@superdoc/mcp 1.2.0-next.5 → 1.2.0-next.6

package/dist/index.js

@@ -31314,7 +31314,7 @@ var init_host = __esm(() => {
 });
 
 // ../../packages/sdk/langs/node/src/runtime/sdk-version.generated.ts
-var SDK_VERSION = "2.15.0-next.5";
+var SDK_VERSION = "2.15.0-next.6";
 
 // ../../packages/sdk/langs/node/src/runtime/embedded-platform.ts
 import { chmodSync, existsSync } from "node:fs";
@@ -134122,6 +134122,13 @@ function parseDirectRunSequence(xml, allowRevisions) {
       originalText += value;
     }
   };
+  const isProofingBoundary = (tag) => {
+    if (tag.uri !== W_NS || tag.local !== "proofErr" || !tag.isSelfClosing)
+      return false;
+    const attributes = Object.values(tag.attributes);
+    const typeAttributes = attributes.filter((attribute) => typeof attribute !== "string" && attribute.uri === W_NS && attribute.local === "type");
+    return typeAttributes.length === 1 && typeof typeAttributes[0] !== "string" && PROOFING_BOUNDARY_TYPES.has(typeAttributes[0].value) && attributes.every((attribute) => typeof attribute !== "string" && (attribute === typeAttributes[0] || attribute.uri === "http://www.w3.org/2000/xmlns/" && (attribute.local === tag.prefix || attribute.local === "w" || tag.prefix === "" && attribute.local === "xmlns") && attribute.value === W_NS));
+  };
   const parser = new import_saxes3.SaxesParser({
     xmlns: true,
     additionalNamespaces: replayAdditionalNamespaces(xml)
@@ -134138,6 +134145,10 @@ function parseDirectRunSequence(xml, allowRevisions) {
       return;
     }
     if (parent === "wrapper") {
+      if (isProofingBoundary(tag)) {
+        stack.push("control");
+        return;
+      }
       rootCount += 1;
       if (tag.uri === W_NS && tag.local === "r") {
         stack.push("run");
@@ -134168,6 +134179,10 @@ function parseDirectRunSequence(xml, allowRevisions) {
       return;
     }
     if (parent === "revision") {
+      if (isProofingBoundary(tag)) {
+        stack.push("control");
+        return;
+      }
       if (tag.uri !== W_NS || tag.local !== "r")
         valid = false;
       activeRevisionRunCount += 1;
@@ -134598,7 +134613,7 @@ function validateRunContents(xml, options) {
   if (withoutBookmarks === null || UNSUPPORTED_XML.test(withoutBookmarks) || RELATIONSHIP_REFERENCE.test(withoutBookmarks)) {
     return { ok: false, reason: "target runs require unsupported structural or relationship replay" };
   }
-  const namespaceResult = validateReplayNamespaces(withoutBookmarks, "target runs use namespace prefixes that cannot be replayed safely");
+  const namespaceResult = validateReplayNamespaces(withoutBookmarks.replace(PROOFING_BOUNDARY_TAG_RE, ""), "target runs use namespace prefixes that cannot be replayed safely");
   if (!namespaceResult.ok)
     return namespaceResult;
   if (!parseDirectRunSequence(withoutBookmarks, false)) {
@@ -134625,7 +134640,7 @@ function validateCompareReplayPayload(payload, options = {}) {
     if (payload.targetRevisionReplay.policy !== "preserve-target-revision-set" || payload.targetRunXml === undefined) {
       return { ok: false, reason: "target revision replay policy is incomplete" };
     }
-    const namespaceResult = validateReplayNamespaces(payload.targetRunXml, "target revision runs use namespace prefixes that cannot be replayed safely");
+    const namespaceResult = validateReplayNamespaces(payload.targetRunXml.replace(PROOFING_BOUNDARY_TAG_RE, ""), "target revision runs use namespace prefixes that cannot be replayed safely");
     if (!namespaceResult.ok)
