scores/batch
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
batch/1.1/old/getInfosReg.php

600 lines
27 KiB
PHP
Raw Normal View History

New branch with new directories structure
2012-10-16 07:44:31 +00:00
#!/usr/bin/php -c/var/www/batch/config/php_batch_sd.ini
<?php
/**
** Attention : Si ajout d'une nouvelle source modifier la table articles !
** ALTER TABLE `articles` CHANGE `source` `source` ENUM( 'A', 'B', 'D', 'E', 'H' ) NOT NULL DEFAULT 'A';
**/
include_once(FWK_PATH.'common/chiffres.php');
include_once(FWK_PATH.'common/dates.php');
include_once(FWK_PATH.'common/ftp.php');
Update from old server
2013-06-19 08:24:49 +00:00
include_once(FWK_PATH.'mail/sendMail.php');
New branch with new directories structure
2012-10-16 07:44:31 +00:00
$iDb=new WDB('presse');
Update from old server
2013-06-19 08:24:49 +00:00
$nbHugin=$nbActus=$nbDiRel=$nbEchos=$nbBusin=0;
$doHugin=$doActus=$doEchos=$doBusin=$doAll=false;
New branch with new directories structure
2012-10-16 07:44:31 +00:00
Update from old server
2013-06-19 08:24:49 +00:00
$strInfoProg='Usage : '.basename($argv[0]). " [OPTION]
Int<EFBFBD>gration des derniers communiqu<EFBFBD>s boursier.
Sans aucun param<EFBFBD>tre, r<EFBFBD>cup<EFBFBD>re les derniers communiqu<EFBFBD>s.
Sinon:
-v Mode bavard
-h Traite uniquement le diffuseur Hugin
-a idem ActuNews
-e idem Les Echos
-b idem BusinessWire
";
$argv=$_SERVER['argv'];
if ($_SERVER['argc']==1) $doAll=true;
else {
for ($i=1; isset($argv[$i]); $i++) {
if (substr($argv[$i],0,1)=='-') {
switch (substr($argv[$i],1,1)) {
case 'h': $doHugin=true; break;
case 'a': $doActus=true; break;
case 'e': $doEchos=true; break;
case 'b': $doBusin=true; break;
case 'v': $modeDebug=true; break;
case '-': die($strInfoProg); break;
default: die('Option '. $argv[$i] . " inconnue !\n"); break;
}
}
}
}
if (!$doHugin && !$doActus && !$doEchos && !$doBusin) $doAll=true;
$strMessage=date('Y/m/d - H:i:s') ." - DEBUT de la r<>cup<75>ration des flux relatifs aux informations r<>glement<6E>es...".EOL;
/*
if ($doHugin || $doAll) {
echo date('Y/m/d - H:i:s') ." - DEBUT de la r<>cup<75>ration des flux Hugin...".EOL;
$ret=ftp_mget(HUGIN_FTP_URL, HUGIN_FTP_USER, HUGIN_FTP_PASS, '*.xml', HUGIN_LOCAL_DIR, true);
if ($ret===false)
die (date ('Y/m/d - H:i:s')." - ERREUR : R<>cup<75>ration des flux Hugin en FTP incorrecte !".EOL);
else {
$message=date ('Y/m/d - H:i:s')." - FIN de la r<>cup<75>ration des flux Hugin en FTP ($ret fichiers r<>cup<75>r<EFBFBD>s).".EOL;
echo $message;
$strMessage.=$message;
}
}
*/
if ($doActus || $doAll) {
echo date('Y/m/d - H:i:s') ." - DEBUT de la r<>cup<75>ration des flux ActusNews...".EOL;
$ret=ftp_mget(ACTUSNEWS_FTP_URL, ACTUSNEWS_FTP_USER, ACTUSNEWS_FTP_PASS, '*.xml', ACTUSNEWS_LOCAL_DIR, true);
if ($ret===false)
die (date ('Y/m/d - H:i:s')." - ERREUR : R<>cup<75>ration des flux ActusNews en FTP incorrecte !".EOL);
else {
$message=date ('Y/m/d - H:i:s')." - FIN de la r<>cup<75>ration des flux ActusNews en FTP ($ret fichiers r<>cup<75>r<EFBFBD>s).".EOL;
echo $message;
$strMessage.=$message;
}
}
/*
New branch with new directories structure
2012-10-16 07:44:31 +00:00
echo date('Y/m/d - H:i:s') ." - DEBUT de la r<>cup<75>ration des flux Di Release...".EOL;
$ret=ftp_mget(DIRELEASE_FTP_URL, DIRELEASE_FTP_USER, DIRELEASE_FTP_PASS, '*.xml', DIRELEASE_LOCAL_DIR, true);
