fix: resolve relative URIs against their base

- re-implement the `PathUtil` class to fix various URI resolution issues - remove the 3-arg `resolveRelativeReference` method - systematically resolve URI references in Content Documents against the base URI - the base URI is initially set to the document path, and overriden with `xml:base` attributes or HTML’s `base` element - add more exhaustive tests for URI normalization/resolution logic What this PR doesn’t do: - cleanup the many uses of paths strings, where we should really use URIs. The `PathUtil` code is mostly a hack, when we should rely on more robust URI normalization/resolution logic from URIs. - implement proper base resolution logic (`xml:base` should be deprecated, and only the first `base` element should be taken into account). See also w3c/epub-specs#1217. Fixes #527
w3c · Jan 22, 2019 · a84e08c · a84e08c
1 parent 0ed1ce3
commit a84e08c
Show file tree

Hide file tree

Showing 27 changed files with 704 additions and 175 deletions.
diff --git a/src/main/java/com/adobe/epubcheck/css/CSSHandler.java b/src/main/java/com/adobe/epubcheck/css/CSSHandler.java
@@ -254,7 +254,7 @@ else if (propertyName.equals("src"))
           if (construct.getType() == CssConstruct.Type.URI)
           {
             fontUri = ((CssURI) construct).toUriString();
-            fontUri = PathUtil.resolveRelativeReference(path, fontUri, null);
+            fontUri = PathUtil.resolveRelativeReference(path, fontUri);
             //check font mimetypes
             String fontMimeType = xrefChecker.getMimeType(fontUri);
             if (fontMimeType != null)
@@ -307,7 +307,7 @@ private void resolveAndRegister(String relativeRef, int line, int col, String co
   {
     if (relativeRef != null && relativeRef.trim().length() > 0)
     {
-      String resolved = PathUtil.resolveRelativeReference(path, relativeRef, null);
+      String resolved = PathUtil.resolveRelativeReference(path, relativeRef);
       xrefChecker.registerReference(path, line + startingLineNumber, col, resolved, XRefChecker.Type.GENERIC);
     }
     else

diff --git a/src/main/java/com/adobe/epubcheck/ctc/EpubCSSCheck.java b/src/main/java/com/adobe/epubcheck/ctc/EpubCSSCheck.java
@@ -176,7 +176,7 @@ static String getEntryFileName(ManifestItem itemEntry, EpubPackage epack)
     String fileToParse;
     if (epack.getPackageMainPath() != null && epack.getPackageMainPath().length() > 0)
     {
-      fileToParse = PathUtil.resolveRelativeReference(epack.getPackageMainFile(), itemEntry.getHref(), null);
+      fileToParse = PathUtil.resolveRelativeReference(epack.getPackageMainFile(), itemEntry.getHref());
     }
     else
     {

diff --git a/src/main/java/com/adobe/epubcheck/ctc/EpubExtLinksCheck.java b/src/main/java/com/adobe/epubcheck/ctc/EpubExtLinksCheck.java
@@ -80,7 +80,7 @@ public boolean validate()
               continue;
             }
 
-            imageFile = PathUtil.resolveRelativeReference(fileToParse, imageFile, null);
+            imageFile = PathUtil.resolveRelativeReference(fileToParse, imageFile);
             int index = imageFile.lastIndexOf("#");
             if (index > 0)
             {

diff --git a/src/main/java/com/adobe/epubcheck/ctc/EpubNCXCheck.java b/src/main/java/com/adobe/epubcheck/ctc/EpubNCXCheck.java
@@ -139,7 +139,7 @@ private void checkNcxDoc(String navDocEntry)
           }
           try
           {
-            path = PathUtil.resolveRelativeReference(navDocEntry, path, null);
+            path = PathUtil.resolveRelativeReference(navDocEntry, path);
           }
           catch (IllegalArgumentException ex)
           {
@@ -176,7 +176,7 @@ private void checkNcxDoc(String navDocEntry)
           if (mi != null)
           {
             String path = mi.getHref();
-            path = PathUtil.resolveRelativeReference(epack.getPackageMainFile(), path,  null);
+            path = PathUtil.resolveRelativeReference(epack.getPackageMainFile(), path);
 