       return namespaceResult;
     const inspected = inspectDirectParagraphRevisionRuns(payload.targetRunXml);
@@ -134645,12 +134660,14 @@ function validateCompareReplayPayload(payload, options = {}) {
   }
   return { ok: true };
 }
-var import_saxes3, UNSUPPORTED_XML, RELATIONSHIP_REFERENCE, W_NS = "http://schemas.openxmlformats.org/wordprocessingml/2006/main", DIRECT_RUN_PROPERTY_ELEMENTS;
+var import_saxes3, UNSUPPORTED_XML, RELATIONSHIP_REFERENCE, W_NS = "http://schemas.openxmlformats.org/wordprocessingml/2006/main", PROOFING_BOUNDARY_TYPES, PROOFING_BOUNDARY_TAG_RE, DIRECT_RUN_PROPERTY_ELEMENTS;
 var init_replayability = __esm(() => {
   init_note_xml_validation();
   import_saxes3 = __toESM(require_saxes(), 1);
   UNSUPPORTED_XML = /<(?:[\w.-]+:)?(?:hyperlink|drawing|pict|object|fldSimple|fldChar|instrText|commentReference|commentRangeStart|commentRangeEnd|permStart|permEnd|sdt|customXml|oMath|oMathPara|footnoteReference|endnoteReference|ins|del|moveFrom|moveTo|pPrChange|rPrChange|sectPr)\b/i;
   RELATIONSHIP_REFERENCE = /\s(?:r:id|r:embed|r:link)\s*=/i;
+  PROOFING_BOUNDARY_TYPES = new Set(["spellStart", "spellEnd", "gramStart", "gramEnd"]);
+  PROOFING_BOUNDARY_TAG_RE = /<(?:[A-Za-z_][\w.-]*:)?proofErr\b(?:[^>"']|"[^"]*"|'[^']*')*\/>/g;
   DIRECT_RUN_PROPERTY_ELEMENTS = new Set([
     "rStyle",
     "rFonts",
@@ -134790,7 +134807,10 @@ function canonicalName(namespace, localName2) {
 }
 function computeRunSignature(runXml) {
   const canonicalXml = canonicalizeXmlForSignature(runXml);
-  const normalized = normalizeRunXmlForSignature(canonicalXml.replace(REVISION_ATTR_RE, "").replace(COMMENT_MARKUP_RE, (markup, name, attributes) => isWordprocessingElement3(name, attributes, canonicalXml) ? "" : markup).replace(NOTE_REFERENCE_MARKUP_RE, (markup, runName, runAttributes, _rPrName, nestedName, nestedAttributes, standaloneName, standaloneAttributes) => {
+  const normalized = normalizeRunXmlForSignature(canonicalXml.replace(REVISION_ATTR_RE, (attribute, offset, xml) => {
+    const precedingTag = xml.slice(xml.lastIndexOf("<", offset), offset);
+    return /^<w:proofErr(?:\s|$)/.test(precedingTag) ? attribute : "";
+  }).replace(REVISION_ELEMENT_RE, (markup) => markup.replace(REVISION_ID_ATTR_RE, "")).replace(COMMENT_MARKUP_RE, (markup, name, attributes) => isWordprocessingElement3(name, attributes, canonicalXml) ? "" : markup).replace(NOTE_REFERENCE_MARKUP_RE, (markup, runName, runAttributes, _rPrName, nestedName, nestedAttributes, standaloneName, standaloneAttributes) => {
     const name = nestedName ?? standaloneName;
     const attributes = nestedAttributes ?? standaloneAttributes;
     if (!isWordprocessingElement3(name, attributes, canonicalXml))
@@ -134817,7 +134837,7 @@ function normalizeRunXmlForSignature(runXml) {
       continue;
     const index = match.index ?? 0;
     if (index > lastIndex) {
-      const interstitial = runXml.slice(lastIndex, index);
+      const interstitial = stripProofingBoundaries(runXml.slice(lastIndex, index));
       if (interstitial.trim().length > 0) {
         pieces.push(interstitial);
         activeTextProperties = "";
@@ -134846,12 +134866,15 @@ function normalizeRunXmlForSignature(runXml) {
   if (!sawRun)
     return runXml;
   if (lastIndex < runXml.length) {
-    const trailing = runXml.slice(lastIndex);
+    const trailing = stripProofingBoundaries(runXml.slice(lastIndex));
     if (trailing.trim().length > 0)
       pieces.push(trailing);
   }
   return pieces.join("");
 }
