Preserve whitespace as it was first specified.

Prevents Textile from collapsing white-space inside
code blocks and removes added extra hard-coded
newlines that only work because they are inside
inline HTML elements, e.g. code.

White-space is trimmed from the end of the text,
because otherwise the last extended block will include the last stranded empty line. This then removes the
need for the trim + hard-coded LF, and allows us
to actually join the blocks with the original separating
whitespace.

This fixes the formatting inside bc, notextile, bq
and pre tags, and/or any other extended blocks.

Closes #111, fixes #109

src/Netcarver/Textile/Parser.php

@@ -1843,18 +1843,30 @@ protected function blocks($text)
     {
         $blocktags = join('|', $this->blocktag_whitelist);
 
-        $text = explode("\n\n", $text);
+        $text = preg_split('/(\n{2,})/', $text, null, PREG_SPLIT_DELIM_CAPTURE);
 
         $tag  = 'p';
         $atts = '';
         $cite = '';
         $graf = '';
         $ext  = '';
         $eat  = false;
+        $whitespace = '';
+        $eatWhitespace = false;
 
         $out  = array();
 
-        foreach ($text as $line) {
+        foreach ($text as $key => $line) {
+
+            // Line is just whitespace, keep it for the next block.
+            if (trim($line) === '') {
+                if ($eatWhitespace === false) {
+                    $whitespace .= $line;
+                }
+                continue;
+            }
+
+            $eatWhitespace = false;
             $anon = 0;
             if (preg_match("/^($blocktags)($this->a$this->c)\.(\.?)(?::(\S+))? (.*)$/Ss", $line, $m)) {
                 // Last block was extended, so close it
@@ -1889,14 +1901,20 @@ protected function blocks($text)
             }
 
             $line = $this->doPBr($line);
-            $line = preg_replace('/<br>/', '<br />', $line);
+            $line = $whitespace . preg_replace('/<br>/', '<br />', $line);
 
             if ($ext && $anon) {
-                $out[count($out)-1] .= "\n".$line;
+                $out[count($out)-1] .= $line;
             } elseif (!$eat) {
                 $out[] = $line;
             }
 
+            if ($eat) {
+                $eatWhitespace = true;
+            } else {
+                $whitespace = '';
+            }
+
             if (!$ext) {
                 $tag  = 'p';
                 $atts = '';
@@ -1910,7 +1928,7 @@ protected function blocks($text)
             $out[count($out)-1] .= $c1;
         }
 
-        return join("\n\n", $out);
+        return join('', $out);
     }
 
     /**
@@ -1985,15 +2003,15 @@ protected function fBlock($m)
             $o2 = "<code>";
             $c2 = "</code>";
             $c1 = "</pre>";
-            $content = $this->shelve($this->rEncodeHTML(rtrim($content, "\n")."\n"));
+            $content = $this->shelve($this->rEncodeHTML($content));
         } elseif ($tag == 'notextile') {
             $content = $this->shelve($content);
             $o1 = '';
             $o2 = '';
             $c1 = '';
             $c2 = '';
         } elseif ($tag == 'pre') {
-            $content = $this->shelve($this->rEncodeHTML(rtrim($content, "\n")."\n"));
+            $content = $this->shelve($this->rEncodeHTML($content));
             $o1 = "<pre$atts>";
             $o2 = '';
             $c2 = '';
@@ -2857,12 +2875,10 @@ protected function cleanWhiteSpace($text)
         $out = preg_replace("/^\xEF\xBB\xBF|\x1A/", '', $text);
         // Replaces CRLF and CR with single LF.
         $out = preg_replace("/\r\n?/", "\n", $out);
-        // Replaces line endings containing only whitespace with single LF.
+        // Removes leading tabs and spaces, if the line is otherwise empty.
         $out = preg_replace("/^[ \t]*\n/m", "\n", $out);
-        // Trim three or more LFs down to 2.
-        $out = preg_replace("/\n{3,}/", "\n\n", $out);
-        // Removes leading blank lines.
-        $out = preg_replace("/^\n*/", "", $out);
+        // Removes leading and ending blank lines.
+        $out = trim($out, "\n");
         return $out;
     }
 

test/fixtures/basic.yaml

@@ -726,8 +726,7 @@ HTML Comments :
     Comment
     --></p>
 
-    <pre><code>&lt;!-- Here is a comment block in a code block. --&gt;
-    </code></pre>
+    <pre><code>&lt;!-- Here is a comment block in a code block. --&gt;</code></pre>
 
 URLs as non-links :
   notes  : Straight URL text sequences should not be converted to a hyperlink.
@@ -1329,10 +1328,21 @@ Code blocks, pre and exclusions :
     <txp:image_info type="caption" />
     </txp:category>
 
+    bc.. Code block
+
+
+
+    with three empty lines
+
     p(regular#somestuff). _Aaaand_, *__relax__*.
+
+    bc.. Paragraph 1
+
+    Paragraph 2
+
+    Paragraph 3
   expect : |
-    <pre class="input"><code>You can&#39;t _*touch*_ *_this_*... Hammer time.
-    </code></pre>
+    <pre class="input"><code>You can&#39;t _*touch*_ *_this_*... Hammer time.</code></pre>
 
     <p>And keep your <code>(expect)mitts _*off*_ *_me_*</code> as well. About the <code>&lt;hr /&gt;</code> tag.</p>
 
@@ -1342,11 +1352,22 @@ Code blocks, pre and exclusions :
     &lt;txp:image class=&quot;prawns&quot; something=&quot;summat else&quot; /&gt;
     &lt;br /&gt;
     &lt;txp:image_info type=&quot;caption&quot; /&gt;
-    &lt;/txp:category&gt;
-    </code></pre>
+    &lt;/txp:category&gt;</code></pre>
+
+    <pre><code>Code block</code>
+
+
+
+    <code>with three empty lines</code></pre>
 
     <p class="regular" id="somestuff"><em>Aaaand</em>, <strong><i>relax</i></strong>.</p>
 
+    <pre><code>Paragraph 1</code>
+
+    <code>Paragraph 2</code>
+
+    <code>Paragraph 3</code></pre>
+
 Basic span modifiers & glyphs:
   input: |
     This is the "start" and 'end'; and I'm stuck in '88!: With smd_g4llery's attributes where hëll_won't__fr33ze_over if I use apostrophes inside odd words and_underscores (Pah).
@@ -1438,8 +1459,7 @@ Make sure these don't get turned into lists... :
     ; Cache-Control: no-store, no-cache, must-revalidate, pre-check=0, post-check=0, max-age=0
     ; Expires: Sat, 24 Jul 2003 05:00:00 GMT
     ; Last-Modified: Wed, 1 Jan 2025 05:00:00 GMT
-    ; Pragma: no-cache
-    </code></pre>
+    ; Pragma: no-cache</code></pre>
 
     <p><strong>123 test</strong></p>