'admin/settings/swish/main', 'title' => 'swish settings',
'callback' => 'swish_settings',
'access' => user_access('administer swish-e'),
'type' => MENU_DEFAULT_LOCAL_TASK, 'weight' => -10
);
$items[] = array(
'path' => 'admin/settings/swish/index', 'title' => 'swish indexer',
'callback' => 'swish_index',
'access' => user_access('administer swish-e'),
'type' => MENU_LOCAL_TASK
);
return $items;
}
/**
* swish_settings callback function
*/
function swish_settings(){
swish_validate_paths();
$form['swish_path'] = array(
'#type' => 'textfield',
'#title' => t('Swish-E Path'),
'#default_value' => variable_get('swish_path', '/usr/local/bin/swish-e'),
'#size' => 60,
'#maxlength' => 120,
'#description' => t('The absolute path of the swish-e binary.'),
);
$form['catdoc_path'] = array(
'#type' => 'textfield',
'#title' => t('M$ Word Filter Path'),
'#default_value' => variable_get('catdoc_path', '/usr/local/bin/catdoc'),
'#size' => 60,
'#maxlength' => 120,
'#description' => t('The absolute path of the M$ Word filter.'),
);
$form['xls2csv_path'] = array(
'#type' => 'textfield',
'#title' => t('M$ Excel Filter Path'),
'#default_value' => variable_get('xls2csv_path', '/usr/local/bin/xls2csv'),
'#size' => 60,
'#maxlength' => 120,
'#description' => t('The absolute path of the M$ Excel filter'),
);
$form['ppthtml_path'] = array(
'#type' => 'textfield',
'#title' => t('M$ Powerpoint Filter Path'),
'#default_value' => variable_get('ppthtml_path', '/usr/local/bin/ppthtml'),
'#size' => 60,
'#maxlength' => 120,
'#description' => t('The absolute path of the M$ Excel filter'),
);
$form['pdftotext_path'] = array(
'#type' => 'textfield',
'#title' => t('PDF Filter Path'),
'#default_value' => variable_get('pdftotext_path', '/usr/local/bin/pdftotext'),
'#size' => 60,
'#maxlength' => 120,
'#description' => t('The absolute path of the PDF filter.'),
);
$form['unrtf_path'] = array(
'#type' => 'textfield',
'#title' => t('RTF Filter Path'),
'#default_value' => variable_get('unrtf_path', '/usr/bin/unrtf'),
'#size' => 60,
'#maxlength' => 120,
'#description' => t('The absolute path of the Rich Text Format filter.'),
);
return $form;
}
/**
* swish_settings callback function
*/
function swish_index() {
if (swish_validate_paths()) {
drupal_set_message('Swish module settings errors exist. Fix it '. l('here.', 'admin/settings/swish'), 'error');
$disable = 'disabled';
}
else {
$op = $_POST['op'];
if ($op) {
$result = _swish_do_index();
if (substr($result[0], 0, 5) == 'err: ' ){
drupal_set_message('Indexing aborted. '.$result[0], 'error');
}
else {
$length = count($result);
//include only the last 3 lines
for ($i = $length - 1 ; $i >= ($length - 3) ; --$i)
$msg .= $result[$i]."
";
drupal_set_message($msg);
}
}
}
$form['index'] = array(
'#type' => 'submit',
'#value' => 'Index Now',
'#attributes' => array('disable' => $disable),
);
$output = drupal_get_form('Swish Indexer', $form);
return print theme('page', $output, 'Swish Indexer');
}
/**
* This will be invoked by the search module.
* Returns the Header, and array of items found, if any.
*/
function swish_search ($op = 'search', $keys = null){
switch ($op) {
case 'name':
return t('uploaded files');
case 'reset':
return;
case 'search' :
$find = array();
$swish_bin = variable_get('swish_path', '/usr/local/bin/swish-e');
$swish_index = getcwd().'\\files\\my_swish_index';
//check if swish-e binary is executable and index file is readable.
//if not, put an entry in the watchdog and just exit quitely.
//we dont want to disturb the end user.
if (!is_executable($swish_bin) && !is_readable($swish_index) && !is_readable($swish_index.'.prop')) {
watchdog('swish', 'Search: Swish binary is not executable or doesnt exist. Or, index files are not readable or do not exist.
Fix it '. l('here.', 'admin/settings/swish'));
return array();
}
$words = '"'.chop(str_replace('\(', '(', str_replace('\)', ')', str_replace('\*','*', escapeshellcmd($keys).' ')))).'"';
$swish_command = variable_get("swish_path","/usr/local/bin/swish-e") . ' -m 50 -f '. getcwd() . '\\files\\my_swish_index -w '.$words;
exec ($swish_command, $results, $return_val);
foreach ($results as $result){
if (substr($result,0,1) == '#' || substr($result,0,1) == '.' ){
//TODO: parse swish result comments
//for now, just continue;
continue;
}
else if (substr($result, 0, 5) == 'err: ' ){
//if swish-e coughs error msgs, just exit and display no
// document found.
watchdog('swish', 'Swish-e Barfed: '.$result.'
Command: '.$swish_command);
return array();
}
else { //if there are results.
