@beyondwork/docx-react-component 1.0.10 → 1.0.12

package/src/validation/compatibility-engine.ts

@@ -7,6 +7,7 @@ import type {
   EditorWarning,
 } from "../core/state/editor-state.ts";
 import type {
+  BlockNode,
   DocumentRootNode,
   InlineNode,
   ParagraphNode,
@@ -221,6 +222,9 @@ function collectPreservationFeatures(
   document: CanonicalDocumentEnvelope,
 ): CompatibilityFeatureEntry[] {
   const entries: CompatibilityFeatureEntry[] = [];
+  const structuredSubPartPaths = collectStructuredSubPartPaths(document);
+
+  entries.push(...collectSubPartFeatures(document));
 
   for (const fragment of listOpaqueFragments(document.preservation as never)) {
     const descriptor = describeOpaqueFragment(fragment);
@@ -242,6 +246,10 @@ function collectPreservationFeatures(
   }
 
   for (const packagePart of listPreservedPackageParts(document.preservation as never)) {
+    if (structuredSubPartPaths.has(packagePart.packagePartName)) {
+      continue;
+    }
+
     entries.push({
       featureEntryId: `feature:package:${packagePart.packagePartName}`,
       featureKey: "unknown-package-parts",
@@ -258,6 +266,206 @@ function collectPreservationFeatures(
   return entries;
 }
 