             if (path != null && !path.equals(tocFileName) && !path.equals(navDocEntry) && !tocLinkSet.contains(path))
             {

diff --git a/src/main/java/com/adobe/epubcheck/ctc/EpubNavCheck.java b/src/main/java/com/adobe/epubcheck/ctc/EpubNavCheck.java
@@ -65,7 +65,7 @@ public boolean validate()
           String fileToParse;
           if (epack.getPackageMainPath() != null && epack.getPackageMainPath().length() > 0)
           {
-            fileToParse = PathUtil.resolveRelativeReference(epack.getPackageMainFile(), navDoc, null);
+            fileToParse = PathUtil.resolveRelativeReference(epack.getPackageMainFile(), navDoc);
           }
           else
           {
@@ -155,7 +155,7 @@ private boolean checkNavDoc(String navDocEntry)
             {
               path = href.substring(0, hash);
             }
-            path = PathUtil.resolveRelativeReference(navDocEntry, path, null);
+            path = PathUtil.resolveRelativeReference(navDocEntry, path);
 
             if (!path.equals("") && !tocLinkSet.contains(path))
             {
@@ -194,7 +194,7 @@ else if (type.equals("landmarks"))
         if (mi != null)
         {
             String path = mi.getHref();
-            path = PathUtil.resolveRelativeReference(epack.getPackageMainFile(), path,  null);
+            path = PathUtil.resolveRelativeReference(epack.getPackageMainFile(), path);
 
             if (path != null && !path.equals(tocFileName) && !path.equals(navDocEntry) && !tocLinkSet.contains(path))
             {

diff --git a/src/main/java/com/adobe/epubcheck/ctc/epubpackage/EpubPackage.java b/src/main/java/com/adobe/epubcheck/ctc/epubpackage/EpubPackage.java
@@ -182,7 +182,7 @@ public String getManifestItemFileName(String entryName)
 
     if (this.getPackageMainPath() != null && this.getPackageMainPath().length() > 0)
     {
-      fileToParse = PathUtil.resolveRelativeReference(this.getPackageMainFile(), entryName, null);
+      fileToParse = PathUtil.resolveRelativeReference(this.getPackageMainFile(), entryName);
     }
     else
     {

diff --git a/src/main/java/com/adobe/epubcheck/dict/SearchKeyMapHandler.java b/src/main/java/com/adobe/epubcheck/dict/SearchKeyMapHandler.java
@@ -45,7 +45,7 @@ private void processRef(String ref)
   {
     if (ref != null && context.xrefChecker.isPresent())
     {
-      ref = PathUtil.resolveRelativeReference(path, ref, null);
+      ref = PathUtil.resolveRelativeReference(path, ref);
       context.xrefChecker.get().registerReference(path, parser.getLineNumber(),
           parser.getColumnNumber(), ref, Type.SEARCH_KEY);
     }

diff --git a/src/main/java/com/adobe/epubcheck/dtbook/DTBookHandler.java b/src/main/java/com/adobe/epubcheck/dtbook/DTBookHandler.java
@@ -113,7 +113,7 @@ else if (name.equals("img"))
       if (href != null)
       {
         // TODO check if dtbook uses xml:base of so set third param
-        href = PathUtil.resolveRelativeReference(path, href, null);
+        href = PathUtil.resolveRelativeReference(path, href);
         xrefChecker.registerReference(path, parser.getLineNumber(), parser.getColumnNumber(), href,
             name.equals("img") ? XRefChecker.Type.IMAGE : XRefChecker.Type.HYPERLINK);
         URI uri = checkURI(href);

diff --git a/src/main/java/com/adobe/epubcheck/ncx/NCXHandler.java b/src/main/java/com/adobe/epubcheck/ncx/NCXHandler.java
@@ -84,7 +84,7 @@ public void startElement()
         String href = e.getAttribute("src");
         if (href != null)
         {
-          href = PathUtil.resolveRelativeReference(path, href, null);
+          href = PathUtil.resolveRelativeReference(path, href);
           if (href.startsWith("http:"))
           {
             parser.getReport().info(path, FeatureEnum.REFERENCE, href);

