@likecoin/epubcheck-ts 0.3.1 → 0.3.2

package/README.md

@@ -6,7 +6,7 @@ A TypeScript port of [EPUBCheck](https://github.com/w3c/epubcheck) - the officia
 [![npm](https://img.shields.io/npm/v/%40likecoin%2Fepubcheck-ts)](https://www.npmjs.com/package/@likecoin/epubcheck-ts)
 [![License](https://img.shields.io/npm/l/%40likecoin%2Fepubcheck-ts)](./LICENSE)
 
-> **Note**: This library is primarily developed for internal use at [3ook.com](https://3ook.com/about) and is built with AI-assisted development. While it has comprehensive test coverage (467 tests) and ~70% feature parity with Java EPUBCheck, it may not be suitable for mission-critical production workloads. For production environments requiring full EPUB validation, consider using the official [Java EPUBCheck](https://github.com/w3c/epubcheck). Contributions and feedback are welcome!
+> **Note**: This library is primarily developed for internal use at [3ook.com](https://3ook.com/about) and is built with AI-assisted development. While it has comprehensive test coverage (476 tests) and ~70% feature parity with Java EPUBCheck, it may not be suitable for mission-critical production workloads. For production environments requiring full EPUB validation, consider using the official [Java EPUBCheck](https://github.com/w3c/epubcheck). Contributions and feedback are welcome!
 
 ## Features
 

package/bin/epubcheck.js

@@ -3,7 +3,7 @@ import { readFile, writeFile } from "node:fs/promises";
 import { parseArgs } from "node:util";
 import { basename } from "node:path";
 const { EpubCheck, toJSONReport } = await import("../dist/index.js");
-const VERSION = "0.3.1";
+const VERSION = "0.3.2";
 const { values, positionals } = parseArgs({
   options: {
     json: { type: "string", short: "j" },

package/bin/epubcheck.ts

@@ -14,7 +14,7 @@ import { basename } from 'node:path';
 // Dynamic import to support both ESM and CJS builds
 const { EpubCheck, toJSONReport } = await import('../dist/index.js');
 
-const VERSION = '0.3.1';
+const VERSION = '0.3.2';
 
