? boost-575386.patch
Index: boost.install
===================================================================
RCS file: /cvs/drupal-contrib/contributions/modules/boost/boost.install,v
retrieving revision 1.2.2.1.2.3.2.49
diff -u -p -r1.2.2.1.2.3.2.49 boost.install
--- boost.install	4 Oct 2009 07:18:41 -0000	1.2.2.1.2.3.2.49
+++ boost.install	6 Oct 2009 03:47:45 -0000
@@ -195,10 +195,17 @@ function boost_schema() {
   $schema['boost_cache'] = array(
     'description' => t('List of the cached page'),
     'fields' => array(
+      'hash' => array(
+        'description' => 'MD5 hash of filename',
+        'type' => 'varchar',
+        'length' => 32,
+        'not null' => TRUE,
+        'default' => '',
+      ),
       'filename' => array(
         'description' => 'Path of the cached file relative to Drupal webroot.',
-        'type' => 'varchar',
-        'length' => 255,
+        'type' => 'text',
+        'size' => 'normal',
         'not null' => TRUE,
         'default' => '',
       ),
@@ -270,10 +277,17 @@ function boost_schema() {
         'default' => 0,
         'description' => 'Average time in milliseconds that the page took to be generated.',
       ),
+      'hash_url' => array(
+        'description' => 'MD5 hash of url',
+        'type' => 'varchar',
+        'length' => 32,
+        'not null' => TRUE,
+        'default' => '',
+      ),
       'url' => array(
         'description' => 'URL of cached page',
-        'type' => 'varchar',
-        'length' => 255,
+        'type' => 'text',
+        'size' => 'normal',
         'not null' => TRUE,
         'default' => '',
       ),
@@ -289,7 +303,7 @@ function boost_schema() {
       'page_arguments' => array('page_arguments'),
       'extension' => array('extension'),
     ),
-    'primary key' => array('filename'),
+    'primary key' => array('hash'),
   );
   $schema['boost_cache_settings'] = array(
     'description' => t('Boost cache settings'),
@@ -367,17 +381,24 @@ function boost_schema() {
         'unsigned' => TRUE,
         'not null' => TRUE
       ),
+      'hash' => array(
+        'description' => 'MD5 hash of url',
+        'type' => 'varchar',
+        'length' => 32,
+        'not null' => TRUE,
+        'default' => '',
+      ),
       'url' => array(
         'description' => 'URL of page',
-        'type' => 'varchar',
-        'length' => 255,
+        'type' => 'text',
+        'size' => 'normal',
         'not null' => TRUE,
         'default' => '',
       ),
     ),
     'primary key' => array('id'),
     'unique keys' => array(
-      'url' => array('url'),
+      'hash' => array('hash'),
     ),
   );
   return $schema;
@@ -695,3 +716,98 @@ function boost_update_6110() {
   drupal_flush_all_caches();
   return array(array('success' => TRUE, 'query' => 'Core Caches Flushed.'));
 }
+
+/**
+ * Update 6111 - Use MD5 hash for DB Keys, allowing for long URL's to be cached
+ */
+function boost_update_6111() {
+  $ret = array();
+  // Add in hash & hash_url columns to boost_cache
+  db_add_field($ret, 'boost_cache', 'hash', array(
+        'description' => 'MD5 hash of filename',
+        'type' => 'varchar',
+        'length' => 32,
+        'not null' => TRUE,
+        'default' => '',
+      ));
+  db_add_field($ret, 'boost_cache', 'hash_url', array(
+        'description' => 'MD5 hash of url',
+        'type' => 'varchar',
+        'length' => 32,
+        'not null' => TRUE,
+        'default' => '',
+      ));
+
+  // Add in hash columns to boost_crawler
+  db_add_field($ret, 'boost_crawler', 'hash', array(
+        'description' => 'MD5 hash of url',
+        'type' => 'varchar',
+        'length' => 32,
+        'not null' => TRUE,
+        'default' => '',
+      ));
+  return $ret;
+}
+
+/**
+ * Update 6112 - Calculate filename MD5 for boost_cache table
+ */
+function boost_update_6112() {
+  $result = db_query("SELECT filename FROM {boost_cache} WHERE hash = ''");
+  while ($filename = db_result($result)) {
+    $hash = md5($filename);
+    db_query("UPDATE {boost_cache} SET hash = '%s' WHERE filename = '%s'", $hash, $filename);