+function stripProofingBoundaries(xml) {
+  return xml.replace(PROOFING_BOUNDARY_MARKUP_RE, "");
+}
 function normalizeRunControls(content, namespaceContext) {
   return [...content.matchAll(RUN_CONTROL_RE)].filter((match) => isWordprocessingElement3(match[1], match[3], namespaceContext)).map((match) => {
     const attributes = [...match[3].matchAll(ATTRIBUTE_RE)].filter((attribute) => attribute[1] !== "xmlns" && !attribute[1].startsWith("xmlns:")).map((attribute) => {
@@ -134884,10 +134907,13 @@ function isWordprocessingElement3(name, attributes, namespaceContext) {
     return true;
   return (declaration[2] ?? declaration[3] ?? "") === WORDPROCESSINGML_NAMESPACE2;
 }
-var REVISION_ATTR_RE, COMMENT_MARKUP_RE, NOTE_REFERENCE_MARKUP_RE, RUN_RE, RUN_PROPERTIES_RE, EMPTY_RUN_PROPERTIES_RE, TEXTISH_RE, RUN_CONTROL_RE, ATTRIBUTE_RE, WORDPROCESSINGML_NAMESPACE2 = "http://schemas.openxmlformats.org/wordprocessingml/2006/main", WORDPROCESSINGML_2010_NAMESPACE = "http://schemas.microsoft.com/office/word/2010/wordml", MARKUP_COMPATIBILITY_NAMESPACE = "http://schemas.openxmlformats.org/markup-compatibility/2006", XML_NAMESPACE = "http://www.w3.org/XML/1998/namespace", XML_TAG_RE;
+var REVISION_ATTR_RE, REVISION_ELEMENT_RE, REVISION_ID_ATTR_RE, COMMENT_MARKUP_RE, PROOFING_BOUNDARY_MARKUP_RE, NOTE_REFERENCE_MARKUP_RE, RUN_RE, RUN_PROPERTIES_RE, EMPTY_RUN_PROPERTIES_RE, TEXTISH_RE, RUN_CONTROL_RE, ATTRIBUTE_RE, WORDPROCESSINGML_NAMESPACE2 = "http://schemas.openxmlformats.org/wordprocessingml/2006/main", WORDPROCESSINGML_2010_NAMESPACE = "http://schemas.microsoft.com/office/word/2010/wordml", MARKUP_COMPATIBILITY_NAMESPACE = "http://schemas.openxmlformats.org/markup-compatibility/2006", XML_NAMESPACE = "http://www.w3.org/XML/1998/namespace", XML_TAG_RE;
 var init_run_signature = __esm(() => {
-  REVISION_ATTR_RE = /\s+(?:w:rsid\w*|w14:paraId|w14:textId|w:id)="[^"]*"/g;
+  REVISION_ATTR_RE = /\s+(?:w:rsid\w*|w14:paraId|w14:textId)="[^"]*"/g;
+  REVISION_ELEMENT_RE = /<w:(?:ins|del|moveFrom|moveTo)\b[^>]*>/g;
+  REVISION_ID_ATTR_RE = /\s+w:id="[^"]*"/g;
   COMMENT_MARKUP_RE = /<((?:[\w.-]+:)?(?:commentRangeStart|commentRangeEnd|commentReference))\b([^>]*)\/>/g;
+  PROOFING_BOUNDARY_MARKUP_RE = /<w:proofErr w:type="(?:spellStart|spellEnd|gramStart|gramEnd)"\/>/g;
   NOTE_REFERENCE_MARKUP_RE = /<((?:[\w.-]+:)?r)\b([^>]*)>(?:<((?:[\w.-]+:)?rPr)\b[^>]*(?:\/>|>[\s\S]*?<\/\3>))?<((?:[\w.-]+:)?(?:footnoteReference|endnoteReference))\b([^>]*)\/><\/\1>|<((?:[\w.-]+:)?(?:footnoteReference|endnoteReference))\b([^>]*)\/>/g;
   RUN_RE = /<((?:[\w.-]+:)?r)\b([^>]*)>([\s\S]*?)<\/\1>/g;
   RUN_PROPERTIES_RE = /<((?:[\w.-]+:)?rPr)\b([^>]*)(?:\/>|>[\s\S]*?<\/\1>)/g;
@@ -219993,15 +220019,31 @@ function canonicalizeCompareBodyRunXml(xml, referenceSteps) {
   let result = "";
   let cursor = 0;
   for (const token of tokens) {
-    const canonicalGap = canonicalizeWordprocessingReplayXml(xml.slice(cursor, token.start));
+    const canonicalGap = canonicalizeProofingRunGap(xml.slice(cursor, token.start));
     if (canonicalGap === null)
       return null;
     result += canonicalGap + token.xml;
     cursor = token.end;
   }
-  const canonicalTail = canonicalizeWordprocessingReplayXml(xml.slice(cursor));
+  const canonicalTail = canonicalizeProofingRunGap(xml.slice(cursor));
   return canonicalTail === null ? null : result + canonicalTail;
 }
+function canonicalizeProofingRunGap(xml) {
+  const markerRe = /<(?:[A-Za-z_][\w.-]*:)?proofErr\b(?:[^>"']|"[^"]*"|'[^']*')*\/>/g;
+  let result = "";
