[github/wallabag/wallabag.git] / inc / poche / pocheCore.php

<?php
/**
 * poche, a read it later open source system
 *
 * @category   poche
 * @author     Nicolas Lœuillet <support@inthepoche.com>
 * @copyright  2013
 * @license    http://www.wtfpl.net/ see COPYING file
 */

function encode_string($string) 
{
    return sha1($string . SALT);
}

function get_external_file($url)
{
    $timeout = 15;
    $useragent = "Mozilla/5.0 (Windows NT 5.1; rv:18.0) Gecko/20100101 Firefox/18.0";

    if  (in_array ('curl', get_loaded_extensions())) {
        # Fetch feed from URL
        $curl = curl_init();
        curl_setopt($curl, CURLOPT_URL, $url);
        curl_setopt($curl, CURLOPT_TIMEOUT, $timeout);
        curl_setopt($curl, CURLOPT_FOLLOWLOCATION, true);
        curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);
        curl_setopt($curl, CURLOPT_HEADER, false);

        # for ssl, do not verified certificate
        curl_setopt($curl, CURLOPT_SSL_VERIFYPEER, FALSE);
        curl_setopt($curl, CURLOPT_AUTOREFERER, TRUE );

        # FeedBurner requires a proper USER-AGENT...
        curl_setopt($curl, CURL_HTTP_VERSION_1_1, true);
        curl_setopt($curl, CURLOPT_ENCODING, "gzip, deflate");
        curl_setopt($curl, CURLOPT_USERAGENT, $useragent);

        $data = curl_exec($curl);
        $httpcode = curl_getinfo($curl, CURLINFO_HTTP_CODE);
        $httpcodeOK = isset($httpcode) and ($httpcode == 200 or $httpcode == 301);
        curl_close($curl);
    } else {
        # create http context and add timeout and user-agent
        $context = stream_context_create(
            array(
                'http' => array(
                    'timeout' => $timeout,
                    'header' => "User-Agent: " . $useragent,
                    'follow_location' => true
                ),
                'ssl' => array(
                    'verify_peer' => false,
                    'allow_self_signed' => true
                )
            )
        );

        # only download page lesser than 4MB
        $data = @file_get_contents($url, false, $context, -1, 4000000); 

        if (isset($http_response_header) and isset($http_response_header[0])) {
            $httpcodeOK = isset($http_response_header) and isset($http_response_header[0]) and ((strpos($http_response_header[0], '200 OK') !== FALSE) or (strpos($http_response_header[0], '301 Moved Permanently') !== FALSE));
        }
    }

    # if response is not empty and response is OK
    if (isset($data) and isset($httpcodeOK) and $httpcodeOK) {

        # take charset of page and get it
        preg_match('#<meta .*charset=.*>#Usi', $data, $meta);

        # if meta tag is found
        if (!empty($meta[0])) {
            preg_match('#charset="?(.*)"#si', $meta[0], $encoding);
            # if charset is found set it otherwise, set it to utf-8
            $html_charset = (!empty($encoding[1])) ? strtolower($encoding[1]) : 'utf-8';
        } else {
            $html_charset = 'utf-8';
            $encoding[1] = '';
        }

        # replace charset of url to charset of page
        $data = str_replace('charset=' . $encoding[1], 'charset=' . $html_charset, $data);

        return $data;
    }
    else {
        return FALSE;
    }
}

function fetch_url_content($url)
{
    $url = base64_decode($url);
    if (pocheTools::isUrl($url)) {
        $url = pocheTools::cleanURL($url);
        $html = Encoding::toUTF8(get_external_file($url));

        # if get_external_file if not able to retrieve HTTPS content, try the same URL with HTTP protocol
        if (!preg_match('!^https?://!i', $url) && (!isset($html) || strlen($html) <= 0)) {
            $url = 'http://' . $url;
            $html = Encoding::toUTF8(get_external_file($url));
        }

        if (function_exists('tidy_parse_string')) {
            $tidy = tidy_parse_string($html, array(), 'UTF8');
            $tidy->cleanRepair();
            $html = $tidy->value;
        }

        $parameters = array();
        if (isset($html) and strlen($html) > 0)
        {
            $readability = new Readability($html, $url);
            $readability->convertLinksToFootnotes = CONVERT_LINKS_FOOTNOTES;
            $readability->revertForcedParagraphElements = REVERT_FORCED_PARAGRAPH_ELEMENTS;

            if($readability->init())
            {
                $content = $readability->articleContent->innerHTML;
                $parameters['title'] = $readability->articleTitle->innerHTML;
                $parameters['content'] = $content;

                return $parameters;
            }
        }
    }
    else {
        #$msg->add('e', _('error during url preparation : the link is not valid'));
        pocheTools::logm($url . ' is not a valid url');
    }

    return FALSE;
}

function get_tpl_file($view)
{
    $tpl_file = 'home.twig';
    switch ($view)
    {
        case 'install':
            $tpl_file = 'install.twig';
            break;
        case 'import';
            $tpl_file = 'import.twig';
            break;
        case 'export':
            $tpl_file = 'export.twig';
            break;
        case 'config':
            $tpl_file = 'config.twig';
            break;
        case 'view':
            $tpl_file = 'view.twig';
            break;
        default:
        break;
    }
    return $tpl_file;
}

function display_view($view, $id = 0)
{
    global $store;

