, selector, or source."
+      "Extraction field descriptors must specify exactly one of c/element, selector, or source."
     );
   }
-  const attribute = raw.attribute === void 0 ? void 0 : normalizeNonEmptyString2("attribute", raw.attribute);
+  const attribute = raw.attr !== void 0 ? normalizeNonEmptyString2("attr", raw.attr) : raw.attribute === void 0 ? void 0 : normalizeNonEmptyString2("attribute", raw.attribute);
   if (hasSource) {
     if (raw.source !== "current_url") {
       throw new Error(`Unsupported extraction source "${String(raw.source)}".`);
@@ -11716,17 +11726,20 @@ function normalizeSchemaField(value) {
       ...attribute === void 0 ? {} : { attribute }
     };
   }
-  const element = Number(raw.element);
-  if (!Number.isInteger(element) || element < 1) {
-    throw new Error(
-      `Extraction field element must be a positive integer, received ${String(raw.element)}.`
-    );
-  }
   return {
-    element,
+    c: normalizeExtractionCounter(raw.c ?? raw.element),
     ...attribute === void 0 ? {} : { attribute }
   };
 }
+function normalizeExtractionCounter(value) {
+  const counter = Number(value);
+  if (!Number.isInteger(counter) || counter < 1) {
+    throw new Error(
+      `Extraction element number must be a positive integer, received ${String(value)}.`
+    );
+  }
+  return counter;
+}
 function normalizeNamespace(namespace) {
   const normalized = String(namespace ?? "default").trim();
   return normalized.length === 0 ? "default" : normalized;
@@ -11757,7 +11770,7 @@ function parseExtractionDescriptorRecord(record) {
       kind: "dom-extraction",
       persist: raw.persist,
       root,
-      ...typeof raw.schemaHash === "string" ? { schemaHash: raw.schemaHash } : {},
+      ...typeof raw.templateHash === "string" ? { templateHash: raw.templateHash } : typeof raw.schemaHash === "string" ? { templateHash: raw.schemaHash } : {},
       ...typeof raw.sourceUrl === "string" ? { sourceUrl: raw.sourceUrl } : {}
     }
   };
@@ -11843,7 +11856,7 @@ var FilesystemOpensteerExtractionDescriptorStore = class {
       kind: "dom-extraction",
       persist: input.persist,
       root: input.root,
-      ...input.schemaHash === void 0 ? {} : { schemaHash: input.schemaHash },
+      ...input.templateHash === void 0 ? {} : { templateHash: input.templateHash },
       ...input.sourceUrl === void 0 ? {} : { sourceUrl: input.sourceUrl }
     };
     const key = persistKey(this.namespace, input.persist);
@@ -11892,7 +11905,7 @@ var MemoryOpensteerExtractionDescriptorStore = class {
       kind: "dom-extraction",
       persist: input.persist,
       root: input.root,
-      ...input.schemaHash === void 0 ? {} : { schemaHash: input.schemaHash },
+      ...input.templateHash === void 0 ? {} : { templateHash: input.templateHash },
       ...input.sourceUrl === void 0 ? {} : { sourceUrl: input.sourceUrl }
     };
     const key = persistKey(this.namespace, input.persist);
@@ -15691,7 +15704,7 @@ function resolveCloudConfig(input = {}) {
 
 // ../runtime-core/package.json
 var package_default = {
-  version: "0.2.3"};
+  version: "0.2.4"};
 
 // ../runtime-core/src/version.ts
 var OPENSTEER_RUNTIME_CORE_VERSION = package_default.version;
@@ -16737,10 +16750,19 @@ var VOID_TAGS = /* @__PURE__ */ new Set([
 // ../runtime-core/src/sdk/snapshot/cleaner.ts
 var STRIP_TAGS = /* @__PURE__ */ new Set(["script", "style", "noscript", "meta", "link", "template"]);
 var TEXT_ATTR_MAX = 150;
-var URL_ATTR_MAX = 500;
-var URL_ATTRS = /* @__PURE__ */ new Set(["href", "src", "srcset"]);
+var SRCSET_ATTR_MAX = 160;
+var MIDDLE_TRUNCATED_URL_ATTRS = /* @__PURE__ */ new Set(["href", "src"]);
 var TEXT_ATTRS = /* @__PURE__ */ new Set(["alt", "title", "aria-label", "placeholder", "value"]);
-var TRUNCATION_SUFFIX = " [truncated]";
+var TRUNCATION_SUFFIX = "...";
+var MIDDLE_TRUNCATION_MARKER = "...";
+var MIDDLE_TRUNCATION_HEAD_MAX = 40;
+var MIDDLE_TRUNCATION_TAIL_MAX = 20;
+var SRCSET_CANDIDATE_HEAD_MAX = 36;
+var SRCSET_CANDIDATE_TAIL_MAX = 12;
+var SRCSET_COMPACT_CANDIDATE_HEAD_MAX = 20;
+var SRCSET_COMPACT_CANDIDATE_TAIL_MAX = 8;
+var SRCSET_FALLBACK_HEAD_MAX = 56;
+var SRCSET_FALLBACK_TAIL_MAX = 20;
 var NOISE_SELECTORS = [
   `[${OPENSTEER_HIDDEN_ATTR}]`,
   "[hidden]",
@@ -16803,19 +16825,260 @@ function truncateValue(value, max) {
   }
   return `${head}${TRUNCATION_SUFFIX}`;
 }
+function takeValueWithinSerializedLengthFromEnd(value, max) {
+  let serializedLength = 0;
+  const chars = [];
+  for (let index = value.length - 1; index >= 0; index -= 1) {
+    const char = value[index];
+    let nextLength = 1;
+    if (char === "&") {
+      nextLength = 5;
+    } else if (char === "<" || char === ">") {
+      nextLength = 4;
