@likecoin/epubcheck-ts 0.6.0 → 0.6.2

package/dist/index.d.cts

@@ -1936,6 +1936,8 @@ interface ValidationContext {
         hasLOV?: boolean;
         hasMicrodata?: boolean;
         hasRDFa?: boolean;
+        sectionCount?: number;
+        tocLinkCount?: number;
     };
 }
 /**

package/dist/index.d.ts

@@ -1936,6 +1936,8 @@ interface ValidationContext {
         hasLOV?: boolean;
         hasMicrodata?: boolean;
         hasRDFa?: boolean;
+        sectionCount?: number;
+        tocLinkCount?: number;
     };
 }
 /**

package/dist/index.js

@@ -1,8 +1,23 @@
-import { XmlDocument, XmlElement } from 'libxml2-wasm';
 import { parse, walk } from 'css-tree';
 import { unzipSync, strFromU8, gunzipSync } from 'fflate';
 
-// src/content/validator.ts
+// src/util/xml-engine.ts
+var engine;
+async function loadXmlEngine() {
+  engine ??= await import('libxml2-wasm');
+}
+function getXmlDocument() {
+  if (!engine) {
+    throw new Error("libxml2-wasm not initialized \u2014 call loadXmlEngine() first");
+  }
+  return engine.XmlDocument;
+}
+function getXmlElement() {
+  if (!engine) {
+    throw new Error("libxml2-wasm not initialized \u2014 call loadXmlEngine() first");
+  }
+  return engine.XmlElement;
+}
 
 // src/messages/messages.ts
 var severityOverrides = /* @__PURE__ */ new Map();
@@ -2691,6 +2706,31 @@ var PROFILE_DC_TYPE = {
   dict: "dictionary",
   preview: "preview"
 };
+var TYPE_TO_PROFILE = {
+  dictionary: "dict",
+  edupub: "edupub",
+  index: "idx",
+  preview: "preview"
+};
+var RESERVED_PREFIX_URIS = {
+  dcterms: "http://purl.org/dc/terms/",
+  marc: "http://id.loc.gov/vocabulary/",
+  media: "http://www.idpf.org/epub/vocab/overlays/#",
+  onix: "http://www.editeur.org/ONIX/book/codelists/current.html#",
+  rendition: "http://www.idpf.org/vocab/rendition/#",
+  schema: "http://schema.org/",
+  xsd: "http://www.w3.org/2001/XMLSchema#",
+  a11y: "http://www.idpf.org/epub/vocab/package/a11y/#"
+};
+function isValidURI(uri) {
+  if (!uri) return false;
+  try {
+    new URL(uri);
+    return true;
+  } catch {
+    return false;
+  }
+}
 var DICTIONARY_TYPE_VALUES = /* @__PURE__ */ new Set([
   "monolingual",
   "bilingual",
@@ -2798,11 +2838,13 @@ var OPFValidator = class {
     this.validatePackageAttributes(context, opfPath);
     this.validateMetadata(context, opfPath);
     if (this.packageDoc.version !== "2.0") {
+      this.validatePrefixDeclarations(context, opfPath, opfXml);
       this.validateMetaPrefixes(context, opfPath, opfXml);
     }
     this.validateLinkElements(context, opfPath);
     this.validateManifest(context, opfPath);
     this.validateSpine(context, opfPath);
+    this.validatePageMap(context, opfPath, opfXml);
     this.validateFallbackChains(context, opfPath);
     this.validateUndeclaredResources(context, opfPath);
     if (this.packageDoc.version === "2.0") {
@@ -2833,6 +2875,7 @@ var OPFValidator = class {
     if (this.packageDoc.version.startsWith("3.")) {
       this.validateAccessibilityMetadata(context, opfPath);
       this.validateProfileDcType(context, opfPath);
+      this.validateDcTypeProfileSwitch(context, opfPath);
       this.validateEdupubMetadata(context, opfPath);
       this.validateDictionaryMetadata(context, opfPath);
       this.validatePreviewMetadata(context, opfPath);
@@ -3036,6 +3079,22 @@ var OPFValidator = class {
       });
     }
   }
+  // Mirrors Java's EPUBProfile.makeTypeCompatible flow.
