function lit_fichier_rss($url_fichier)
{
global $opac_curl_available, $pmb_curl_timeout;
$res = "";
if ($opac_curl_available) {
$timeout = $pmb_curl_timeout * 1 ? $pmb_curl_timeout * 1 : 5;
$ch = curl_init($url_fichier);
curl_setopt($ch, CURLOPT_HEADER, false);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, $timeout);
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
configurer_proxy_curl($ch, $url_fichier);
$res = curl_exec($ch);
curl_close($ch);
} else {
$fp = fopen($url_fichier, "r");
if ($fp) {
while (!feof($fp)) {
$res .= fread($fp, 2048);
}
fclose($fp);
}
}
return $res;
}
function lit_fichier_rss($url_fichier)
{
global $opac_curl_available;
$res = "";
if ($opac_curl_available) {
$ch = curl_init($url_fichier);
curl_setopt($ch, CURLOPT_HEADER, false);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
configurer_proxy_curl($ch);
$res = curl_exec($ch);
curl_close($ch);
} else {
$fp = fopen($url_fichier, "r");
if ($fp) {
while (!feof($fp)) {
$res .= fread($fp, 2048);
}
fclose($fp);
}
}
return $res;
}
function search($source_id, $query, $search_id)
{
global $base_path, $charset;
$params = $this->get_source_params($source_id);
$this->fetch_global_properties();
if ($params["PARAMETERS"]) {
//Affichage du formulaire avec $params["PARAMETERS"]
$vars = unserialize($params["PARAMETERS"]);
foreach ($vars as $key => $val) {
global ${$key};
${$key} = $val;
}
}
if (!isset($url)) {
$this->error_message = $this->msg["nasa_ads_unconfigured"];
$this->error = 1;
return;
}
$boolsearch = "";
foreach ($query as $element) {
$boolsearch .= implode(" ", $element->values) . " ";
}
$boolsearch = str_replace(" ", "+", $boolsearch);
$addr = $url . "/cgi-bin/basic_connect?version=1&qsearch=" . rawurlencode($boolsearch);
$ch = curl_init();
// configuration des options CURL
curl_setopt($ch, CURLOPT_URL, $addr);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
configurer_proxy_curl($ch, $addr);
$cexec = curl_exec($ch);
if (!$cexec) {
$this->error = true;
$this->error_message = "Can't get Lob answer : " . curl_error($ch);
} else {
if (strtoupper($charset) != "UTF-8") {
$cexec = utf8_decode($cexec);
}
$sep_notices = "<tr><td colspan=6><HR></td></tr>";
$query = get_field_betwen_2sep($cexec, $sep_notices, "</table>");
$notices_html = explode($sep_notices, $query);
foreach ($notices_html as $notice_html) {
$data_notice = array();
$tab_tr = explode("</tr>", $notice_html);
$tab_td_0 = explode("</td>", $tab_tr[0]);
$tab_td_1 = explode("</td>", $tab_tr[1]);
$id = get_field_betwen_2sep($notice_html, "value=\"", "\"");
$data_notice["id"] = $id;
$titre = get_field_from_sep($tab_td_1[3], ">");
$authors = get_field_from_sep($tab_td_1[1], ">");
$date = get_field_from_sep($tab_td_0[4], ">");
$links = explode("</a>", $tab_td_0[5]);
foreach ($links as $link) {
$doc_type = substr($link, strlen($link) - 1);
$doc_link = get_field_betwen_2sep($link, "href=\"", "\"");
if ($doc_link) {
$data_notice["doc_links"][$doc_type]["link"] = $doc_link;
$data_notice["doc_links"][$doc_type]["label"] = $this->msg["nasa_ads_doc_" . strtolower($doc_type)];
}
}
$url_notice = "http://adsabs.harvard.edu/abs/{$id}";
$data_notice["Url_notice"] = $url_notice;
$ch_notice = curl_init();
// configuration des options CURL
curl_setopt($ch_notice, CURLOPT_URL, $url_notice);
curl_setopt($ch_notice, CURLOPT_RETURNTRANSFER, true);
configurer_proxy_curl($ch_notice, $url_notice);
$cexec_notice = curl_exec($ch_notice);
if (!$cexec_notice) {
$this->error = true;
$this->error_message = "Can't get Lob answer : " . curl_error($ch_notice);
} else {
if (strtoupper($charset) != "UTF-8") {
$cexec_notice = utf8_decode($cexec_notice);
}
$cexec_notice = get_field_betwen_2sep($cexec_notice, "Translate This Page</a></strong>", "<form method");
$data_notice["Title"] = get_field_betwen_2sep($cexec_notice, "Title:</b></td><td><br></td><td align=\"left\" valign=\"top\">", "</td></tr>", 1);
$Authors = get_field_betwen_2sep($cexec_notice, "Authors:</b></td><td><br></td><td align=\"left\" valign=\"top\">", "</td></tr>");
$aut_list = explode("</a>", $Authors);
foreach ($aut_list as $aut) {
$autor = get_field_from_sep($aut, ">", 1);
if ($autor) {
$data_notice["Authors"][] = $autor;
}
}