+    } else if (char === '"') {
+      nextLength = 6;
+    }
+    if (serializedLength + nextLength > max) {
+      break;
+    }
+    chars.push(char);
+    serializedLength += nextLength;
+  }
+  return chars.reverse().join("");
+}
+function truncateValueInMiddle(value, headMax, tailMax, marker = MIDDLE_TRUNCATION_MARKER) {
+  const markerLength = getSerializedLength(marker);
+  const max = headMax + markerLength + tailMax;
+  if (getSerializedLength(value) <= max) {
+    return value;
+  }
+  const head = takeValueWithinSerializedLength(value, headMax).replace(/\s+$/u, "");
+  const tail = takeValueWithinSerializedLengthFromEnd(value, tailMax).replace(/^\s+/u, "");
+  if (head.length === 0) {
+    return tail.length === 0 ? marker : `${marker}${tail}`;
+  }
+  if (tail.length === 0) {
+    return `${head}${marker}`;
+  }
+  return `${head}${marker}${tail}`;
+}
 function getAttrLimit(attr) {
-  if (URL_ATTRS.has(attr)) {
-    return URL_ATTR_MAX;
+  if (attr === "srcset") {
+    return SRCSET_ATTR_MAX;
   }
   if (TEXT_ATTRS.has(attr)) {
     return TEXT_ATTR_MAX;
   }
   return void 0;
 }
+function shouldBoundAttr(attr) {
+  return MIDDLE_TRUNCATED_URL_ATTRS.has(attr) || getAttrLimit(attr) !== void 0;
+}
 function setBoundedAttr(el, attr, value) {
+  if (MIDDLE_TRUNCATED_URL_ATTRS.has(attr)) {
+    el.attr(
+      attr,
+      truncateValueInMiddle(value, MIDDLE_TRUNCATION_HEAD_MAX, MIDDLE_TRUNCATION_TAIL_MAX)
+    );
+    return;
+  }
   const limit = getAttrLimit(attr);
+  if (attr === "srcset" && limit !== void 0) {
+    el.attr(attr, truncateSrcsetValue(value, limit));
+    return;
+  }
   el.attr(attr, limit === void 0 ? value : truncateValue(value, limit));
 }
+function truncateSrcsetValue(value, max) {
+  if (getSerializedLength(value) <= max) {
+    return value;
+  }
+  const candidates = parseSrcsetCandidates2(value);
+  if (candidates.length === 0) {
+    return truncateValueInMiddle(value, SRCSET_FALLBACK_HEAD_MAX, SRCSET_FALLBACK_TAIL_MAX);
+  }
+  for (const [headMax, tailMax, includeBest] of [
+    [SRCSET_CANDIDATE_HEAD_MAX, SRCSET_CANDIDATE_TAIL_MAX, true],
+    [SRCSET_COMPACT_CANDIDATE_HEAD_MAX, SRCSET_COMPACT_CANDIDATE_TAIL_MAX, true],
+    [SRCSET_COMPACT_CANDIDATE_HEAD_MAX, SRCSET_COMPACT_CANDIDATE_TAIL_MAX, false]
+  ]) {
+    const compact = buildTruncatedSrcsetValue(candidates, headMax, tailMax, includeBest);
+    if (getSerializedLength(compact) <= max) {
+      return compact;
+    }
+  }
+  return truncateValueInMiddle(value, SRCSET_FALLBACK_HEAD_MAX, SRCSET_FALLBACK_TAIL_MAX);
+}
+function buildTruncatedSrcsetValue(candidates, headMax, tailMax, includeBest) {
+  const kept = getPreferredSrcsetCandidateIndices(candidates, includeBest);
+  const parts = [];
+  let previousIndex;
+  for (const candidateIndex of kept) {
+    if (previousIndex !== void 0 && candidateIndex - previousIndex > 1) {
+      parts.push(MIDDLE_TRUNCATION_MARKER);
+    }
+    parts.push(formatSrcsetCandidate(candidates[candidateIndex], headMax, tailMax));
+    previousIndex = candidateIndex;
+  }
+  return parts.join(", ");
+}
+function getPreferredSrcsetCandidateIndices(candidates, includeBest) {
+  if (candidates.length === 0) {
+    return [];
+  }
+  const kept = /* @__PURE__ */ new Set([0, candidates.length - 1]);
+  if (includeBest) {
+    kept.add(pickBestSrcsetCandidateIndex(candidates));
+  }
+  return [...kept].filter((index) => index >= 0 && index < candidates.length).sort((a, b) => a - b);
+}
+function pickBestSrcsetCandidateIndex(candidates) {
+  let bestWidthIndex = -1;
+  let bestWidth = -1;
+  let bestDensityIndex = -1;
+  let bestDensity = -1;
+  for (let index = 0; index < candidates.length; index += 1) {
+    const candidate = candidates[index];
+    if (typeof candidate.width === "number" && Number.isFinite(candidate.width) && candidate.width > bestWidth) {
+      bestWidth = candidate.width;
+      bestWidthIndex = index;
+    }
+    if (typeof candidate.density === "number" && Number.isFinite(candidate.density) && candidate.density > bestDensity) {
+      bestDensity = candidate.density;
+      bestDensityIndex = index;
+    }
+  }
+  if (bestWidthIndex >= 0) {
+    return bestWidthIndex;
+  }
+  if (bestDensityIndex >= 0) {
+    return bestDensityIndex;
+  }
+  return candidates.length - 1;
+}
+function formatSrcsetCandidate(candidate, headMax, tailMax) {
+  const url = truncateValueInMiddle(candidate.url, headMax, tailMax);
+  return candidate.descriptorText ? `${url} ${candidate.descriptorText}` : url;
+}
+function parseSrcsetCandidates2(raw) {
+  const text = raw.trim();
+  if (!text) {
+    return [];
+  }
+  const out = [];
+  let index = 0;
+  while (index < text.length) {
+    index = skipSrcsetSeparators(text, index);
