This commit is contained in:
2021-07-13 21:25:04 -03:00
5 changed files with 90 additions and 36 deletions

0
.gitignore vendored Executable file
View File

View File

@@ -0,0 +1,43 @@
Sdocument = document.getElementsByClassName("blob js-post-images-container")[0]
big_blank_space = Sdocument.getElementsByClassName('base-wrapper-image')
console.log(big_blank_space);
ad = Sdocument.getElementsByClassName('desvio-container')
console.log(ad);
imgs = Sdocument.querySelectorAll('img')
console.log(imgs)
for (i = 0; i < ad.length; i++) {
ad[i].innerHTML = ''
}
for (let i = 0; i < big_blank_space.length; i++) {
big_blank_space[i].style.paddingTop = ""
}
for (let i = 0; i < imgs.length; i++) {
if (imgs[i].src === "") {
imgs[i].src = imgs[i].dataset.sfSrc
}
}
console.log(Sdocument);
links = document.getElementsByClassName('article-links')
links[0].remove()
paragraphs = document.getElementsByTagName('p')
for (i = 0; i < paragraphs.length; i++) {
if (paragraphs[i].innerHTML.includes("En Xataka |")) {
paragraphs[i].remove()
}
}
script = document.getElementById('script-estructurator');
script.remove()

View File

@@ -18,7 +18,6 @@
<div class="card-container">
<?php
require 'mySQLconnect.php';
require 'news_logic.php';
check_news();

0
news/00_ids_done.txt Normal file
View File

View File

@@ -6,41 +6,26 @@ ini_set('display_errors', 1);
function check_news() {
$news = glob('./news/*');
print_r($news);
echo '<br>';
foreach ($news as $new) {
if ($new !== './news/00_news_done.txt'){
echo '<br>';
echo 'NEW: ' . $new;
echo '<br>';
foreach ($news as $news_unique) {
if ($news_unique !== './news/00_news_done.txt' && $news_unique !== './news/00_ids_done.txt' ){
$news_done = file_get_contents('./news/00_news_done.txt');
echo $news_done;
echo '<br>';
if (!strstr($news_done, $new)) {
$page = know_page($new); //! HACER FUNCION
echo $page;
$author = know_author($page, $new);
echo $author;
//$create_entry_in_DB($new, $author); //! HACER FUNCION
//file_put_contents($news_done, $new . '/n', FILE_APPEND);
if (!strstr($news_done, $news_unique)) {
$page = know_page($news_unique);
$author = know_author($page, $news_unique);
create_entry_in_DB($news_unique, $page, $author); //! HACER FUNCION
}
/*if ($new is in $done) {
/*if ($news_unique is in $done) {
}*/
}
}
}
function know_page($new) {
switch ($new) {
function know_page($news_unique) {
switch ($news_unique) {
case './news/xataka.html'|| './news/Xataka.html':
return 'Xataka';
break;
@@ -51,8 +36,8 @@ function know_page($new) {
}
}
function know_author($page, $new) {
$content = file_get_contents($new);
function know_author($page, $news_unique) {
$content = file_get_contents($news_unique);
switch ($page) {
case 'Xataka':
@@ -77,16 +62,43 @@ function get_string_between($string, $start, $end){
//! CREDITS: https://stackoverflow.com/questions/5696412/how-to-get-a-substring-between-two-strings-in-phphttps://stackoverflow.com/questions/5696412/how-to-get-a-substring-between-two-strings-in-php
}
/*
switch ($variable) {
case 'value':
# code...
break;
function create_entry_in_DB($news_unique, $page, $author) {
require 'mySQLconnect.php';
$content = file_get_contents($news_unique);
switch ($page) {
case 'Xataka':
$title = get_string_between($content, '<h1><span>', '</span></h1>');
$pre_icon = get_string_between($content, '<div class="base-wrapper-image" style="padding-top: 61.38%;">', '</div>');
$icon = get_string_between($pre_icon, 'src=', ' ');
$inner_HTML = get_string_between($content, '<div class="article-content">', '<div class="article-content-outer">') . '<script id="script-estructurator" src="3lqzK81oyJW4C+q8OXEsRs7xuJco4Gz9ewZc993eBZwfxOqs3ToZOJ9KYmX5v0IEG83ds9TcRSvHyhztvNs9KyucmzRo7IxfonPGF+PFg99QZn3EOfTul3GeCApquf6/5WS70jg66hp3mYWfcpK5B5kbJWIF/NhXHUusw2jtsrw7MsZ0J3TzL0s/g9UZhj30/LtiHKDBL2nWtFVCo/MiOZcfRmMyFSi6QhJnoi7Ri5GcVHym6tCAUGXiPaAWEmikxfosgrUDyjUp4hCdos9jFEQO+G7DE50h3dKWIEKlrVPaDbygJA9d47TEvcSq7FTD1f3PnTeibUV+VBIi4ZgRpHrlk45FBUKvdxeGquoAvApW3734L0.js"></script>';
$frist_p = strip_tags(get_string_between($inner_HTML, '<p>', '</p>'));
break;
default:
echo 0;
return 0;
break;
}
$insert_news = $mySQLconnect -> prepare('insert into noticias (title, content, icon_route, page, author, frist_paragraph) values (?, ?, ?, ?, ?, ?)');
$insert_news -> bindParam(1, $title, PDO::PARAM_STR);
$insert_news -> bindParam(2, $inner_HTML, PDO::PARAM_STR);
$insert_news -> bindParam(3, $icon, PDO::PARAM_STR);
$insert_news -> bindParam(4, $page, PDO::PARAM_STR);
$insert_news -> bindParam(5, $author, PDO::PARAM_STR);
$insert_news -> bindParam(6, $frist_p, PDO::PARAM_STR);
default:
# code...
break;
$insert_news -> execute();
$ids_done = file_get_contents('./news/00_ids_done.txt');
$num = $ids_done + 1;
file_put_contents('./news/00_ids_done.txt', $num);
rename($news_unique,"./news/$num.html");
file_put_contents('./news/00_news_done.txt', "./news/$num.html", FILE_APPEND);
}
*/
?>