if ($ret===false)
die (date ('Y/m/d - H:i:s')." - ERREUR : R<>cup<75>ration des flux Di Release en FTP incorrecte !".EOL);
Update from old server
2013-06-19 08:24:49 +00:00
else {
$message=date ('Y/m/d - H:i:s')." - FIN de la r<>cup<75>ration des flux Di Release en FTP ($ret fichiers r<>cup<75>r<EFBFBD>s).".EOL;
echo $message;
$strMessage.=$message;
}*/
New branch with new directories structure
2012-10-16 07:44:31 +00:00
Update from old server
2013-06-19 08:24:49 +00:00
if ($doBusin || $doAll) {
echo date('Y/m/d - H:i:s') ." - DEBUT de la r<>cup<75>ration des flux BusinessWire...".EOL;
$dh = opendir(BUSINESSWIRE_LOCAL_DIR_INCOMING);
$ret=0;
while (false !== ($filename = readdir($dh))) {
if ($filename<>'.' && $filename<>'..' && substr($filename, -4)=='.xml')
if (!file_exists(BUSINESSWIRE_LOCAL_DIR . $filename) && filesize(BUSINESSWIRE_LOCAL_DIR_INCOMING . $filename)>0) {
copy(BUSINESSWIRE_LOCAL_DIR_INCOMING . $filename,BUSINESSWIRE_LOCAL_DIR . $filename);
$ret++;
}
New branch with new directories structure
2012-10-16 07:44:31 +00:00
}
Update from old server
2013-06-19 08:24:49 +00:00
$message=date ('Y/m/d - H:i:s')." - FIN de la r<>cup<75>ration des flux BusinessWire ($ret fichiers copi<70>s).".EOL;
echo $message;
$strMessage.=$message;
New branch with new directories structure
2012-10-16 07:44:31 +00:00
}
Update from old server
2013-06-19 08:24:49 +00:00
if ($doEchos || $doAll) {
echo date('Y/m/d - H:i:s') ." - DEBUT de la r<>cup<75>ration des flux Les Echos...".EOL;
$ret=ftp_mget(LESECHOS_FTP_URL, LESECHOS_FTP_USER, LESECHOS_FTP_PASS, 'syndication/*.xml', LESECHOS_LOCAL_DIR, true);
if ($ret===false)
die (date ('Y/m/d - H:i:s')." - ERREUR : R<>cup<75>ration des flux Les Echos en FTP incorrecte !".EOL);
else {
$message=date ('Y/m/d - H:i:s')." - FIN de la r<>cup<75>ration des flux Les Echos en FTP ($ret fichiers r<>cup<75>r<EFBFBD>s).".EOL;
echo $message;
$strMessage.=$message;
}
}
New branch with new directories structure
2012-10-16 07:44:31 +00:00
/**
** INTEGRATION DES COMMUNIQUES "HUGIN"
**/
Update from old server
2013-06-19 08:24:49 +00:00
if ($doHugin || $doAll) {
echo date('Y/m/d - H:i:s') ." - DEBUT de l'int<6E>gration des flux Hugin...".EOL;
$tabFichier=array();
$dh = opendir(HUGIN_LOCAL_DIR);
while (false !== ($filename = readdir($dh))) {
if ($filename<>'.' && $filename<>'..' && substr($filename, -4)=='.xml')
$tabFichier[] = $filename;
New branch with new directories structure
2012-10-16 07:44:31 +00:00
}
Update from old server
2013-06-19 08:24:49 +00:00
foreach ($tabFichier as $k => $nomFichier) {
$tabTmp=file(HUGIN_LOCAL_DIR.$nomFichier);
if (!$tabTmp) die(date ('Y/m/d - H:i:s')." - ERREUR : Fichier local $nomFichier inexistant !".EOL);
if (stripos($tabTmp[0], 'ISO-8859-1')>0) $encoding='ISO-8859-1';
elseif (stripos($tabTmp[0], 'UTF-8')>0) $encoding='UTF-8';
else die(date ('Y/m/d - H:i:s')." - ERREUR : Encoding ".$tabTmp[0]." non g<>r<EFBFBD> !".EOL);
$dom = @new DomDocument2();
@$dom->load(HUGIN_LOCAL_DIR.$nomFichier);
$pressReleaseId=@$dom->getValueFromTag('PRESS_RELEASE_ID');
$tabInsert=array( 'companyId'=>$dom->getValueFromTag('COMPANY_ID'),
'companyName'=>$dom->getValueFromTag('COMPANY_NAME'),
'companyIsin'=>$dom->getValueFromTag('COMPANY_ISIN'),
