@beyondwork/docx-react-component 1.0.18 → 1.0.19

package/src/io/ooxml/parse-footnotes.ts

@@ -4,6 +4,9 @@ import type {
   FootnoteDefinition,
   InlineNode,
   ParagraphNode,
+  TableCellNode,
+  TableNode,
+  TableRowNode,
   TextMark,
 } from "../../model/canonical-document.ts";
 
@@ -137,11 +140,17 @@ function parseNoteElement(
     if (name === "p") {
       blocks.push(parseParagraphElement(child));
     } else if (name === "tbl") {
-      blocks.push({
-        type: "opaque_block",
-        fragmentId: `fragment:note-tbl-${noteId}`,
-        warningId: `warning:note-opaque-table`,
-      });
+      // Simple tables (no revisions, fields, or nested tables) are promoted
+      // to supported-roundtrip; structurally risky tables stay opaque.
+      if (isSimpleSecondaryStoryTable(child)) {
+        blocks.push(parseSimpleTableElement(child));
+      } else {
+        blocks.push({
+          type: "opaque_block",
+          fragmentId: `fragment:note-tbl-${noteId}`,
+          warningId: `warning:note-opaque-table`,
+        });
+      }
     } else {
       blocks.push({
         type: "opaque_block",
@@ -175,7 +184,7 @@ function parseParagraphElement(pElement: XmlElementNode): ParagraphNode {
     } else if (name === "bookmarkStart" || name === "bookmarkEnd") {
       children.push(parseBookmarkElement(child));
     } else if (name === "fldSimple") {
-      children.push(parseFieldElement(child));
+      pushFieldNode(children, child, "simple");
     }
   }
 
@@ -226,8 +235,8 @@ function parseRunElement(rElement: XmlElementNode): InlineNode[] {
       }
     } else if (name === "bookmarkStart" || name === "bookmarkEnd") {
       nodes.push(parseBookmarkElement(child));
-    } else if (name === "fldChar" || name === "instrText") {
-      nodes.push(parseFieldElement(child));
+    } else if (name === "instrText") {
+      pushFieldNode(nodes, child, "complex");
     }
   }
 
@@ -275,22 +284,30 @@ function parseBookmarkElement(
   };
 }
 
-function parseFieldElement(element: XmlElementNode): Extract<InlineNode, { type: "field" }> {
-  const rawFieldType =
-    element.attributes["w:fldCharType"] ??
-    element.attributes.fldCharType ??
-    localName(element.name);
-  const fieldType: "simple" | "complex" = rawFieldType === "complex" ? "complex" : "simple";
-  const instruction =
-    element.attributes["w:instr"] ??
-    element.attributes.instr ??
-    extractTextContent(element);
-  return {
+function pushFieldNode(
+  nodes: InlineNode[],
+  element: XmlElementNode,
+  fieldType: "simple" | "complex",
+): void {
+  const instruction = readFieldInstruction(element);
+  if (!instruction) {
+    return;
+  }
+
+  nodes.push({
     type: "field",
     fieldType,
     instruction,
     children: [],
-  };
+  });
+}
+
+function readFieldInstruction(element: XmlElementNode): string | undefined {
+  const instruction =
+    element.attributes["w:instr"] ??
+    element.attributes.instr ??
+    extractTextContent(element);
+  return instruction.trim().length > 0 ? instruction : undefined;
 }
 
 function parseRunProperties(rElement: XmlElementNode): TextMark[] {
@@ -353,6 +370,181 @@ function localName(name: string): string {
   return idx >= 0 ? name.slice(idx + 1) : name;
 }
 
