@daisy/epub-utils - npm Package Compare versions

Comparing version 1.2.7-alpha.3 to 1.2.7-alpha.4

lib/epub-parse.js

@@ -50,2 +50,26 @@ // input: unzipped book directory
 
+  // Select the ToC
+  const select = xpath.useNamespaces({
+    html: 'http://www.w3.org/1999/xhtml',
+    epub: 'http://www.idpf.org/2007/ops'
+  });
+
+  const sPageList = select('//html:nav[@epub:type="page-list"]', doc);
+  const hasPageList = sPageList.length > 0;
+
+  let pageListHrefs = undefined;
+  if (hasPageList) {
+    const arr1 = select('descendant::html:a/@href', sPageList[0]);
+    pageListHrefs = arr1.map(o => decodeURI(o.nodeValue));
+    // console.log(arr1.length, JSON.stringify(pageListHrefs, null, 4));
+  }
+
+  let tocHrefs = undefined;
+  const sTOC = select('//html:nav[@epub:type="toc"]/html:ol', doc);
+  if (sTOC[0]) {
+    const arr2 = select('descendant::html:a/@href', sTOC[0]);
+    tocHrefs = arr2.map(o => decodeURI(o.nodeValue));
+    // console.log(arr2.length, JSON.stringify(tocHrefs, null, 4));
+  }
+
   // Remove all links
@@ -60,13 +84,9 @@ const aElems = doc.getElementsByTagNameNS('http://www.w3.org/1999/xhtml', 'a');
 
-  // Select the ToC
-  const select = xpath.useNamespaces({
-    html: 'http://www.w3.org/1999/xhtml',
-    epub: 'http://www.idpf.org/2007/ops'
-  });
-  const toc = select('//html:nav' + '[@epub:type="toc"]/html:ol', doc);
-  const tocHTML = new XMLSerializer().serializeToString(toc[0]);
-  const hasPageList = select('//html:nav' + '[@epub:type="page-list"]', doc).length > 0;
+  const tocHTML = new XMLSerializer().serializeToString(sTOC[0]);
+  // console.log(tocHTML);
 
   return {
     src: path.relative(epubDir, fullpath),
+    pageListHrefs,
+    tocHrefs,
     tocHTML,
@@ -78,2 +98,8 @@ hasPageList
 function addMeta(name, value, meta) {
+  name = name.trim();
+  value = value.trim();
+  if (!name || !value) {
+    // empty strings => bail out
+    return;
+  }
   if (!meta[name]) {
@@ -110,2 +136,4 @@ meta[name] = value;
 function addLink(rel, href, link) {
+  rel = rel.trim();
+  href = href.trim();
   if (!link[rel]) {
@@ -142,5 +170,7 @@ link[rel] = href;
       reject(new Error("Package document not found."));
+      return;
     }
     this.packageDoc = {
-      src: path.relative(epubDir, packageDocPath)
+      src: path.relative(epubDir, packageDocPath),
+      path: packageDocPath
     };
@@ -156,3 +186,9 @@ this.parseData(packageDocPath, epubDir);
   const select = xpath.useNamespaces({ opf: 'http://www.idpf.org/2007/opf',
-    dc: 'http://purl.org/dc/elements/1.1/' });
+    dc: 'http://purl.org/dc/elements/1.1/',
+    xml: 'http://www.w3.org/XML/1998/namespace'
+  });
+
+  const langAttr = select('/opf:package/@xml:lang', doc)[0];
+  this.opfLang = langAttr ? langAttr.nodeValue : undefined;
+
   this.metadata = parseMetadata(doc, select);
@@ -172,4 +208,7 @@ this.links = parseLinks(doc, select);
         spineItem.filepath = path.join(path.dirname(packageDocPath), spineItem.relpath);
-        spineItem.title = this.parseContentDocTitle(spineItem.filepath);
 