+  validateDcTypeProfileSwitch(context, opfPath) {
+    if (!this.packageDoc) return;
+    for (const dc of this.packageDoc.dcElements) {
+      if (dc.name !== "type") continue;
+      const inferred = TYPE_TO_PROFILE[dc.value.trim().toLowerCase()];
+      if (inferred && inferred !== context.options.profile) {
+        pushMessage(context.messages, {
+          id: MessageId.OPF_064,
+          message: `OPF declares type "${dc.value.trim().toLowerCase()}"; consider validating using the "${inferred}" profile.`,
+          location: { path: opfPath }
+        });
+        return;
+      }
+    }
+  }
   /**
    * Build lookup maps for manifest items
    */
@@ -3052,6 +3111,13 @@ var OPFValidator = class {
    */
   validatePackageAttributes(context, opfPath) {
     if (!this.packageDoc) return;
+    if (this.packageDoc.isLegacyOebps12) {
+      pushMessage(context.messages, {
+        id: MessageId.OPF_047,
+        message: "OPF file is using OEBPS 1.2 syntax allowing backwards compatibility.",
+        location: { path: opfPath }
+      });
+    }
     if (this.packageDoc.versionDeclared === false) {
       pushMessage(context.messages, {
         id: MessageId.OPF_001,
@@ -3976,14 +4042,24 @@ var OPFValidator = class {
       const resolvedPath = resolvePath(opfPath, basePathNoQuery);
       const resolvedPathDecoded = basePathDecodedNoQuery !== basePathNoQuery ? resolvePath(opfPath, basePathDecodedNoQuery) : resolvedPath;
       const fileExists = context.files.has(resolvedPath) || context.files.has(resolvedPathDecoded);
-      const inManifest = this.manifestByHref.has(basePathNoQuery) || this.manifestByHref.has(basePathDecodedNoQuery);
-      if (!fileExists && !inManifest) {
+      const manifestItem = this.manifestByHref.get(basePathNoQuery) ?? this.manifestByHref.get(basePathDecodedNoQuery);
+      if (!fileExists && !manifestItem) {
         pushMessage(context.messages, {
           id: MessageId.RSC_007w,
           message: `Referenced resource "${resolvedPath}" could not be found in the EPUB`,
           location: { path: opfPath }
         });
       }
+      if (manifestItem) {
+        const inSpine = this.packageDoc.spine.some((ref) => ref.idref === manifestItem.id);
+        if (!inSpine) {
+          pushMessage(context.messages, {
+            id: MessageId.OPF_067,
+            message: `Resource "${manifestItem.href}" is referenced as a link but is also declared as a manifest item.`,
+            location: { path: opfPath }
+          });
+        }
+      }
     }
   }
   /**
@@ -4265,6 +4341,56 @@ var OPFValidator = class {
       }
     }
   }
+  // Mirrors Java's PrefixDeclarationParser + VocabUtil.parsePrefixDeclaration,
+  // but emits only the four main IDs (not Java's OPF-004a..f sub-codes).
