@beyondwork/docx-react-component 1.0.1 → 1.0.3

package/src/io/export/serialize-footnotes.ts

@@ -0,0 +1,217 @@
+import type {
+  FootnoteCollection,
+  FootnoteDefinition,
+  InlineNode,
+  ParagraphNode,
+  TextMark,
+} from "../../model/canonical-document.ts";
+
+export const WORD_FOOTNOTES_CONTENT_TYPE =
+  "application/vnd.openxmlformats-officedocument.wordprocessingml.footnotes+xml";
+export const WORD_ENDNOTES_CONTENT_TYPE =
+  "application/vnd.openxmlformats-officedocument.wordprocessingml.endnotes+xml";
+
+const W_NS = `xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main"`;
+const R_NS = `xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships"`;
+
+/**
+ * Serialize the footnotes portion of a FootnoteCollection to footnotes.xml.
+ * Includes the required separator and continuation-separator stubs.
+ */
+export function serializeFootnotesXml(collection: FootnoteCollection): string {
+  const entries = Object.values(collection.footnotes).sort(compareNoteIds);
+  const body = [
+    serializeSeparatorStub("footnote", "-1", "separator"),
+    serializeSeparatorStub("footnote", "0", "continuationSeparator"),
+    ...entries.map((entry) => serializeNoteDefinition("footnote", entry)),
+  ].join("");
+
+  return [
+    `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>`,
+    `<w:footnotes ${W_NS} ${R_NS}>${body}</w:footnotes>`,
+  ].join("\n");
+}
+
+/**
+ * Serialize the endnotes portion of a FootnoteCollection to endnotes.xml.
+ * Includes the required separator and continuation-separator stubs.
+ */
+export function serializeEndnotesXml(collection: FootnoteCollection): string {
+  const entries = Object.values(collection.endnotes).sort(compareNoteIds);
+  const body = [
+    serializeSeparatorStub("endnote", "-1", "separator"),
+    serializeSeparatorStub("endnote", "0", "continuationSeparator"),
+    ...entries.map((entry) => serializeNoteDefinition("endnote", entry)),
+  ].join("");
+
+  return [
+    `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>`,
+    `<w:endnotes ${W_NS} ${R_NS}>${body}</w:endnotes>`,
+  ].join("\n");
+}
+
+// ---- Internal serialization ----
+
+function serializeSeparatorStub(
+  kind: "footnote" | "endnote",
+  id: string,
+  type: "separator" | "continuationSeparator",
+): string {
+  const tag = kind === "footnote" ? "w:footnote" : "w:endnote";
+  return `<${tag} w:type="${type}" w:id="${id}"><w:p/></${tag}>`;
+}
+
+function serializeNoteDefinition(
+  kind: "footnote" | "endnote",
+  definition: FootnoteDefinition,
+): string {
+  const tag = kind === "footnote" ? "w:footnote" : "w:endnote";
+  const blocks = definition.blocks
+    .map((block) => {
+      if (block.type === "paragraph") {
+        return serializeParagraph(block);
+      }
+      // opaque_block: emit empty paragraph
+      return `<w:p/>`;
+    })
+    .join("");
+
+  const body = blocks || `<w:p><w:r><w:t></w:t></w:r></w:p>`;
+  return `<${tag} w:id="${escapeAttribute(definition.noteId)}">${body}</${tag}>`;
+}
+
+function serializeParagraph(paragraph: ParagraphNode): string {
+  let xml = "<w:p>";
+
+  const propertiesXml = buildParagraphPropertiesXml(paragraph);
+  if (propertiesXml) {
+    xml += propertiesXml;
+  }
+
+  const childrenXml = paragraph.children
+    .map((child) => serializeInlineNode(child))
+    .join("");
+  xml += childrenXml || "<w:r><w:t></w:t></w:r>";
+  xml += "</w:p>";
+
+  return xml;
+}
+
+function buildParagraphPropertiesXml(paragraph: ParagraphNode): string {
+  const parts: string[] = [];
+
+  if (paragraph.styleId) {
+    parts.push(`<w:pStyle w:val="${escapeAttribute(paragraph.styleId)}"/>`);
+  }
+  if (paragraph.alignment) {
+    parts.push(`<w:jc w:val="${escapeAttribute(paragraph.alignment)}"/>`);
+  }
+
+  return parts.length > 0 ? `<w:pPr>${parts.join("")}</w:pPr>` : "";
+}
+
+function serializeInlineNode(node: InlineNode): string {
+  switch (node.type) {
+    case "text": {