diff --git a/src/main/java/com/adobe/epubcheck/opf/OPFHandler.java b/src/main/java/com/adobe/epubcheck/opf/OPFHandler.java
@@ -279,7 +279,7 @@ else if (name.equals("item"))
           {
             try
             {
-              href = PathUtil.resolveRelativeReference(path, href, null);
+              href = PathUtil.resolveRelativeReference(path, href);
             } catch (IllegalArgumentException ex)
             {
               report
@@ -333,7 +333,7 @@ else if (name.equals("reference"))
         {
           try
           {
-            href = PathUtil.resolveRelativeReference(path, href, null);
+            href = PathUtil.resolveRelativeReference(path, href);
             context.xrefChecker.get().registerReference(path, parser.getLineNumber(),
                 parser.getColumnNumber(), href, XRefChecker.Type.GENERIC);
           } catch (IllegalArgumentException ex)

diff --git a/src/main/java/com/adobe/epubcheck/opf/OPFHandler30.java b/src/main/java/com/adobe/epubcheck/opf/OPFHandler30.java
@@ -398,7 +398,7 @@ private void processLink(XMLElement e)
     {
       try
       {
-        href = PathUtil.resolveRelativeReference(path, href, null);
+        href = PathUtil.resolveRelativeReference(path, href);
       } catch (IllegalArgumentException ex)
       {
         report.message(MessageId.OPF_010,

diff --git a/src/main/java/com/adobe/epubcheck/ops/OPSHandler.java b/src/main/java/com/adobe/epubcheck/ops/OPSHandler.java
@@ -73,11 +73,8 @@ int getColumnNumber()
     }
   }
 
-  /**
-   * null unless head/base or xml:base is given
-   */
-  protected URI base;
-
+  protected String base;
+  protected String baseScheme;
   protected final ValidationContext context;
   protected final XMLParser parser;
   protected final String path;
@@ -100,6 +97,7 @@ public OPSHandler(ValidationContext context, XMLParser parser)
   {
     this.context = context;
     this.path = context.path;
+    this.base = path;
     this.xrefChecker = context.xrefChecker;
     this.report = context.report;
     this.parser = parser;
@@ -111,7 +109,7 @@ protected void checkPaint(XMLElement e, String attr)
     if (xrefChecker.isPresent() && paint != null && paint.startsWith("url(") && paint.endsWith(")"))
     {
       String href = paint.substring(4, paint.length() - 1);
-      href = PathUtil.resolveRelativeReference(path, href, base == null ? null : base.toString());
+      href = PathUtil.resolveRelativeReference(base, href);
       xrefChecker.get().registerReference(path, parser.getLineNumber(), parser.getColumnNumber(),
           href, XRefChecker.Type.SVG_PAINT);
     }
@@ -122,7 +120,7 @@ protected void checkImage(XMLElement e, String attrNS, String attr)
     String href = e.getAttributeNS(attrNS, attr);
     if (xrefChecker.isPresent() && href != null)
     {
-      href = PathUtil.resolveRelativeReference(path, href, base == null ? null : base.toString());
+      href = PathUtil.resolveRelativeReference(base, href);
       xrefChecker.get().registerReference(path, parser.getLineNumber(), parser.getColumnNumber(),
           href, XRefChecker.Type.IMAGE);
     }
@@ -133,7 +131,7 @@ protected void checkObject(XMLElement e, String attrNS, String attr)
     String href = e.getAttributeNS(attrNS, attr);
     if (xrefChecker.isPresent() && href != null)
     {
-      href = PathUtil.resolveRelativeReference(path, href, base == null ? null : base.toString());
+      href = PathUtil.resolveRelativeReference(base, href);
       xrefChecker.get().registerReference(path, parser.getLineNumber(), parser.getColumnNumber(),
           href, XRefChecker.Type.OBJECT);
     }
@@ -146,7 +144,7 @@ protected void checkLink(XMLElement e, String attrNS, String attr)
     if (xrefChecker.isPresent() && href != null && rel != null
         && rel.toLowerCase(Locale.ROOT).contains("stylesheet"))
     {
-      href = PathUtil.resolveRelativeReference(path, href, base == null ? null : base.toString());
+      href = PathUtil.resolveRelativeReference(base, href);
       xrefChecker.get().registerReference(path, parser.getLineNumber(), parser.getColumnNumber(),
           href, XRefChecker.Type.STYLESHEET);
 