+        const o = this.parseContentDocTitleAndIds(spineItem.filepath);
+        spineItem.title = o.titleText;
+        spineItem.targetIDs = o.docIds;
+
         // does encodeURI() as per https://tools.ietf.org/html/rfc3986#section-3.3 in a nutshell: encodeURI(`file://${tmpFile}`).replace(/[?#]/g, encodeURIComponent)
@@ -187,2 +226,12 @@ spineItem.url = fileUrl(spineItem.filepath);
         }
+
+        const moAttr = manifestItem[0].getAttribute('media-overlay');
+        const smilManifestItem = select(`/opf:package/opf:manifest/opf:item[@id='${moAttr}']`, doc);
+        if (smilManifestItem.length > 0) {
+          spineItem.mediaOverlay = {};
+          spineItem.mediaOverlay.smilRelPath = decodeURI(smilManifestItem[0].getAttribute('href'));
+          spineItem.mediaOverlay.smilFilePath = path.join(path.dirname(packageDocPath), spineItem.mediaOverlay.smilRelPath);
+          // spineItem.mediaOverlay.smilUrl = fileUrl(spineItem.mediaOverlay.smilFilePath);
+          spineItem.mediaOverlay.smilRefs = this.parseSmilRefs(spineItem.mediaOverlay.smilFilePath);
+        }
       } else if (!this.hasSVGContentDocuments && 'image/svg+xml' === contentType) {
@@ -201,2 +250,6 @@ winston.warn('The SVG Content Documents in this EPUB will be ignored.');
 
+    this.navDoc.relpath = navDocPath;
+    this.navDoc.filepath = navDocFullPath;
+    this.navDoc.url = fileUrl(this.navDoc.filepath);
+
     if (spineContainsNavDoc) {
@@ -211,5 +264,11 @@ if (spineContainsNavDoc.filepath !== navDocFullPath) {
       spi.filepath = navDocFullPath;
-      spi.title = this.parseContentDocTitle(spi.filepath);
+
+      const o = this.parseContentDocTitleAndIds(spi.filepath);
+      spi.title = o.titleText;
+      spi.targetIDs = o.docIds;
+
       spi.url = fileUrl(spi.filepath);
 
+      spi.notInReadingOrder = true;
+
       this.contentDocs.push(spi);
@@ -224,4 +283,30 @@ }
 
-EpubParser.prototype.parseContentDocTitle = function (filepath) {
+EpubParser.prototype.parseSmilRefs = function (filepath) {
   const content = fs.readFileSync(filepath).toString();
+  const doc = new DOMParser({ errorHandler }).parseFromString(content, 'application/xml');
+  const select = xpath.useNamespaces({ smil: "http://www.w3.org/ns/SMIL", epub: "http://www.idpf.org/2007/ops" });
+
+  const arr = select('//smil:text[@src]', doc);
+  let smilRefs = arr.map(o => {
+    let epubType = o.parentNode ? o.parentNode.getAttributeNS('http://www.idpf.org/2007/ops', 'type') : undefined;
+    if (epubType) {
+      epubType = epubType.trim();
+    }
+    if (!epubType) {
+      epubType = undefined;
+    }
+
+    const src = o.getAttribute("src");
+    return {
+      src,
+      full: path.join(path.dirname(filepath), src),
+      epubType
+    };
+  });
+  // console.log(arr.length, JSON.stringify(smilRefs, null, 4));
+
+  return smilRefs;
+};
+EpubParser.prototype.parseContentDocTitleAndIds = function (filepath) {
+  const content = fs.readFileSync(filepath).toString();
   // not application/xhtml+xml because:
@@ -234,7 +319,23 @@ // https://github.com/jindw/xmldom/pull/208
   const title = select('/html:html/html:head/html:title/text()', doc);
-  if (title.length > 0) {
-    return title[0].nodeValue;
-  } else {
-    return "";
-  }
+  let titleText = title.length > 0 ? title[0].nodeValue : "";
+
+  const arr = select('//*[@id]', doc);
+  let docIds = arr.map(o => {
+    let epubType = o.getAttributeNS('http://www.idpf.org/2007/ops', 'type');
+    if (epubType) {
+      epubType = epubType.trim();
+    }
+    if (!epubType) {
+      epubType = undefined;
+    }
+    return {
+      id: o.getAttribute("id"),
+      epubType
+    };
+  });
+  // console.log(arr.length, JSON.stringify(docIds, null, 4));
+
+  return {
+    titleText, docIds
+  };
 };
@@ -241,0 +342,0 @@

package.json

 {
   "name": "@daisy/epub-utils",
-  "version": "1.2.7-alpha.3",
+  "version": "1.2.7-alpha.4",
   "engines": {
@@ -31,3 +31,3 @@ "node": ">=12.0.0",
     "tmp": "^0.2.1",
-    "winston": "^3.7.2",
+    "winston": "^3.8.1",
     "xmldom": "^0.6.0",
@@ -34,0 +34,0 @@ "xpath": "^0.0.32"

No alert changes

Improved metrics

Total package byte prevSize

18513

increased by21.2%

Lines of code

465

increased by22.05%

Dependency changes

• Updatedwinston@^3.8.1