From 88ed237f7c80b00cb6b5060fed64b97a4372b23e Mon Sep 17 00:00:00 2001
From: alecpl <alec@alec.pl>
Date: Thu, 4 Jun 2009 10:27:19 +0000
Subject: - speed up plain text messages parsing (up to 60%)

---
 program/include/rcube_string_replacer.php | 24 ++++++++++++-----
 program/steps/mail/func.inc               | 43 +++++++++++++++++--------------
 2 files changed, 40 insertions(+), 27 deletions(-)

(limited to 'program')

diff --git a/program/include/rcube_string_replacer.php b/program/include/rcube_string_replacer.php
index fe082a583..064a723a2 100644
--- a/program/include/rcube_string_replacer.php
+++ b/program/include/rcube_string_replacer.php
@@ -28,10 +28,20 @@
 class rcube_string_replacer
 {
   public static $pattern = '/##str_replacement\[([0-9]+)\]##/';
-
+  public $mailto_pattern;
+  public $link_pattern;
   private $values = array();
 
 
+  function __construct()
+  {
+    $url_chars = 'a-z0-9_\-\+\*\$\/&%=@#:;';
+    $url_chars_within = '\?\.~,!';
+
+    $this->link_pattern = "/([\w]+:\/\/|\Wwww\.)([a-z0-9\-\.]+[a-z]{2,4}([$url_chars$url_chars_within]*[$url_chars])?)/i";
+    $this->mailto_pattern = "/([a-z0-9][a-z0-9\-\.\+\_]*@[a-z0-9]([a-z0-9\-][.]?)*[a-z0-9]\\.[a-z]{2,5})/i";
+  }
+
   /**
    * Add a string to the internal list
    *
@@ -64,13 +74,13 @@ class rcube_string_replacer
     $i = -1;
     $scheme = strtolower($matches[1]);
 
-    if ($scheme == 'http' || $scheme == 'https' || $scheme == 'ftp') {
-      $url = $matches[1] . '://' . $matches[2];
-      $i = $this->add(html::a(array('href' => $url, 'target' => "_blank"), Q($url)));
+    if ($scheme == 'http://' || $scheme == 'https://' || $scheme == 'ftp://') {
+      $url = $matches[1] . $matches[2];
+      $i = $this->add(html::a(array('href' => $url, 'target' => '_blank'), Q($url)));
     }
-    else if ($matches[2] == 'www.') {
-      $url = $matches[2] . $matches[3];
-      $i = $this->add($matches[1] . html::a(array('href' => 'http://' . $url, 'target' => "_blank"), Q($url)));
+    else if (preg_match('/^(\W)www\.$/', $matches[1], $m)) {
+      $url = 'www.' . $matches[2];
+      $i = $this->add($m[1] . html::a(array('href' => 'http://' . $url, 'target' => '_blank'), Q($url)));
     }
 
     return $i >= 0 ? $this->get_replacement($i) : '';
diff --git a/program/steps/mail/func.inc b/program/steps/mail/func.inc
index 25f1d81c5..9b8398ecd 100644
--- a/program/steps/mail/func.inc
+++ b/program/steps/mail/func.inc
@@ -776,50 +776,54 @@ function rcmail_print_body($part, $p = array())
   // free some memory (hopefully)
   unset($data['body']);
 
-
   // plaintext postprocessing
   if ($part->ctype_secondary == 'plain') {
     // make links and email-addresses clickable
     $replacements = new rcube_string_replacer;
     
-    $url_chars = 'a-z0-9_\-\+\*\$\/&%=@#:;';
-    $url_chars_within = '\?\.~,!';
-
     // search for patterns like links and e-mail addresses
-    $body = preg_replace_callback("/([\w]+):\/\/([a-z0-9\-\.]+[a-z]{2,4}([$url_chars$url_chars_within]*[$url_chars])?)/i", array($replacements, 'link_callback'), $body);
-    $body = preg_replace_callback("/([^\/:]|\s)(www\.)([a-z0-9\-]{2,}[a-z]{2,4}([$url_chars$url_chars_within]*[$url_chars])?)/i", array($replacements, 'link_callback'), $body);
-    $body = preg_replace_callback('/([a-z0-9][a-z0-9\-\.\+\_]*@[a-z0-9]([a-z0-9\-][.]?)*[a-z0-9]\\.[a-z]{2,5})/i', array($replacements, 'mailto_callback'), $body);
+    $body = preg_replace_callback($replacements->link_pattern, array($replacements, 'link_callback'), $body);
+    $body = preg_replace_callback($replacements->mailto_pattern, array($replacements, 'mailto_callback'), $body);
 
     // split body into single lines
     $a_lines = preg_split('/\r?\n/', $body);
+    $q_lines = array();
     $quote_level = 0;
 
-    // colorize quoted parts
-    for ($n=0; $n < count($a_lines); $n++) {
-      $line = $a_lines[$n];
-      $quotation = '';
+    // find/mark quoted lines...
+    for ($n=0, $cnt=count($a_lines); $n < $cnt; $n++) {
       $q = 0;
     
-      if (preg_match('/^(>+\s*)+/', $line, $regs)) {
-        $q    = strlen(preg_replace('/\s/', '', $regs[0]));
-        $line = substr($line, strlen($regs[0]));
+      if ($a_lines[$n][0] == '>' && preg_match('/^(>+\s*)+/', $a_lines[$n], $regs)) {
+        $q = strlen(preg_replace('/\s/', '', $regs[0]));
+	$a_lines[$n] = substr($a_lines[$n], strlen($regs[0]));
 
         if ($q > $quote_level)
-          $quotation = str_repeat('<blockquote>', $q - $quote_level);
+          $q_lines[$n]['quote'] = $q - $quote_level;
         else if ($q < $quote_level)
-          $quotation = str_repeat("</blockquote>", $quote_level - $q);
+          $eq_lines[$n]['endquote'] = $quote_level - $q;
       }
       else if ($quote_level > 0)
-        $quotation = str_repeat("</blockquote>", $quote_level);
+        $q_lines[$n]['endquote'] = $quote_level;
 
       $quote_level = $q;
-      $a_lines[$n] = $quotation . Q($line, 'replace', false);  // htmlquote plaintext
     }
 
+    // quote plain text
+    $body = Q(join("\n", $a_lines), 'replace', false);
+
+    // ... colorize quoted lines
+    $a_lines = preg_split('/\n/', $body);
+    foreach ($q_lines as $i => $q)
+      if ($q['quote'])
+        $a_lines[$i] = str_repeat('<blockquote>', $q['quote']) . $a_lines[$i];
+      else if ($q['endquote'])
+        $a_lines[$i] = str_repeat('</blockquote>', $q['endquote']) . $a_lines[$i];
+
     // insert the links for urls and mailtos
     $body = $replacements->resolve(join("\n", $a_lines));
   }
-  
+
   // allow post-processing of the message body
   $data = $RCMAIL->plugins->exec_hook('message_part_after', array('type' => $part->ctype_secondary, 'body' => $body) + $data);
 
@@ -990,7 +994,6 @@ function rcmail_message_body($attrib)
   else
     $out .= html::div('message-part', html::tag('pre', array(), Q($MESSAGE->body)));
 
-
   $ctype_primary = strtolower($MESSAGE->structure->ctype_primary);
   $ctype_secondary = strtolower($MESSAGE->structure->ctype_secondary);
 
-- 
cgit v1.2.3