+    if (index >= text.length) {
+      break;
+    }
+    const urlToken = readSrcsetUrlToken(text, index);
+    index = urlToken.nextIndex;
+    const url = urlToken.value.trim();
+    if (!url) {
+      continue;
+    }
+    index = skipSrcsetWhitespace(text, index);
+    const descriptors = [];
+    while (index < text.length && text[index] !== ",") {
+      const descriptorToken = readSrcsetDescriptorToken(text, index);
+      if (!descriptorToken.value) {
+        index = descriptorToken.nextIndex;
+        continue;
+      }
+      descriptors.push(descriptorToken.value);
+      index = descriptorToken.nextIndex;
+      index = skipSrcsetWhitespace(text, index);
+    }
+    if (index < text.length && text[index] === ",") {
+      index += 1;
+    }
+    let width = null;
+    let density = null;
+    for (const descriptor of descriptors) {
+      const token = descriptor.trim().toLowerCase();
+      if (!token) {
+        continue;
+      }
+      const widthMatch = token.match(/^(\d+)w$/);
+      if (widthMatch) {
+        const parsed = Number.parseInt(widthMatch[1], 10);
+        if (Number.isFinite(parsed)) {
+          width = parsed;
+        }
+        continue;
+      }
+      const densityMatch = token.match(/^(\d*\.?\d+)x$/);
+      if (densityMatch) {
+        const parsed = Number.parseFloat(densityMatch[1]);
+        if (Number.isFinite(parsed)) {
+          density = parsed;
+        }
+      }
+    }
+    out.push({
+      url,
+      descriptorText: descriptors.join(" "),
+      width,
+      density
+    });
+  }
+  return out;
+}
+function skipSrcsetWhitespace(value, index) {
+  let cursor = index;
+  while (cursor < value.length && /\s/u.test(value[cursor])) {
+    cursor += 1;
+  }
+  return cursor;
+}
+function skipSrcsetSeparators(value, index) {
+  let cursor = skipSrcsetWhitespace(value, index);
+  while (cursor < value.length && value[cursor] === ",") {
+    cursor += 1;
+    cursor = skipSrcsetWhitespace(value, cursor);
+  }
+  return cursor;
+}
+function readSrcsetUrlToken(value, index) {
+  let cursor = index;
+  let out = "";
+  const isDataUrl = value.slice(index, index + 5).toLowerCase().startsWith("data:");
+  while (cursor < value.length) {
+    const char = value[cursor];
+    if (/\s/u.test(char)) {
+      break;
+    }
+    if (char === "," && !isDataUrl) {
+      break;
+    }
+    out += char;
+    cursor += 1;
+  }
+  if (isDataUrl && out.endsWith(",") && cursor < value.length) {
+    out = out.slice(0, -1);
+  }
+  return {
+    value: out,
+    nextIndex: cursor
+  };
+}
+function readSrcsetDescriptorToken(value, index) {
+  let cursor = skipSrcsetWhitespace(value, index);
+  let out = "";
+  while (cursor < value.length) {
+    const char = value[cursor];
+    if (char === "," || /\s/u.test(char)) {
+      break;
+    }
+    out += char;
+    cursor += 1;
+  }
+  return {
+    value: out.trim(),
+    nextIndex: cursor
+  };
+}
 function removeNoise($) {
   for (const tag of STRIP_TAGS) {
     $(tag).remove();
@@ -16840,38 +17103,68 @@ function markInlineSelfHiddenFallback($) {
   });
 }
 function pruneSelfHiddenNodes($) {
-  const nodes = [];
-  $(`[${OPENSTEER_SELF_HIDDEN_ATTR}]`).each(function collectSelfHiddenNodes() {
-    nodes.push($(this));
-  });
-  nodes.sort((left, right) => right.parents().length - left.parents().length);
-  for (const el of nodes) {
-    if (!el[0]) {
+  for (const node of getElementsInReverseDocumentOrder($)) {
+    if (node.attribs?.[OPENSTEER_SELF_HIDDEN_ATTR] === void 0) {
       continue;
     }
+    const el = $(node);
     el.contents().each(function removeSelfHiddenText() {
       if (this.type === "text") {
         $(this).remove();
       }
     });
-    if (el.children().length === 0) {
+    if (!hasElementChildren(node)) {
       el.remove();
     }
   }
 }
-function hasDirectText($, el) {
-  return el.contents().filter(function hasDirectNodeText() {
-    return this.type === "text" && $(this).text().trim() !== "";
-  }).length > 0;
+function getChildNodes(node) {
+  return node?.children ?? [];
+}
+function isElementLikeNode(node) {
+  return node?.type === "tag" || node?.type === "script" || node?.type === "style";
+}
+function hasDirectText(node) {
+  if (!node) {
+    return false;
+  }
+  for (const child of getChildNodes(node)) {
+    if (child.type === "text" && (child.data || "").trim() !== "") {
+      return true;
+    }
+  }
+  return false;
+}
+function hasElementChildren(node) {
+  if (!node) {
+    return false;
+  }
+  for (const child of getChildNodes(node)) {
+    if (isElementLikeNode(child)) {
+      return true;
+    }
+  }
+  return false;
 }
-function hasTextDeep(el) {
-  return el.text().trim().length > 0;