+  validatePrefixDeclarations(context, opfPath, opfXml) {
+    const stripped = stripXmlComments(opfXml);
+    const match = /<package[^>]*\sprefix\s*=\s*["']([^"']*)["']/.exec(stripped);
+    if (!match) return;
+    const raw = match[1] ?? "";
+    if (raw !== raw.trim()) {
+      pushMessage(context.messages, {
+        id: MessageId.OPF_004,
+        message: "The value of the prefix attribute has leading or trailing whitespace.",
+        location: { path: opfPath }
+      });
+    }
+    const parts = raw.trim().split(/\s+/).filter(Boolean);
+    for (let i = 0; i < parts.length; ) {
+      const token = parts[i] ?? "";
+      if (token.endsWith(":") && token.length > 1) {
+        const prefix = token.slice(0, -1);
+        const uri = parts[i + 1];
+        if (!uri || uri.endsWith(":")) {
+          pushMessage(context.messages, {
+            id: MessageId.OPF_005,
+            message: `The prefix "${prefix}" is declared but no URI is bound to it.`,
+            location: { path: opfPath }
+          });
+          i += 1;
+          continue;
+        }
+        if (!isValidURI(uri)) {
+          pushMessage(context.messages, {
+            id: MessageId.OPF_006,
+            message: `The value "${uri}" bound to prefix "${prefix}" is not a valid URI.`,
+            location: { path: opfPath }
+          });
+        }
+        const reservedUri = RESERVED_PREFIX_URIS[prefix];
+        if (reservedUri !== void 0 && reservedUri !== uri) {
+          pushMessage(context.messages, {
+            id: MessageId.OPF_007,
+            message: `The prefix "${prefix}" is reserved and must not be re-declared.`,
+            location: { path: opfPath }
+          });
+        }
+        i += 2;
+      } else {
+        i += 1;
+      }
+    }
+  }
   /**
    * RSC-005: all id attributes on elements in the OPF document must be unique.
    * Mirrors Java's id-unique.sch / opf.sch opf_idAttrUnique pattern, which
@@ -4273,16 +4399,7 @@ var OPFValidator = class {
    */
   validateMetaPrefixes(context, opfPath, opfXml) {
     if (!this.packageDoc) return;
-    const RESERVED = /* @__PURE__ */ new Set([
-      "dcterms",
-      "marc",
-      "onix",
-      "schema",
-      "xsd",
-      "a11y",
-      "media",
-      "rendition"
-    ]);
+    const RESERVED = new Set(Object.keys(RESERVED_PREFIX_URIS));
     const declared = new Set(Object.keys(this.packageDoc.prefixes ?? {}));
     const reported = /* @__PURE__ */ new Set();
     const reportIfUndeclared = (prefix) => {
@@ -4489,6 +4606,26 @@ var OPFValidator = class {
       }
     }
   }
+  validatePageMap(context, opfPath, opfXml) {
+    if (!this.packageDoc) return;
+    const stripped = stripXmlComments(opfXml);
+    const m = /<spine\b[^>]*\spage-map\s*=\s*["']([^"']*)["']/.exec(stripped);
+    if (!m) return;
+    const pageMapId = (m[1] ?? "").trim();
+    pushMessage(context.messages, {
+      id: MessageId.OPF_062,
+      message: `Found Adobe page-map attribute on spine element (page-map="${pageMapId}")`,
+      location: { path: opfPath }
+    });
+    if (!pageMapId) return;
+    if (!this.manifestById.has(pageMapId)) {
+      pushMessage(context.messages, {
+        id: MessageId.OPF_063,
+        message: `The Adobe page-map item "${pageMapId}" was not found in the manifest`,
+        location: { path: opfPath }
+      });
+    }
+  }
   /**
    * Validate fallback chains
    */
@@ -4960,7 +5097,7 @@ var SKMValidator = class {
     const content = typeof data === "string" ? data : new TextDecoder().decode(data);
     let doc = null;
     try {
-      doc = XmlDocument.fromString(content);
+      doc = getXmlDocument().fromString(content);
     } catch {
       pushMessage(context.messages, {
         id: MessageId.RSC_016,
@@ -5197,7 +5334,7 @@ var SMILValidator = class {
     const content = typeof data === "string" ? data : new TextDecoder().decode(data);
     let doc = null;
     try {
-      doc = XmlDocument.fromString(content);
+      doc = getXmlDocument().fromString(content);
     } catch {
       pushMessage(context.messages, {