+// ---- Simple secondary-story table support ----
+
+const RISKY_TABLE_ELEMENT_NAMES = new Set([
+  "ins",
+  "del",
+  "moveFrom",
+  "moveTo",
+  "tblPrChange",
+  "trPrChange",
+  "tcPrChange",
+  "rPrChange",
+  "pPrChange",
+  "sectPrChange",
+  "fldSimple",
+  "fldChar",
+  "instrText",
+  "sdt",
+  "customXml",
+]);
+
+function isSimpleSecondaryStoryTable(tblElement: XmlElementNode): boolean {
+  return !containsRiskyElement(tblElement);
+}
+
+function containsRiskyElement(element: XmlElementNode): boolean {
+  for (const child of element.children) {
+    if (child.type !== "element") {
+      continue;
+    }
+    const name = localName(child.name);
+    if (RISKY_TABLE_ELEMENT_NAMES.has(name)) {
+      return true;
+    }
+    if (name === "tbl") {
+      return true;
+    }
+    if (containsRiskyElement(child)) {
+      return true;
+    }
+  }
+  return false;
+}
+
+function parseSimpleTableElement(tblElement: XmlElementNode): TableNode {
+  let gridColumns: number[] = [];
+  const rows: TableRowNode[] = [];
+  let propertiesXml: string | undefined;
+  let styleId: string | undefined;
+
+  for (const child of tblElement.children) {
+    if (child.type !== "element") continue;
+    const name = localName(child.name);
+
+    if (name === "tblPr") {
+      propertiesXml = serializeElementToXml(child);
+      const pStyle = findChildElementOptional(child, "tblStyle");
+      styleId = pStyle?.attributes["w:val"] ?? pStyle?.attributes.val;
+    } else if (name === "tblGrid") {
+      gridColumns = readGridColumns(child);
+    } else if (name === "tr") {
+      rows.push(parseSimpleTableRow(child));
+    }
+  }
+
+  return {
+    type: "table",
+    ...(styleId ? { styleId } : {}),
+    ...(propertiesXml ? { propertiesXml } : {}),
+    gridColumns,
+    rows,
+  };
+}
+
+function readGridColumns(tblGrid: XmlElementNode): number[] {
+  const columns: number[] = [];
+  for (const child of tblGrid.children) {
+    if (child.type !== "element") continue;
+    if (localName(child.name) === "gridCol") {
+      const w = child.attributes["w:w"] ?? child.attributes.w ?? "0";
+      columns.push(Number.parseInt(w, 10) || 0);
+    }
+  }
+  return columns;
+}
+
+function parseSimpleTableRow(trElement: XmlElementNode): TableRowNode {
+  const cells: TableCellNode[] = [];
+  let propertiesXml: string | undefined;
+
+  for (const child of trElement.children) {
+    if (child.type !== "element") continue;
+    const name = localName(child.name);
+
+    if (name === "trPr") {
+      propertiesXml = serializeElementToXml(child);
+    } else if (name === "tc") {
+      cells.push(parseSimpleTableCell(child));
+    }
+  }
+
+  return {
+    type: "table_row",
+    ...(propertiesXml ? { propertiesXml } : {}),
+    cells,
+  };
+}
+
+function parseSimpleTableCell(tcElement: XmlElementNode): TableCellNode {
+  const children: BlockNode[] = [];
+  let propertiesXml: string | undefined;
+  let gridSpan: number | undefined;
+  let verticalMerge: "restart" | "continue" | undefined;
+
+  for (const child of tcElement.children) {
+    if (child.type !== "element") continue;
+    const name = localName(child.name);
+
+    if (name === "tcPr") {
+      propertiesXml = serializeElementToXml(child);
+      const gsEl = findChildElementOptional(child, "gridSpan");
+      const gsVal = gsEl?.attributes["w:val"] ?? gsEl?.attributes.val;
+      if (gsVal) gridSpan = Number.parseInt(gsVal, 10) || undefined;
+
+      const vmEl = findChildElementOptional(child, "vMerge");
+      if (vmEl) {
+        const vmVal = vmEl.attributes["w:val"] ?? vmEl.attributes.val ?? "continue";
+        verticalMerge = vmVal === "restart" ? "restart" : "continue";
+      }
+    } else if (name === "p") {
+      children.push(parseParagraphElement(child));
+    }
+  }
+
+  return {
+    type: "table_cell",
+    ...(propertiesXml ? { propertiesXml } : {}),
+    ...(gridSpan ? { gridSpan } : {}),
+    ...(verticalMerge ? { verticalMerge } : {}),
+    children: children.length > 0 ? children : [{ type: "paragraph", children: [] }],
+  };
+}
+
+function serializeElementToXml(element: XmlElementNode): string {
+  const attrs = Object.entries(element.attributes)
+    .map(([key, value]) => ` ${key}="${escapeXmlAttribute(value)}"`)
+    .join("");
+  const children = element.children
+    .map((child) => {
+      if (child.type === "text") {
+        return escapeXmlText(child.text);
+      }
+      return serializeElementToXml(child);
+    })
+    .join("");
+  if (children.length === 0) {
+    return `<${element.name}${attrs}/>`;
+  }
+  return `<${element.name}${attrs}>${children}</${element.name}>`;
+}
+
+function escapeXmlAttribute(text: string): string {
+  return text
+    .replace(/&/g, "&amp;")
+    .replace(/"/g, "&quot;")
+    .replace(/</g, "&lt;")
+    .replace(/>/g, "&gt;");
+}
+
+function escapeXmlText(text: string): string {
+  return text
+    .replace(/&/g, "&amp;")
+    .replace(/</g, "&lt;")
+    .replace(/>/g, "&gt;");
+}
+
 // ---- Minimal XML parser (same pattern as parse-numbering.ts) ----
 