+      const properties = buildRunPropertiesXml(node.marks);
+      const preserve = requiresPreservedSpace(node.text)
+        ? ` xml:space="preserve"`
+        : "";
+      return `<w:r>${properties}<w:t${preserve}>${escapeXml(node.text)}</w:t></w:r>`;
+    }
+    case "tab":
+      return "<w:r><w:tab/></w:r>";
+    case "hard_break":
+      return "<w:r><w:br/></w:r>";
+    case "footnote_ref": {
+      const refElement =
+        node.noteKind === "footnote"
+          ? `<w:footnoteReference w:id="${escapeAttribute(node.noteId)}"/>`
+          : `<w:endnoteReference w:id="${escapeAttribute(node.noteId)}"/>`;
+      const styleVal =
+        node.noteKind === "footnote"
+          ? "FootnoteReference"
+          : "EndnoteReference";
+      return `<w:r><w:rPr><w:rStyle w:val="${styleVal}"/></w:rPr>${refElement}</w:r>`;
+    }
+    case "opaque_inline":
+      return "";
+    case "hyperlink": {
+      return node.children
+        .map((child) => {
+          if (child.type === "text") {
+            const preserve = requiresPreservedSpace(child.text)
+              ? ` xml:space="preserve"`
+              : "";
+            return `<w:r><w:t${preserve}>${escapeXml(child.text)}</w:t></w:r>`;
+          }
+          if (child.type === "tab") return "<w:r><w:tab/></w:r>";
+          if (child.type === "hard_break") return "<w:r><w:br/></w:r>";
+          return "";
+        })
+        .join("");
+    }
+    default:
+      return "";
+  }
+}
+
+function buildRunPropertiesXml(marks: TextMark[] | undefined): string {
+  if (!marks || marks.length === 0) {
+    return "";
+  }
+
+  const parts: string[] = [];
+  for (const mark of marks) {
+    switch (mark.type) {
+      case "bold":
+        parts.push("<w:b/>");
+        break;
+      case "italic":
+        parts.push("<w:i/>");
+        break;
+      case "underline":
+        parts.push("<w:u w:val=\"single\"/>");
+        break;
+      case "strikethrough":
+        parts.push("<w:strike/>");
+        break;
+      case "doubleStrikethrough":
+        parts.push("<w:dstrike/>");
+        break;
+      default:
+        break;
+    }
+  }
+
+  return parts.length > 0 ? `<w:rPr>${parts.join("")}</w:rPr>` : "";
+}
+
+function compareNoteIds(
+  left: FootnoteDefinition,
+  right: FootnoteDefinition,
+): number {
+  return Number.parseInt(left.noteId, 10) - Number.parseInt(right.noteId, 10);
+}
+
+function requiresPreservedSpace(text: string): boolean {
+  return (
+    text.length > 0 &&
+    (text[0] === " " || text[text.length - 1] === " " || text.includes("  "))
+  );
+}
+
+function escapeXml(text: string): string {
+  return text
+    .replace(/&/g, "&amp;")
+    .replace(/</g, "&lt;")
+    .replace(/>/g, "&gt;");
+}
+
+function escapeAttribute(value: string): string {
+  return value
+    .replace(/&/g, "&amp;")
+    .replace(/</g, "&lt;")
+    .replace(/>/g, "&gt;")
+    .replace(/"/g, "&quot;");
+}

package/src/io/export/serialize-headers-footers.ts

@@ -0,0 +1,200 @@
+import type {
+  FooterDocument,
+  HeaderDocument,
+  InlineNode,
+  ParagraphNode,
+  TextMark,
+} from "../../model/canonical-document.ts";
+
+export const WORD_HEADER_CONTENT_TYPE =
+  "application/vnd.openxmlformats-officedocument.wordprocessingml.header+xml";
+export const WORD_FOOTER_CONTENT_TYPE =
+  "application/vnd.openxmlformats-officedocument.wordprocessingml.footer+xml";
+
+const W_NS = `xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main"`;
+const R_NS = `xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships"`;
+
+/**
+ * Serialize a HeaderDocument into a headerN.xml string.
+ */
+export function serializeHeaderXml(header: HeaderDocument): string {
+  const body = serializeBlocks(header.blocks);
+  return [
+    `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>`,
+    `<w:hdr ${W_NS} ${R_NS}>${body}</w:hdr>`,
+  ].join("\n");
+}
+
+/**
+ * Serialize a FooterDocument into a footerN.xml string.