         id: MessageId.RSC_016,
@@ -6107,7 +6244,7 @@ var ContentValidator = class {
     const svgContent = new TextDecoder().decode(svgData);
     let doc;
     try {
-      doc = XmlDocument.fromString(svgContent);
+      doc = getXmlDocument().fromString(svgContent);
       this.extractAndRegisterIDs(path, doc.root, registry);
     } catch (e) {
       pushMessage(context.messages, {
@@ -6125,7 +6262,7 @@ var ContentValidator = class {
     const svgContent = new TextDecoder().decode(svgData);
     let doc;
     try {
-      doc = XmlDocument.fromString(svgContent);
+      doc = getXmlDocument().fromString(svgContent);
     } catch {
       return;
     }
@@ -6171,7 +6308,7 @@ var ContentValidator = class {
     const svgContent = new TextDecoder().decode(svgData);
     let doc;
     try {
-      doc = XmlDocument.fromString(svgContent);
+      doc = getXmlDocument().fromString(svgContent);
     } catch {
       return;
     }
@@ -6521,7 +6658,7 @@ var ContentValidator = class {
     }
     let doc = null;
     try {
-      doc = XmlDocument.fromString(content);
+      doc = getXmlDocument().fromString(content);
     } catch (error) {
       if (error instanceof Error) {
         const { message, line, column } = this.parseLibxmlError(error.message);
@@ -7212,6 +7349,9 @@ var ContentValidator = class {
     const docDir = path.includes("/") ? path.substring(0, path.lastIndexOf("/")) : "";
     const opfDir = context.opfPath?.includes("/") ? context.opfPath.substring(0, context.opfPath.lastIndexOf("/")) : "";
     const tocAnchors = tocNav.find(".//html:a[@href]", HTML_NS);
+    if (context.contentFeatures) {
+      context.contentFeatures.tocLinkCount = (context.contentFeatures.tocLinkCount ?? 0) + tocAnchors.length;
+    }
     const tocLinks = [];
     for (const anchor of tocAnchors) {
       const href = this.getAttribute(anchor, "href")?.trim();
@@ -8334,6 +8474,10 @@ var ContentValidator = class {
     if (!features.hasRDFa && root.get(".//*[@property]")) {
       features.hasRDFa = true;
     }
+    if (context.options.profile === "edupub") {
+      const sections = root.find(".//html:body//html:section", XHTML_NS);
+      features.sectionCount = (features.sectionCount ?? 0) + sections.length;
+    }
   }
   validateImages(context, path, root) {
     const packageDoc = context.packageDocument;
@@ -8599,6 +8743,7 @@ var ContentValidator = class {
       }
       return Number.parseInt(el.name.substring(1), 10);
     };
+    const XmlElement = getXmlElement();
     const directElementChildren = (parent) => {
       const out = [];
       let n = parent.firstChild;
@@ -10207,11 +10352,13 @@ function toJSONReport(result) {
     2
   );
 }
+
+// src/ncx/validator.ts
 var NCXValidator = class {
   validate(context, ncxContent, ncxPath, registry) {
     let doc = null;
     try {
-      doc = XmlDocument.fromString(ncxContent);
+      doc = getXmlDocument().fromString(ncxContent);
     } catch (error) {
       if (error instanceof Error) {
         pushMessage(context.messages, {
@@ -10269,6 +10416,13 @@ var NCXValidator = class {
       });
       return;
     }
+    if (uidContent !== uidContent.trim()) {
+      pushMessage(context.messages, {
+        id: MessageId.NCX_004,
+        message: "NCX dtb:uid meta content has leading or trailing whitespace.",
+        location: { path, line: uidElement.line }
+      });
+    }
     context.ncxUid = uidContent.trim();
   }
   checkNavMap(context, root, path) {
@@ -10898,8 +11052,8 @@ var OCFValidator = class {
       zip = ZipReader.open(context.data);
     } catch (error) {
       pushMessage(context.messages, {
-        id: MessageId.PKG_001,
-        message: `Failed to open EPUB file: ${error instanceof Error ? error.message : "Unknown error"}`
+        id: MessageId.PKG_004,
+        message: `Failed to open EPUB ZIP: ${error instanceof Error ? error.message : "Unknown error"}`
       });
       return;
     }
@@ -10933,8 +11087,8 @@ var OCFValidator = class {
     const compressionInfo = zip.getMimetypeCompressionInfo();
     if (compressionInfo === null) {