$data_notice["Affiliation"] = get_field_betwen_2sep($cexec_notice, "Affiliation:</b></td><td><br></td><td align=\"left\" valign=\"top\">", "</td></tr>", 1);
$data_notice["Publication"] = get_field_betwen_2sep($cexec_notice, "Publication:</b></td><td><br></td><td align=\"left\" valign=\"top\">", "</td></tr>");
$data_notice["Publication_Date"] = get_field_betwen_2sep($cexec_notice, "Publication Date:</b></td><td><br></td><td align=\"left\" valign=\"top\">", "</td></tr>");
$Origin = get_field_betwen_2sep($cexec_notice, "Origin:</b></td><td><br></td><td align=\"left\" valign=\"top\">", "</td></tr>", 1);
$data_notice["Origin"] = get_field_betwen_2sep($Origin, "\">", "</a>");
$data_notice["Keywords"] = get_field_betwen_2sep($cexec_notice, "Keywords:</b></td><td><br></td><td align=\"left\" valign=\"top\">", "</td></tr>");
$DOI = get_field_betwen_2sep($cexec_notice, "DOI:</b></td><td><br></td><td align=\"left\" valign=\"top\">", "</td></tr>");
$data_notice["DOI"] = get_field_betwen_2sep($DOI, "\">", "</a>", 1);
$Bibliographic_Code = get_field_betwen_2sep($cexec_notice, "Bibliographic Code:</b></td><td><br></td><td align=\"left\" valign=\"top\">", "</td></tr>");
$data_notice["Bibliographic_Code"] = get_field_betwen_2sep($Bibliographic_Code, "href=\"", "\"");
$data_notice["Abstract"] = get_field_betwen_2sep($cexec_notice, "Abstract</h3>", "<hr>", 1, "<p><sup><sub>");
$order = array("\r\n", "\n", "\r");
$data_notice["Abstract"] = str_replace($order, " ", $data_notice["Abstract"]);
// print"<pre>";print_r ($data_notice);print"</pre>";
$this->rec_record($this->notice_2_uni($data_notice), $source_id, $search_id);
if ($nb++ > 20) {
break;
}
}
curl_close($ch_notice);
}
}
curl_close($ch);
}
function get_data($url)
{
//Remise à zéro des erreurs
$this->error = false;
$this->error_message = "";
//Initialisation de la ressource
$ch = curl_init();
// configuration des options CURL
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_WRITEFUNCTION, array(&$this, "parse_xml"));
curl_setopt($ch, CURLOPT_HEADERFUNCTION, array(&$this, "verif_header"));
curl_setopt($ch, CURLOPT_HEADER, 0);
curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
if ($this->time_out) {
curl_setopt($ch, CURLOPT_TIMEOUT, $this->time_out);
}
//Réinitialisation du "retry_after"
$this->retry_after = "";
//Explosion des arguments de la requète pour ceux qui ne respectent pas la norme !!
$query = substr($url, strpos($url, "?") + 1);
$query = explode("&", $query);
for ($i = 0; $i < count($query); $i++) {
if (strpos($query[$i], "operation") !== false) {
$operation = substr($query[$i], 9);
break;
}
}
//Initialisation du parser
$this->xml_parser = xml_parser_create("utf-8");
xml_parser_set_option($this->xml_parser, XML_OPTION_CASE_FOLDING, 0);
xml_parser_set_option($this->xml_parser, XML_OPTION_SKIP_WHITE, 1);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
configurer_proxy_curl($ch, $url);
$n_try = 0;
$data = $cexec = curl_exec($ch);
while ($cexec && $this->retry_after && $n_try < 3) {
$n_try++;
sleep((int) $this->retry_after * 1);
$this->retry_after = "";
$data = $cexec = curl_exec($ch);
}
if (!$cexec) {
$this->error = true;
$this->error_message = curl_error($ch);
}
xml_parser_free($this->xml_parser);
$this->xml_parser = "";
curl_close($ch);
if ($this->error) {
$this->error_message .= " - " . $url;
unset($s);
return;
}
$this->data = $data;
}
function analyse_response($url, $rcallback = "")
{
//Remise à zéro des erreurs
$this->error = false;
$this->error_message = "";
//remise à zero des enregistrements
if ($url != $this->next_request) {
$this->records = array();
}
$this->next_request = "";
$this->rtoken = "";
//Initialisation de la ressource
$this->remainder = '';
$ch = curl_init();
// configuration des options CURL
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_WRITEFUNCTION, array(&$this, "parse_xml"));
curl_setopt($ch, CURLOPT_HEADERFUNCTION, array(&$this, "verif_header"));
curl_setopt($ch, CURLOPT_HEADER, 0);
curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
if ($this->time_out) {
curl_setopt($ch, CURLOPT_TIMEOUT, $this->time_out);
}
//Réinitialisation du "retry_after"
$this->retry_after = "";
configurer_proxy_curl($ch, $url);
//Explosion des arguments de la requête pour ceux qui ne respectent pas la norme !!