+  }
+  return array(array('success' => TRUE, 'query' => 'Filenames hashed'));
+}
+
+/**
+ * Update 6113 - Calculate url MD5 for boost_cache table
+ */
+function boost_update_6113() {
+  $result = db_query("SELECT url FROM {boost_cache} WHERE hash_url = ''");
+  while ($url = db_result($result)) {
+    $hash = md5($url);
+    db_query("UPDATE {boost_cache} SET hash_url = '%s' WHERE url = '%s'", $hash, $url);
+  }
+  return array(array('success' => TRUE, 'query' => 'URLs hashed'));
+}
+
+/**
+ * Update 6114 - Calculate url MD5 for boost_crawler table
+ */
+function boost_update_6114() {
+  $result = db_query("SELECT url FROM {boost_crawler} WHERE hash = ''");
+  while ($url = db_result($result)) {
+    $hash = md5($url);
+    db_query("UPDATE {boost_crawler} SET hash = '%s' WHERE url = '%s'", $hash, $url);
+  }
+  return array(array('success' => TRUE, 'query' => 'URLs hashed'));
+}
+
+/**
+ * Update 6115 - Change PK
+ */
+function boost_update_6115() {
+  $ret = array();
+  // Edit filename column & set PK
+  db_drop_primary_key($ret, 'boost_cache');
+  db_change_field($ret, 'boost_cache', 'filename', 'filename', array(
+        'type' => 'text',
+        'size' => 'normal',
+        'not null' => TRUE,
+        'default' => ''
+      ));
+  db_add_primary_key($ret, 'boost_cache', array('hash'));
+  // Edit URL column & set unique key
+  db_drop_unique_key($ret, 'boost_crawler', 'url');
+  db_change_field($ret, 'boost_crawler', 'url', 'url', array(
+        'type' => 'text',
+        'size' => 'normal',
+        'not null' => TRUE,
+        'default' => ''
+      ));
+  db_add_unique_key($ret, 'boost_crawler', 'hash', array('hash'));
+  return $ret;
+}
+
Index: boost.module
===================================================================
RCS file: /cvs/drupal-contrib/contributions/modules/boost/boost.module,v
retrieving revision 1.3.2.2.2.5.2.165
diff -u -p -r1.3.2.2.2.5.2.165 boost.module
--- boost.module	4 Oct 2009 07:24:50 -0000	1.3.2.2.2.5.2.165
+++ boost.module	6 Oct 2009 03:47:46 -0000
@@ -462,7 +462,7 @@ function boost_expire_node($node) {
       $filenames = array_merge($filenames, boost_get_db_term($tid));
     }
     foreach ($filenames as $filename) {
-      boost_cache_kill($filename);
+      boost_cache_kill($filename['filename'], $filename['hash']);
     }
   }
 
@@ -1295,17 +1295,24 @@ function boost_cache_expire($path, $wild
  * Deletes cached page from file system
  *
  * @param $filename
- *   Name of cached file; primary key in database
+ *   Name of cached file.
+ * @param $hash
+ *   Primary key in database; filename hash
  */
-function boost_cache_kill($filename) {
+function boost_cache_kill($filename, $hash = '') {
   if (BOOST_IGNORE_FLUSH < 3 && strstr($filename, BOOST_FILE_PATH)) {
-    db_query("UPDATE {boost_cache} SET expire = 0 WHERE filename = '%s'", $filename);
-    if (file_exists($filename)) {
-      @unlink($filename);
+    if ($hash == '') {
+      $hash = md5($filename);
     }
-    $gz_filename = str_replace(BOOST_FILE_PATH, BOOST_GZIP_FILE_PATH, $filename) . BOOST_GZIP_EXTENSION;
-    if (file_exists($gz_filename)) {
-      @unlink($gz_filename);
+    if ($hash != '') {
+      db_query("UPDATE {boost_cache} SET expire = 0 WHERE hash = '%s'", $hash);
+      if (file_exists($filename)) {
+        @unlink($filename);
+      }
+      $gz_filename = str_replace(BOOST_FILE_PATH, BOOST_GZIP_FILE_PATH, $filename) . BOOST_GZIP_EXTENSION;
+      if (file_exists($gz_filename)) {
+        @unlink($gz_filename);
+      }
     }
   }
 }
@@ -1317,9 +1324,9 @@ function boost_cache_kill($filename) {
  */
 function boost_cache_db_expire() {
   if (BOOST_IGNORE_FLUSH < 2) {
-    $result = db_query('SELECT filename FROM {boost_cache} WHERE expire BETWEEN 1 AND %d', BOOST_TIME);
+    $result = db_query('SELECT filename, hash FROM {boost_cache} WHERE expire BETWEEN 1 AND %d', BOOST_TIME);
     while ($boost = db_fetch_array($result)) {
-      boost_cache_kill($boost['filename']);
+      boost_cache_kill($boost['filename'], $boost['hash']);
     }
     if (BOOST_FLUSH_DIR) {
       // TO-DO: del empty dirs.