+function hasTextDeepNode(node) {
+  if (!node) {
+    return false;
+  }
+  if (node.type === "text") {
+    return (node.data || "").trim() !== "";
+  }
+  for (const child of getChildNodes(node)) {
+    if (hasTextDeepNode(child)) {
+      return true;
+    }
+  }
+  return false;
 }
 function hasActionLabel(attrs) {
   return typeof attrs["aria-label"] === "string" && attrs["aria-label"].trim() !== "" || typeof attrs["aria-labelledby"] === "string" && attrs["aria-labelledby"].trim() !== "" || typeof attrs["aria-describedby"] === "string" && attrs["aria-describedby"].trim() !== "" || typeof attrs.title === "string" && attrs.title.trim() !== "" || typeof attrs.placeholder === "string" && attrs.placeholder.trim() !== "" || typeof attrs.value === "string" && attrs.value.trim() !== "";
 }
 function unwrapActionNode($, el) {
-  if (hasTextDeep(el)) {
+  if (hasTextDeepNode(el[0])) {
     if (el.prev().length > 0) {
       el.before(" ");
     }
@@ -16892,7 +17185,7 @@ function stripToAttrs(el, keep) {
     if (typeof value !== "string") {
       continue;
     }
-    if (getAttrLimit(attr) !== void 0) {
+    if (shouldBoundAttr(attr)) {
       setBoundedAttr(el, attr, value);
     }
   }
@@ -16910,6 +17203,9 @@ function restoreBoundedAttr(el, attr, value) {
 function deduplicateImages(html) {
   const seen = /* @__PURE__ */ new Set();
   return html.replace(/<img\b([^>]*)>/gi, (full, attrContent) => {
+    if (/\bc\s*=/.test(attrContent)) {
+      return full;
+    }
     const srcMatch = attrContent.match(/\bsrc\s*=\s*(["']?)(.*?)\1/);
     const srcsetMatch = attrContent.match(/\bsrcset\s*=\s*(["'])(.*?)\1/);
     let src = null;
@@ -16928,59 +17224,155 @@ function deduplicateImages(html) {
     return full;
   });
 }
-function isPreservedImageElement($, el) {
-  const tag = (el[0]?.tagName || "").toLowerCase();
+function hasAttribute2(node, attr) {
+  return node?.attribs?.[attr] !== void 0;
+}
+function hasPictureAncestor(node) {
+  let current = node?.parent;
+  while (current) {
+    if (isElementLikeNode(current) && (current.tagName || "").toLowerCase() === "picture") {
+      return true;
+    }
+    current = current.parent;
+  }
+  return false;
+}
+function pictureHasPreservedDescendant(node) {
+  if (!node) {
+    return false;
+  }
+  for (const child of getChildNodes(node)) {
+    if (!isElementLikeNode(child)) {
+      continue;
+    }
+    const tag = (child.tagName || "").toLowerCase();
+    if (tag === "img") {
+      return true;
+    }
+    if (tag === "source" && typeof child.attribs?.src === "string" && child.attribs.src.trim() !== "") {
+      return true;
+    }
+    if (tag === "source" && typeof child.attribs?.srcset === "string" && child.attribs.srcset.trim() !== "") {
+      return true;
+    }
+    if (pictureHasPreservedDescendant(child)) {
+      return true;
+    }
+  }
+  return false;
+}
+function isPreservedImageElement(node) {
+  const tag = (node?.tagName || "").toLowerCase();
   if (tag === "img") {
     return true;
   }
   if (tag === "picture") {
-    const hasImg = el.find("img").length > 0;
-    const hasSource = el.find("source[src], source[srcset]").length > 0;
-    return hasImg || hasSource;
+    return pictureHasPreservedDescendant(node);
   }
   if (tag === "source") {
-    const inPicture = el.parents("picture").length > 0;
-    const hasSrc = el.attr("src") != null && el.attr("src").trim() !== "" || el.attr("srcset") != null && el.attr("srcset").trim() !== "";
+    const inPicture = hasPictureAncestor(node);
+    const hasSrc = typeof node?.attribs?.src === "string" && node.attribs.src.trim() !== "" || typeof node?.attribs?.srcset === "string" && node.attribs.srcset.trim() !== "";
     return inPicture && hasSrc;
   }
   return false;
 }
+function getElementsInReverseDocumentOrder($) {
+  return $.root().find("*").toArray().reverse().filter((node) => node.type === "tag");
+}
+function getNodeDepth(node) {
+  let depth = 0;
+  let current = node.parent;
+  while (current) {
+    depth++;
+    current = current.parent;
+  }
+  return depth;
+}
+function getElementsByDepthDescending($) {
+  const elements = $.root().find("*").toArray().filter((node) => node.type === "tag");
+  const depths = /* @__PURE__ */ new Map();
+  for (const el of elements) {
+    depths.set(el, getNodeDepth(el));
+  }
+  return elements.sort((a, b) => (depths.get(b) ?? 0) - (depths.get(a) ?? 0));
+}
 function flattenExtractionTree($) {
-  const flatten = (root) => {
-    root.find("*").each(function flattenNode() {
-      const el = $(this);
-      const node = el[0];
-      if (!node) {
-        return;
-      }
-      const tag = (node.tagName || "").toLowerCase();
-      if (ROOT_TAGS.has(tag) || isBoundaryTag(tag)) {