'companySiren'=>$dom->getValueFromTag('COMPANY_SIREN'),
'companyRic'=>$dom->getValueFromTag('COMPANY_RIC'),
'companyLogoUrl'=>$dom->getValueFromTag('URL1'),
'companyWebSite'=>$dom->getValueFromTag('URL2'),
'companyProfile'=>$dom->getValueFromTag('URL3'),
'companyAnnualReport'=>$dom->getValueFromTag('URL4'),
'companyInfoActionnaires'=>$dom->getValueFromTag('URL5'),
'pressReleaseId'=>$pressReleaseId,
'pressReleaseDate'=>$dom->getValueFromTag('PRESS_RELEASE_DATE_TIME'),
'pressReleaseTitle'=>$dom->getValueFromTag('PRESS_RELEASE_TITLE'),
'pressReleaseText'=>$dom->getValueFromTag('TEXT_FORMAT'),
'pressReleaseHtml'=>$dom->getValueFromTag('HTML_FORMAT'),
'pressReleaseAttachments'=>$dom->getValueFromTag('ATTACHMENTS_URL1'),
'pressReleaseUrl'=>$dom->getValueFromTag('PRESS_RELEASE_URL_CNG'),
'indexAll'=>$dom->getValueFromTag('INDEX_ALL'),
'indexTheme'=>$dom->getValueFromTag('INDEX_THEME'),
'indexSector'=>$dom->getValueFromTag('INDEX_SECTOR'),
'indexCountry'=>$dom->getValueFromTag('INDEX_COUNTRY'),
'indexLanguage'=>$dom->getValueFromTag('INDEX_LANGUAGE'),
'indexMarketPlace'=>$dom->getValueFromTag('INDEX_MARKET_PLACE'),
'indexQuoteInd'=>$dom->getValueFromTag('INDEX_QUOTE_INDICATOR'),
'source'=>'H',
);
$ret=$iDb->insert('articles', $tabInsert);
if (!$ret && $iDb->getLastErrorNum()<>1062) {
echo date ('Y/m/d - H:i:s')." - ERREUR ". $iDb->getLastError() . EOL;
print_r($tabInsert);
die();
} elseif ($iDb->getLastErrorNum()<>1062) {
echo date ('Y/m/d - H:i:s')." - Communiqu<71> n<>$pressReleaseId enregistr<74> avec succ<63>s.".EOL;
$nbHugin++;
}
unset($dom);
}
$message=date('Y/m/d - H:i:s') ." - FIN de l'int<6E>gration des $nbHugin communiqu<71>s Hugin.".EOL;
echo $message;
$strMessage.=$message;
New branch with new directories structure
2012-10-16 07:44:31 +00:00
}
/**
** INTEGRATION DES COMMUNIQUES "ACTUSNEWS"
**/
Update from old server
2013-06-19 08:24:49 +00:00
if ($doActus || $doAll) {
echo date('Y/m/d - H:i:s') ." - DEBUT de l'int<6E>gration des flux ActusNews...".EOL;
$tabFichier=array();
$dh = opendir(ACTUSNEWS_LOCAL_DIR);
while (false !== ($filename = readdir($dh))) {
if ($filename<>'.' && $filename<>'..' && substr($filename, -4)=='.xml')
$tabFichier[] = $filename;
}
foreach ($tabFichier as $k => $nomFichier) {
$tabTmp=file(ACTUSNEWS_LOCAL_DIR.$nomFichier);
if (!$tabTmp) die(date ('Y/m/d - H:i:s')." - ERREUR : Fichier local $nomFichier inexistant !".EOL);
New branch with new directories structure
2012-10-16 07:44:31 +00:00
Update from old server
2013-06-19 08:24:49 +00:00
if (stripos($tabTmp[0], 'ISO-8859-1')>0) $encoding='ISO-8859-1';
elseif (stripos($tabTmp[0], 'UTF-8')>0) $encoding='UTF-8';
else die(date ('Y/m/d - H:i:s')." - ERREUR : Encoding ".$tabTmp[0]." non g<>r<EFBFBD> !".EOL);
foreach ($tabTmp as $i_ligne => $ligne) {
if (stripos($ligne, '<communique ')!== false) {//<communique
//echo date ('Y/m/d - H:i:s')." - Premi<6D>re ligne du communiqu<71>...".EOL;
if (stripos($ligne, 'langue="FR"')>0) {
//echo date ('Y/m/d - H:i:s')." - Communiqu<71> en fran<61>ais ($ligne).".EOL;
$french=true;
} else {
//echo date ('Y/m/d - H:i:s')." - Langue non int<6E>gr<67>e en base ($ligne) !".EOL;
$french=false;
break;
}
New branch with new directories structure
2012-10-16 07:44:31 +00:00
break;
}