+  let cursor = 0;
+  for (const marker of xml.matchAll(markerRe)) {
+    const index = marker.index ?? 0;
+    const before = canonicalizeWordprocessingReplayXml(xml.slice(cursor, index));
+    const canonicalMarker = canonicalizeWordprocessingReplayXml(marker[0]);
+    if (before === null || canonicalMarker === null)
+      return null;
+    result += before + canonicalMarker;
+    cursor = index + marker[0].length;
+  }
+  const trailing = canonicalizeWordprocessingReplayXml(xml.slice(cursor));
+  return trailing === null ? null : result + trailing;
+}
 function collectProtectedReferenceTokens(xml, referenceSteps) {
   const tokens = [];
   for (let index = 0;index < referenceSteps.length; index += 1) {
@@ -239472,7 +239514,7 @@ function toRoman(ordinal) {
 }
 function extractParagraphText(paraXml) {
   let out = "";
-  for (const match of paraXml.matchAll(/<[\w.-]+:t\b[^>]*>([\s\S]*?)<\/[\w.-]+:t>/g)) {
+  for (const match of paraXml.matchAll(/<[\w.-]+:t\b[^>]*?(?<!\/)>([\s\S]*?)<\/[\w.-]+:t>/g)) {
     out += decodeXmlText(match[1] ?? "");
   }
   return out;
@@ -407429,7 +407471,7 @@ function parseRun2(runXml, textTagLocalName = "t", inheritedNamespaceBindings =
         atomKind: INLINE_ATOM_KIND.noteRef,
         id: attrLocal2(next.openTagInner, "id"),
         xml: next.xml,
-        meta: { kind: local },
+        meta: { kind: local, runAttributes: attributes },
         visibleLength: 0
       });
     } else if (local === "sym" || local === "noBreakHyphen") {
@@ -407440,7 +407482,7 @@ function parseRun2(runXml, textTagLocalName = "t", inheritedNamespaceBindings =
         atomKind: INLINE_ATOM_KIND.preserved,
         id: null,
         xml: next.xml,
-        meta: { tag: local, container: "run-child" },
+        meta: { tag: local, container: "run-child", runAttributes: attributes },
         visibleLength: 0
       });
     }
@@ -420672,7 +420714,9 @@ function serializeBlockXml(block, extraAtoms = [], context = { inherited: CANONI
       const atomXml = atom.kind === "synthetic-comment-marker" ? `<${w}:${atom.meta.marker} ${w}:id="${xmlEscape(String(atom.meta.commentId))}"/>` : atom.xml;
       const attributedBreakXml = runBackedAttributedBreakXml(atom);
       if (atomEmissionMode(atom) === "run-child") {
-        sb.push(serializeInlineRunAtom(attributedBreakXml ?? atomXml, runAttributes, w));
+        const atomRunAttributes = atom.meta.runAttributes;
+        const attributes = atomRunAttributes && typeof atomRunAttributes === "object" && !Array.isArray(atomRunAttributes) ? atomRunAttributes : runAttributes;
+        sb.push(serializeInlineRunAtom(attributedBreakXml ?? atomXml, attributes, w));
       } else {
         sb.push(atomXml);
       }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@superdoc/mcp",
-  "version": "1.2.0-next.5",
+  "version": "1.2.0-next.6",
   "description": "Model Context Protocol server for SuperDoc, giving an AI agent tools to read and edit .docx files.",
   "type": "module",
   "license": "AGPL-3.0",
@@ -24,13 +24,13 @@
   ],
   "dependencies": {
     "@modelcontextprotocol/sdk": "^1.26.0",
-    "@superdoc/sdk": "2.15.0-next.5",
+    "@superdoc/sdk": "2.15.0-next.6",
     "debug": "^4.4.3",
     "zod": "^4.3.6"
   },
   "devDependencies": {
-    "superdoc": "2.18.0-next.5",
-    "@superdoc/docx-engine": "0.17.0-next.6",
+    "superdoc": "2.18.0-next.6",
+    "@superdoc/docx-engine": "0.17.0-next.7",
     "@superdoc/document-api": "0.1.0-alpha.0",
     "@types/bun": "^1.3.8",
     "@types/node": "22.19.2",