@beyondwork/docx-react-component 1.0.58 → 1.0.60

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

@@ -1,5 +1,7 @@
 import type { CommentThread } from "../../review/store/comment-store.ts";
 import { createImportedCommentThread } from "../../review/store/comment-thread.ts";
+import { parseXmlWithOffsets as parseXml } from "./xml-parser.ts";
+import { localName, readStringAttr } from "./xml-attr-helpers.ts";
 
 interface XmlElementNode {
   type: "element";
@@ -293,7 +295,7 @@ function parseCommentDefinitions(
     return [];
   }
 
-  const root = parseXml(commentsXml);
+  const root = parseXml(commentsXml) as XmlElementNode;
   const commentsElement = findChildElement(root, "comments");
   const definitions: ImportedCommentDefinition[] = [];
   let order = 0;
@@ -303,16 +305,16 @@ function parseCommentDefinitions(
       continue;
     }
 
-    const commentId = child.attributes["w:id"] ?? child.attributes.id;
+    const commentId = readStringAttr(child, "w:id");
     if (!commentId) {
       continue;
     }
 
-    const authorId = child.attributes["w:author"] ?? child.attributes.author;
+    const authorId = readStringAttr(child, "w:author");
     const createdAt = normalizeImportedTimestamp(
-      child.attributes["w:date"] ?? child.attributes.date,
+      readStringAttr(child, "w:date"),
     );
-    const initials = child.attributes["w:initials"] ?? child.attributes.initials;
+    const initials = readStringAttr(child, "w:initials");
     const paragraphNodes = child.children.filter(
       (node): node is XmlElementNode => node.type === "element" && localName(node.name) === "p",
     );
@@ -359,7 +361,7 @@ function parseCommentExtensions(xml: string | undefined): Map<string, CommentExt
     return new Map();
   }
 
-  const root = parseXml(xml);
+  const root = parseXml(xml) as XmlElementNode;
   const commentsElement = findChildElement(root, "commentsEx");
   const extensions = new Map<string, CommentExtensionRecord>();
 
@@ -393,7 +395,7 @@ function parseCommentDurableIds(xml: string | undefined): Map<string, string> {
     return new Map();
   }
 
-  const root = parseXml(xml);
+  const root = parseXml(xml) as XmlElementNode;
   const commentsIdsElement = findChildElement(root, "commentsIds");
   const durableIds = new Map<string, string>();
 
@@ -417,7 +419,7 @@ function parsePeopleAuthors(xml: string | undefined): string[] {
     return [];
   }
 
-  const root = parseXml(xml);
+  const root = parseXml(xml) as XmlElementNode;
   const peopleElement = findChildElement(root, "people");
   const authors = new Set<string>();
 
@@ -445,7 +447,7 @@ function extractRootTag(xml: string | undefined, localTagName: string): string |
 }
 
 function parseCommentAnchors(documentXml: string): Map<string, CommentAnchorBounds> {
-  const root = parseXml(documentXml);
+  const root = parseXml(documentXml) as XmlElementNode;
   const documentElement = findChildElement(root, "document");
   const bodyElement = findChildElement(documentElement, "body");
   const anchors = new Map<string, CommentAnchorBounds>();
@@ -562,7 +564,7 @@ function walkInlineNode(
 
   switch (localName(node.name)) {
     case "commentRangeStart": {
-      const commentId = node.attributes["w:id"] ?? node.attributes.id;
+      const commentId = readStringAttr(node, "w:id");
       if (commentId) {
         const bounds = ensureCommentAnchor(anchors, commentId);
         bounds.start = getCursor();
@@ -571,7 +573,7 @@ function walkInlineNode(
       return;
     }
     case "commentRangeEnd": {
-      const commentId = node.attributes["w:id"] ?? node.attributes.id;
+      const commentId = readStringAttr(node, "w:id");
       if (commentId) {
         const bounds = ensureCommentAnchor(anchors, commentId);
         bounds.end = getCursor();
@@ -580,7 +582,7 @@ function walkInlineNode(
       return;
     }
     case "commentReference": {
-      const commentId = node.attributes["w:id"] ?? node.attributes.id;
+      const commentId = readStringAttr(node, "w:id");
       if (commentId) {
         const bounds = ensureCommentAnchor(anchors, commentId);
         bounds.referenceAt = getCursor();
@@ -674,103 +676,6 @@ function toDetachedRange(anchor: CommentAnchorBounds): { from: number; to: numbe
   };
 }
 