 function parseXml(xml: string): XmlElementNode {

package/src/io/ooxml/parse-headers-footers.ts

@@ -4,6 +4,9 @@ import type {
   HeaderFooterVariant,
   InlineNode,
   ParagraphNode,
+  TableCellNode,
+  TableNode,
+  TableRowNode,
   TextMark,
 } from "../../model/canonical-document.ts";
 
@@ -13,6 +16,7 @@ export interface ParsedHeaderFooterReference {
   variant: HeaderFooterVariant;
   relationshipId: string;
   kind: "header" | "footer";
+  sectionIndex?: number;
 }
 
 export interface ParsedHeaderFooterDocument {
@@ -83,6 +87,7 @@ function collectSectPrReferences(
   element: XmlElementNode,
   refs: ParsedHeaderFooterReference[],
 ): void {
+  let sectionIndex = 0;
   for (const child of element.children) {
     if (child.type !== "element") {
       continue;
@@ -91,14 +96,16 @@ function collectSectPrReferences(
     const name = localName(child.name);
 
     if (name === "sectPr") {
-      extractSectPrRefs(child, refs);
+      // Body-level sectPr is the final section
+      extractSectPrRefs(child, refs, sectionIndex);
     } else if (name === "p") {
-      // Check paragraph properties for sectPr
+      // Check paragraph properties for sectPr (non-final section break)
       const pPr = findChildElementOptional(child, "pPr");
       if (pPr) {
         const sectPr = findChildElementOptional(pPr, "sectPr");
         if (sectPr) {
-          extractSectPrRefs(sectPr, refs);
+          extractSectPrRefs(sectPr, refs, sectionIndex);
+          sectionIndex++;
         }
       }
     }
@@ -108,6 +115,7 @@ function collectSectPrReferences(
 function extractSectPrRefs(
   sectPr: XmlElementNode,
   refs: ParsedHeaderFooterReference[],
+  sectionIndex: number,
 ): void {
   for (const child of sectPr.children) {
     if (child.type !== "element") {
@@ -133,7 +141,7 @@ function extractSectPrRefs(
           (ref) => ref.relationshipId === relationshipId && ref.kind === kind,
         );
         if (!alreadyAdded) {
-          refs.push({ variant, relationshipId, kind });
+          refs.push({ variant, relationshipId, kind, sectionIndex });
         }
       }
     }
@@ -172,12 +180,17 @@ function parseHdrFtrXml(
     if (name === "p") {
       blocks.push(parseParagraphElement(child));
     } else if (name === "tbl") {
-      // Table in header/footer: store as opaque to preserve fidelity
-      blocks.push({
-        type: "opaque_block",
-        fragmentId: "fragment:hdrftr-tbl",
-        warningId: "warning:hdrftr-opaque-table",
-      });
+      // Simple tables (no revisions, fields, or nested tables) are promoted
+      // to supported-roundtrip; structurally risky tables stay opaque.
+      if (isSimpleSecondaryStoryTable(child)) {
+        blocks.push(parseSimpleTableElement(child));
+      } else {
+        blocks.push({
+          type: "opaque_block",
+          fragmentId: "fragment:hdrftr-tbl",
+          warningId: "warning:hdrftr-opaque-table",
+        });
+      }
     } else {
       // Other block-level elements: treat as opaque
       blocks.push({
@@ -218,7 +231,7 @@ function parseParagraphElement(pElement: XmlElementNode): ParagraphNode {
     } else if (name === "bookmarkStart" || name === "bookmarkEnd") {
       children.push(parseBookmarkElement(child));
     } else if (name === "fldSimple") {
-      children.push(parseFieldElement(child));
+      pushFieldNode(children, child, "simple");
     }
   }
 
@@ -278,8 +291,8 @@ function parseRunElement(rElement: XmlElementNode): InlineNode[] {
       }
     } else if (name === "bookmarkStart" || name === "bookmarkEnd") {
       nodes.push(parseBookmarkElement(child));
-    } else if (name === "fldChar" || name === "instrText") {
-      nodes.push(parseFieldElement(child));
+    } else if (name === "instrText") {
+      pushFieldNode(nodes, child, "complex");
     }
   }
 
@@ -327,22 +340,30 @@ function parseBookmarkElement(
   };
 }
 
-function parseFieldElement(element: XmlElementNode): Extract<InlineNode, { type: "field" }> {
-  const rawFieldType =
-    element.attributes["w:fldCharType"] ??
-    element.attributes.fldCharType ??
-    localName(element.name);
-  const fieldType: "simple" | "complex" = rawFieldType === "complex" ? "complex" : "simple";
-  const instruction =
-    element.attributes["w:instr"] ??
-    element.attributes.instr ??
-    extractTextContent(element);
-  return {
+function pushFieldNode(
+  nodes: InlineNode[],
+  element: XmlElementNode,
+  fieldType: "simple" | "complex",
+): void {
+  const instruction = readFieldInstruction(element);
+  if (!instruction) {
+    return;
+  }
+
+  nodes.push({
     type: "field",
     fieldType,
     instruction,
     children: [],
-  };
+  });
+}
+
+function readFieldInstruction(element: XmlElementNode): string | undefined {
+  const instruction =
+    element.attributes["w:instr"] ??
+    element.attributes.instr ??
+    extractTextContent(element);
+  return instruction.trim().length > 0 ? instruction : undefined;
 }
 
 function parseRunProperties(rElement: XmlElementNode): TextMark[] {
@@ -418,6 +439,189 @@ function localName(name: string): string {
   return separatorIndex >= 0 ? name.slice(separatorIndex + 1) : name;
 }
 