$k = strpos($result,' '); //first space
$i = strpos($result,' "');
$j = strpos ($result, '" ');
$rank = substr($result, 0, $k);
$basename = basename(substr($result, $k+1, $i-$k-1));
$l = strlen($basename);
$filetype = substr($basename, $l-3, 3 );
$title = substr($result, $i + 2, $j -$i -2 );
$file_size = substr($result, $j+2);
$extra = array(
'rank' => $rank,
'file_size' => $file_size,
'file_type' => $filetype );
$link = file_create_url ("files/".$basename);
$find[] = array('link' => $link, 'title' => $title, 'extra' => $extra);
}
}
return $find;
}
}
/**
* Overrides the _search_item hook.
* Returns a nice html output of each found item.
*/
function swish_search_item($item){
global $base_url;
$extra = $item['extra'];
$output = '
.'.png)
'. $item['title'] .' ';
$output .= '
Rank: '. $extra['rank'] .' File Size: '
.round($extra['file_size']/1024, 1) .' KB
', $results));
}
}
/**
* The SWISH-E indexer. Calls the swish-e indexer commandline via exec.
* Returns the standard output of swish-e.
*/
function _swish_do_index(){
//build an_on_the_fly config file for the swish indexer in temp folder
//and do the swish-e index with this config file.
//when done. remove the config file.
$getcwd = getcwd();
$tmpfname = tempnam(variable_get('file_directory_temp', '/tmp'), 'swish');
$handle = fopen($tmpfname, 'w');
//Include our site-wide configuration settings:
fwrite($handle, "IncludeConfigFile $getcwd\\modules\\swish\\conf\\common.conf\n");
//Index the only this directory
$index_dir = "$getcwd\\".variable_get('file_directory_path', 'files');
fwrite($handle,"IndexDir $index_dir\n");
//Ignore these words. could be put in the settings in the future.
fwrite($handle, "IgnoreWords file: $getcwd\\modules\\swish\\conf\\stopwords\\english.txt\n");
//Dont index these directories/files. could be put in the settings in the future.
fwrite($handle, "FileRules pathname contains stopwords CVS database images includes misc modules scripts themes tmp uploadimage\n");
// And don't index php modules inc, etc. could be put in the settings in the future.
fwrite($handle, "FileRules filename contains .php .inc .module .sql index.\n");
//based on the admin settings of swish, include and index the ff document types.
//the filter programs option for now are for catdoc, xls2csv, pdftotext, unrtf
$indexonly = 'IndexOnly .txt';
$filefilter = "\n";
if (($filter = variable_get('catdoc_path', '/usr/local/bin/catdoc')) ){
$indexonly .= ' .doc';
//FileFilter .doc /usr/local/bin/catdoc "-s8859-1 -d8859-1 %p"
$filefilter .= 'FileFilter .doc '.$filter. ' "-s8859-1 -d8859-1 %p"'. "\n";
}
if ( ($filter = variable_get('xls2csv_path', '/usr/local/bin/xls2csv')) ){
$indexonly .= ' .xls';
$filefilter .= 'FileFilter .xls '.$filter.' "-fw %p"'. "\n";
}
if ( ($filter = variable_get('ppthtml_path', '/usr/local/bin/ppthtml')) ){
$indexonly .= ' .ppt';
$filefilter .= 'FileFilter .ppt '.$filter.' "%p"'. "\n";
}
if ( ($filter = variable_get('pdftotext_path', '/usr/local/bin/pdftotext')) ){
$indexonly .= ' .pdf';
$filefilter .= 'FileFilter .pdf '.$filter.' "%p -"'. "\n";
}
if ( ($filter = variable_get("unrtf_path","/usr/bin/unrtf")) ) {
$indexonly .= ' .rtf';
$filefilter .= 'FileFilter .rtf '.$filter.' "%p"'. "\n";
}
$indexonly .= "\n";
$filefilter .= "\n";
fwrite ($handle, $indexonly.$filefilter);
//assemble the swish-e index command, swish-e -c conf/myconfig.conf -f myindex
$swish_indx_cmd = variable_get('swish_path', '/usr/local/bin/swish-e');
$swish_indx_cmd .= " -c $tmpfname ";
//save the index file in files/my_swish_index
$swish_indx_cmd .= " -f $index_dir\\my_swish_index";
//ok to make sure escape this shell command
exec ($swish_indx_cmd, $results, $rv);
//although there's a tweak on how to make swish-e to do incremental indexing,
//it's complicated. we'll just rely on swish-e fast indexing feature!
fclose($handle);
//delete the temp conf file
unlink($tmpfname);
return $results;
}
/**
* Check the various paths needed by the swish-e engine before performing any swish operation
*/
function swish_validate_paths(){
$file_paths = array(
'swish_path' => 'SWISH-E PATH',
'catdoc_path' => 'catdoc, M$ Word filter',
'xls2csv_path' => 'xls2csv, M$ Excel filter',
'ppthtml_path' => 'ppthtml, M$ Powerpoint filter',
'pdftotext_path' => 'pdftotext, PDF filter',
'unrtf_path' => 'unrtf, RTF filter'
);
$drupal_files_folder = array(
'file_directory_path' => 'Files ',
'file_directory_temp' => 'Temp'
);
$errors = false;
foreach ($file_paths as $name => $label){
$value = variable_get($name, '');
if ($value && !is_executable($value) ){
form_set_error($name, "$label does not exist or not executable.");
$errors = true;
}
}
foreach ($drupal_files_folder as $name => $label) {
if (!is_writable(variable_get($name, ''))) {
drupal_set_message("$label folder does not exist or not writable. Fixed it ". l('here.', 'admin/settings'), 'error');
$errors = true;
}
}
return $errors;
}
?>