-        return;
-      }
-      if (isPreservedImageElement($, el)) {
-        return;
-      }
-      if (tag === "a") {
-        el.children().each(function flattenAnchorChild() {
-          flatten($(this));
-        });
-        return;
-      }
-      const hasText = hasDirectText($, el);
-      if (hasText) {
-        return;
-      }
-      if (el.children().length === 0) {
-        el.remove();
-        return;
+  for (const node of getElementsInReverseDocumentOrder($)) {
+    const el = $(node);
+    const tag = (node.tagName || "").toLowerCase();
+    if (ROOT_TAGS.has(tag) || isBoundaryTag(tag) || isPreservedImageElement(node)) {
+      continue;
+    }
+    if (tag === "a" || hasDirectText(node)) {
+      continue;
+    }
+    if (!hasElementChildren(node)) {
+      el.remove();
+      continue;
+    }
+    el.replaceWith(el.contents());
+  }
+}
+function hasMarkedAncestor(el, attr) {
+  let current = el[0]?.parent;
+  while (current) {
+    if (!isElementLikeNode(current)) {
+      return false;
+    }
+    if (current.attribs?.[attr] !== void 0) {
+      return true;
+    }
+    current = current.parent;
+  }
+  return false;
+}
+function isIndicatorImage(node) {
+  return (node?.tagName || "").toLowerCase() === "img" && (hasAttribute2(node, "alt") || hasAttribute2(node, "src") || hasAttribute2(node, "srcset"));
+}
+function isIndicatorPictureSource(node) {
+  return (node?.tagName || "").toLowerCase() === "source" && hasPictureAncestor(node) && (hasAttribute2(node, "src") || hasAttribute2(node, "srcset"));
+}
+function isSemanticIndicator(node) {
+  const tag = (node?.tagName || "").toLowerCase();
+  if (tag === "svg") {
+    return true;
+  }
+  return hasAttribute2(node, "aria-label") || hasAttribute2(node, "title") || hasAttribute2(node, "data-icon") || node?.attribs?.role === "img";
+}
+function findIndicatorDescendant(root) {
+  if (!root) {
+    return void 0;
+  }
+  let firstImage;
+  let firstSource;
+  let firstSemantic;
+  const visit = (node) => {
+    if (!isElementLikeNode(node)) {
+      return false;
+    }
+    if (isIndicatorImage(node)) {
+      firstImage = node;
+      return true;
+    }
+    if (firstSource === void 0 && isIndicatorPictureSource(node)) {
+      firstSource = node;
+    }
+    if (firstSemantic === void 0 && isSemanticIndicator(node)) {
+      firstSemantic = node;
+    }
+    for (const child of getChildNodes(node)) {
+      if (visit(child)) {
+        return true;
       }
-      el.children().each(function flattenChild() {
-        flatten($(this));
-      });
-      el.replaceWith(el.contents());
-    });
+    }
+    return false;
   };
-  flatten($.root());
+  for (const child of getChildNodes(root)) {
+    if (visit(child)) {
+      return firstImage;
+    }
+  }
+  return firstImage ?? firstSource ?? firstSemantic;
 }
 function serializeForExtraction($, root) {
   const lines = [];
@@ -17049,7 +17441,7 @@ function serializeForExtraction($, root) {
     lines.push(`${"  ".repeat(depth)}</${tagName}>`);
   }
   traverse(root, 0);
-  return lines.join("\n");
+  return lines.map((l) => l.trim()).filter((l) => l.length > 0).join("");
 }
 function isClickable($, el, context) {
   if (context.hasPreMarked) {
@@ -17141,7 +17533,11 @@ function cleanForExtraction(html) {
     }
   });
   flattenExtractionTree($clean);
-  return deduplicateImages(serializeForExtraction($clean, $clean.root()[0]));
+  const root = $clean.root()[0];
+  if (root === void 0) {
+    return "";
+  }
+  return deduplicateImages(serializeForExtraction($clean, root));
 }
 function cleanForAction(html) {
   if (!html.trim()) {
@@ -17167,22 +17563,12 @@ function cleanForAction(html) {
   $(`[${clickableMark}]`).each(function markIndicators() {
     const el = $(this);
     const wrapperAttrs = el.attr() || {};
-    if (hasTextDeep(el) || hasActionLabel(wrapperAttrs)) {
-      return;
-    }
-    const imageIndicator = el.find("img[alt], img[src], img[srcset]").first();
-    if (imageIndicator.length) {
-      imageIndicator.attr(indicatorMark, "1");
-      return;
-    }
-    const pictureSourceIndicator = el.find("picture source[src], picture source[srcset]").first();
-    if (pictureSourceIndicator.length) {
-      pictureSourceIndicator.attr(indicatorMark, "1");
+    if (hasTextDeepNode(el[0]) || hasActionLabel(wrapperAttrs)) {
       return;
     }
-    const semanticIndicator = el.find('[aria-label], [title], [data-icon], [role="img"], svg').first();
-    if (semanticIndicator.length) {
-      semanticIndicator.attr(indicatorMark, "1");
+    const indicatorNode = findIndicatorDescendant(el[0]);
+    if (indicatorNode !== void 0) {
+      $(indicatorNode).attr(indicatorMark, "1");
     }
   });