+// ---- Simple secondary-story table support ----
+
+/**
+ * Revision-bearing, field-bearing, or structurally risky elements that
+ * disqualify a secondary-story table from supported-roundtrip.
+ */
+const RISKY_TABLE_ELEMENT_NAMES = new Set([
+  "ins",
+  "del",
+  "moveFrom",
+  "moveTo",
+  "tblPrChange",
+  "trPrChange",
+  "tcPrChange",
+  "rPrChange",
+  "pPrChange",
+  "sectPrChange",
+  "fldSimple",
+  "fldChar",
+  "instrText",
+  "sdt",
+  "customXml",
+]);
+
+function isSimpleSecondaryStoryTable(tblElement: XmlElementNode): boolean {
+  return !containsRiskyElement(tblElement);
+}
+
+function containsRiskyElement(element: XmlElementNode): boolean {
+  for (const child of element.children) {
+    if (child.type !== "element") {
+      continue;
+    }
+    const name = localName(child.name);
+    if (RISKY_TABLE_ELEMENT_NAMES.has(name)) {
+      return true;
+    }
+    // Nested tables remain risky
+    if (name === "tbl") {
+      return true;
+    }
+    if (containsRiskyElement(child)) {
+      return true;
+    }
+  }
+  return false;
+}
+
+function parseSimpleTableElement(tblElement: XmlElementNode): TableNode {
+  let gridColumns: number[] = [];
+  const rows: TableRowNode[] = [];
+  let propertiesXml: string | undefined;
+  let styleId: string | undefined;
+
+  for (const child of tblElement.children) {
+    if (child.type !== "element") continue;
+    const name = localName(child.name);
+
+    if (name === "tblPr") {
+      propertiesXml = serializeElementToXml(child);
+      const pStyle = findChildElementOptional(child, "tblStyle");
+      styleId = pStyle?.attributes["w:val"] ?? pStyle?.attributes.val;
+    } else if (name === "tblGrid") {
+      gridColumns = readGridColumns(child);
+    } else if (name === "tr") {
+      rows.push(parseSimpleTableRow(child));
+    }
+  }
+
+  return {
+    type: "table",
+    ...(styleId ? { styleId } : {}),
+    ...(propertiesXml ? { propertiesXml } : {}),
+    gridColumns,
+    rows,
+  };
+}
+
+function readGridColumns(tblGrid: XmlElementNode): number[] {
+  const columns: number[] = [];
+  for (const child of tblGrid.children) {
+    if (child.type !== "element") continue;
+    if (localName(child.name) === "gridCol") {
+      const w = child.attributes["w:w"] ?? child.attributes.w ?? "0";
+      columns.push(Number.parseInt(w, 10) || 0);
+    }
+  }
+  return columns;
+}
+
+function parseSimpleTableRow(trElement: XmlElementNode): TableRowNode {
+  const cells: TableCellNode[] = [];
+  let propertiesXml: string | undefined;
+
+  for (const child of trElement.children) {
+    if (child.type !== "element") continue;
+    const name = localName(child.name);
