Preserve DOCX table column spans

src/io/wordOffice.test.ts

@@ -349,6 +349,22 @@ describe("wordOffice helpers", () => {
     ]);
   });
 
+  it("preserves HTML table colspan as DOCX gridSpan", async () => {
+    const html = '<table><tr><td colspan="2">Merged</td><td>Tail</td></tr></table>';
+    const blocks = htmlToWordBlocks(html);
+    const blob = await exportDocxBlob("Table colspan.docx", html);
+    const zip = await JSZip.loadAsync(await blob.arrayBuffer());
+    const documentXml = (await zip.file("word/document.xml")?.async("string")) ?? "";
+
+    expect(blocks).toEqual([
+      {
+        type: "table",
+        rows: [[{ text: "Merged", runs: [{ text: "Merged" }], colSpan: 2 }, "Tail"]]
+      }
+    ]);
+    expect(documentXml).toMatch(/<w:gridSpan\b[^>]*w:val="2"[^>]*\/>/);
+  });
+
   it("сохраняет пустые ячейки и строки таблицы для DOCX", async () => {
     const html = `
       <table>

src/io/wordOffice.ts

@@ -25,7 +25,7 @@ export interface WordListItem {
   runs: WordInlineRun[];
 }
 
-export type WordTableCell = string | { text: string; runs: WordInlineRun[] };
+export type WordTableCell = string | { text: string; runs: WordInlineRun[]; colSpan?: number };
 
 export interface WordImageBlock {
   type: "image";
@@ -406,9 +406,7 @@ function wordBlockToDocxChildren(block: WordBlock, docx: DocxModule) {
           new docx.TableRow({
             children: row.map(
               (cell) =>
-                new docx.TableCell({
+                new docx.TableCell(wordTableCellToDocxOptions(cell, docx))
-                  children: wordTableCellToDocxParagraphs(cell, docx)
-                })
             )
           })
       )
@@ -416,6 +414,14 @@ function wordBlockToDocxChildren(block: WordBlock, docx: DocxModule) {
   ];
 }
 
+function wordTableCellToDocxOptions(cell: WordTableCell, docx: DocxModule) {
+  const colSpan = typeof cell === "string" ? undefined : cell.colSpan;
+  return {
+    children: wordTableCellToDocxParagraphs(cell, docx),
+    ...(colSpan ? { columnSpan: colSpan } : {})
+  };
+}
+
 function wordTableCellToDocxParagraphs(cell: WordTableCell, docx: DocxModule) {
   const runs = typeof cell === "string" ? [{ text: cell }] : cell.runs;
   return [
@@ -662,7 +668,8 @@ function tableCellValue(cell: Element): WordTableCell {
   const runs = elementInlineRuns(cell);
   const text = inlineRunsText(runs);
   const hasFormattedRuns = runs.some((run) => Object.keys(cleanInlineFormat(run)).length > 0);
-  return hasFormattedRuns ? { text, runs } : text;
+  const colSpan = normalizedTableSpan(cell.getAttribute("colspan"));
+  return hasFormattedRuns || colSpan ? { text, runs, ...(colSpan ? { colSpan } : {}) } : text;
 }
 
 function paragraphElementToBlocks(element: Element, options: { quote?: boolean } = {}): WordBlock[] {
@@ -1122,7 +1129,9 @@ function orderedDocxTableToHtml(
             })
             .filter(Boolean);
           const content = paragraphs.length > 0 ? paragraphs.join("") : escapeHtml(orderedDocxCellText(cell));
-          return content ? `<td>${content}</td>` : "<td></td>";
+          const colSpan = orderedDocxTableCellColspan(cell);
+          const attributes = colSpan ? ` colspan="${colSpan}"` : "";
+          return content ? `<td${attributes}>${content}</td>` : `<td${attributes}></td>`;
         })
         .join("");
 
@@ -1133,6 +1142,12 @@ function orderedDocxTableToHtml(
   return { html: rows.length > 0 ? `<table><tbody>${rows.join("")}</tbody></table>` : "", hasImage, hasLineBreak, hasPageBreak, hasTab };
 }
 
+function orderedDocxTableCellColspan(cell: unknown) {
+  const properties = firstOrderedElement(orderedElementChildren(cell), "w:tcPr");
+  const gridSpan = firstOrderedElement(orderedElementChildren(properties), "w:gridSpan");
+  return normalizedTableSpan(docxAttribute(orderedElementAttributes(gridSpan), "w:val") || docxAttribute(orderedElementAttributes(gridSpan), "val"));
+}
+
 function orderedDocxCellText(cell: unknown) {
   return orderedDirectElements(cell, "w:p")
     .flatMap((paragraph) => orderedDocxParagraphInlineRecords(paragraph, {}))
@@ -1998,6 +2013,16 @@ function numericAttribute(element: Element, name: string) {
   return Number.isFinite(value) && value > 0 ? Math.round(value) : undefined;
 }
 
+function normalizedTableSpan(value?: string | null) {
+  const raw = value?.trim() ?? "";
+  if (!/^\d+$/.test(raw)) {
+    return undefined;
+  }
+
+  const span = Number.parseInt(raw, 10);
+  return Number.isSafeInteger(span) && span > 1 ? Math.min(span, 63) : undefined;
+}
+
 function cssPixelValue(style: string, property: "width" | "height") {
   const match = new RegExp(`${property}\\s*:\\s*(\\d+(?:\\.\\d+)?)px`, "i").exec(style);
   if (!match) {

src/io/wordOfficeImport.test.ts

@@ -312,6 +312,25 @@ describe("wordOffice DOCX import", () => {
     ]);
   });
 
+  it("imports DOCX horizontal merged table cells as HTML colspan", async () => {
+    const blob = await exportDocxBlob(
+      "Table colspan.docx",
+      '<table><tr><td colspan="2">Merged</td><td>Tail</td></tr></table>'
+    );
+    const file = new File([blob], "Table colspan.docx", {
+      type: "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+    });
+    const imported = await importDocxFile(file);
+
+    expect(imported.html).toContain('<td colspan="2"><p>Merged</p></td>');
+    expect(htmlToWordBlocks(imported.html)).toEqual([
+      {
+        type: "table",
+        rows: [[{ text: "Merged", runs: [{ text: "Merged" }], colSpan: 2 }, "Tail"]]
+      }
+    ]);
+  });
+
   it("imports DOCX bullet and numbered lists from OOXML numbering", async () => {
     const blob = await exportDocxBlob(
       "Lists.docx",