 // Parse command line arguments
 const { values, positionals } = parseArgs({

package/dist/index.cjs

@@ -1758,9 +1758,27 @@ var ContentValidator = class {
       } else if (item.mediaType === "text/css" && refValidator) {
         const fullPath = opfDir ? `${opfDir}/${item.href}` : item.href;
         this.validateCSSDocument(context, fullPath, opfDir, refValidator);
+      } else if (item.mediaType === "image/svg+xml" && registry) {
+        const fullPath = opfDir ? `${opfDir}/${item.href}` : item.href;
+        this.extractSVGIDs(context, fullPath, registry);
       }
     }
   }
+  extractSVGIDs(context, path, registry) {
+    const svgData = context.files.get(path);
+    if (!svgData) {
+      return;
+    }
+    const svgContent = new TextDecoder().decode(svgData);
+    let doc;
+    try {
+      doc = libxml2Wasm.XmlDocument.fromString(svgContent);
+      this.extractAndRegisterIDs(path, doc.root, registry);
+    } catch {
+    } finally {
+      doc?.dispose();
+    }
+  }
   validateCSSDocument(context, path, opfDir, refValidator) {
     const cssData = context.files.get(path);
     if (!cssData) {
@@ -1958,9 +1976,11 @@ var ContentValidator = class {
         this.extractAndRegisterIDs(path, root, registry);
       }
       if (refValidator && opfDir !== void 0) {
-        this.extractAndRegisterHyperlinks(path, root, opfDir, refValidator);
+        this.extractAndRegisterHyperlinks(context, path, root, opfDir, refValidator);
         this.extractAndRegisterStylesheets(path, root, opfDir, refValidator);
         this.extractAndRegisterImages(path, root, opfDir, refValidator);
+        this.extractAndRegisterMathMLAltimg(path, root, opfDir, refValidator);
+        this.extractAndRegisterScripts(path, root, opfDir, refValidator);
         this.extractAndRegisterCiteAttributes(path, root, opfDir, refValidator);
         this.extractAndRegisterMediaElements(path, root, opfDir, refValidator);
       }
@@ -2425,12 +2445,20 @@ var ContentValidator = class {
       }
     }
   }
-  extractAndRegisterHyperlinks(path, root, opfDir, refValidator) {
+  extractAndRegisterHyperlinks(context, path, root, opfDir, refValidator) {
     const docDir = path.includes("/") ? path.substring(0, path.lastIndexOf("/")) : "";
     const links = root.find(".//html:a[@href]", { html: "http://www.w3.org/1999/xhtml" });
     for (const link of links) {
       const href = this.getAttribute(link, "href");
-      if (!href) continue;
+      if (href === null) continue;
+      if (href.trim() === "") {
+        pushMessage(context.messages, {
+          id: MessageId.HTM_045,
+          message: "Encountered empty href",
+          location: { path, line: link.line }
+        });
+        continue;
+      }
       const line = link.line;
       if (href.startsWith("http://") || href.startsWith("https://")) {
         continue;
@@ -2570,7 +2598,8 @@ var ContentValidator = class {
     const docDir = path.includes("/") ? path.substring(0, path.lastIndexOf("/")) : "";
     const images = root.find(".//html:img[@src]", { html: "http://www.w3.org/1999/xhtml" });
     for (const img of images) {
-      const src = this.getAttribute(img, "src");
+      const imgElem = img;
+      const src = this.getAttribute(imgElem, "src");
       if (!src) continue;
       const line = img.line;
       if (src.startsWith("http://") || src.startsWith("https://")) {
@@ -2580,15 +2609,26 @@ var ContentValidator = class {
           type: "image" /* IMAGE */,
           location: { path, line }
         });
-        continue;
+      } else {
+        const resolvedPath = this.resolveRelativePath(docDir, src, opfDir);
+        const hashIndex = resolvedPath.indexOf("#");
+        const targetResource = hashIndex >= 0 ? resolvedPath.slice(0, hashIndex) : resolvedPath;
+        const fragment = hashIndex >= 0 ? resolvedPath.slice(hashIndex + 1) : void 0;
+        const ref = {
+          url: src,
+          targetResource,
+          type: "image" /* IMAGE */,
+          location: { path, line }
+        };
+        if (fragment) {
+          ref.fragment = fragment;
+        }
+        refValidator.addReference(ref);
+      }
+      const srcset = this.getAttribute(imgElem, "srcset");
+      if (srcset) {
+        this.parseSrcset(srcset, docDir, opfDir, path, line, refValidator);
       }
-      const resolvedPath = this.resolveRelativePath(docDir, src, opfDir);
-      refValidator.addReference({
-        url: src,
-        targetResource: resolvedPath,
-        type: "image" /* IMAGE */,
-        location: { path, line }
-      });
     }
     let svgImages = [];
     try {
@@ -2618,12 +2658,19 @@ var ContentValidator = class {
         continue;
       }
       const resolvedPath = this.resolveRelativePath(docDir, href, opfDir);
-      refValidator.addReference({
+      const hashIndex = resolvedPath.indexOf("#");
+      const targetResource = hashIndex >= 0 ? resolvedPath.slice(0, hashIndex) : resolvedPath;
+      const fragment = hashIndex >= 0 ? resolvedPath.slice(hashIndex + 1) : void 0;
+      const svgImgRef = {
         url: href,
-        targetResource: resolvedPath,
+        targetResource,