   $(`[${clickableMark}]`).each(function removeEmptyClickable() {
@@ -17192,7 +17578,7 @@ function cleanForAction(html) {
     if (NATIVE_INTERACTIVE_TAGS.has(tag) || tag === "a") {
       return;
     }
-    if (el.children().length > 0 || hasDirectText($, el)) {
+    if (hasElementChildren(node) || hasDirectText(node)) {
       return;
     }
     const wrapperAttrs = el.attr() || {};
@@ -17218,46 +17604,31 @@ function cleanForAction(html) {
       current = ancestor.parent();
     }
   });
-  let changed = true;
-  while (changed) {
-    changed = false;
-    const nodes = [];
-    $("*").each(function collectNodes() {
-      nodes.push($(this));
-    });
-    nodes.sort((left, right) => right.parents().length - left.parents().length);
-    for (const el of nodes) {
-      const node = el[0];
-      if (!node) {
-        continue;
-      }
-      const tag = (node.tagName || "").toLowerCase();
-      if (ROOT_TAGS.has(tag) || isBoundaryTag(tag)) {
-        continue;
-      }
-      if (el.attr(clickableMark) !== void 0 || el.attr(indicatorMark) !== void 0) {
-        continue;
-      }
-      const insideClickable = el.parents(`[${clickableMark}]`).length > 0;
-      const preserveBranch = el.attr(branchMark) !== void 0;
-      const hasContent = el.children().length > 0 || hasDirectText($, el);
-      if (insideClickable || preserveBranch) {
-        if (!hasContent) {
-          el.remove();
-        } else {
-          unwrapActionNode($, el);
-        }
-        changed = true;
-        continue;
-      }
+  for (const node of getElementsByDepthDescending($)) {
+    const el = $(node);
+    const tag = (node.tagName || "").toLowerCase();
+    if (ROOT_TAGS.has(tag) || isBoundaryTag(tag)) {
+      continue;
+    }
+    if (el.attr(clickableMark) !== void 0 || el.attr(indicatorMark) !== void 0) {
+      continue;
+    }
+    const insideClickable = hasMarkedAncestor(el, clickableMark);
+    const preserveBranch = el.attr(branchMark) !== void 0;
+    const hasContent = hasElementChildren(node) || hasDirectText(node);
+    if (insideClickable || preserveBranch) {
       if (!hasContent) {
         el.remove();
-        changed = true;
-        continue;
+      } else {
+        unwrapActionNode($, el);
       }
-      unwrapActionNode($, el);
-      changed = true;
+      continue;
+    }
+    if (!hasContent) {
+      el.remove();
+      continue;
     }
+    unwrapActionNode($, el);
   }
   $.root().find("*").contents().each(function normalizeActionTextNodes() {
     if (this.type !== "text") {
@@ -17337,21 +17708,7 @@ function cleanForAction(html) {
       OPENSTEER_SPARSE_COUNTER_ATTR
     ]);
     if (clickable) {
-      for (const attr of [
-        "href",
-        "role",
-        "type",
-        "title",
-        "placeholder",
-        "value",
-        "aria-label",
-        "aria-labelledby",
-        "aria-describedby",
-        "aria-expanded",
-        "aria-pressed",
-        "aria-selected",
-        "aria-haspopup"
-      ]) {
+      for (const attr of ["href", "role", "type", "title", "placeholder", "value", "aria-label"]) {
         keep.add(attr);
       }
     }
@@ -17971,9 +18328,9 @@ function renderNode(snapshot, node, nodesById, snapshotsByDocumentRef, snapshotI
   const snapshotAttributes = normalizeNodeAttributes(node.attributes);
   const authoredAttributes = stripInternalSnapshotAttributes(snapshotAttributes);
   const attributes = [...authoredAttributes];
-  const subtreeHidden = hasAttribute2(snapshotAttributes, OPENSTEER_HIDDEN_ATTR) || isLikelySubtreeHidden(node);
-  const selfHidden = !subtreeHidden && (hasAttribute2(snapshotAttributes, OPENSTEER_SELF_HIDDEN_ATTR) || isLikelySelfHidden(node, nodesById));
-  const interactive = !subtreeHidden && !selfHidden && (hasAttribute2(snapshotAttributes, OPENSTEER_INTERACTIVE_ATTR) || isLikelyInteractive(tagName, node, authoredAttributes));
+  const subtreeHidden = hasAttribute3(snapshotAttributes, OPENSTEER_HIDDEN_ATTR) || isLikelySubtreeHidden(node);
+  const selfHidden = !subtreeHidden && (hasAttribute3(snapshotAttributes, OPENSTEER_SELF_HIDDEN_ATTR) || isLikelySelfHidden(node, nodesById));
+  const interactive = !subtreeHidden && !selfHidden && (hasAttribute3(snapshotAttributes, OPENSTEER_INTERACTIVE_ATTR) || isLikelyInteractive(tagName, node, authoredAttributes));
   if (interactive) {
     attributes.push({ name: OPENSTEER_INTERACTIVE_ATTR, value: "1" });
   }
@@ -18291,7 +18648,7 @@ function parseOpacity(value) {
   const parsed = Number.parseFloat(value);
   return Number.isFinite(parsed) ? parsed : Number.NaN;
 }
-function hasAttribute2(attributes, name) {
+function hasAttribute3(attributes, name) {
   const normalizedName = name.toLowerCase();