@@ -1466,7 +1473,7 @@ function boost_db_prep($filename, $exten
  * Puts boost info into database.
  *
  * @param $filename
- *   Name of cached file; primary key in database
+ *   Name of cached file; hash of this is primary key in database
  * @param $expire
  *   Expiration time
  * @param $lifetime
@@ -1485,9 +1492,11 @@ function boost_db_prep($filename, $exten
 function boost_put_db($filename, $expire, $lifetime, $push, $router_item, $timer, $timer_average, $extension) {
   global $base_root;
   $url = $base_root . request_uri();
-  db_query("UPDATE {boost_cache} SET expire = %d, lifetime = %d, push = %d, page_callback = '%s', page_arguments = '%s', timer = %d, timer_average = %d, base_dir = '%s', page_id = %d, extension = '%s', url = '%s' WHERE filename = '%s'", $expire, $lifetime, $push, $router_item['page_callback'], $router_item['page_arguments'], $timer, $timer_average, BOOST_FILE_PATH, $router_item['page_id'], $extension, $url, $filename);
+  $hash = md5($filename);
+  $hash_url = md5($url);
+  db_query("UPDATE {boost_cache} SET expire = %d, lifetime = %d, push = %d, page_callback = '%s', page_arguments = '%s', timer = %d, timer_average = %d, base_dir = '%s', page_id = %d, extension = '%s', url = '%s', filename = '%s', hash_url = '%s' WHERE hash = '%s'", $expire, $lifetime, $push, $router_item['page_callback'], $router_item['page_arguments'], $timer, $timer_average, BOOST_FILE_PATH, $router_item['page_id'], $extension, $url, $filename, $hash_url, $hash);
   if (!db_affected_rows()) {
-    db_query("INSERT INTO {boost_cache} (filename, expire, lifetime, push, page_callback, page_arguments, timer, timer_average, base_dir, page_id, extension, url) VALUES ('%s', %d, %d, %d, '%s', '%s', %d, %d, '%s', %d, '%s', '%s')", $filename, $expire, $lifetime, $push, $router_item['page_callback'], $router_item['page_arguments'], $timer, $timer_average, BOOST_FILE_PATH, $router_item['page_id'], $extension, $url);
+    db_query("INSERT INTO {boost_cache} (hash, filename, expire, lifetime, push, page_callback, page_arguments, timer, timer_average, base_dir, page_id, extension, url) VALUES ('%s', '%s', '%s', %d, %d, %d, '%s', '%s', %d, %d, '%s', %d, '%s', '%s')", $hash, $hash_url, $filename, $expire, $lifetime, $push, $router_item['page_callback'], $router_item['page_arguments'], $timer, $timer_average, BOOST_FILE_PATH, $router_item['page_id'], $extension, $url);
   }
 }
 
@@ -1495,19 +1504,18 @@ function boost_put_db($filename, $expire
  * Removes info from database. Use on 404 or 403.
  *
  * @param $filename
- *   Name of cached file; primary key in database
+ *   Name of cached file; hash of this is primary key in database
  */
 function boost_remove_db($filename) {
   if (strstr($filename, BOOST_FILE_PATH)) {
-    db_query("DELETE FROM {boost_cache} WHERE filename = '%s'", $filename);
+    $hash = md5($filename);
+    db_query("DELETE FROM {boost_cache} WHERE hash = '%s'", $hash);
   }
 }
 
 /**
  * Puts boost info into database.