@@ -189,7 +187,7 @@ protected void checkSymbol(XMLElement e, String attrNS, String attr)
     String href = e.getAttributeNS(attrNS, attr);
     if (xrefChecker.isPresent() && href != null)
     {
-      href = PathUtil.resolveRelativeReference(path, href, base == null ? null : base.toString());
+      href = PathUtil.resolveRelativeReference(base, href);
       xrefChecker.get().registerReference(path, parser.getLineNumber(), parser.getColumnNumber(),
           href, XRefChecker.Type.SVG_SYMBOL);
     }
@@ -245,7 +243,7 @@ else if (".".equals(href))
      * solution to issue 155
      */
     if (URISchemes.contains(uri.getScheme())
-        || (null != base && URISchemes.contains(base.getScheme())))
+        || (null != base && URISchemes.contains(baseScheme)))
     {
       return;
     }
@@ -260,7 +258,7 @@ else if (uri.getScheme() != null)
 
     try
     {
-      href = PathUtil.resolveRelativeReference(path, href, base == null ? null : base.toString());
+      href = PathUtil.resolveRelativeReference(base, href);
     } catch (IllegalArgumentException err)
     {
       report.message(MessageId.OPF_010,
@@ -299,7 +297,9 @@ public void startElement()
     String baseTest = e.getAttributeNS(XMLConstants.XML_NS_URI, "base");
     if (baseTest != null)
     {
-      base = checkURI(baseTest);
+      URI baseURI = checkURI(baseTest);
+      baseScheme = baseURI.getScheme();
+      base = PathUtil.resolveRelativeReference(path, baseURI.toString());
     }
 
     if (!epubTypeInUse)
@@ -366,7 +366,9 @@ else if (name.equals("link"))
         }
         else if (name.equals("base"))
         {
-          base = checkURI(e.getAttribute("href"));
+          URI baseURI = checkURI(e.getAttribute("href"));
+          baseScheme = baseURI.getScheme();
+          base = PathUtil.resolveRelativeReference(path, baseURI.toString());
         }
         else if (name.equals("style"))
         {

diff --git a/src/main/java/com/adobe/epubcheck/ops/OPSHandler30.java b/src/main/java/com/adobe/epubcheck/ops/OPSHandler30.java
@@ -386,8 +386,8 @@ protected void processVideo(XMLElement e)
     String posterMimeType = null;
     if (xrefChecker.isPresent() && posterSrc != null)
     {
-      posterMimeType = xrefChecker.get().getMimeType(PathUtil.resolveRelativeReference(path,
-          posterSrc, base == null ? null : base.toString()));
+      posterMimeType = xrefChecker.get().getMimeType(PathUtil.resolveRelativeReference(base,
+          posterSrc));
     }
 
     if (posterMimeType != null && !OPFChecker.isBlessedImageType(posterMimeType))
@@ -446,7 +446,7 @@ protected void processSrc(String name, String src)
       }
       else
       {
-        src = PathUtil.resolveRelativeReference(path, src, base == null ? null : base.toString());
+        src = PathUtil.resolveRelativeReference(base, src);
       }
 
       XRefChecker.Type refType;
@@ -499,7 +499,7 @@ protected void processObject(XMLElement e)
     if (data != null)
     {
       processSrc(e.getName(), data);
-      data = PathUtil.resolveRelativeReference(path, data, base == null ? null : base.toString());
+      data = PathUtil.resolveRelativeReference(base, data);
     }
 
     if (type != null && data != null && xrefChecker.isPresent()

diff --git a/src/main/java/com/adobe/epubcheck/overlay/OverlayHandler.java b/src/main/java/com/adobe/epubcheck/overlay/OverlayHandler.java
@@ -99,7 +99,7 @@ private void processRef(String ref, XRefChecker.Type type)
   {
     if (ref != null && context.xrefChecker.isPresent())
     {
-      ref = PathUtil.resolveRelativeReference(path, ref, null);
+      ref = PathUtil.resolveRelativeReference(path, ref);
       if (type == XRefChecker.Type.AUDIO)
       {
         String mimeType = context.xrefChecker.get().getMimeType(ref);