-function parseXml(xml: string): XmlElementNode {
-  const root: XmlElementNode = {
-    type: "element",
-    name: "#document",
-    attributes: {},
-    children: [],
-    start: 0,
-    end: xml.length,
-  };
-  const stack: XmlElementNode[] = [root];
-  const tokenPattern =
-    /<!--[\s\S]*?-->|<\?[\s\S]*?\?>|<!DOCTYPE[\s\S]*?>|<!\[CDATA\[[\s\S]*?\]\]>|<[^>]+>|[^<]+/gu;
-
-  for (const match of xml.matchAll(tokenPattern)) {
-    const token = match[0] ?? "";
-    const start = match.index ?? 0;
-    const end = start + token.length;
-
-    if (token.startsWith("<?") || token.startsWith("<!DOCTYPE") || token.startsWith("<!--")) {
-      continue;
-    }
-
-    if (token.startsWith("<![CDATA[")) {
-      const text = token.slice(9, -3);
-      stack[stack.length - 1]?.children.push({
-        type: "text",
-        text,
-        start,
-        end,
-      });
-      continue;
-    }
-
-    if (token.startsWith("</")) {
-      const node = stack.pop();
-      if (!node) {
-        throw new Error("Malformed XML: unexpected closing tag.");
-      }
-      node.end = end;
-      continue;
-    }
-
-    if (token.startsWith("<")) {
-      const selfClosing = /\/>$/.test(token);
-      const tagBody = token.slice(1, token.length - (selfClosing ? 2 : 1)).trim();
-      const { name, attributes } = parseTag(tagBody);
-      const node: XmlElementNode = {
-        type: "element",
-        name,
-        attributes,
-        children: [],
-        start,
-        end,
-      };
-      stack[stack.length - 1]?.children.push(node);
-      if (!selfClosing) {
-        stack.push(node);
-      }
-      continue;
-    }
-
-    const text = decodeXmlText(token);
-    if (text.length > 0) {
-      stack[stack.length - 1]?.children.push({
-        type: "text",
-        text,
-        start,
-        end,
-      });
-    }
-  }
-
-  if (stack.length !== 1) {
-    throw new Error("Malformed XML: unclosed tag.");
-  }
-
-  return root;
-}
-
-function parseTag(tagBody: string): { name: string; attributes: Record<string, string> } {
-  const whitespaceIndex = tagBody.search(/\s/u);
-  const name = whitespaceIndex === -1 ? tagBody : tagBody.slice(0, whitespaceIndex);
-  const rawAttributes = whitespaceIndex === -1 ? "" : tagBody.slice(whitespaceIndex + 1);
-  const attributes: Record<string, string> = {};
-  const pattern = /([A-Za-z_][A-Za-z0-9:._-]*)\s*=\s*("([^"]*)"|'([^']*)')/gu;
-
-  for (const match of rawAttributes.matchAll(pattern)) {
-    const key = match[1];
-    const value = match[3] ?? match[4] ?? "";
-    if (key) {
-      attributes[key] = decodeXmlText(value);
-    }
-  }
-
-  return { name, attributes };
-}
-
 function findChildElement(node: XmlElementNode, name: string): XmlElementNode {
   const match = node.children.find(
     (child): child is XmlElementNode =>
@@ -784,36 +689,3 @@ function findChildElement(node: XmlElementNode, name: string): XmlElementNode {
   return match;
 }
 
-function localName(name: string): string {
-  const index = name.indexOf(":");
-  return index === -1 ? name : name.slice(index + 1);
-}
-
-function decodeXmlText(text: string): string {
-  return text.replace(
-    /&(?:#x([0-9A-Fa-f]+)|#([0-9]+)|([A-Za-z]+));/gu,
-    (_, hex, dec, named) => {
-      if (hex) {
-        return String.fromCodePoint(Number.parseInt(hex, 16));
-      }
-      if (dec) {
-        return String.fromCodePoint(Number.parseInt(dec, 10));
-      }
-
-      switch (named) {
-        case "amp":
-          return "&";
-        case "lt":
-          return "<";
-        case "gt":
-          return ">";
-        case "quot":
-          return "\"";
-        case "apos":
-          return "'";
-        default:
-          return `&${named};`;
-      }
-    },
-  );
-}

package/src/io/ooxml/parse-complex-content.ts

@@ -13,6 +13,8 @@ import type { OpcRelationship } from "./part-manifest.ts";
 import { normalizePartPath, resolveRelationshipTarget } from "./part-manifest.ts";
 import { parseChartSpace } from "./chart/parse-chart-space.ts";
 import type { ChartModel } from "./chart/types.ts";
+import { parseXml } from "./xml-parser.ts";
+import { localName } from "./xml-attr-helpers.ts";
 
 export interface InlineMediaPart {
   path: string;
@@ -260,21 +262,7 @@ function isSmartArtUri(uri: string): boolean {
   );
 }
 