$query = substr($url, strpos($url, "?") + 1);
$query = explode("&", $query);
for ($i = 0; $i < count($query); $i++) {
if (strpos($query[$i], "verb") !== false) {
$verb = substr($query[$i], 5);
break;
}
}
//Initialisation de l'environnement d'état du parser
$s = new oai_parser($rcallback, $this->charset);
//Si le verb est affecté, on prérempli histoire d'aider un peu... :-)
if ($verb) {
$s->verb = $verb;
}
//Initialisation du parser
$this->xml_parser = xml_parser_create("utf-8");
xml_set_object($this->xml_parser, $s);
xml_parser_set_option($this->xml_parser, XML_OPTION_CASE_FOLDING, 0);
xml_parser_set_option($this->xml_parser, XML_OPTION_SKIP_WHITE, 1);
xml_set_element_handler($this->xml_parser, "oai_startElement", "oai_endElement");
xml_set_character_data_handler($this->xml_parser, "oai_charElement");
$n_try = 0;
$cexec = curl_exec($ch);
while ($cexec && $this->retry_after && $n_try < 3) {
$n_try++;
sleep((int) $this->retry_after * 1);
$this->retry_after = "";
$cexec = curl_exec($ch);
}
if (!$cexec) {
$this->error = true;
$this->error_message = curl_error($ch);
}
xml_parser_free($this->xml_parser);
$this->xml_parser = "";
curl_close($ch);
if ($this->error) {
$this->error_message .= " - " . $url;
unset($s);
return;
}
//Affectation des éléments de réponse
if (stripos($this->charset, 'iso-8859-1') !== false) {
$c = true;
} else {
$c = false;
}
//Test de l'url base
if ($this->clean_base_url) {
$p = strpos($s->tree[1][1]["CHAR"], "?");
if ($p !== false) {
$s->tree[1][1]["CHAR"] = substr($s->tree[1][1]["CHAR"], 0, $p);
}
}
$this->response_date = $c ? utf8_decode($s->tree[1][0]["CHAR"]) : $s->tree[1][0]["CHAR"];
$this->url_base = $c ? utf8_decode($s->tree[1][1]["CHAR"]) : $s->tree[1][1]["CHAR"];
$this->request["URL_BASE"] = $c ? utf8_decode($s->tree[1][1]["CHAR"]) : $s->tree[1][1]["CHAR"];
foreach ($s->tree[1][1]["ATTRIB"] as $key => $val) {
if ($key != "resumptionToken") {
$this->request["ATTRIBS"][$key] = $c ? utf8_decode($val) : $val;
}
}
$this->verb = $c ? utf8_decode($s->tree[1][1]["ATTRIB"]["verb"]) : $s->tree[1][1]["ATTRIB"]["verb"];
$this->rtoken = $s->rtoken;
if ($s->tree[1][2]["NAME"] == "error") {
$this->error = true;
$this->error_message = "OAI Error, the server tell : " . $s->tree[1][2]["ATTRIB"]["code"] . " : " . $s->tree[1][2]["CHAR"];
$this->error_oai_code = $s->tree[1][2]["ATTRIB"]["code"];
}
//Si c'est la requête identify
if ($this->verb == "Identify") {
$this->records[0] = $c ? utf8_decode($s->cur_elt) : $s->cur_elt;
} else {
if (!$rcallback) {
for ($i = 0; $i < count($s->records); $i++) {
$this->records[] = $c ? utf8_decode($s->records[$i]) : $s->records[$i];
}
}
}
//Si on a un resumptionToken
if (is_array($this->rtoken) && $this->rtoken["token"]) {
$t_nr = explode('?', $this->request['URL_BASE']);
$this->next_request = $t_nr[0] . "?verb=" . $s->verb . "&resumptionToken=" . rawurlencode($this->rtoken["token"]);
}
//Supression de l'environnement d'état !
unset($s);
}
function maj_entrepot($source_id, $callback_progress = "", $recover = false, $recover_env = "")
{
global $base_path, $charset;
$this->n_recu = 0;
$this->callback_progress = $callback_progress;
$params = $this->get_source_params($source_id);
$this->fetch_global_properties();
if ($params["PARAMETERS"]) {
//Affichage du formulaire avec $params["PARAMETERS"]
$vars = unserialize($params["PARAMETERS"]);
foreach ($vars as $key => $val) {
global ${$key};
${$key} = $val;
}
}
if (!isset($url)) {
$this->error_message = $this->msg["toutapprendre_unconfigured"];
$this->error = 1;
return;
}
//Recherche de la dernière date...