  *
- * @param $filename
- *   Name of cached file; primary key in database
  * @param $expire
  *   Expiration time
  * @param $lifetime
@@ -1543,16 +1551,8 @@ function boost_put_settings_db($lifetime
 /**
  * Removes info from boost database.
  *
- * @param $filename
- *   Name of cached file; primary key in database
- * @param $expire
- *   Expiration time
- * @param $lifetime
- *   Default lifetime
- * @param $push
- *   Pre-cache this file
- * @param $router_item
- *   Array containing page_callback, page_arguments & page_id.
+ * @param $csid
+ *   Cache Settings primary ID
  */
 function boost_remove_settings_db($csid) {
   db_query("DELETE FROM {boost_cache_settings} WHERE csid = %d", $csid);
@@ -1565,6 +1565,8 @@ function boost_remove_settings_db($csid)
  *   Default lifetime
  * @param $push
  *   Pre-cache this file
+ * @param $scope
+ *   At what level does this effect cache expiration
  */
 function boost_set_db_page_settings($lifetime, $push, $scope) {
   $router_item = isset($GLOBALS['_boost_router_item']) ? $GLOBALS['_boost_router_item'] : _boost_get_menu_router();
@@ -1596,7 +1598,8 @@ function boost_set_db_page_settings($lif
  *   Filename to be looked up in the database
  */
 function boost_get_db($filename) {
-  return db_fetch_array(db_query_range("SELECT * FROM {boost_cache} WHERE filename = '%s'", $filename, 0, 1));
+  $hash = md5($filename);
+  return db_fetch_array(db_query("SELECT * FROM {boost_cache} WHERE hash = '%s'", $hash));
 }
 
 /**
@@ -1624,9 +1627,9 @@ function boost_get_settings_db($router_i
  */
 function boost_get_db_term($term) {
   $filenames = array();
-  $result = db_query("SELECT filename FROM {boost_cache} WHERE expire > 0 AND page_id = %d AND page_callback = 'taxonomy'", $term);
+  $result = db_query("SELECT filename, hash FROM {boost_cache} WHERE expire > 0 AND page_id = %d AND page_callback = 'taxonomy'", $term);
   while ($filename = db_fetch_array($result)) {
-    $filenames[] = $filename['filename'];
+    $filenames[] = $filename;
   }
   return $filenames;
 }
@@ -2005,12 +2008,13 @@ function _boost_change_extension($old, $
   // make sure we are in the webroot
   chdir(dirname($_SERVER['SCRIPT_FILENAME']));
 
-  $result = db_query("SELECT filename FROM {boost_cache} WHERE extension = '%s'", $old);
-  while ($filename = db_result($result)) {
+  $result = db_query("SELECT filename, hash FROM {boost_cache} WHERE extension = '%s'", $old);
+  while ($filename = db_fetch_array($result)) {
     // change extension
-    $new_filename = _boost_kill_file_extension($filename) . $new;
+    $new_filename = _boost_kill_file_extension($filename['filename']) . $new;
+    $hash = md5($new_filename);
     // update database
-    db_query("UPDATE {boost_cache} SET filename = '%s', extension = '%s' WHERE filename = '%s'", $new_filename, $new, $filename);
+    db_query("UPDATE {boost_cache} SET filename = '%s', extension = '%s', hash = '%s' WHERE hash = '%s'", $new_filename, $new, $hash, $filename['hash']);
 
     // update files, normal & gzip
     if (file_exists($filename)) {
@@ -2785,18 +2789,18 @@ function boost_crawler_add_to_table($pus
   if ($total > $loaded) {
     if ($push_setting) {
       if ($expire && BOOST_LOOPBACK_BYPASS) {
-        db_query_range("INSERT INTO {boost_crawler} (url) SELECT url FROM {boost_cache} WHERE push != 0 AND extension = '%s' AND expire BETWEEN 0 AND %d", $extension, BOOST_TIME, $loaded, $count);
+        db_query_range("INSERT INTO {boost_crawler} (url, hash) SELECT url, hash_url FROM {boost_cache} WHERE push != 0 AND extension = '%s' AND expire BETWEEN 0 AND %d", $extension, BOOST_TIME, $loaded, $count);