-// ---- Minimal XML parser (local, self-contained) ----
-
-interface XmlElementNode {
-  type: "element";
-  name: string;
-  attributes: Record<string, string>;
-  children: XmlNode[];
-}
-
-interface XmlTextNode {
-  type: "text";
-  text: string;
-}
-
-type XmlNode = XmlElementNode | XmlTextNode;
+import type { XmlElementNode } from "./xml-element.ts";
 
 function findFirstChild(node: XmlElementNode, local: string): XmlElementNode | undefined {
   for (const child of node.children) {
@@ -295,95 +283,4 @@ function findFirstDescendant(node: XmlElementNode, local: string): XmlElementNod
   return undefined;
 }
 
-function localName(name: string): string {
-  const i = name.indexOf(":");
-  return i >= 0 ? name.slice(i + 1) : name;
-}
-
-function parseXml(xml: string): XmlElementNode {
-  const root: XmlElementNode = { type: "element", name: "__root__", attributes: {}, children: [] };
-  const stack: XmlElementNode[] = [root];
-  let cursor = 0;
 
-  while (cursor < xml.length) {
-    if (xml.startsWith("<!--", cursor)) {
-      const end = xml.indexOf("-->", cursor);
-      cursor = end >= 0 ? end + 3 : xml.length;
-      continue;
-    }
-    if (xml.startsWith("<?", cursor)) {
-      const end = xml.indexOf("?>", cursor);
-      cursor = end >= 0 ? end + 2 : xml.length;
-      continue;
-    }
-    if (xml[cursor] !== "<") {
-      const nextTag = xml.indexOf("<", cursor);
-      const end = nextTag >= 0 ? nextTag : xml.length;
-      const text = xml.slice(cursor, end);
-      if (text.length > 0) {
-        stack[stack.length - 1]?.children.push({ type: "text", text });
-      }
-      cursor = end;
-      continue;
-    }
-    if (xml[cursor + 1] === "/") {
-      const end = xml.indexOf(">", cursor);
-      stack.pop();
-      cursor = end + 1;
-      continue;
-    }
-    const tagEnd = findTagEnd(xml, cursor);
-    const tagBody = xml.slice(cursor + 1, tagEnd);
-    const selfClosing = /\/\s*$/.test(tagBody);
-    const { name, attributes } = parseTag(tagBody.replace(/\/\s*$/, "").trim());
-    const element: XmlElementNode = { type: "element", name, attributes, children: [] };
-    stack[stack.length - 1]?.children.push(element);
-    if (!selfClosing) stack.push(element);
-    cursor = tagEnd + 1;
-  }
-
-  return root;
-}
-
-function parseTag(tagBody: string): { name: string; attributes: Record<string, string> } {
-  let cursor = 0;
-  while (cursor < tagBody.length && /\s/.test(tagBody[cursor] ?? "")) cursor++;
-  const nameStart = cursor;
-  while (cursor < tagBody.length && !/\s/.test(tagBody[cursor] ?? "")) cursor++;
-  const name = tagBody.slice(nameStart, cursor);
-  const attributes: Record<string, string> = {};
-
-  while (cursor < tagBody.length) {
-    while (cursor < tagBody.length && /\s/.test(tagBody[cursor] ?? "")) cursor++;
-    if (cursor >= tagBody.length) break;
-    const keyStart = cursor;
-    while (cursor < tagBody.length && !/[\s=]/.test(tagBody[cursor] ?? "")) cursor++;
-    const key = tagBody.slice(keyStart, cursor);
-    while (cursor < tagBody.length && /\s/.test(tagBody[cursor] ?? "")) cursor++;
-    if (tagBody[cursor] !== "=") { attributes[key] = ""; continue; }
-    cursor++;
-    while (cursor < tagBody.length && /\s/.test(tagBody[cursor] ?? "")) cursor++;
-    const quote = tagBody[cursor];
-    if (quote !== `"` && quote !== `'`) break;
-    cursor++;
-    const valueStart = cursor;
-    while (cursor < tagBody.length && tagBody[cursor] !== quote) cursor++;
-    attributes[key] = tagBody.slice(valueStart, cursor);
-    cursor++;
-  }
-
-  return { name, attributes };
-}
-
-function findTagEnd(xml: string, start: number): number {
-  let cursor = start + 1;
-  let quote: string | null = null;
-  while (cursor < xml.length) {
-    const c = xml[cursor];
-    if (quote) { if (c === quote) quote = null; cursor++; continue; }
-    if (c === `"` || c === `'`) { quote = c; cursor++; continue; }
-    if (c === ">") return cursor;
-    cursor++;
-  }
-  return xml.length - 1;
-}