$requete = "select unix_timestamp(max(date_import)) from entrepot_source_" . $source_id . " where 1;";
$resultat = pmb_mysql_query($requete);
if (pmb_mysql_num_rows($resultat)) {
$last_date = pmb_mysql_result($resultat, 0, 0);
if ($last_date) {
$last_date += 3600 * 24;
}
}
$ch = curl_init();
$addr = $url;
// configuration des options CURL
curl_setopt($ch, CURLOPT_URL, $addr);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
configurer_proxy_curl($ch, $addr);
$xml = curl_exec($ch);
if ($charset == 'utf-8') {
$xml = preg_replace('/[\\x00-\\x08\\x10\\x0B\\x0C\\x0E-\\x19\\x7F]' . '|[\\x00-\\x7F][\\x80-\\xBF]+' . '|([\\xC0\\xC1]|[\\xF0-\\xFF])[\\x80-\\xBF]*' . '|[\\xC2-\\xDF]((?![\\x80-\\xBF])|[\\x80-\\xBF]{2,})' . '|[\\xE0-\\xEF](([\\x80-\\xBF](?![\\x80-\\xBF]))|(?![\\x80-\\xBF]{2})|[\\x80-\\xBF]{3,})/S', '?', $xml);
}
@ini_set("zend.ze1_compatibility_mode", "0");
$this->dom = new DomDocument();
$this->dom->encoding = $charset;
if (!@$this->dom->loadXML($xml)) {
return 0;
}
$cours = $this->dom->getElementsByTagName('cours');
foreach ($cours as $cour) {
$data_notice = array();
if ($cour->childNodes->length) {
foreach ($cour->childNodes as $i) {
if ($charset != 'utf-8') {
$val = utf8_decode($i->nodeValue);
} else {
$val = $i->nodeValue;
}
$data_notice[$i->nodeName] = $val;
}
}
$this->rec_record($this->notice_2_uni($data_notice), $source_id);
}
curl_close($ch);
return $this->n_recu;
}
function maj_entrepot($source_id, $callback_progress = "", $recover = false, $recover_env = "")
{
global $base_path, $charset;
$this->n_recu = 0;
$this->callback_progress = $callback_progress;
$params = $this->get_source_params($source_id);
$this->fetch_global_properties();
if ($params["PARAMETERS"]) {
//Affichage du formulaire avec $params["PARAMETERS"]
$vars = unserialize($params["PARAMETERS"]);
foreach ($vars as $key => $val) {
global ${$key};
${$key} = $val;
}
}
if (!isset($url)) {
$this->error_message = $this->msg["emploi_unconfigured"];
$this->error = 1;
return;
}
//Recherche de la derni�re date...
$requete = "select unix_timestamp(max(date_import)) from entrepot_source_" . $source_id . " where 1;";
$resultat = mysql_query($requete);
if (mysql_num_rows($resultat)) {
$last_date = mysql_result($resultat, 0, 0);
if ($last_date) {
$last_date += 3600 * 24;
}
}
$ch = curl_init();
//$addr="http://travail-emploi.gouv.fr/etudes-recherche-statistiques-de,76/etudes-et-recherche,77/publications-dares,98/dares-analyses-dares-indicateurs,102/";
$addr = $url;
// configuration des options CURL
curl_setopt($ch, CURLOPT_URL, $addr);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
configurer_proxy_curl($ch);
$html = curl_exec($ch);
if (!$html) {
$sortir = 1;
} else {
// if (strtoupper($charset)!="UTF-8") $html=utf8_decode($html);
$notice_list = $this->get_field_betwen_2sep($html, "<h2 class=\"smaller\">", "</ul>");
$notice_list = $this->get_field_betwen_2sep($notice_list, "<ul>", "<ul>");
$sep_notices = "</li>\n";
$notices_html = explode($sep_notices, $notice_list);
$nb = 0;
//print printr($notices_html);
foreach ($notices_html as $notice_html) {
$data_notice = array();
$link = $this->get_field_betwen_2sep($notice_html, "<a href=\"", "\">");
$link = "http://travail-emploi.gouv.fr/" . $link;
curl_setopt($ch, CURLOPT_URL, $link);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
configurer_proxy_curl($ch);
$html_notice_complete = curl_exec($ch);
// http://travail-emploi.gouv.fr/IMG/pdf/2012-035.pdf
if (!$html_notice_complete) {
continue;
}
$data_notice["Url_notice"] = $link;
$html_notice_complete = $this->get_field_from_sep($html_notice_complete, "<div class=\"gris clearfix\">");
$data_notice["Publication_Date"] = $this->get_field_betwen_2sep($html_notice_complete, "<span class=\"date\">", "</span>");