    $tpl_vars = array();

    switch ($view)
    {
        case 'install':
            pocheTools::logm('install mode');
            break;
        case 'import';
            pocheTools::logm('import mode');
            break;
        case 'export':
            $entries = $store->retrieveAll();
            $tpl->assign('export', pocheTools::renderJson($entries));
            $tpl->draw('export');
            pocheTools::logm('export view');
            break;
        case 'config':
            pocheTools::logm('config view');
            break;
        case 'view':
            $entry = $store->retrieveOneById($id);
            if ($entry != NULL) {
                pocheTools::logm('view link #' . $id);
                $content = $entry['content'];
                if (function_exists('tidy_parse_string')) {
                    $tidy = tidy_parse_string($content, array('indent'=>true, 'show-body-only' => true), 'UTF8');
                    $tidy->cleanRepair();
                    $content = $tidy->value;
                }
                $tpl_vars = array(
                    'entry' => $entry,
                    'content' => $content,
                );
            }
            else {
                pocheTools::logm('error in view call : entry is NULL');
            }
            break;
        default: # home view
            $entries = $store->getEntriesByView($view);
            $tpl_vars = array(
                'entries' => $entries,
            );
            break;
    }

    return $tpl_vars;
}

/**
 * Call action (mark as fav, archive, delete, etc.)
 */
function action_to_do($action, $url, $id = 0)
{
    global $store;

