fix: support CSS logical combination pseudo-classes

rdeltour · rdeltour · commit 5635807bc0f9 · 2022-11-27T21:20:15.000+01:00
This commit update the CSS grammar parser to support funcitonal pseudo- classes taking selector listrs as argument. In effect, this impacts the parsing of the following pseudo-classes: - `:is()` (taking a forgiving selector list) - `:not()` (taking a selector list) - `:where()` (taking a forgiving selector list) - `:has()` (taking a forgiving relative selector list) Fixes #1289, Fixes #1354
diff --git a/src/main/java/org/idpf/epubcheck/util/css/CssGrammar.java b/src/main/java/org/idpf/epubcheck/util/css/CssGrammar.java
@@ -823,6 +823,7 @@ public CssSimpleSelectorSequence createSimpleSelectorSequence(final CssToken sta
       while (next.type != CssToken.Type.S
           && !MATCH_COMMA.apply(next)
           && !MATCH_OPENBRACE.apply(next)
+          && !MATCH_CLOSEPAREN.apply(next)
           && !MATCH_COMBINATOR_CHAR.apply(next))
       {
         seqItem = createSimpleSelector(iter.next(FILTER_NONE), iter, err);
@@ -838,9 +839,89 @@ public CssSimpleSelectorSequence createSimpleSelectorSequence(final CssToken sta
     }
 
     /**
-     * Create one item in a simple selector sequence. If creation fails,
-     * errors are issued, and null is returned. On return, the iterator
-     * will return the next token after the constructs last token.
+     * With start inparam being the first significant token in a selector, build
+     * the selector group (aka comma separated selectors), expected return when
+     * iter.last is '{'. On error, issue to errorlistener, and return (caller
+     * will forward).
+     *
+     * @return A syntactically valid CssSelector list, or null if fail.
+     * @throws CssException
+     */
+
+    public List<CssSelector> createSelectorList(CssToken start, CssTokenIterator iter,
+        CssErrorHandler err)
+      throws CssException
+    {
+      return createSelectorList(start, iter, err, false, false, MATCH_OPENBRACE);
+    }
+    
+    private List<CssSelector> createSelectorList(CssToken start, CssTokenIterator iter,
+        CssErrorHandler err, boolean forgiving, boolean relative, Predicate<CssToken> endMatcher)
+      throws CssException
+    {
+
+      List<CssSelector> selectors = Lists.newArrayList();
+      boolean end = false;
+      while (true)
+      { // comma loop
+        CssSelector selector = new CssSelector(start.location);
+        int selectorIndex = iter.index();
+        while (true)
+        { // combinator loop
+          if (!relative || iter.index() > selectorIndex)
+          {
+            CssSimpleSelectorSequence seq = createSimpleSelectorSequence(start, iter,
+                (forgiving) ? ForgivingErrorHandler.INSTANCE : err);
+            if (seq == null)
+            {
+              // errors already issued
+              return null;
+            }
+            selector.components.add(seq);
+            start = iter.next();
+          }
+          int idx = iter.index();
+          if (endMatcher.apply(start))
+          {
+            end = true;
+            break;
+          }
+          if (MATCH_COMMA.apply(start))
+          {
+            break;
+          }
+
+          CssSelectorCombinator comb = createCombinator(start, iter, err);
+          if (comb != null)
+          {
+            selector.components.add(comb);
+            start = iter.next();
+          }
+          else if (iter.list.get(idx - 1).type == CssToken.Type.S)
+          {
+            selector.components.add(new CssSelectorCombinator(' ', start.location));
+          }
+          else
+          {
+            err.error(new CssGrammarException(GRAMMAR_UNEXPECTED_TOKEN, iter.last.location,
+                messages.getLocale(), iter.last.chars));
+            return null;
+          }
+        } // combinator loop
+        selectors.add(selector);
+        if (end)
+        {
+          break;
+        }
+        start = iter.next();
+      } // comma loop
+      return selectors;
+    }
+
+    /**
+     * Create one item in a simple selector sequence. If creation fails, errors
+     * are issued, and null is returned. On return, the iterator will return the
+     * next token after the constructs last token.
      */
     CssConstruct createSimpleSelector(final CssToken start, final CssTokenIterator iter,
         final CssErrorHandler err) throws
@@ -971,13 +1052,21 @@ else if (next.type == CssToken.Type.FUNCTION)
         //pseudo: type_selector | universal | HASH | class | attrib | pseudo
 