$data_notice["Title"] = $this->get_field_betwen_2sep($html_notice_complete, "<h1>", "</h1>");
$data_notice["Abstract"] = $this->get_field_betwen_2sep($html_notice_complete, "<div class=\"texteencadre-spip spip\">", "<div class=\"listdoc\">", 1, 1);
$data_notice["Abstract"] = str_replace("’", "'", $data_notice["Abstract"]);
$zone_docnum = $this->get_field_from_sep($html_notice_complete, " spip_documents spip_lien_ok\">");
$link_docnum = $this->get_field_betwen_2sep($zone_docnum, "<a href=\"", "\" ");
if ($link_docnum) {
$data_notice["doc_links"][0]["link"] = "http://travail-emploi.gouv.fr/" . $link_docnum;
$data_notice["id"] = $this->get_field_betwen_2sep($link_docnum, "IMG/pdf/", ".pdf");
}
// printr ($data_notice);
if ($this->rec_record($this->notice_2_uni($data_notice), $source_id)) {
// notice d�j� en entrepos, on ne va pas chercher les suivantes
break;
}
}
}
curl_close($ch);
return $this->n_recu;
}
function search($source_id, $query, $search_id)
{
global $charset;
global $opac_curl_proxy;
global $base_path;
$this->error = false;
$this->error_message = "";
$params = $this->get_source_params($source_id);
$this->fetch_global_properties();
if ($params["PARAMETERS"]) {
//Affichage du formulaire avec $params["PARAMETERS"]
$vars = unserialize($params["PARAMETERS"]);
foreach ($vars as $key => $val) {
global ${$key};
${$key} = $val;
}
}
if (!$max_return) {
$max_return = 100;
}
//Construction de la requête
$boolsearch = "";
for ($i = 0; $i < count($query); $i++) {
$term = $query[$i];
$chaine = "";
switch ($term->ufield) {
case "200\$a":
$chaine = "(TI=" . str_replace("*", "%", $term->values[0]) . " or ST=" . str_replace("*", "%", $term->values[0]) . ")";
break;
case "010\$a":
$chaine = "IS=" . str_replace("*", "%", $term->values[0]);
break;
case "7XX":
$chaine = "AU=" . str_replace("*", "%", $term->values[0]);
break;
case "210\$c":
$chaine = "PU=" . str_replace("*", "%", $term->values[0]);
break;
case "210\$d":
$chaine = "PY=" . str_replace("*", "%", $term->values[0]);
break;
case "300":
case "327":
case "330":
case "3XX":
$chaine = "ME=" . str_replace("*", "%", $term->values[0]);
break;
case "60X":
$chaine = "FT=" . str_replace("*", "%", $term->values[0]);
break;
case "XXX":
$chaine = "(TI=" . str_replace("*", "%", $term->values[0]) . " or ST=" . str_replace("*", "%", $term->values[0]) . " or AU=" . str_replace("*", "%", $term->values[0]) . " or FT=" . str_replace("*", "%", $term->values[0]) . " or ME=" . str_replace("*", "%", $term->values[0]) . " or IS=" . str_replace("*", "%", $term->values[0]) . " or PU=" . str_replace("*", "%", $term->values[0]) . ")";
break;
}
if ($chaine != "" && $i != 0) {
switch ($term->inter) {
case "and":
$boolsearch .= " and ";
break;
case "or":
$boolsearch .= " or ";
break;
case "ex":
$boolsearch .= " and not ";
break;
}
}
$boolsearch .= $chaine;
}
$boolthemes = "";
if (count($themes)) {
for ($i = 0; $i < count($themes); $i++) {
if ($i != 0) {
$boolthemes .= " or ";
}
$boolthemes .= "FT=" . $themes[$i];
}
$boolthemes = "(" . $boolthemes . ")";
}
if ($boolthemes) {
$boolsearch = "(" . $boolsearch . ") and " . $boolthemes;
}
$nb_pages = floor($max_return / 40);
$stop = false;
for ($i = 0; $i < $nb_pages; $i++) {
$addr = "http://sun2.lehmanns.de/cgi-bin/work/xmlboolsearch?mode=xmlboolsearch&titcount={$max_return}&titnext=" . $i * 40 . "&boolstring=" . rawurlencode($boolsearch);
//Initialisation de la ressource
$ch = curl_init();
// configuration des options CURL
curl_setopt($ch, CURLOPT_URL, $addr);
if ($params["TIMEOUT"]) {
curl_setopt($ch, CURLOPT_TIMEOUT, (int) $params["TIMEOUT"]);
}
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
configurer_proxy_curl($ch);
$cexec = curl_exec($ch);
if (!$cexec) {
$this->error = true;
$this->error_message = "Can't get Lob answer : " . curl_error($ch);
break;