+ */
+export function serializeFooterXml(footer: FooterDocument): string {
+  const body = serializeBlocks(footer.blocks);
+  return [
+    `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>`,
+    `<w:ftr ${W_NS} ${R_NS}>${body}</w:ftr>`,
+  ].join("\n");
+}
+
+// ---- Internal serialization ----
+
+function serializeBlocks(
+  blocks: HeaderDocument["blocks"] | FooterDocument["blocks"],
+): string {
+  if (blocks.length === 0) {
+    // A valid header/footer must have at least one paragraph
+    return `<w:p><w:r><w:t></w:t></w:r></w:p>`;
+  }
+
+  return blocks
+    .map((block) => {
+      if (block.type === "paragraph") {
+        return serializeParagraph(block);
+      }
+      // opaque_block: emit empty paragraph to preserve structure
+      return `<w:p/>`;
+    })
+    .join("");
+}
+
+function serializeParagraph(paragraph: ParagraphNode): string {
+  let xml = "<w:p>";
+
+  const propertiesXml = buildParagraphPropertiesXml(paragraph);
+  if (propertiesXml) {
+    xml += propertiesXml;
+  }
+
+  const childrenXml = paragraph.children
+    .map((child) => serializeInlineNode(child))
+    .join("");
+  xml += childrenXml || "<w:r><w:t></w:t></w:r>";
+  xml += "</w:p>";
+
+  return xml;
+}
+
+function buildParagraphPropertiesXml(paragraph: ParagraphNode): string {
+  const parts: string[] = [];
+
+  if (paragraph.styleId) {
+    parts.push(`<w:pStyle w:val="${escapeAttribute(paragraph.styleId)}"/>`);
+  }
+  if (paragraph.alignment) {
+    parts.push(`<w:jc w:val="${escapeAttribute(paragraph.alignment)}"/>`);
+  }
+
+  return parts.length > 0 ? `<w:pPr>${parts.join("")}</w:pPr>` : "";
+}
+
+function serializeInlineNode(node: InlineNode): string {
+  switch (node.type) {
+    case "text": {
+      const properties = buildRunPropertiesXml(node.marks);
+      const preserve = requiresPreservedSpace(node.text)
+        ? ` xml:space="preserve"`
+        : "";
+      return `<w:r>${properties}<w:t${preserve}>${escapeXml(node.text)}</w:t></w:r>`;
+    }
+    case "tab":
+      return "<w:r><w:tab/></w:r>";
+    case "hard_break":
+      return "<w:r><w:br/></w:r>";
+    case "footnote_ref": {
+      const refElement =
+        node.noteKind === "footnote"
+          ? `<w:footnoteReference w:id="${escapeAttribute(node.noteId)}"/>`
+          : `<w:endnoteReference w:id="${escapeAttribute(node.noteId)}"/>`;
+      return `<w:r><w:rPr><w:rStyle w:val="${node.noteKind === "footnote" ? "FootnoteReference" : "EndnoteReference"}"/></w:rPr>${refElement}</w:r>`;
+    }
+    case "opaque_inline":
+      // Cannot reproduce opaque inline content without original XML; emit empty
+      return "";
+    case "hyperlink": {
+      const childrenXml = node.children
+        .map((child) => {
+          switch (child.type) {
+            case "text": {
+              const properties = buildRunPropertiesXml(undefined);
+              const preserve = requiresPreservedSpace(child.text)
+                ? ` xml:space="preserve"`
+                : "";
+              return `<w:r>${properties}<w:t${preserve}>${escapeXml(child.text)}</w:t></w:r>`;
+            }
+            case "tab":
+              return "<w:r><w:tab/></w:r>";
+            case "hard_break":
+              return "<w:r><w:br/></w:r>";
+            default:
+              return "";
+          }
+        })
+        .join("");
+      // Hyperlinks in headers/footers typically use bookmark anchors or external URLs.
+      // Emit as a plain run since we don't retain the relationship ID here.
+      return childrenXml;
+    }
+    case "image":
+    case "field":
+    case "bookmark_start":
+    case "bookmark_end":
+    case "column_break":
+    case "symbol":
+      // These node types are not parsed from headers/footers by parse-headers-footers.ts
+      return "";
+  }
+}
+
+function buildRunPropertiesXml(marks: TextMark[] | undefined): string {
+  if (!marks || marks.length === 0) {
+    return "";
+  }
+
+  const parts: string[] = [];
+  for (const mark of marks) {
+    switch (mark.type) {
+      case "bold":
+        parts.push("<w:b/>");
+        break;
+      case "italic":
+        parts.push("<w:i/>");
+        break;
+      case "underline":
+        parts.push("<w:u w:val=\"single\"/>");
+        break;
+      case "strikethrough":
+        parts.push("<w:strike/>");
+        break;
+      case "doubleStrikethrough":
+        parts.push("<w:dstrike/>");
+        break;
+      default:
+        // Other mark types not parsed from headers/footers
+        break;
+    }
+  }
+
+  return parts.length > 0 ? `<w:rPr>${parts.join("")}</w:rPr>` : "";
+}
+
+function requiresPreservedSpace(text: string): boolean {
+  return (
+    text.length > 0 &&
+    (text[0] === " " || text[text.length - 1] === " " || text.includes("  "))
+  );
+}
+
+function escapeXml(text: string): string {
+  return text
+    .replace(/&/g, "&amp;")
+    .replace(/</g, "&lt;")
+    .replace(/>/g, "&gt;");
+}
+
+function escapeAttribute(value: string): string {
+  return value
+    .replace(/&/g, "&amp;")
+    .replace(/</g, "&lt;")
+    .replace(/>/g, "&gt;")
+    .replace(/"/g, "&quot;");
+}