    switch ($action)
    {
        case 'add':
            if($parametres_url = fetch_url_content($url)) {
                if ($store->add($url, $parametres_url['title'], $parametres_url['content'])) {
                    pocheTools::logm('add link ' . $url);
                    $last_id = $store->getLastId();
                    if (DOWNLOAD_PICTURES) {
                        $content = filtre_picture($parametres_url['content'], $url, $last_id);
                    }
                    #$msg->add('s', _('the link has been added successfully'));
                }
                else {
                    #$msg->add('e', _('error during insertion : the link wasn\'t added'));
                    pocheTools::logm('error during insertion : the link wasn\'t added');
                }
            }
            else {
                #$msg->add('e', _('error during url preparation : the link wasn\'t added'));
                pocheTools::logm('error during content fetch');
            }
            break;
        case 'delete':
            if ($store->deleteById($id)) {
                if (DOWNLOAD_PICTURES) {
                    remove_directory(ABS_PATH . $id);
                }
                #$msg->add('s', _('the link has been deleted successfully'));
                pocheTools::logm('delete link #' . $id);
            }
            else {
                #$msg->add('e', _('the link wasn\'t deleted'));
                pocheTools::logm('error : can\'t delete link #' . $id);
            }
            break;
        case 'toggle_fav' :
            $store->favoriteById($id);
            pocheTools::logm('mark as favorite link #' . $id);
            break;
        case 'toggle_archive' :
            $store->archiveById($id);
            pocheTools::logm('archive link #' . $id);
            break;
        default:
            break;
    }
}
Commit	Line	Data
a4565e88 NL	1	<?php
	2	/**
	3	* poche, a read it later open source system
	4	*
	5	* @category poche
	6	* @author Nicolas Lœuillet <support@inthepoche.com>
	7	* @copyright 2013
	8	* @license http://www.wtfpl.net/ see COPYING file
	9	*/
	10
	11	function encode_string($string)
	12	{
	13	return sha1($string . SALT);
	14	}
	15
	16	function get_external_file($url)
	17	{
	18	$timeout = 15;
	19	$useragent = "Mozilla/5.0 (Windows NT 5.1; rv:18.0) Gecko/20100101 Firefox/18.0";
	20
	21	if (in_array ('curl', get_loaded_extensions())) {
	22	# Fetch feed from URL
	23	$curl = curl_init();
	24	curl_setopt($curl, CURLOPT_URL, $url);
	25	curl_setopt($curl, CURLOPT_TIMEOUT, $timeout);
	26	curl_setopt($curl, CURLOPT_FOLLOWLOCATION, true);
	27	curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);
	28	curl_setopt($curl, CURLOPT_HEADER, false);
	29
	30	# for ssl, do not verified certificate
	31	curl_setopt($curl, CURLOPT_SSL_VERIFYPEER, FALSE);
	32	curl_setopt($curl, CURLOPT_AUTOREFERER, TRUE );
	33
	34	# FeedBurner requires a proper USER-AGENT...
	35	curl_setopt($curl, CURL_HTTP_VERSION_1_1, true);
	36	curl_setopt($curl, CURLOPT_ENCODING, "gzip, deflate");
	37	curl_setopt($curl, CURLOPT_USERAGENT, $useragent);
	38
	39	$data = curl_exec($curl);
	40	$httpcode = curl_getinfo($curl, CURLINFO_HTTP_CODE);
	41	$httpcodeOK = isset($httpcode) and ($httpcode == 200 or $httpcode == 301);
	42	curl_close($curl);
	43	} else {
	44	# create http context and add timeout and user-agent
	45	$context = stream_context_create(
	46	array(
	47	'http' => array(
	48	'timeout' => $timeout,
	49	'header' => "User-Agent: " . $useragent,
	50	'follow_location' => true
	51	),
	52	'ssl' => array(
	53	'verify_peer' => false,
	54	'allow_self_signed' => true
	55	)
	56	)
	57	);
	58
	59	# only download page lesser than 4MB
	60	$data = @file_get_contents($url, false, $context, -1, 4000000);
	61
	62	if (isset($http_response_header) and isset($http_response_header[0])) {
	63	$httpcodeOK = isset($http_response_header) and isset($http_response_header[0]) and ((strpos($http_response_header[0], '200 OK') !== FALSE) or (strpos($http_response_header[0], '301 Moved Permanently') !== FALSE));
	64	}
65	}
66
67	# if response is not empty and response is OK
68	if (isset($data) and isset($httpcodeOK) and $httpcodeOK) {
69
70	# take charset of page and get it
71	preg_match('#<meta .charset=.>#Usi', $data, $meta);
72
73	# if meta tag is found
74	if (!empty($meta[0])) {
75	preg_match('#charset="?(.*)"#si', $meta[0], $encoding);
76	# if charset is found set it otherwise, set it to utf-8
77	$html_charset = (!empty($encoding[1])) ? strtolower($encoding[1]) : 'utf-8';
78	} else {
79	$html_charset = 'utf-8';
80	$encoding[1] = '';
81	}
82
83	# replace charset of url to charset of page
84	$data = str_replace('charset=' . $encoding[1], 'charset=' . $html_charset, $data);
85
86	return $data;
87	}
88	else {
89	return FALSE;
90	}
91	}
92
93	function fetch_url_content($url)
94	{
95	$url = base64_decode($url);
161395d7 NL	96	if (pocheTools::isUrl($url)) {
161395d7 NL	97	$url = pocheTools::cleanURL($url);
a4565e88 NL	98	$html = Encoding::toUTF8(get_external_file($url));
	99
	100	# if get_external_file if not able to retrieve HTTPS content, try the same URL with HTTP protocol