         type: "image" /* IMAGE */,
         location: { path, line }
-      });
+      };
+      if (fragment) {
+        svgImgRef.fragment = fragment;
+      }
+      refValidator.addReference(svgImgRef);
     }
     const videos = root.find(".//html:video[@poster]", { html: "http://www.w3.org/1999/xhtml" });
     for (const video of videos) {
@@ -2648,6 +2695,58 @@ var ContentValidator = class {
       });
     }
   }
+  extractAndRegisterMathMLAltimg(path, root, opfDir, refValidator) {
+    const docDir = path.includes("/") ? path.substring(0, path.lastIndexOf("/")) : "";
+    const mathElements = root.find(".//math:math[@altimg]", {
+      math: "http://www.w3.org/1998/Math/MathML"
+    });
+    for (const mathElem of mathElements) {
+      const altimg = this.getAttribute(mathElem, "altimg");
+      if (!altimg) continue;
+      const line = mathElem.line;
+      if (altimg.startsWith("http://") || altimg.startsWith("https://")) {
+        refValidator.addReference({
+          url: altimg,
+          targetResource: altimg,
+          type: "image" /* IMAGE */,
+          location: { path, line }
+        });
+        continue;
+      }
+      const resolvedPath = this.resolveRelativePath(docDir, altimg, opfDir);
+      refValidator.addReference({
+        url: altimg,
+        targetResource: resolvedPath,
+        type: "image" /* IMAGE */,
+        location: { path, line }
+      });
+    }
+  }
+  extractAndRegisterScripts(path, root, opfDir, refValidator) {
+    const docDir = path.includes("/") ? path.substring(0, path.lastIndexOf("/")) : "";
+    const scripts = root.find(".//html:script[@src]", { html: "http://www.w3.org/1999/xhtml" });
+    for (const script of scripts) {
+      const src = this.getAttribute(script, "src");
+      if (!src) continue;
+      const line = script.line;
+      if (src.startsWith("http://") || src.startsWith("https://")) {
+        refValidator.addReference({
+          url: src,
+          targetResource: src,
+          type: "generic" /* GENERIC */,
+          location: { path, line }
+        });
+        continue;
+      }
+      const resolvedPath = this.resolveRelativePath(docDir, src, opfDir);
+      refValidator.addReference({
+        url: src,
+        targetResource: resolvedPath,
+        type: "generic" /* GENERIC */,
+        location: { path, line }
+      });
+    }
+  }
   /**
    * Extract cite attribute references from blockquote, q, ins, del elements
    * These need to be validated as RSC-007 if the referenced resource is missing
@@ -2798,6 +2897,32 @@ var ContentValidator = class {
       }
     }
   }
+  parseSrcset(srcset, docDir, opfDir, path, line, refValidator) {
+    const entries = srcset.split(",");
+    for (const entry of entries) {
+      const trimmed = entry.trim();
+      if (!trimmed) continue;
+      const url = trimmed.split(/\s+/)[0];
+      if (!url) continue;
+      const location = line !== void 0 ? { path, line } : { path };
+      if (url.startsWith("http://") || url.startsWith("https://")) {
+        refValidator.addReference({
+          url,
+          targetResource: url,
+          type: "image" /* IMAGE */,
+          location
+        });
+      } else {
+        const resolvedPath = this.resolveRelativePath(docDir, url, opfDir);
+        refValidator.addReference({
+          url,
+          targetResource: resolvedPath,
+          type: "image" /* IMAGE */,
+          location
+        });
+      }
+    }
+  }
   resolveRelativePath(docDir, href, _opfDir) {
     const hrefWithoutFragment = href.split("#")[0] ?? href;
     const fragment = href.includes("#") ? href.split("#")[1] : "";
@@ -4838,11 +4963,19 @@ var ReferenceValidator = class {
     }
     if (isDataURL(url)) {
       if (this.version.startsWith("3.")) {
-        pushMessage(context.messages, {
-          id: MessageId.RSC_029,
-          message: "Data URLs are not allowed in EPUB 3",
-          location: reference.location
-        });
+        const forbiddenDataUrlTypes = [
+          "hyperlink" /* HYPERLINK */,
+          "nav-toc-link" /* NAV_TOC_LINK */,
+          "nav-pagelist-link" /* NAV_PAGELIST_LINK */,
+          "cite" /* CITE */
+        ];
+        if (forbiddenDataUrlTypes.includes(reference.type)) {
+          pushMessage(context.messages, {
+            id: MessageId.RSC_029,
+            message: "Data URLs are not allowed in this context",
+            location: reference.location
+          });
+        }
       }
       return;
     }
@@ -4982,6 +5115,14 @@ var ReferenceValidator = class {
       });
       return;
     }
+    if (reference.type === "image" /* IMAGE */ && resource?.mimeType !== "image/svg+xml") {
+      pushMessage(context.messages, {
+        id: MessageId.RSC_009,
+        message: `Fragment identifier used on a non-SVG image resource: ${resourcePath}#${fragment}`,
+        location: reference.location
+      });
+      return;
+    }
     if (resource?.mimeType === "image/svg+xml") {
       const hasSVGView = fragment.includes("svgView(") || fragment.includes("viewBox(");
       if (hasSVGView && reference.type === "hyperlink" /* HYPERLINK */) {