--- modules/node/node.pages.inc.orig 2010-09-19 01:27:49.000000000 -0700
+++ modules/node/node.pages.inc 2010-09-19 01:36:19.000000000 -0700
@@ -263,7 +263,13 @@
function node_body_field(&$node, $label, $word_count) {
// Check if we need to restore the teaser at the beginning of the body.
- $include = !isset($node->teaser) || ($node->teaser == substr($node->body, 0, strlen($node->teaser)));
+ $include = TRUE;
+ if (isset($node->teaser)) {
+ // remove all the tags at the end because if they are closing tags
+ // then they were likely added by node_teaser().
+ $teaser = preg_replace('/(<[^>]+>)+$/', '', $node->teaser);
+ $include = $teaser == substr($node->body, 0, strlen($teaser));
+ }
$form = array(
'#after_build' => array('node_teaser_js', 'node_teaser_include_verify'));
--- modules/node/node.module.orig 2010-09-19 01:28:14.000000000 -0700
+++ modules/node/node.module 2010-09-19 01:34:47.000000000 -0700
@@ -340,61 +340,141 @@
// If the delimiter has not been specified, try to split at paragraph or
// sentence boundaries.
+ $filter_newline = isset($filters['filter/1']);
+ $len = strlen($body);
- // The teaser may not be longer than maximum length specified. Initial slice.
- $teaser = truncate_utf8($body, $size);
+ $p = 0;
+ $l = 0;
+ $s = array(); // stack
+ while ($p < $len && $l < $size) {
+ $last_tag = FALSE;
+ $o = strpos($body, '<', $p);
+ if ($o === FALSE) {
+ // no more tags till the end
+ $a = drupal_strlen(substr($body, $p, $len - $p)); // UTF-8 length
+ $n = $len;
+ }
+ else {
+ // count characters between previous position and
+ // beginning of tag
+ $a = drupal_strlen(substr($body, $p, $o - $p)); // UTF-8 length
- // Store the actual length of the UTF8 string -- which might not be the same
- // as $size.
- $max_rpos = strlen($teaser);
+ ++$o; // skip the '<'
+ $n = strpos($body, '>', $o);
- // How much to cut off the end of the teaser so that it doesn't end in the
- // middle of a paragraph, sentence, or word.
- // Initialize it to maximum in order to find the minimum.
- $min_rpos = $max_rpos;
+ if ($body[$o] == '/') {
+ // closing tag, pop the opening tag too
+ array_pop($s);
+ }
+ elseif ($body[$n - 1] != '/') { // skip empty tags
+ // opening tag, save its name on the stack so we can close it later
+ $end_name = strpos($body, ' ', $o);
+ if ($end_name === FALSE || $end_name > $n) {
+ $end_name = $n;
+ }
+ $tag_name = substr($body, $o, $end_name - $o);
+ switch ($tag_name) { // ignore empty tags that were not properly closed
+ case 'br':
+ case 'hr':
+ case 'img':
+ case 'input':
+ break;
- // Store the reverse of the teaser. We use strpos on the reversed needle and
- // haystack for speed and convenience.
- $reversed = strrev($teaser);
+ default:
+ $s[] = $tag_name;
+ $last_tag = TRUE;
+ break;
- // Build an array of arrays of break points grouped by preference.
- $break_points = array();
+ }
+ }
- // A paragraph near the end of sliced teaser is most preferable.
- $break_points[] = array('' => 0);
+ // skip the tag now (we assume properly opening/closing tag boundaries!)
+ if ($n === FALSE) {
+ // last tag not closed or it wasn't a tag?!
+ $n = $len;
+ }
+ else {
+ ++$n; // skip the '>' character
+ }
+ }
- // If no complete paragraph then treat line breaks as paragraphs.
- $line_breaks = array('
' => 6, '
' => 4);
- // Newline only indicates a line break if line break converter
- // filter is present.
- if (isset($filters['filter/1'])) {
- $line_breaks["\n"] = 1;
- }
- $break_points[] = $line_breaks;
+ // any characters to add the to result?
+ if ($a) {
+ if ($l + $a >= $size) {
+ // the last tag did not make it in
+ if ($last_tag) {
+ array_pop($s);
+ }
+ // we've got more than we want to, search for a break point
+ $o = $p + $size - $l;
+ if ($body[$o] != ' ') while ($o > $p) {
+ switch ($body[$o - 1]) {
+ case "\xD8": // "\xD8\x9F" == arabic '?' (right to left)
+ if (!isset($body[$o]) || $body[$o] != "\x9F") {
+ // no the right sequence
+ break;
+ }
+ if ($o + 1 == $len || $body[$o + 1] == ' ') {
+ // found a break-point
+ break 2;
+ }
+ if ($body[$o + 1] == '"') {
+ $o += 2;
+ break 2;
+ }
+ break;
- // If the first paragraph is too long, split at the end of a sentence.
- $break_points[] = array('. ' => 1, '! ' => 1, '? ' => 1, '。' => 0, '؟ ' => 1);
+ case '.':
+ case '!':
+ case '?':
+ if ($o == $len || $body[$o] == ' ') {
+ // found a break-point
+ break 2;
+ }
+ if ($body[$o] == '"') {
+ ++$o;
+ break 2;
+ }
+ break;
- // Iterate over the groups of break points until a break point is found.
- foreach ($break_points as $points) {
- // Look for each break point, starting at the end of the teaser.
- foreach ($points as $point => $offset) {
- // The teaser is already reversed, but the break point isn't.
- $rpos = strpos($reversed, strrev($point));
- if ($rpos !== FALSE) {
- $min_rpos = min($rpos + $offset, $min_rpos);
+ case "\n":
+ if (!$filter_newline) {
+ break;
+ }
+ case ' ':
+ // found and remove the space (not that we ignore no-break spaces since we're not supposed to break there)
+ --$o;
+ break 2;
+
+ //case ... add support for other UTF-8 spaces?
+
+ case "\xE3":
+ // found the CJK ideographic full stop?
+ if (isset($body[$o + 1]) && $body[$o] == "\x80" && $body[$o + 1] == "\x82") {
+ // keep this character in full
+ $o += 2;
+ break 2;
+ }
+ break;
+
+ }
+ --$o;
+ }
+ $p = $o;
+ break;
}
+ $l += $a;
}
- // If a break point was found in this group, slice and return the teaser.
- if ($min_rpos !== $max_rpos) {
- // Don't slice with length 0. Length must be <0 to slice from RHS.
- return ($min_rpos === 0) ? $teaser : substr($teaser, 0, 0 - $min_rpos);
- }
+ $p = $n;
}
- // If a break point was not found, still return a teaser.
- return $teaser;
+ $result = substr($body, 0, $p);
+ while (!empty($s)) {
+ $result .= '';
+ }
+
+ return $result;
}
/**