	101	if (!preg_match('!^https?://!i', $url) && (!isset($html) \|\| strlen($html) <= 0)) {
	102	$url = 'http://' . $url;
	103	$html = Encoding::toUTF8(get_external_file($url));
	104	}
	105
	106	if (function_exists('tidy_parse_string')) {
	107	$tidy = tidy_parse_string($html, array(), 'UTF8');
	108	$tidy->cleanRepair();
	109	$html = $tidy->value;
	110	}
	111
	112	$parameters = array();
	113	if (isset($html) and strlen($html) > 0)
	114	{
	115	$readability = new Readability($html, $url);
	116	$readability->convertLinksToFootnotes = CONVERT_LINKS_FOOTNOTES;
	117	$readability->revertForcedParagraphElements = REVERT_FORCED_PARAGRAPH_ELEMENTS;
	118
	119	if($readability->init())
	120	{
	121	$content = $readability->articleContent->innerHTML;
	122	$parameters['title'] = $readability->articleTitle->innerHTML;
	123	$parameters['content'] = $content;
	124
	125	return $parameters;
	126	}
	127	}
	128	}
	129	else {
	130	#$msg->add('e', _('error during url preparation : the link is not valid'));
161395d7	131	pocheTools::logm($url . ' is not a valid url');
a4565e88 NL	132	}
	133
	134	return FALSE;
	135	}
	136
2b840e0c NL	137	function get_tpl_file($view)
	138	{
	139	$tpl_file = 'home.twig';
	140	switch ($view)
	141	{
	142	case 'install':
	143	$tpl_file = 'install.twig';
	144	break;
	145	case 'import';
	146	$tpl_file = 'import.twig';
	147	break;
	148	case 'export':
	149	$tpl_file = 'export.twig';
	150	break;
	151	case 'config':
	152	$tpl_file = 'config.twig';
	153	break;
	154	case 'view':
	155	$tpl_file = 'view.twig';
	156	break;
	157	default:
	158	break;
	159	}
	160	return $tpl_file;
	161	}
	162
8cbb2a88	163	function display_view($view, $id = 0)
a4565e88	164	{
2b840e0c	165	global $store;
a4565e88	166
8cbb2a88 NL	167	$tpl_vars = array();
8cbb2a88 NL	168
a4565e88 NL	169	switch ($view)
a4565e88 NL	170	{
5ffe5cf5	171	case 'install':
161395d7	172	pocheTools::logm('install mode');
5ffe5cf5 NL	173	break;
5ffe5cf5 NL	174	case 'import';
161395d7	175	pocheTools::logm('import mode');
5ffe5cf5	176	break;
a4565e88 NL	177	case 'export':
a4565e88 NL	178	$entries = $store->retrieveAll();
161395d7	179	$tpl->assign('export', pocheTools::renderJson($entries));
a4565e88	180	$tpl->draw('export');
161395d7	181	pocheTools::logm('export view');
a4565e88 NL	182	break;
a4565e88 NL	183	case 'config':
161395d7	184	pocheTools::logm('config view');
a4565e88 NL	185	break;
	186	case 'view':
	187	$entry = $store->retrieveOneById($id);
a4565e88	188	if ($entry != NULL) {
161395d7	189	pocheTools::logm('view link #' . $id);
a4565e88 NL	190	$content = $entry['content'];
	191	if (function_exists('tidy_parse_string')) {
	192	$tidy = tidy_parse_string($content, array('indent'=>true, 'show-body-only' => true), 'UTF8');
	193	$tidy->cleanRepair();
	194	$content = $tidy->value;
	195	}
3ba5f81b NL	196	$tpl_vars = array(
	197	'entry' => $entry,
	198	'content' => $content,
	199	);
a4565e88 NL	200	}
a4565e88 NL	201	else {
161395d7	202	pocheTools::logm('error in view call : entry is NULL');
a4565e88	203	}
a4565e88 NL	204	break;
	205	default: # home view
	206	$entries = $store->getEntriesByView($view);
8cbb2a88 NL	207	$tpl_vars = array(
	208	'entries' => $entries,
	209	);
a4565e88 NL	210	break;
a4565e88 NL	211	}
8cbb2a88 NL	212
8cbb2a88 NL	213	return $tpl_vars;
a4565e88 NL	214	}
	215
	216	/**
5ffe5cf5	217	* Call action (mark as fav, archive, delete, etc.)
a4565e88 NL	218	*/
	219	function action_to_do($action, $url, $id = 0)
	220	{
5ffe5cf5	221	global $store;
a4565e88 NL	222
	223	switch ($action)
	224	{
	225	case 'add':
	226	if($parametres_url = fetch_url_content($url)) {
	227	if ($store->add($url, $parametres_url['title'], $parametres_url['content'])) {
161395d7	228	pocheTools::logm('add link ' . $url);
a4565e88 NL	229	$last_id = $store->getLastId();
	230	if (DOWNLOAD_PICTURES) {
	231	$content = filtre_picture($parametres_url['content'], $url, $last_id);
	232	}
	233	#$msg->add('s', _('the link has been added successfully'));
	234	}
	235	else {
	236	#$msg->add('e', _('error during insertion : the link wasn\'t added'));
161395d7	237	pocheTools::logm('error during insertion : the link wasn\'t added');
a4565e88 NL	238	}
	239	}
	240	else {
	241	#$msg->add('e', _('error during url preparation : the link wasn\'t added'));
161395d7	242	pocheTools::logm('error during content fetch');
a4565e88 NL	243	}
	244	break;
	245	case 'delete':
	246	if ($store->deleteById($id)) {
	247	if (DOWNLOAD_PICTURES) {
	248	remove_directory(ABS_PATH . $id);
	249	}
	250	#$msg->add('s', _('the link has been deleted successfully'));
161395d7	251	pocheTools::logm('delete link #' . $id);
a4565e88 NL	252	}
	253	else {
	254	#$msg->add('e', _('the link wasn\'t deleted'));
161395d7	255	pocheTools::logm('error : can\'t delete link #' . $id);
a4565e88 NL	256	}
	257	break;
	258	case 'toggle_fav' :
	259	$store->favoriteById($id);
161395d7	260	pocheTools::logm('mark as favorite link #' . $id);
a4565e88 NL	261	break;
	262	case 'toggle_archive' :
	263	$store->archiveById($id);
161395d7	264	pocheTools::logm('archive link #' . $id);
a4565e88 NL	265	break;
	266	default:
	267	break;
	268	}
	269	}