       }
       else {
-        db_query_range("INSERT INTO {boost_crawler} (url) SELECT url FROM {boost_cache} WHERE push != 0 AND extension = '%s' AND expire = 0", $extension, $loaded, $count);
+        db_query_range("INSERT INTO {boost_crawler} (url, hash) SELECT url, hash_url FROM {boost_cache} WHERE push != 0 AND extension = '%s' AND expire = 0", $extension, $loaded, $count);
       }
     }
     else {
       if ($expire && BOOST_LOOPBACK_BYPASS) {
-        db_query_range("INSERT INTO {boost_crawler} (url) SELECT url FROM {boost_cache} WHERE push = 1 AND extension = '%s' AND expire BETWEEN 0 AND %d", $extension, BOOST_TIME, $loaded, $count);
+        db_query_range("INSERT INTO {boost_crawler} (url, hash) SELECT url, hash_url FROM {boost_cache} WHERE push = 1 AND extension = '%s' AND expire BETWEEN 0 AND %d", $extension, BOOST_TIME, $loaded, $count);
       }
       else {
-        db_query_range("INSERT INTO {boost_crawler} (url) SELECT url FROM {boost_cache} WHERE push = 1 AND extension = '%s' AND expire = 0", $extension, $loaded, $count);
+        db_query_range("INSERT INTO {boost_crawler} (url, hash) SELECT url, hash_url FROM {boost_cache} WHERE push = 1 AND extension = '%s' AND expire = 0", $extension, $loaded, $count);
       }
     }
     variable_set('boost_crawler_loaded_count' . $extension, $loaded + $count);
@@ -2866,12 +2870,14 @@ function boost_crawler_add_alias_to_tabl
     return TRUE;
   }
   $count = 1000;
-  $total = db_query("SELECT COUNT(*) FROM {url_alias}");
+  $total = db_result(db_query("SELECT COUNT(*) FROM {url_alias}"));
   $loaded = variable_get('boost_crawler_loaded_count_alias', 0);
   if ($total > $loaded) {
     $list = db_query_range("SELECT dst FROM {url_alias}", $loaded, $count);
     while ($url = db_result($list)) {
-      @db_query("INSERT INTO {boost_crawler} (url) VALUES ('%s')", $base_url . '/' . $url);
+      $url = $base_url . '/' . $url;
+      $hash = md5($url);
+      @db_query("INSERT INTO {boost_crawler} (url, hash) VALUES ('%s', '%s')", $url, $hash);
     }
     variable_set('boost_crawler_loaded_count_alias', $loaded + $count);
     return FALSE;
@@ -2896,16 +2902,16 @@ function boost_crawler_total_count() {
  */
 function boost_crawler_verify($expire) {
   if ($expire && BOOST_LOOPBACK_BYPASS) {
-    $list = db_query("SELECT bcrawler.url FROM {boost_cache} bcache INNER JOIN {boost_crawler} bcrawler ON bcache.url=bcrawler.url WHERE bcache.expire BETWEEN 0 AND %d", BOOST_TIME);
+    $list = db_query("SELECT bcrawler.url, bcrawler.hash FROM {boost_cache} bcache INNER JOIN {boost_crawler} bcrawler ON bcache.hash_url=bcrawler.hash WHERE bcache.expire BETWEEN 0 AND %d", BOOST_TIME);
   }
   else {
-    $list = db_query("SELECT bcrawler.url FROM {boost_cache} bcache INNER JOIN {boost_crawler} bcrawler ON bcache.url=bcrawler.url WHERE bcache.expire = 0");
+    $list = db_query("SELECT bcrawler.url, bcrawler.hash FROM {boost_cache} bcache INNER JOIN {boost_crawler} bcrawler ON bcache.hash_url=bcrawler.hash WHERE bcache.expire = 0");
   }
   db_query('TRUNCATE {boost_crawler}');
   variable_set('boost_crawler_position', 0);
   $recrawl = FALSE;
-  while ($url = db_result($list)) {
-    db_query("INSERT INTO {boost_crawler} (url) VALUES ('%s')", $url);
+  while ($url = db_fetch_array($list)) {
+    db_query("INSERT INTO {boost_crawler} (url, hash) VALUES ('%s', '%s')", $url['url'], $url['hash']);
     $recrawl = TRUE;
   }
   return $recrawl;