Update from old server
2013-06-19 08:24:49 +00:00
}
if ($french) {
//echo date ('Y/m/d - H:i:s')." - Chargement du communiqu<71> ...".EOL;
$dom = @new DomDocument2();
$dom->load(ACTUSNEWS_LOCAL_DIR.$nomFichier);
$pressReleaseId=$dom->getValueFromTag('id');
$tabInsert=array( //'companyId'=>$dom->getValueFromTag('COMPANY_ID'),
'companyName'=>trim($dom->getValueFromTag('raisonsociale')),
'companyIsin'=>trim($dom->getValueFromTag('codeisin')),
//'companySiren'=>$dom->getValueFromTag('COMPANY_SIREN'),
'companyRic'=>trim($dom->getValueFromTag('code_reuters')),
'companyLogoUrl'=>trim($dom->getValueFromTag('logo')),
'companyWebSite'=>trim($dom->getValueFromTag('site')),
//'companyProfile'=>$dom->getValueFromTag('URL3'),
//'companyAnnualReport'=>$dom->getValueFromTag('URL4'),
'companyInfoActionnaires'=>trim($dom->getValueFromTag('site_investisseur')),
'pressReleaseId'=>$pressReleaseId,
'pressReleaseDate'=>trim($dom->getValueFromTag('miseajour')),
'pressReleaseTitle'=>trim($dom->getValueFromTag('titre')),
'pressReleaseText'=>trim(strtr(html_entity_decode(strip_tags($dom->getValueFromTag('contenu'))),array('&rsquo;'=>'<27>'))),
'pressReleaseHtml'=>trim($dom->getValueFromTag('contenu')),
'pressReleaseAttachments'=>trim($dom->getValueFromTag('pdf')),
//'pressReleaseUrl'=>$dom->getValueFromTag('PRESS_RELEASE_URL_CNG'),
/*'indexAll'=>$dom->getValueFromTag('INDEX_ALL'),
'indexTheme'=>$dom->getValueFromTag('INDEX_THEME'),
'indexCountry'=>$dom->getValueFromTag('INDEX_COUNTRY'),
'indexLanguage'=>$dom->getValueFromTag('INDEX_LANGUAGE'),
*/
'indexQuoteInd'=>trim($dom->getValueFromTag('indice')),
'indexSector'=>trim($dom->getValueFromTag('secteur')),
'indexMarketPlace'=>trim($dom->getValueFromTag('marche_cotation')),
'source'=>'A',
'companyBloomberg'=>trim($dom->getValueFromTag('code_bloomberg')),
'companyMnemo'=>trim($dom->getValueFromTag('mnemo')),
'companyNbTitles'=>trim($dom->getValueFromTag('nb_de_titres')),
'companyFootsie'=>trim($dom->getValueFromTag('footsie')),
);
$ret=$iDb->insert('articles', $tabInsert);
if (!$ret && $iDb->getLastErrorNum()<>1062) {
echo date ('Y/m/d - H:i:s')." - ERREUR ". $iDb->getLastError() . EOL;
print_r($tabInsert);
die();
} elseif ($iDb->getLastErrorNum()<>1062) {
echo date ('Y/m/d - H:i:s')." - Communiqu<71> n<>$pressReleaseId enregistr<74> avec succ<63>s.".EOL;
$nbActus++;
}
unset($dom);
New branch with new directories structure
2012-10-16 07:44:31 +00:00
}
}
Update from old server
2013-06-19 08:24:49 +00:00
$message=date('Y/m/d - H:i:s') ." - FIN de l'int<6E>gration des $nbActus communiqu<71>s ActusNews.".EOL;
echo $message;
$strMessage.=$message;
New branch with new directories structure
2012-10-16 07:44:31 +00:00
}
/**
** INTEGRATION DES COMMUNIQUES "DIRELEASE"
**/
Update from old server
2013-06-19 08:24:49 +00:00
/*echo date('Y/m/d - H:i:s') ." - DEBUT de l'int<6E>gration des flux DiRelease...".EOL;
New branch with new directories structure
2012-10-16 07:44:31 +00:00
$tabFichier=array();
$dh = opendir(DIRELEASE_LOCAL_DIR);
while (false !== ($filename = readdir($dh))) {
if ($filename<>'.' && $filename<>'..' && substr($filename, -4)=='.xml')
$tabFichier[] = $filename;
}
/** @todo Pour DiRelease, Ne pas utiliser le parser XML mais uniquement des expr. r<EFBFBD>guli<EFBFBD>re car XML de merde.
