quenousimporte
/
presse
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Compare commits

base: quenousimporte:cb59059a2b5c44ed56563a254f9807a52a860c13
Branches Tags
quenousimporte:main
...
compare: quenousimporte:e5e7ab27e30e8e35ce9540a8e772806aea0c8e32
Branches Tags
quenousimporte:main

2 Commits
cb59059a2b ... e5e7ab27e3

Author SHA1 Message Date
quenousimporte e5e7ab27e3 refactor mp 2024-08-26 18:15:17 +02:00
quenousimporte b97f730205 prepare mp refactor home and rss 2024-08-26 18:10:14 +02:00
1 changed files with 62 additions and 116 deletions
Show Stats Expand all files Collapse all files

176
epub.php
View File

@ -17,6 +17,65 @@
require 'settings.php'; require 'settings.php';
require 'TPEpubCreator.php'; require 'TPEpubCreator.php';
function add_mp_article($url, $context, $epub)
{
$article = file_get_contents($url, false, $context);
$doc = new DOMDocument();
$doc->loadHTML($article);
$finder = new DomXPath($doc);
$title = $finder->query('//h1')->item(0)->textContent;
$author = $finder->query('//a[contains(@href, "biographie")]')->item(0)->textContent;
$date = $finder->query('//time')->item(1)->textContent;
$summary = $finder->query('//p[contains(@class, "news__heading__top__intro")]')->item(0)->textContent;
// strip images
$toremove = $finder->query('//svg');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$toremove = $finder->query('//figure');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$toremove = $finder->query('//span[@class="screen-reader-only"]');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$result = '<h1>' . $title . '</h1>';
$result .= '<p>' . $author . '</p>';
$result .= '<p>' . $date . '</p>';
$result .= '<p><b>' . $summary . '</b></p>';
$nodes = $finder->query('//div[contains(@class, "paywall-restricted-content")]');
if (!$nodes->length)
{
// articles accès libre
$nodes = $finder->query('//div[contains(@class, "news__body__center__article")]');
}
if (!$nodes->length)
{
echo 'warning: could not get content of "' . $title . '"<br>';
}
else
{
$node = $nodes->item(0);
$innerHTML = '';
foreach ($node->childNodes as $childNode){
$innerHTML .= $childNode->ownerDocument->saveHTML($childNode);
}
$result .= '<div>' . strip_tags($innerHTML, '<p><b><h2><i>') . '</div>';
$epub->AddPage($result, false, $title);
}
}
function get_mp_cookie() function get_mp_cookie()
{ {
global $mp_user; global $mp_user;
@ -325,62 +384,7 @@
foreach ($items as $item) foreach ($items as $item)
{ {
$title = $item->title; add_mp_article($item->link, $context, $epub);
$category = $item->xpath('dc:subject')[0];
$author = $item->xpath('dc:creator')[0];
$summary = $item->description;
$article = file_get_contents($item->link, false, $context);
$doc = new DOMDocument();
$doc->loadHTML($article);
$finder = new DomXPath($doc);
// strip images
$toremove = $finder->query('//svg');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$toremove = $finder->query('//figure');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$toremove = $finder->query('//span[@class="screen-reader-only"]');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$result = '<h1>' . $title . '</h1>';
$result .= '<p>' . $author . '</p>';
$result .= '<p>' . $item->pubDate . '</p>';
$result .= '<p><b>' . $summary . '</b></p>';
$nodes = $finder->query('//div[contains(@class, "paywall-restricted-content")]');
if (!$nodes->length)
{
// articles accès libre
$nodes = $finder->query('//div[contains(@class, "news__body__center__article")]');
}
if (!$nodes->length)
{
echo 'warning: could not get content of "' . $title . '"<br>';
}
else
{
$node = $nodes->item(0);
$innerHTML = '';
foreach ($node->childNodes as $childNode){
$innerHTML .= $childNode->ownerDocument->saveHTML($childNode);
}
$result .= '<div>' . strip_tags($innerHTML, '<p><b><h2><i>') . '</div>';
$epub->AddPage($result, false, $title);
}
} }
write_epub($epub); write_epub($epub);
@ -414,63 +418,7 @@
foreach ($links as $link) foreach ($links as $link)
{ {
$url = 'https://www.mediapart.fr' . $link->value; add_mp_article('https://www.mediapart.fr' . $link->value, $context, $epub);
$article = file_get_contents($url, false, $context);
$doc = new DOMDocument();
$doc->loadHTML($article);
$finder = new DomXPath($doc);
$title = $finder->query('//h1')->item(0)->textContent;
$author = $finder->query('//a[contains(@href, "biographie")]')->item(0)->textContent;
$date = $finder->query('//time')->item(1)->textContent;
$summary = $finder->query('//p[contains(@class, "news__heading__top__intro")]')->item(0)->textContent;
// strip images
$toremove = $finder->query('//svg');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$toremove = $finder->query('//figure');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$toremove = $finder->query('//span[@class="screen-reader-only"]');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$result = '<h1>' . $title . '</h1>';
$result .= '<p>' . $author . '</p>';
$result .= '<p>' . $date . '</p>';
$result .= '<p><b>' . $summary . '</b></p>';
$nodes = $finder->query('//div[contains(@class, "paywall-restricted-content")]');
if (!$nodes->length)
{
// articles accès libre
$nodes = $finder->query('//div[contains(@class, "news__body__center__article")]');
}
if (!$nodes->length)
{
echo 'warning: could not get content of "' . $title . '"<br>';
}
else
{
$node = $nodes->item(0);
$innerHTML = '';
foreach ($node->childNodes as $childNode){
$innerHTML .= $childNode->ownerDocument->saveHTML($childNode);
}
$result .= '<div>' . strip_tags($innerHTML, '<p><b><h2><i>') . '</div>';
$epub->AddPage($result, false, $title);
}
} }
write_epub($epub); write_epub($epub);
@ -501,8 +449,6 @@
foreach ($items as $item) foreach ($items as $item)
{ {
$title = $item->title; $title = $item->title;
$category = $item->xpath('dc:subject')[0];
$author = $item->xpath('dc:creator')[0]; $author = $item->xpath('dc:creator')[0];
$summary = $item->description; $summary = $item->description;