         CssConstruct func;
-        if (Ascii.toLowerCase(next.getChars()).startsWith("not"))
-        {
-          func = createNegationPseudo(tk, iter, err);
-        }
-        else
+        switch (Ascii.toLowerCase(next.getChars()))
         {
+        case "not(":
+          func = createFunctionalSelectorListPseudo(tk, iter, err, false, false);
+          break;
+        case "is(":
+        case "where(":
+          func = createFunctionalSelectorListPseudo(tk, iter, err, true, false);
+          break;
+        case "has(":
+          func = createFunctionalSelectorListPseudo(tk, iter, err, true, true);
+          break;
+        default:
           func = createFunctionalPseudo(tk, iter, MATCH_OPENBRACE, err);
+          break;
         }
 
         if (func == null)
@@ -1021,25 +1110,28 @@ CssConstruct createFunctionalPseudo(final CssToken start,
       return function;
     }
 
-    CssConstruct createNegationPseudo(final CssToken start,
-        final CssTokenIterator iter, final CssErrorHandler err) throws
-        CssException
+    CssConstruct createFunctionalSelectorListPseudo(final CssToken start,
+        final CssTokenIterator iter, final CssErrorHandler err, boolean forgiving, boolean relative)
+      throws CssException
     {
 
       String name = start.getChars().substring(0, start.getChars().length() - 1);
 
       CssFunction negation = new CssFunction(name, start.location);
 
       CssToken tk = iter.next();
-      CssConstruct cc = createSimpleSelector(tk, iter, err);
-      if (cc == null || !ContextRestrictions.PSEUDO_NEGATION.apply(cc))
+      List<CssSelector> selectors = createSelectorList(tk, iter, err, forgiving, relative,
+          MATCH_CLOSEPAREN);
+      if (selectors == null)
       {
         return null;
       }
       else
       {
-        negation.components.add(cc);
-        iter.next();
+        for (CssSelector selector : selectors)
+        {
+          negation.components.add(selector);
+        }
       }
       return negation;
     }
diff --git a/src/main/java/org/idpf/epubcheck/util/css/CssParser.java b/src/main/java/org/idpf/epubcheck/util/css/CssParser.java
@@ -40,16 +40,13 @@
 import org.idpf.epubcheck.util.css.CssGrammar.CssConstructFactory;
 import org.idpf.epubcheck.util.css.CssGrammar.CssDeclaration;
 import org.idpf.epubcheck.util.css.CssGrammar.CssSelector;
-import org.idpf.epubcheck.util.css.CssGrammar.CssSelectorCombinator;
 import org.idpf.epubcheck.util.css.CssGrammar.CssSelectorConstructFactory;
-import org.idpf.epubcheck.util.css.CssGrammar.CssSimpleSelectorSequence;
 import org.idpf.epubcheck.util.css.CssToken.CssTokenConsumer;
 import org.idpf.epubcheck.util.css.CssTokenList.CssTokenIterator;
 import org.idpf.epubcheck.util.css.CssTokenList.PrematureEOFException;
 
 import com.adobe.epubcheck.util.Messages;
 import com.google.common.base.Predicate;
-import com.google.common.collect.Lists;
 
 /**
  * A CSS parser.
@@ -227,14 +224,20 @@ private void handleRuleSet(CssToken start, final CssTokenIterator iter, final Cs
     char errChar = '{';
     try
     {
-      List<CssSelector> selectors = handleSelectors(start, iter, err);
+      List<CssSelector> selectors = cssSelectorFactory.createSelectorList(start, iter, err);
       errChar = '}';
       if (selectors == null)
       {
         // handleSelectors() has issued errors, we forward
         iter.next(MATCH_CLOSEBRACE);
         return;
       }
+      if (MATCH_CLOSEPAREN.apply(iter.last)) {
+        err.error(new CssGrammarException(GRAMMAR_UNEXPECTED_TOKEN, iter.last.location,
+            messages.getLocale(), iter.last.chars));
+        iter.next(MATCH_CLOSEBRACE);
+        return;
+      }
       if (debug)
       {
         checkState(iter.last.getChar() == '{');
@@ -453,78 +456,6 @@ private boolean handlePropertyValue(CssDeclaration declaration, CssToken start,
     }
   }
 