       pushMessage(messages, {
-        id: MessageId.PKG_006,
-        message: "Could not read ZIP header",
+        id: MessageId.PKG_003,
+        message: "Unable to read EPUB file header, likely corrupted",
         location: { path: "mimetype" }
       });
       return;
@@ -11961,11 +12115,11 @@ var RelaxNGValidator = class extends BaseSchemaValidator {
     try {
       const libxml2 = await import('libxml2-wasm');
       const LibRelaxNGValidator = libxml2.RelaxNGValidator;
-      const { XmlDocument: XmlDocument5 } = libxml2;
-      const doc = XmlDocument5.fromString(xml);
+      const { XmlDocument } = libxml2;
+      const doc = XmlDocument.fromString(xml);
       try {
         const schemaContent = await loadSchema(schemaPath);
-        const schemaDoc = XmlDocument5.fromString(schemaContent);
+        const schemaDoc = XmlDocument.fromString(schemaContent);
         try {
           const validator = LibRelaxNGValidator.fromDoc(schemaDoc);
           try {
@@ -12175,6 +12329,7 @@ var EpubCheck = class _EpubCheck {
    */
   async check(data, filename) {
     const startTime = performance.now();
+    await loadXmlEngine();
     const context = {
       data,
       options: this.options,
@@ -12199,7 +12354,7 @@ var EpubCheck = class _EpubCheck {
       await this.runPipeline(context);
     } catch (error) {
       pushMessage(context.messages, {
-        id: MessageId.PKG_025,
+        id: MessageId.PKG_008,
         message: error instanceof Error ? error.message : "Unknown validation error"
       });
     } finally {
@@ -12216,6 +12371,7 @@ var EpubCheck = class _EpubCheck {
    */
   async checkExpanded(files) {
     const startTime = performance.now();
+    await loadXmlEngine();
     const context = {
       data: new Uint8Array(0),
       options: this.options,
@@ -12241,7 +12397,7 @@ var EpubCheck = class _EpubCheck {
       await this.runPipeline(context);
     } catch (error) {
       pushMessage(context.messages, {
-        id: MessageId.PKG_025,
+        id: MessageId.PKG_008,
         message: error instanceof Error ? error.message : "Unknown validation error"
       });
     } finally {
@@ -12260,6 +12416,7 @@ var EpubCheck = class _EpubCheck {
   async checkSingleFile(data, filename) {
     const startTime = performance.now();
     const mode = this.options.mode;
+    await loadXmlEngine();
     const context = {
       data: new Uint8Array(0),
       options: this.options,
@@ -12308,7 +12465,7 @@ var EpubCheck = class _EpubCheck {
       }
     } catch (error) {
       pushMessage(context.messages, {
-        id: MessageId.PKG_025,
+        id: MessageId.PKG_008,
         message: error instanceof Error ? error.message : "Unknown validation error"
       });
     } finally {
@@ -12358,6 +12515,15 @@ var EpubCheck = class _EpubCheck {
     const profile = context.options.profile;
     const opfPath = context.opfPath ?? "";
     if (profile === "edupub") {
+      const sectionCount = features.sectionCount ?? 0;
+      const tocLinkCount = features.tocLinkCount ?? 0;
+      if (sectionCount > 0 && sectionCount !== tocLinkCount) {
+        pushMessage(context.messages, {
+          id: MessageId.NAV_004,
+          message: "The Navigation Document should contain the full hierarchy of headings in the document for EDUPUB.",
+          location: { path: opfPath }
+        });
+      }
       if (features.hasPageBreak && !features.hasPageList) {
         pushMessage(context.messages, {
           id: MessageId.NAV_003,
@@ -12698,7 +12864,14 @@ var EpubCheck = class _EpubCheck {
         message: "For maximum compatibility, use only lowercase characters for the EPUB file extension.",
         location: { path: filename }
       });
+      return;
     }
+    const isEpub2 = context.version.startsWith("2");
+    pushMessage(context.messages, {
+      id: isEpub2 ? MessageId.PKG_017 : MessageId.PKG_024,
+      message: `EPUB file has an uncommon extension "${extension}".`,
+      location: { path: filename }
+    });
   }
   /**
    * Build a filtered report from validation context