XInclude by xml:id and automatic <xi:fallback/>

configure.php

@@ -19,9 +19,11 @@
   +----------------------------------------------------------------------+
 */
 
-error_reporting(-1);
-$cvs_id = '$Id$';
+ini_set( 'display_errors' , 1 );
+ini_set( 'display_startup_errors' , 1 );
+error_reporting( E_ALL );
 
+$cvs_id = '$Id$';
 echo "configure.php: $cvs_id\n";
 
 const RNG_SCHEMA_DIR = __DIR__ . DIRECTORY_SEPARATOR . 'docbook' . DIRECTORY_SEPARATOR . 'docbook-v5.2-os' . DIRECTORY_SEPARATOR . 'rng' . DIRECTORY_SEPARATOR;
@@ -774,29 +776,173 @@ function getFileModificationHistory(): array {
 }
 echo "done.\n";
 
-echo "Running XInclude/XPointer... ";
-$status = $dom->xinclude();
-if ($status === -1) {
-    echo "failed.\n";
-} else {
-    /* For some dumb reason when no substitution are made it returns false instead of 0... */
-    $status = (int) $status;
-    echo "done. Performed $status XIncludes\n";
-}
-flush();
 
-if ( $ac['XPOINTER_REPORTING'] == 'yes' || $ac['LANG'] == 'en' )
-{
+{   // XInclude/XPointer
+
+    echo "Running XInclude/XPointer... ";
+
+    // Always report failures on doc-en.
+    // Translations can --disable-xpointer-reporting.
+    // Failures on doc-en are fatal.
+
+    $report = $ac['XPOINTER_REPORTING'] == 'yes' || $ac['LANG'] == 'en';
+    $print = $ac['STDERR_TO_STDOUT'] == 'yes' ? STDOUT : STDERR;
+    $fatal = $ac['LANG'] == 'en';
+    $xpath = new DOMXPath( $dom );
+
+    // xinclude by xml:id pre run
+
+    $nodes = $xpath->query( "//*[xiparent]" );
+    if ( $nodes->length > 0 )
+    {
+        echo "XML contains elements with 'xiparent' attribute. XInclude by xml:id will fail.\n";
+        exit(-1);
+    }
+
+    $stx = chr( 1 ); // invalid attribute chars
+    $etx = chr( 2 ); // cannot be used in text
+
+    $nodes = $xpath->query( "//*[@xpointer]" );
+    foreach( $nodes as $xinclude )
+    {
+        $id = $xinclude->getAttribute( "xpointer" );
+        $target = $dom->getElementById( $id );
+        if ( $target == null )
+            continue;
+        $parent = $target->parentNode;
+        if ( $parent->nodeType != XML_ELEMENT_NODE )
+            continue;
+        // creates an 'xiparent' attribute on parent of xi:include xpointer target,
+        // so these targered elements are not to be changed on later fixup.
+        $mark = $parent->getAttribute( "xiparent" );
+        $mark .= "{$stx}{$id}{$etx}";
+        $parent->setAttribute( "xiparent" , $mark );
+    }
+
+    // automatic xi:fallback
+
+    $nodes = $xpath->query( "//*[local-name()='include']" );
+    foreach( $nodes as $node )
+    {
+        $target = $node->getAttribute( "xpointer" );
+        $failback = null;
+
+        $elements = $node->getElementsByTagName( "fallback" );
+        if ( count( $elements ) == 0 )
+        {
+            $failback = $node->ownerDocument->createElementNS( "http://www.w3.org/2001/XInclude" , "fallback" );
+            $node->append( $failback );
+        }
+        else
+            $failback = $elements[0];
+
+        $comment = $failback->ownerDocument->createComment( "xi:fallback $target" );
+        $failback->append( $comment );
+
+        unset( $failback ); // otherwise core dumps on PHP 8.1.2-1ubuntu2.19
+        unset( $comment );  // otherwise core dumps on PHP 8.1.2-1ubuntu2.19
+    }
+
+    // run
+
+    $ret = $dom->xinclude();
+
     $errors = libxml_get_errors();
-    $output = ( $ac['STDERR_TO_STDOUT'] == 'yes' ) ? STDOUT : STDERR;
-    if ( count( $errors ) > 0 )
+    libxml_clear_errors();
+
+    $status = $ret === -1 ? "failed" : "done";
+    $intret = $ret === -1 ? 0 : (int) $ret;
+    echo "$status. Performed $intret XIncludes.\n";
+
+    // failures
+
+    if ( $report )
     {
-        fprintf( $output , "\n");
         foreach( $errors as $error )
-            fprintf( $output , "{$error->message}\n");
-        if ( $ac['LANG'] == 'en' )
-            errors_are_bad(1);
+            fprintf( $print , "  Failed xi:include at line:   %s\n" , $error->line );
+
+        $nodes = $xpath->query( "//comment()" );
+        foreach( $nodes as $node )
+        {
+            $payload = $node->data;
+            if ( str_starts_with( $payload , "xi:fallback " ) )
+            {
+                $payload = trim( substr( $payload , 12 ) );
+                fprintf( $print , "  Used xi:fallback for target: $payload\n" );
+            }
+        }
+
+        if ( count( $errors ) > 0 )
+        {
+            fprintf( $print , "\n\n");
+            if ( $fatal )
+                errors_are_bad( 1 );
+        }
     }
+
+    // xinclude by xml:id post run fixup
+
+    $nodes = $xpath->query( "//*[@xml:id]" );
+    $count = []; // id : count
+
+    foreach( $nodes as $node )
+    {
+        $id = $node->getAttribute( "xml:id" );
+        isset( $count[$id] ) ? $count[$id]++ : $count[$id] = 1;
+    }
+
+    foreach( $nodes as $node )
+    {
+        $parent = $node->parentNode;
+        if ( $parent->nodeType != XML_ELEMENT_NODE )
+            continue;
+
+        $id = $node->getAttribute( "xml:id" );
+        $mark = "$stx" . $id . $etx;
+        $list = $parent->getAttribute( "xiparent" );
+        if ( str_contains( $list , $mark ) )
+        {
+            if ( $count[ $id ] > 1 )
+            {
+                $node->removeAttribute( "xml:id" );
+                $count[ $id ]--;
+            }
+            else
+                echo "\tStrange, xml:id was to be over deleted: $id\n";
+        }
+    }
+
+    $list = array();
+    $nodes = $xpath->query( "//*[@xml:id]" );
+    foreach( $nodes as $node )
+    {
+        $id = $node->getAttribute( "xml:id" );
+        if ( in_array( $id , $list ) )
+        {
+            // may happen while migratating xpointers from xpaths to xml:ids
+            echo "  Random removing duplicated xml:id: $id\n";
+            $node->removeAttribute( "xml:id" );
+        }
+        $list[] = $id;
+    }
+
+    $nodes = $xpath->query( "//*[@xiparent]" );
+    foreach( $nodes as $node )
+        $node->removeAttribute( "xiparent" );
+
+    flush();
+}
+
+function getNodePathWithIds( DOMNode|null $node ) : string
+{
+    if ( $node == null || $node->nodeType == XML_DOCUMENT_NODE )
+        return "";
+    $path = getNodePathWithIds( $node->parentNode );
+    $path .= "/{$node->nodeName}";
+    $id = $node->getAttribute( "xml:id" );
+    if ( $id != "" )
+        $path .= "($id)";
+    return $path;
 }
 
 echo "Validating {$ac["INPUT_FILENAME"]}... ";