+
+    if (name === "trPr") {
+      propertiesXml = serializeElementToXml(child);
+    } else if (name === "tc") {
+      cells.push(parseSimpleTableCell(child));
+    }
+  }
+
+  return {
+    type: "table_row",
+    ...(propertiesXml ? { propertiesXml } : {}),
+    cells,
+  };
+}
+
+function parseSimpleTableCell(tcElement: XmlElementNode): TableCellNode {
+  const children: BlockNode[] = [];
+  let propertiesXml: string | undefined;
+  let gridSpan: number | undefined;
+  let verticalMerge: "restart" | "continue" | undefined;
+
+  for (const child of tcElement.children) {
+    if (child.type !== "element") continue;
+    const name = localName(child.name);
+
+    if (name === "tcPr") {
+      propertiesXml = serializeElementToXml(child);
+      const gsEl = findChildElementOptional(child, "gridSpan");
+      const gsVal = gsEl?.attributes["w:val"] ?? gsEl?.attributes.val;
+      if (gsVal) gridSpan = Number.parseInt(gsVal, 10) || undefined;
+
+      const vmEl = findChildElementOptional(child, "vMerge");
+      if (vmEl) {
+        const vmVal = vmEl.attributes["w:val"] ?? vmEl.attributes.val ?? "continue";
+        verticalMerge = vmVal === "restart" ? "restart" : "continue";
+      }
+    } else if (name === "p") {
+      children.push(parseParagraphElement(child));
+    }
+  }
+
+  return {
+    type: "table_cell",
+    ...(propertiesXml ? { propertiesXml } : {}),
+    ...(gridSpan ? { gridSpan } : {}),
+    ...(verticalMerge ? { verticalMerge } : {}),
+    children: children.length > 0 ? children : [{ type: "paragraph", children: [] }],
+  };
+}
+
+/**
+ * Serialize an XmlElementNode back to XML string for propertiesXml preservation.
+ */
+function serializeElementToXml(element: XmlElementNode): string {
+  const attrs = Object.entries(element.attributes)
+    .map(([key, value]) => ` ${key}="${escapeXmlAttribute(value)}"`)
+    .join("");
+  const children = element.children
+    .map((child) => {
+      if (child.type === "text") {
+        return escapeXmlText(child.text);
+      }
+      return serializeElementToXml(child);
+    })
+    .join("");
+  if (children.length === 0) {
+    return `<${element.name}${attrs}/>`;
+  }
+  return `<${element.name}${attrs}>${children}</${element.name}>`;
+}
+
+function escapeXmlAttribute(text: string): string {
+  return text
+    .replace(/&/g, "&amp;")
+    .replace(/"/g, "&quot;")
+    .replace(/</g, "&lt;")
+    .replace(/>/g, "&gt;");
+}
+
+function escapeXmlText(text: string): string {
+  return text
+    .replace(/&/g, "&amp;")
+    .replace(/</g, "&lt;")
+    .replace(/>/g, "&gt;");
+}
+
 // ---- Minimal XML parser (same pattern as parse-numbering.ts) ----
 