} else {
if (strpos($cexec, "<LBook>") !== false) {
$unixml = $this->apply_xsl_to_xml($cexec, file_get_contents($base_path . "/admin/connecteurs/in/lehmanns/xslt/lehmanns2uni.xsl"));
$this->rec_records($unixml, $source_id, $search_id);
} else {
$stop = true;
}
}
curl_close($ch);
if ($stop) {
break;
}
}
}
function maj_entrepot($source_id, $callback_progress = "", $recover = false, $recover_env = "")
{
global $base_path, $charset;
$this->n_recu = 0;
$this->callback_progress = $callback_progress;
$params = $this->get_source_params($source_id);
$this->fetch_global_properties();
if ($params["PARAMETERS"]) {
//Affichage du formulaire avec $params["PARAMETERS"]
$vars = unserialize($params["PARAMETERS"]);
foreach ($vars as $key => $val) {
global ${$key};
${$key} = $val;
}
}
if (!isset($url)) {
$this->error_message = $this->msg["oecd_unconfigured"];
$this->error = 1;
return;
}
//Recherche de la derni�re date...
$requete = "select unix_timestamp(max(date_import)) from entrepot_source_" . $source_id . " where 1;";
$resultat = mysql_query($requete);
if (mysql_num_rows($resultat)) {
$last_date = mysql_result($resultat, 0, 0);
if ($last_date) {
$last_date += 3600 * 24;
}
}
$ch = curl_init();
// http://www.oecd-ilibrary.org/fr/emploi/livres/2012
// http://www.oecd-ilibrary.org/fr/questionssociales/livres/2012
$addr = $url;
$sortir = 0;
// pour sortir du while!
$page = 0;
do {
// configuration des options CURL
curl_setopt($ch, CURLOPT_URL, $addr);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
configurer_proxy_curl($ch);
$html = curl_exec($ch);
if (!$html) {
$sortir = 1;
} else {
if (strtoupper($charset) != "UTF-8") {
$html = utf8_decode($html);
}
$notice_list = $this->get_field_betwen_2sep($html, "</thead>\n<tbody>", "</tbody>\n</table>");
$sep_notices = "</tr>\n<tr>";
$notices_html = explode($sep_notices, $notice_list);
//print printr($notices_html);
foreach ($notices_html as $notice_html) {
$data_notice = array();
$type = $this->get_field_betwen_2sep($notice_html, "\"type nowrap box3\"><strong>", " </strong>");
if ($type == "Livre") {
$link = $this->get_field_betwen_2sep($notice_html, "<strong>\n<a href=\"", "\" title=\"");
$link = "http://www.oecd-ilibrary.org" . $link;
$data_notice["Url_notice"] = $link;
$zone_title = $this->get_field_betwen_2sep($notice_html, "</ul>\n<strong>\n", "</strong>");
$data_notice["Title"] = $this->get_field_betwen_2sep($zone_title, "rel=\"\"><span>", "</span>", 1);
$data_notice["Abstract"] = $this->get_field_betwen_2sep($notice_html, "class=\"abstract \"><span>", "</span>");
$date_zone = $this->get_field_betwen_2sep($notice_html, "nowrap box2\">\n", " ");
$data_notice["Publication_Date"] = $this->get_field_betwen_2sep($date_zone, "nowrap box2\">\n", "\n");
$data_notice["numberofpages"] = $this->get_field_betwen_2sep($notice_html, " Pages: ", "\n");
$data_notice["Authors"][0] = $this->get_field_betwen_2sep($notice_html, "</a></strong><br />\n", ", Pages: ");
$zone_docnum = $this->get_field_betwen_2sep($notice_html, "<li class=\"last\">\n", "</li>");
$link_docnum = $this->get_field_betwen_2sep($zone_docnum, "<a href=\"", "\" ");
if ($link_docnum) {
$data_notice["doc_links"][0]["link"] = "http://www.oecd-ilibrary.org" . $link_docnum;
}
$data_notice["id"] = $this->get_field_betwen_2sep($data_notice["Url_notice"], "_", ";jsessionid");
//print $notice_html;
} else {
continue;
}
// printr ($data_notice);
if ($this->rec_record($this->notice_2_uni($data_notice), $source_id)) {
// notice d�j� en entrepos, on ne va pas chercher les suivantes
break;
}
/* if($nb++>20){
$sortir=1; break;
}*/
}
}
if (!$sortir) {
$sortir = 1;
$next_page_link = "";
$page_zone = $this->get_field_betwen_2sep($html, "bobby-inline pager", "</ul>");
if ($page_zone) {