   return attributes.some((attribute) => attribute.name.toLowerCase() === normalizedName);
 }
@@ -20252,12 +20609,12 @@ var OpensteerSessionRuntime = class {
         async (timeout) => {
           let descriptor2;
           let data;
-          if (input.schema !== void 0) {
-            assertValidOpensteerExtractionSchemaRoot(input.schema);
+          if (input.template !== void 0) {
+            assertValidOpensteerExtractionTemplateRoot(input.template);
             const fieldTargets = await timeout.runStep(
               () => compileOpensteerExtractionFieldTargets({
                 pageRef,
-                schema: input.schema,
+                template: input.template,
                 dom: this.requireDom()
               })
             );
@@ -20289,7 +20646,7 @@ var OpensteerSessionRuntime = class {
                 () => descriptors.write({
                   persist,
                   root: payload,
-                  schemaHash: canonicalJsonString(input.schema),
+                  templateHash: canonicalJsonString(input.template),
                   sourceUrl: pageInfo.url
                 })
               );
@@ -20349,7 +20706,7 @@ var OpensteerSessionRuntime = class {
         artifacts,
         data: {
           ...input.persist === void 0 ? {} : { persist: input.persist },
-          ...descriptor?.payload.schemaHash === void 0 ? {} : { schemaHash: descriptor.payload.schemaHash },
+          ...descriptor?.payload.templateHash === void 0 ? {} : { templateHash: descriptor.payload.templateHash },
           data: output.data
         },
         context: buildRuntimeTraceContext({
@@ -24659,142 +25016,6 @@ function screenshotMediaType(format2) {
       return "image/webp";
   }
 }
-
-// ../runtime-core/src/sdk/semantic-dispatch.ts
-async function dispatchSemanticOperation(runtime, operation, input, options = {}) {
-  switch (operation) {
-    case "session.open":
-      return runtime.open(input, options);
-    case "page.list":
-      return runtime.listPages(
-        input,
-        options
-      );
-    case "page.new":
-      return runtime.newPage(input, options);
-    case "page.activate":
-      return runtime.activatePage(
-        input,
-        options
-      );
-    case "page.close":
-      return runtime.closePage(
-        input,
-        options
-      );
-    case "page.goto":
-      return runtime.goto(input, options);
-    case "page.evaluate":
-      return runtime.evaluate(
-        input,
-        options
-      );
-    case "page.add-init-script":
-      return runtime.addInitScript(
-        input,
-        options
-      );
-    case "page.snapshot":
-      return runtime.snapshot(
-        input,
-        options
-      );
-    case "dom.click":
-      return runtime.click(input, options);
-    case "dom.hover":
-      return runtime.hover(input, options);
-    case "dom.input":
-      return runtime.input(input, options);
-    case "dom.scroll":
-      return runtime.scroll(input, options);
-    case "dom.extract":
-      return runtime.extract(input, options);
-    case "network.query":
-      return runtime.queryNetwork(
-        input,
-        options
-      );
-    case "network.detail":
-      return runtime.getNetworkDetail(
-        input,
-        options
-      );
-    case "interaction.capture":
-      return runtime.captureInteraction(
-        input,
-        options
-      );
-    case "interaction.get":
-      return runtime.getInteraction(
-        input,
-        options
-      );
-    case "interaction.diff":
-      return runtime.diffInteraction(
-        input,
-        options
-      );
-    case "interaction.replay":
-      return runtime.replayInteraction(
-        input,
-        options
-      );
-    case "artifact.read":
-      return runtime.readArtifact(
-        input,
-        options
-      );
-    case "session.cookies":
-      return runtime.getCookies(
-        input,
-        options
-      );
-    case "session.storage":
-      return runtime.getStorageSnapshot(
-        input,
-        options
-      );
-    case "session.state":
-      return runtime.getBrowserState(
-        input,
-        options
-      );
-    case "session.fetch":
-      return runtime.fetch(input, options);
-    case "scripts.capture":
-      return runtime.captureScripts(
-        input,
-        options
-      );
-    case "scripts.beautify":
-      return runtime.beautifyScript(
-        input,
-        options
-      );
-    case "scripts.deobfuscate":
-      return runtime.deobfuscateScript(
-        input,
-        options
-      );
-    case "scripts.sandbox":
-      return runtime.sandboxScript(
-        input,
-        options
-      );
-    case "captcha.solve":
-      return runtime.solveCaptcha(