** Prendre tous les ISIN et rechercher le SIREN.
Update from old server
2013-06-19 08:24:49 +00:00
**
New branch with new directories structure
2012-10-16 07:44:31 +00:00
foreach ($tabFichier as $k => $nomFichier) {
$strTmp=file_get_contents(DIRELEASE_LOCAL_DIR.$nomFichier);
if (stripos($strTmp, 'ISO-8859-1')>0) $encoding='ISO-8859-1';
elseif (stripos($strTmp, 'UTF-8')>0) $encoding='UTF-8';
else die(date ('Y/m/d - H:i:s')." - ERREUR : Encoding non g<>r<EFBFBD> !".EOL);
if (stripos($strTmp, 'langue="FR"')>0) {
echo date ('Y/m/d - H:i:s')." - Chargement du communiqu<71> en fran<61>ais ($ligne).".EOL;
preg_match('/<id>(.*)<\/id>/isU',$strTmp,$matches);
$pressReleaseId=trim(@$matches[1]);
preg_match('/<raisonsociale>(.*)<\/raisonsociale>/isU',$strTmp,$matches);
$raisonsociale=trim(@$matches[1]);
preg_match('/<codeisin>(.*)<\/codeisin>/isU',$strTmp,$matches);
$codeisin=trim(@$matches[1]);
preg_match('/<code_reuters>(.*)<\/code_reuters>/isU',$strTmp,$matches);
$code_reuters=trim(@$matches[1]);
preg_match('/<logo>(.*)<\/logo>/isU',$strTmp,$matches);
$logo=trim(@$matches[1]);
preg_match('/<site>(.*)<\/site>/isU',$strTmp,$matches);
$site=trim(@$matches[1]);
preg_match('/<site_investisseur>(.*)<\/site_investisseur>/isU',$strTmp,$matches);
$site_investisseur=trim(@$matches[1]);
preg_match('/<logo>(.*)<\/logo>/isU',$strTmp,$matches);
$logo=trim(preg_replace('/\s/','',@$matches[1]));
preg_match('/<miseajour>(.*)<\/miseajour>/isU',$strTmp,$matches);
$miseajour=trim(@$matches[1]);
preg_match('/<titre>(.*)<\/titre>/isU',$strTmp,$matches);
$titre=trim(@$matches[1]);
preg_match('/<contenu>(.*)<\/contenu>/isU',$strTmp,$matches);
$contenu=trim(strtr(@$matches[1],array('<![CDATA['=>'', ']]>'=>'')));
preg_match('/<pdf>(.*)<\/pdf>/isU',$strTmp,$matches);
$pdf=preg_replace('/\s/','',@$matches[1]);
preg_match('/<indice>(.*)<\/indice>/isU',$strTmp,$matches);
$indice=trim(@$matches[1]);
preg_match('/<secteur>(.*)<\/secteur>/isU',$strTmp,$matches);
$secteur=trim(@$matches[1]);
preg_match('/<marche_cotation>(.*)<\/marche_cotation>/isU',$strTmp,$matches);
$marche_cotation=trim(@$matches[1]);
preg_match('/<code_bloomberg>(.*)<\/code_bloomberg>/isU',$strTmp,$matches);
$code_bloomberg=trim(@$matches[1]);
preg_match('/<mnemo>(.*)<\/mnemo>/isU',$strTmp,$matches);
$mnemo=trim(@$matches[1]);
preg_match('/<nb_de_titres>(.*)<\/nb_de_titres>/isU',$strTmp,$matches);
$nb_de_titres=trim(@$matches[1]);
preg_match('/<footsie>(.*)<\/footsie>/isU',$strTmp,$matches);
$footsie=trim(@$matches[1]);
$tabInsert=array( //'companyId'=>$dom->getValueFromTag('COMPANY_ID'),
'companyName'=>$raisonsociale,
'companyIsin'=>$codeisin,
//'companySiren'=>$dom->getValueFromTag('COMPANY_SIREN'),
'companyRic'=>$code_reuters,
'companyLogoUrl'=>$logo,
'companyWebSite'=>$site,
//'companyProfile'=>$dom->getValueFromTag('URL3'),
//'companyAnnualReport'=>$dom->getValueFromTag('URL4'),
'companyInfoActionnaires'=>$site_investisseur,
'pressReleaseId'=>$pressReleaseId,
'pressReleaseDate'=>$miseajour,
'pressReleaseTitle'=>$titre,
'pressReleaseText'=>trim(strtr(html_entity_decode(strip_tags($contenu)),array('&rsquo;'=>'<27>'))),
'pressReleaseHtml'=>$contenu,
'pressReleaseAttachments'=>$pdf,
'indexQuoteInd'=>$indice,
'indexSector'=>$secteur,
'indexMarketPlace'=>$marche_cotation,
'source'=>'D',
'companyBloomberg'=>$code_bloomberg,
'companyMnemo'=>$mnemo,
'companyNbTitles'=>$nb_de_titres,
'companyFootsie'=>$footsie,
);
$ret=$iDb->insert('articles', $tabInsert);
if (!$ret && $iDb->getLastErrorNum()<>1062) {
echo date ('Y/m/d - H:i:s')." - ERREUR ". $iDb->getLastError() . EOL;
print_r($tabInsert);
die();
} elseif ($iDb->getLastErrorNum()<>1062) {
echo date ('Y/m/d - H:i:s')." - Communiqu<71> n<>$pressReleaseId enregistr<74> avec succ<63>s.".EOL;