$next_page_link = $this->get_field_betwen_2sep($page_zone, "<li>|\n<a href=\"", "\" title=\"next page\"");
if ($next_page_link) {
$addr = "http://www.oecd-ilibrary.org" . $next_page_link;
///print $adr;
$sortir = 0;
}
}
}
if ($page++ > 20) {
$sortir = 1;
}
} while (!$sortir);
curl_close($ch);
return $this->n_recu;
}
function request($method, $url, $vars = array())
{
$this->handle = curl_init();
# Set some default CURL options
if ($this->timeout) {
curl_setopt($this->handle, CURLOPT_CONNECTTIMEOUT, $this->timeout);
curl_setopt($this->handle, CURLOPT_TIMEOUT, $this->timeout);
}
curl_setopt($this->handle, CURLOPT_COOKIEFILE, $this->cookie_file);
curl_setopt($this->handle, CURLOPT_COOKIEJAR, $this->cookie_file);
@curl_setopt($this->handle, CURLOPT_FOLLOWLOCATION, true);
curl_setopt($this->handle, CURLOPT_HEADER, true);
curl_setopt($this->handle, CURLOPT_POSTFIELDS, is_array($vars) ? http_build_query($vars, '', '&') : $vars);
curl_setopt($this->handle, CURLOPT_REFERER, $this->referer);
curl_setopt($this->handle, CURLOPT_RETURNTRANSFER, true);
curl_setopt($this->handle, CURLOPT_URL, str_replace(" ", "%20", preg_replace("/#.*\$/", "", $url)));
/*On supprime ce qui suit le # car c'est une ancre pour le navigateur et avec on consière la validation fausse alors qu'elle est bonne
*On remplace les espaces par %20 pour la même raison
*/
curl_setopt($this->handle, CURLOPT_USERAGENT, $this->user_agent);
if ($this->limit) {
curl_setopt($this->handle, CURLOPT_WRITEFUNCTION, array(&$this, 'getBodyOverflow'));
}
if ($this->save_file_name) {
$this->header_detect = 0;
curl_setopt($this->handle, CURLOPT_WRITEFUNCTION, array(&$this, 'saveBodyInFile'));
}
configurer_proxy_curl($this->handle, str_replace(" ", "%20", preg_replace("/#.*\$/", "", $url)));
# Format custom headers for this request and set CURL option
$headers = array();
foreach ($this->headers as $key => $value) {
$headers[] = $key . ': ' . $value;
}
curl_setopt($this->handle, CURLOPT_HTTPHEADER, $headers);
# Determine the request method and set the correct CURL option
switch ($method) {
case 'GET':
curl_setopt($this->handle, CURLOPT_HTTPGET, true);
break;
case 'POST':
curl_setopt($this->handle, CURLOPT_POST, true);
break;
default:
curl_setopt($this->handle, CURLOPT_CUSTOMREQUEST, $method);
}
# Set any custom CURL options
foreach ($this->options as $option => $value) {
curl_setopt($this->handle, constant('CURLOPT_' . str_replace('CURLOPT_', '', strtoupper($option))), $value);
}
$this->body_overflow = "";
$response = curl_exec($this->handle);
if ($this->limit) {
$response = $this->body_overflow;
}
if ($response) {
$response = new CurlResponse($response);
} else {
$this->error = curl_errno($this->handle) . ' - ' . curl_error($this->handle);
}
curl_close($this->handle);
return $response;
}
function fetch_and_record_images($isbn)
{
if (!is_dir($this->image_folder)) {
return;
}
$result = array("thumb" => "", "front" => "", "back" => "");
$ch = curl_init();
curl_setopt($ch, CURLOPT_HEADER, 0);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
$folder = $this->image_folder;
$folder_url = $this->image_folder_url;
if (!file_exists($folder . "/" . $isbn . "_thumb.jpg")) {
$url = str_replace("!!isbn!!", $isbn, $this->image_thumb_url);
curl_setopt($ch, CURLOPT_URL, $url);
configurer_proxy_curl($ch, $url);
$buffer = curl_exec($ch);
if (!curl_error($ch)) {
file_put_contents($folder . "/" . $isbn . "_thumb.jpg", $buffer);
$result['thumb'] = $folder_url . "/" . $isbn . "_thumb.jpg";
}
} else {
$result['thumb'] = $folder_url . "/" . $isbn . "_thumb.jpg";
}
if (!file_exists($folder . "/" . $isbn . "_front.jpg")) {
$url = str_replace("!!isbn!!", $isbn, $this->image_front);
curl_setopt($ch, CURLOPT_URL, $url);
configurer_proxy_curl($ch, $url);
$buffer = curl_exec($ch);
if (!curl_error($ch)) {
file_put_contents($folder . "/" . $isbn . "_front.jpg", $buffer);