-        input,
-        options
-      );
-    case "computer.execute":
-      return runtime.computerExecute(
-        input,
-        options
-      );
-    case "session.close":
-      return runtime.close(options);
-  }
-}
 var OpensteerSemanticRestError = class extends Error {
   opensteerError;
   statusCode;
@@ -25798,39 +26019,6 @@ var OpensteerRuntime = class extends OpensteerSessionRuntime {
     );
   }
 };
-var OpensteerSessionRuntime2 = class extends OpensteerSessionRuntime {
-  constructor(options) {
-    const rootPath = options.rootPath ?? path10__default.default.resolve(options.rootDir ?? process.cwd());
-    const cleanupRootOnClose = options.cleanupRootOnClose ?? false;
-    const engineName = options.engineName ?? DEFAULT_OPENSTEER_ENGINE;
-    assertSupportedEngineOptions({
-      engineName,
-      ...options.browser === void 0 ? {} : { browser: options.browser },
-      ...options.context === void 0 ? {} : { context: options.context }
-    });
-    super(
-      buildSharedRuntimeOptions({
-        name: options.name,
-        ...options.rootDir === void 0 ? {} : { rootDir: options.rootDir },
-        rootPath,
-        ...options.environment === void 0 ? {} : { environment: options.environment },
-        ...options.browser === void 0 ? {} : { browser: options.browser },
-        ...options.launch === void 0 ? {} : { launch: options.launch },
-        ...options.context === void 0 ? {} : { context: options.context },
-        engineName,
-        ...options.engine === void 0 ? {} : { engine: options.engine },
-        ...options.engineFactory === void 0 ? {} : { engineFactory: options.engineFactory },
-        ...options.policy === void 0 ? {} : { policy: options.policy },
-        ...options.descriptorStore === void 0 ? {} : { descriptorStore: options.descriptorStore },
-        ...options.extractionDescriptorStore === void 0 ? {} : { extractionDescriptorStore: options.extractionDescriptorStore },
-        cleanupRootOnClose,
-        ...options.observability === void 0 ? {} : { observability: options.observability },
-        ...options.observationSessionId === void 0 ? {} : { observationSessionId: options.observationSessionId },
-        ...options.observationSink === void 0 ? {} : { observationSink: options.observationSink }
-      })
-    );
-  }
-};
 function buildSharedRuntimeOptions(input) {
   const ownership = resolveOwnership(input.browser);
   const engineFactory = input.engineFactory ?? ((factoryOptions) => new OpensteerBrowserManager({
@@ -26098,9 +26286,6 @@ var Opensteer = class {
       await delay3(pollIntervalMs);
     }
   }
-  async snapshot(mode = "action") {
-    return (await this.runtime.snapshot({ mode })).html;
-  }
   async cookies(domain) {
     return new SessionCookieJar(
       await this.runtime.getCookies(domain === void 0 ? {} : { domain })
@@ -26251,7 +26436,6 @@ function delay3(ms) {
   return new Promise((resolve4) => setTimeout(resolve4, ms));
 }
 
-exports.CloudSessionProxy = CloudSessionProxy;
 exports.DEFAULT_OPENSTEER_ENGINE = DEFAULT_OPENSTEER_ENGINE;
 exports.DEFERRED_MATCH_ATTR_KEYS = DEFERRED_MATCH_ATTR_KEYS;
 exports.ElementPathError = ElementPathError;
@@ -26264,8 +26448,6 @@ exports.Opensteer = Opensteer;
 exports.OpensteerAttachAmbiguousError = OpensteerAttachAmbiguousError;
 exports.OpensteerBrowserManager = OpensteerBrowserManager;
 exports.OpensteerCloudClient = OpensteerCloudClient;
-exports.OpensteerRuntime = OpensteerRuntime;
-exports.OpensteerSessionRuntime = OpensteerSessionRuntime2;
 exports.STABLE_PRIMARY_ATTR_KEYS = STABLE_PRIMARY_ATTR_KEYS;
 exports.assertProviderSupportsEngine = assertProviderSupportsEngine;
 exports.buildArrayFieldPathCandidates = buildArrayFieldPathCandidates;
@@ -26285,7 +26467,6 @@ exports.createDomRuntime = createDomRuntime;
 exports.createFilesystemOpensteerWorkspace = createFilesystemOpensteerWorkspace;
 exports.createObservationStore = createObservationStore;
 exports.createOpensteerExtractionDescriptorStore = createOpensteerExtractionDescriptorStore;
-exports.createOpensteerSemanticRuntime = createOpensteerSemanticRuntime;
 exports.defaultFallbackPolicy = defaultFallbackPolicy;
 exports.defaultPolicy = defaultPolicy;
 exports.defaultRetryPolicy = defaultRetryPolicy;
@@ -26293,7 +26474,6 @@ exports.defaultSettlePolicy = defaultSettlePolicy;
 exports.defaultTimeoutPolicy = defaultTimeoutPolicy;
 exports.delayWithSignal = delayWithSignal;
 exports.discoverLocalCdpBrowsers = discoverLocalCdpBrowsers;
-exports.dispatchSemanticOperation = dispatchSemanticOperation;
 exports.hashDomDescriptorPersist = hashDomDescriptorPersist;
 exports.inspectCdpEndpoint = inspectCdpEndpoint;
 exports.isCurrentUrlField = isCurrentUrlField;