-  /**
-   * With start inparam being the first significant token in a selector, build
-   * the selector group (aka comma separated selectors), expected return when
-   * iter.last is '{'. On error, issue to errorlistener, and return
-   * (caller will forward).
-   *
-   * @return A syntactically valid CssSelector list, or null if fail.
-   * @throws CssException
-   */
-  private List<CssSelector> handleSelectors(CssToken start, CssTokenIterator iter,
-      CssErrorHandler err) throws
-      CssException
-  {
-
-    List<CssSelector> selectors = Lists.newArrayList();
-    boolean end = false;
-    while (true)
-    { // comma loop
-      CssSelector selector = new CssSelector(start.location);
-      while (true)
-      { //combinator loop
-        CssSimpleSelectorSequence seq = cssSelectorFactory.createSimpleSelectorSequence(start, iter,
-            err);
-        if (seq == null)
-        {
-          //errors already issued
-          return null;
-        }
-        selector.components.add(seq);
-        int idx = iter.index();
-        start = iter.next();
-        if (MATCH_OPENBRACE.apply(start))
-        {
-          end = true;
-          break;
-        }
-        if (MATCH_COMMA.apply(start))
-        {
-          break;
-        }
-
-        CssSelectorCombinator comb = cssSelectorFactory.createCombinator(start, iter, err);
-        if (comb != null)
-        {
-          selector.components.add(comb);
-          start = iter.next();
-        }
-        else if (iter.list.get(idx + 1).type == CssToken.Type.S)
-        {
-          selector.components.add(new CssSelectorCombinator(' ', start.location));
-        }
-        else
-        {
-          err.error(new CssGrammarException(GRAMMAR_UNEXPECTED_TOKEN, iter.last.location,
-              messages.getLocale(), iter.last.chars));
-          return null;
-        }
-      } //combinator loop
-      selectors.add(selector);
-      if (end)
-      {
-        break;
-      }
-      if (debug)
-      {
-        checkState(MATCH_COMMA.apply(start));
-      }
-      start = iter.next();
-    } // comma loop
-    return selectors;
-  }
-
   /**
    * With start token required to be an ATKEYWORD, collect at-rule parameters if
    * any, and if the at-rule has a block, invoke those handlers.
@@ -759,23 +690,6 @@ public boolean apply(CssConstruct cc)
             ;
       }
     };
-
-    /**
-     * A context restriction for elements inside a negation pseudo.
-     */
-    static final Predicate<CssConstruct> PSEUDO_NEGATION = new Predicate<CssConstruct>()
-    {
-      public boolean apply(final CssConstruct cc)
-      {
-        checkNotNull(cc);
-        return cc.type == CssConstruct.Type.TYPE_SELECTOR
-            || cc.type == CssConstruct.Type.HASHNAME
-            || cc.type == CssConstruct.Type.CLASSNAME
-            || (cc.type == CssConstruct.Type.ATTRIBUTE_SELECTOR)
-            || (cc.type == CssConstruct.Type.PSEUDO)
-            ;
-      }
-    };
   }
 
 }
diff --git a/src/main/java/org/idpf/epubcheck/util/css/ForgivingErrorHandler.java b/src/main/java/org/idpf/epubcheck/util/css/ForgivingErrorHandler.java
@@ -0,0 +1,17 @@
+package org.idpf.epubcheck.util.css;
+
+import org.idpf.epubcheck.util.css.CssExceptions.CssException;
+
+public final class ForgivingErrorHandler implements CssErrorHandler
+{
+
+  public static final ForgivingErrorHandler INSTANCE = new ForgivingErrorHandler();
+
+  @Override
+  public void error(CssException e)
+    throws CssException
+  {
+    // do nothing
+  }
+
+}
diff --git a/src/test/resources/epub3/00-minimal/minimal.feature b/src/test/resources/epub3/00-minimal/minimal.feature
@@ -8,7 +8,6 @@
     Given EPUB test files located at '/epub3/00-minimal/files/'
     And EPUBCheck with default settings
 
-
   Scenario: Verify a minimal expanded EPUB
     When checking EPUB 'minimal'
     Then no errors or warnings are reported
diff --git a/src/test/resources/epub3/06-content-document/files/content-css-selectors-valid/EPUB/style.css b/src/test/resources/epub3/06-content-document/files/content-css-selectors-valid/EPUB/style.css
@@ -29,3 +29,23 @@ p > span:nth-child(even) {
 div[epub|type = "chapter"] {
   padding: 2em;
 }
+
+:active {
+  font-size: 1em;
+}
+
+:is(h1,h2,h3) {
+  font-size: 1em;
+}
+
+:not(nav[epub|type="landmarks"]) {
+  font-size: 1em;
+}
+
+:not(nav.class) {
+  font-size: 1em;
+}
+
+:has(>img) {
+  font-size: 1em;
+}