$result['front'] = $folder_url . "/" . $isbn . "_front.jpg";
}
} else {
$result['front'] = $folder_url . "/" . $isbn . "_front.jpg";
}
if (!file_exists($folder . "/" . $isbn . "_back.jpg")) {
$url = str_replace("!!isbn!!", $isbn, $this->image_back);
curl_setopt($ch, CURLOPT_URL, $url);
configurer_proxy_curl($ch, $url);
$buffer = curl_exec($ch);
if (!curl_error($ch)) {
file_put_contents($folder . "/" . $isbn . "_back.jpg", $buffer);
$result['back'] = $folder_url . "/" . $isbn . "_back.jpg";
}
} else {
$result['back'] = $folder_url . "/" . $isbn . "_back.jpg";
}
curl_close($ch);
return $result;
}
function search($source_id, $query, $search_id)
{
global $base_path, $charset;
$params = $this->get_source_params($source_id);
$this->fetch_global_properties();
if ($params["PARAMETERS"]) {
//Affichage du formulaire avec $params["PARAMETERS"]
$vars = unserialize($params["PARAMETERS"]);
foreach ($vars as $key => $val) {
global ${$key};
${$key} = $val;
}
}
if (!isset($url)) {
$this->error_message = $this->msg["emploi_unconfigured"];
$this->error = 1;
return;
}
$boolsearch = "";
foreach ($query as $element) {
if ($boolsearch) {
$boolsearch .= " ";
}
$boolsearch .= implode(" ", $element->values);
}
$boolsearch = str_replace(" ", "+", $boolsearch);
//$addr=$url."/search?value21=true&value22=true&discontin=factbooks&value1=".rawurlencode($boolsearch)."&option1=titleAbstract&option18=sort&site=fr&form_name=quick&option21=discontinued&option22=excludeKeyTableEditions&option19=content_type&value19=books";
$ch = curl_init();
//$addr="http://travail-emploi.gouv.fr/etudes-recherche-statistiques-de,76/etudes-et-recherche,77/publications-dares,98/dares-analyses-dares-indicateurs,102/";
$addr = $url;
// configuration des options CURL
curl_setopt($ch, CURLOPT_URL, $addr);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
configurer_proxy_curl($ch);
$html = curl_exec($ch);
if (!$html) {
$sortir = 1;
} else {
// if (strtoupper($charset)!="UTF-8") $html=utf8_decode($html);
// print $html;
$notice_list = $this->get_field_betwen_2sep($html, "<h2 class=\"smaller\">", "</ul>");
$notice_list = $this->get_field_betwen_2sep($notice_list, "<ul>", "<ul>");
$sep_notices = "</li>\n";
$notices_html = explode($sep_notices, $notice_list);
//print printr($notices_html);
foreach ($notices_html as $notice_html) {
$data_notice = array();
$link = $this->get_field_betwen_2sep($notice_html, "<a href=\"", "\">");
$link = "http://travail-emploi.gouv.fr/" . $link;
curl_setopt($ch, CURLOPT_URL, $link);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
configurer_proxy_curl($ch);
$html_notice_complete = curl_exec($ch);
// http://travail-emploi.gouv.fr/IMG/pdf/2012-035.pdf
if (!$html_notice_complete) {
continue;
}
$data_notice["Url_notice"] = $link;
$html_notice_complete = $this->get_field_from_sep($html_notice_complete, "<div class=\"gris clearfix\">");
$data_notice["Publication_Date"] = $this->get_field_betwen_2sep($html_notice_complete, "<span class=\"date\">", "</span>");
$data_notice["Title"] = $this->get_field_betwen_2sep($html_notice_complete, "<h1>", "</h1>");
$data_notice["Abstract"] = $this->get_field_betwen_2sep($html_notice_complete, "<div class=\"texteencadre-spip spip\">", "</div>", 1);
$zone_docnum = $this->get_field_from_sep($html_notice_complete, " spip_documents spip_lien_ok\">");
$link_docnum = $this->get_field_betwen_2sep($zone_docnum, "<a href=\"", "\" ");
if ($link_docnum) {
$data_notice["doc_links"][0]["link"] = "http://travail-emploi.gouv.fr/" . $link_docnum;
$data_notice["id"] = $this->get_field_betwen_2sep($link_docnum, "IMG/pdf/", ".pdf");
}
// printr ($data_notice);
$this->rec_record($this->notice_2_uni($data_notice), $source_id, $search_id);
if ($nb++ > 20) {
break;
}
}
}
curl_close($ch);
}