Update from old server
2013-06-19 08:24:49 +00:00
$nbDiRel++;
New branch with new directories structure
2012-10-16 07:44:31 +00:00
}
unset($dom);
}
}
Update from old server
2013-06-19 08:24:49 +00:00
$message=date('Y/m/d - H:i:s') ." - FIN de l'int<6E>gration des $nbDiRel communiqu<71>s DiRelease.".EOL;
echo $message;
$strMessage.=$message;
*/
New branch with new directories structure
2012-10-16 07:44:31 +00:00
/**
** INTEGRATION DES COMMUNIQUES "BUSINESSWIRE"
**/
Update from old server
2013-06-19 08:24:49 +00:00
if ($doBusin || $doAll) {
echo date('Y/m/d - H:i:s') ." - DEBUT de l'int<6E>gration des flux BusinessWire...".EOL;
$tabFichier=array();
$dh = opendir(BUSINESSWIRE_LOCAL_DIR);
while (false !== ($filename = readdir($dh))) {
if ($filename<>'.' && $filename<>'..' && substr($filename, -4)=='.xml')
$tabFichier[] = $filename;
New branch with new directories structure
2012-10-16 07:44:31 +00:00
}
Update from old server
2013-06-19 08:24:49 +00:00
foreach ($tabFichier as $k => $nomFichier) {
$tabTmp=file(BUSINESSWIRE_LOCAL_DIR.$nomFichier);
$strTmp=implode("\n", $tabTmp);
if (!$tabTmp) die(date ('Y/m/d - H:i:s')." - ERREUR : Fichier local ".BUSINESSWIRE_LOCAL_DIR."$nomFichier inexistant !".EOL);
/*if (stripos($tabTmp[0], 'ISO-8859-1')>0) $encoding='ISO-8859-1';
elseif (stripos($tabTmp[0], 'UTF-8')>0) $encoding='UTF-8';
elseif (stripos($tabTmp[0], '<?xml version = \'1.0\'?>')) $encoding='?';
// <?xml version = '1.0'?>
else die(date ('Y/m/d - H:i:s')." - ERREUR : Encoding ".$tabTmp[0]." non g<>r<EFBFBD> pour le fichier ".BUSINESSWIRE_LOCAL_DIR."$nomFichier !".EOL);*/
/*<DateAndTime>20070906T190800+0000</DateAndTime>
<NewsService FormalName="Business Wire"/>
<NewsProduct FormalName="BUSINESS WIRE"/>
</NewsEnvelope>
<NewsItem>
<Identification>
<NewsIdentifier>
<ProviderId>businesswire.com</ProviderId>
<DateId>20010714</DateId>
<>20070906006073</NewsItemId>
<RevisionId PreviousRevision="0" Update="N">1</RevisionId>
<PublicIdentifier>urn:newsml:businesswire.com:20010714:20070906006073:1</PublicIdentifier>
</NewsIdentifier>
</Identification>
<NewsManagement>
<NewsItemType FormalName="News"/>
<FirstCreated>20070906T190800+0000</FirstCreated>
<ThisRevisionCreated>20070906T190800+0000</ThisRevisionCreated>
<Status FormalName="Usable"/>
<AssociatedWith NewsItem="businesswire.com:20010714:242009"/>
</NewsManagement>
<NewsComponent>
<BasisForChoice Rank="1">./NewsComponent/DescriptiveMetadata/Language</BasisForChoice>
<NewsLines>
<CopyrightLine>Copyright Business Wire 2007</CopyrightLine>
</NewsLines>
<AdministrativeMetadata>
<Source>
<Party FormalName="CIT Group Inc."/>
</Source>
<Contributor>
<Comment FormalName="BWoffices">NY</Comment>
<Party FormalName="DB" Scheme="BWEditor"/>
</Contributor>
</AdministrativeMetadata>
<DescriptiveMetadata>
<Language FormalName="fr"/>
<Genre FormalName="Release"/>
</DescriptiveMetadata>
<Metadata>
<MetadataType FormalName="BWKeywords"/>
<Property FormalName="BWCountryKeywords" Value="United States"/>
<Property FormalName="BWRegionKeywords" Value="Europe"/>
<Property FormalName="BWRegionKeywords" Value="North America"/>
<Property FormalName="BWIndustryKeywords" Value="Manufacturing"/>
<Property FormalName="BWIndustryKeywords" Value="Aerospace"/>
<Property FormalName="BWCategoryKeywords" Value="Product/Service"/>
<Property FormalName="BWIndustryKeywords" Value="Professional Services"/>
<Property FormalName="BWIndustryKeywords" Value="Finance"/>
<Property FormalName="BWStateKeywords" Value="New York"/>
<Property FormalName="BWCountryKeywords" Value="Ireland"/>
</Metadata>
<Metadata>
<MetadataType FormalName="Securities Identifier"/>
<Property FormalName="" Value="CIT"/>
<Property FormalName="Exchange" Value="NYSE"/>