+function collectStructuredSubPartPaths(
+  document: CanonicalDocumentEnvelope,
+): ReadonlySet<string> {
+  const subParts = document.subParts;
+  if (!subParts) {
+    return new Set<string>();
+  }
+
+  const paths = new Set<string>();
+
+  for (const header of subParts.headers ?? []) {
+    paths.add(header.partPath);
+  }
+
+  for (const footer of subParts.footers ?? []) {
+    paths.add(footer.partPath);
+  }
+
+  if (subParts.footnoteCollection) {
+    paths.add("/word/footnotes.xml");
+    paths.add("/word/endnotes.xml");
+  }
+
+  if (subParts.theme) {
+    paths.add("/word/theme/theme1.xml");
+  }
+
+  return paths;
+}
+
+function collectSubPartFeatures(
+  document: CanonicalDocumentEnvelope,
+): CompatibilityFeatureEntry[] {
+  const subParts = document.subParts;
+  if (!subParts) {
+    return [];
+  }
+
+  const entries: CompatibilityFeatureEntry[] = collectLossySubPartFeatures(subParts);
+  const hasHeaderFooterContent = (subParts.headers?.length ?? 0) > 0 || (subParts.footers?.length ?? 0) > 0;
+  const noteCount =
+    Object.keys(subParts.footnoteCollection?.footnotes ?? {}).length +
+    Object.keys(subParts.footnoteCollection?.endnotes ?? {}).length;
+
+  if (hasHeaderFooterContent) {
+    entries.push({
+      featureEntryId: "feature:subparts:headers-footers",
+      featureKey: "headers-footers",
+      featureClass: "preserve-only",
+      message: "Headers and footers are preserved through structured sub-part ownership.",
+      details: {
+        headerCount: subParts.headers?.length ?? 0,
+        footerCount: subParts.footers?.length ?? 0,
+      },
+    });
+  }
+
+  if (noteCount > 0) {
+    entries.push({
+      featureEntryId: "feature:subparts:notes",
+      featureKey: "notes",
+      featureClass: "preserve-only",
+      message: "Footnotes and endnotes are preserved through structured sub-part ownership.",
+      details: {
+        footnoteCount: Object.keys(subParts.footnoteCollection?.footnotes ?? {}).length,
+        endnoteCount: Object.keys(subParts.footnoteCollection?.endnotes ?? {}).length,
+      },
+    });
+  }
+
+  if (subParts.theme) {
+    entries.push({
+      featureEntryId: "feature:subparts:theme",
+      featureKey: "unknown-package-parts",
+      featureClass: "preserve-only",
+      message: "Theme metadata is preserved through structured sub-part ownership.",
+      details: {
+        surface: "theme-subpart",
+        themeName: subParts.theme.name,
+      },
+    });
+  }
+
+  return entries;
+}
+
+function collectLossySubPartFeatures(
+  subParts: NonNullable<CanonicalDocumentEnvelope["subParts"]>,
+): CompatibilityFeatureEntry[] {
+  const entries: CompatibilityFeatureEntry[] = [];
+
+  const headerFooterLossy = [
+    ...(subParts.headers ?? []).flatMap((header) =>
+      collectLossyBlocks(header.blocks, `header:${header.partPath}`),
+    ),
+    ...(subParts.footers ?? []).flatMap((footer) =>
+      collectLossyBlocks(footer.blocks, `footer:${footer.partPath}`),
+    ),
+  ];
+  if (headerFooterLossy.length > 0) {
+    entries.push({
+      featureEntryId: "feature:subparts:headers-footers-lossy",
+      featureKey: "headers-footers-lossy",
+      featureClass: "unsupported-fatal",
+      message: "Headers or footers contain content the current sub-part serializer cannot re-emit safely.",
+      details: {
+        issues: headerFooterLossy,
+      },
+    });
+  }
+
+  const noteLossy = [
+    ...Object.values(subParts.footnoteCollection?.footnotes ?? {}).flatMap((note) =>
+      collectLossyBlocks(note.blocks, `footnote:${note.noteId}`),
+    ),
+    ...Object.values(subParts.footnoteCollection?.endnotes ?? {}).flatMap((note) =>
+      collectLossyBlocks(note.blocks, `endnote:${note.noteId}`),
+    ),
+  ];
+  if (noteLossy.length > 0) {
+    entries.push({
+      featureEntryId: "feature:subparts:notes-lossy",
+      featureKey: "notes-lossy",
+      featureClass: "unsupported-fatal",
+      message: "Footnotes or endnotes contain content the current sub-part serializer cannot re-emit safely.",
+      details: {
+        issues: noteLossy,
+      },
+    });
+  }
+
+  return entries;
+}
+
+function collectLossyBlocks(
+  blocks: readonly BlockNode[],
+  surface: string,
+): string[] {
+  const issues: string[] = [];
+
+  for (const block of blocks) {
+    if (block.type !== "paragraph") {
+      issues.push(`${surface}:${block.type}`);
+      continue;
+    }
+
+    if (
+      block.numbering !== undefined ||
+      block.spacing !== undefined ||
+      block.indentation !== undefined ||
+      block.tabStops !== undefined ||
+      block.keepNext !== undefined ||
+      block.keepLines !== undefined ||
+      block.outlineLevel !== undefined ||
+      block.pageBreakBefore !== undefined ||
+      block.widowControl !== undefined ||
+      block.borders !== undefined ||
+      block.shading !== undefined ||
+      block.bidi !== undefined ||
+      block.suppressLineNumbers !== undefined ||
+      block.cnfStyle !== undefined
+    ) {
+      issues.push(`${surface}:paragraph-properties`);
+    }
+
+    for (const child of block.children) {
+      issues.push(...collectLossyInlineContent(child, surface));
+    }
+  }
+
+  return [...new Set(issues)];
+}
+
+function collectLossyInlineContent(
+  node: InlineNode,
+  surface: string,
+): string[] {
+  switch (node.type) {
+    case "text": {
+      const unsupportedMarks = (node.marks ?? [])
+        .filter(
+          (mark) =>
+            mark.type !== "bold" &&
+            mark.type !== "italic" &&
+            mark.type !== "underline" &&
+            mark.type !== "strikethrough" &&
+            mark.type !== "doubleStrikethrough",
+        )
+        .map((mark) => `${surface}:mark:${mark.type}`);
+      return unsupportedMarks;
+    }
+    case "tab":
+    case "hard_break":
+    case "footnote_ref":
+      return [];
+    default:
+      return [`${surface}:${node.type}`];
+  }
+}
+
 function collectDiagnosticWarnings(
   document: CanonicalDocumentEnvelope,
 ): EditorWarning[] {