 function parseXml(xml: string): XmlElementNode {

package/src/io/ooxml/parse-inline-media.ts

@@ -14,6 +14,8 @@ export interface ParsedInlineMedia {
   contentType?: string;
   filename: string;
   altText?: string;
+  widthEmu?: number;
+  heightEmu?: number;
   display?: "inline" | "floating";
   floating?: {
     horizontalPosition?: {
@@ -85,6 +87,11 @@ export function parseInlineMediaXml(
     const altText = readAltText(docProperties);
     const floating = anchor ? readFloatingProperties(anchor) : undefined;
 
+    // Read extent dimensions (wp:extent cx/cy in EMUs)
+    const extent = findFirstDescendant(container, "extent");
+    const widthEmu = extent ? readEmuAttribute(extent, "cx") : undefined;
+    const heightEmu = extent ? readEmuAttribute(extent, "cy") : undefined;
+
     media.push({
       type: "image",
       mediaId: `media:${packagePartName.slice(1)}`,
@@ -93,6 +100,8 @@ export function parseInlineMediaXml(
       ...(mediaPart ? { contentType: mediaPart.contentType } : {}),
       filename: packagePartName.slice(packagePartName.lastIndexOf("/") + 1),
       ...(altText ? { altText } : {}),
+      ...(widthEmu !== undefined ? { widthEmu } : {}),
+      ...(heightEmu !== undefined ? { heightEmu } : {}),
       ...(anchor ? { display: "floating" as const } : {}),
       ...(floating ? { floating } : {}),
     });
@@ -238,6 +247,13 @@ function readOptionalAttribute(node: XmlElementNode, name: string): string | und
     ?? node.attributes[name];
 }
 
+function readEmuAttribute(node: XmlElementNode, name: string): number | undefined {
+  const value = node.attributes[name] ?? node.attributes[`wp:${name}`];
+  if (value === undefined) return undefined;
+  const parsed = Number.parseInt(value, 10);
+  return Number.isFinite(parsed) && parsed >= 0 ? parsed : undefined;
+}
+
 function readBooleanAttribute(node: XmlElementNode, name: string): boolean | undefined {
   const value = readOptionalAttribute(node, name);
   if (value === undefined) {