<Property FormalName="ISIN" Value="US1255811085"/>
<Property FormalName="SlugLine Display Order" Value="1"/>
</Metadata>
<NewsComponent>
<BasisForChoice Rank="1">./NewsComponent/Role</BasisForChoice>
<NewsLines>
<>CIT consolide son engagement envers l'<EFBFBD>conomie irlandaise et agrandit son centre op<EFBFBD>rationnel de Dublin</HeadLine>*/
$dom = @new DomDocument2();
$dom->load(BUSINESSWIRE_LOCAL_DIR.$nomFichier);
$pressReleaseId=$dom->getValueFromTag('NewsItemId');
preg_match('/<Property FormalName="ISIN" Value="(.*)"\/>/i',$strTmp,$matches);
$isin=@$matches[1];
preg_match('/<Property FormalName="Exchange" Value="(.*)"\/>/i',$strTmp,$matches);
$exchange=@$matches[1];
preg_match('/<Property FormalName="Ticker Symbol" Value="(.*)"\/>/i',$strTmp,$matches);
$mnemo=@$matches[1];
preg_match('/<NewsLineType FormalName="ClickThru URL"(?:.*)<NewsLineText>(.*)<\/NewsLineText>/isU',$strTmp,$matches);
$siteWeb=@$matches[1];
preg_match('/<NewsComponent>(?:.*)<Role FormalName="Logo"\/>(?:.*)<ContentItem Duid="(?:.*)" Href="(.*)">(?:.*)<\/NewsComponent>/isU',$strTmp,$matches);
$logo=urldecode(@$matches[1]);
preg_match('/<NewsComponent>(?:.*)<Role FormalName="Body"\/>(?:.*)<ContentItem (?:.*)<Format FormalName="XHTML"\/>(?:.*)<DataContent>(.*)<\/DataContent>/isU',$strTmp,$matches);
$bodyHtml=utf8_decode(@$matches[1]);
preg_match('/<NewsComponent>(?:.*)<Role FormalName="Body"\/>(?:.*)<ContentItem (?:.*)<Format FormalName="BW-Text"\/>(?:.*)<DataContent>(.*)<\/DataContent>/isU',$strTmp,$matches);
$bodyTxt=utf8_decode(@$matches[1]);
preg_match_all('/<Property FormalName="BWIndustryKeywords" Value="(.*)"\/>/iU',$strTmp,$matches);
$industry=@implode(';',@$matches[1]);
preg_match_all('/<Property FormalName="BWCategoryKeywords" Value="(.*)"\/>/iU',$strTmp,$matches);
$category=@implode(';',@$matches[1]);
preg_match_all('/<Property FormalName="BWCountryKeywords" Value="(.*)"\/>/iU',$strTmp,$matches);
$country=@implode(';',@$matches[1]);
preg_match('/<Language FormalName="(.*)"\/>/i',$strTmp,$matches);
$language=@$matches[1];
$tabInsert=array( //'companyId'=>$dom->getValueFromTag('COMPANY_ID'),
'companyName'=>$dom->getValueFromTag('SlugLine'),
'companyIsin'=>$isin,
'companyMnemo'=>$mnemo,
'companyWebSite'=>$siteWeb,
'companyLogoUrl'=>$logo,
/*'companySiren'=>$dom->getValueFromTag('COMPANY_SIREN'),
'companyRic'=>$dom->getValueFromTag('COMPANY_RIC'),
'companyProfile'=>$dom->getValueFromTag('URL3'),
'companyAnnualReport'=>$dom->getValueFromTag('URL4'),
'companyInfoActionnaires'=>$dom->getValueFromTag('URL5'),*/
'pressReleaseId'=>$pressReleaseId,
'pressReleaseDate'=>substr(str_replace('T','',$dom->getValueFromTag('DateAndTime')),0,14),// 20070906T190800+0000
'pressReleaseTitle'=>$dom->getValueFromTag('HeadLine'),
'pressReleaseText'=>$bodyTxt,
'pressReleaseHtml'=>$bodyHtml,
'indexMarketPlace'=>$exchange,
'indexTheme'=>$category,
'indexSector'=>$industry,
'indexCountry'=>$country,
'indexLanguage'=>$language,
/*
'pressReleaseAttachments'=>$dom->getValueFromTag('ATTACHMENTS_URL1'),
'pressReleaseUrl'=>$dom->getValueFromTag('PRESS_RELEASE_URL_CNG'),
'indexAll'=>$dom->getValueFromTag('INDEX_ALL'),
'indexQuoteInd'=>$dom->getValueFromTag('INDEX_QUOTE_INDICATOR'),*/
'source'=>'B',
);
New branch with new directories structure
2012-10-16 07:44:31 +00:00
$ret=$iDb->insert('articles', $tabInsert);
if (!$ret && $iDb->getLastErrorNum()<>1062) {
echo date ('Y/m/d - H:i:s')." - ERREUR ". $iDb->getLastError() . EOL;
print_r($tabInsert);
die();
} elseif ($iDb->getLastErrorNum()<>1062) {
echo date ('Y/m/d - H:i:s')." - Communiqu<71> n<>$pressReleaseId enregistr<74> avec succ<63>s.".EOL;
Update from old server
2013-06-19 08:24:49 +00:00
$nbBusin++;
New branch with new directories structure
2012-10-16 07:44:31 +00:00
}
Update from old server
2013-06-19 08:24:49 +00:00
unset($dom);
}
$message=date('Y/m/d - H:i:s') ." - FIN de l'int<6E>gration des $nbBusin communiqu<71>s BusinessWire.".EOL;
echo $message;
$strMessage.=$message;
}
/**
** INTEGRATION DES COMMUNIQUES "LES ECHOS"
**/
if ($doEchos || $doAll) {
echo date('Y/m/d - H:i:s') ." - DEBUT de l'int<6E>gration des flux Les Echos...".EOL;
$tabFichier=array();
$dh = opendir(LESECHOS_LOCAL_DIR.'syndication/');
while (false !== ($filename = readdir($dh))) {
if ($filename<>'.' && $filename<>'..' && substr($filename, -4)=='.xml')
$tabFichier[] = $filename;
}
foreach ($tabFichier as $k => $nomFichier) {
$tabTmp=file(LESECHOS_LOCAL_DIR.'syndication/'.$nomFichier);
if (!$tabTmp) die(date ('Y/m/d - H:i:s')." - ERREUR : Fichier local $nomFichier inexistant !".EOL);
if (stripos($tabTmp[0], 'ISO-8859-1')>0) $encoding='ISO-8859-1';
elseif (stripos($tabTmp[0], 'UTF-8')>0) $encoding='UTF-8';
else die(date ('Y/m/d - H:i:s')." - ERREUR : Encoding ".$tabTmp[0]." non g<>r<EFBFBD> !".EOL);
$dom = @new DomDocument2();
$dom->load(LESECHOS_LOCAL_DIR.'syndication/'.$nomFichier);
$pressReleaseId=$dom->getValueFromTag('CODE');
$language=strtoupper($dom->getValueFromTag('PRESS_RELEASE_LANGUAGE'));
$tabInsert=array( //'companyId'=>$dom->getValueFromTag('COMPANY_ID'),
'companyName'=>$dom->getValueFromTag('COMPANY_NAME'),
'companyIsin'=>$dom->getValueFromTag('COMPANY_ISIN'),
//'companySiren'=>$dom->getValueFromTag('COMPANY_SIREN'),
//'companyRic'=>$dom->getValueFromTag('COMPANY_RIC'),
//'companyLogoUrl'=>$dom->getValueFromTag('URL1'),
//'companyWebSite'=>$dom->getValueFromTag('URL2'),
//'companyProfile'=>$dom->getValueFromTag('URL3'),
//'companyAnnualReport'=>$dom->getValueFromTag('URL4'),
//'companyInfoActionnaires'=>$dom->getValueFromTag('URL5'),
'pressReleaseId'=>$pressReleaseId,
'pressReleaseDate'=>WDate::dateT('d/m/Y', 'Y-m-d', $dom->getValueFromTag('PRESS_RELEASE_PUBDATE')).' '.
$dom->getValueFromTag('PRESS_RELEASE_PUBTIME'),
'pressReleaseTitle'=>$dom->getValueFromTag('PRESS_RELEASE_TITLE'),
'pressReleaseText'=>$dom->getValueFromTag('TEXT_FORMAT'),
'pressReleaseHtml'=>$dom->getValueFromTag('HTML_FORMAT'),
'pressReleaseAttachments'=>$dom->getValueFromTag('LINK_ORIGINAL'),
'pressReleaseUrl'=>$dom->getValueFromTag('LINK'),
//'indexAll'=>$dom->getValueFromTag('INDEX_ALL'),
'indexTheme'=>$dom->getValueFromTag('PRESS_RELEASE_THEME'),
//'indexSector'=>$dom->getValueFromTag('INDEX_SECTOR'),
//'indexCountry'=>$dom->getValueFromTag('INDEX_COUNTRY'),
'indexLanguage'=>$language,
//'indexMarketPlace'=>$dom->getValueFromTag('INDEX_MARKET_PLACE'),
//'indexQuoteInd'=>$dom->getValueFromTag('INDEX_QUOTE_INDICATOR'),
'source'=>'E',
);
if ($language=='FR') {
$ret=$iDb->insert('articles', $tabInsert);
if (!$ret && $iDb->getLastErrorNum()<>1062) {
echo date ('Y/m/d - H:i:s')." - ERREUR ". $iDb->getLastError() . EOL;
print_r($tabInsert);
die();
} elseif ($iDb->getLastErrorNum()<>1062) {
echo date ('Y/m/d - H:i:s')." - Communiqu<71> n<>$pressReleaseId enregistr<74> avec succ<63>s.".EOL;
$nbEchos++;
}
} else
echo date ('Y/m/d - H:i:s')." - Langue du communiqu<71> non int<6E>gr<67>e en base ($language).".EOL;
unset($dom);
}
$message=date('Y/m/d - H:i:s') ." - FIN de l'int<6E>gration des $nbEchos communiqu<71>s Les Echos.".EOL;
echo $message;
$strMessage.=$message;
New branch with new directories structure
2012-10-16 07:44:31 +00:00
}
Update from old server
2013-06-19 08:24:49 +00:00
sendMail('webmaster@scores-decisions.com', 'support@scores-decisions.com,ylenaour@scores-decisions.com', "Int<EFBFBD>gration des informations r<>glement<6E>es", $strMessage);
New branch with new directories structure
2012-10-16 07:44:31 +00:00
die();
?>
Reference in New Issue Copy Permalink