Soporte para Infobae añadido

2025-10-13 00:32:45 +00:00 · 2021-07-16 21:09:07 -03:00
parent 00ff3b2ad9
commit e60588e577
8 changed files with 152 additions and 4 deletions
--- a/control-files/00_ids_done.txt
+++ b/control-files/00_ids_done.txt
@@ -1 +1 @@
-7
+8
--- a/control-files/00_news_done.txt
+++ b/control-files/00_news_done.txt
@@ -1 +1 @@
- ./news/1.html ./news/2.html ./news/3.html ./news/4.html ./news/5.html ./news/6.html ./news/7.html
+ ./news/1.html ./news/2.html ./news/3.html ./news/4.html ./news/5.html ./news/6.html ./news/7.html ./news/8.html
--- a/index.php
+++ b/index.php
@@ -40,6 +40,7 @@
        $title = $news[1];
        $frist_p = $news[2];
        $icon = $news[3];
+        $source = $news[4];

        echo
            "<a href='news.php?id=$id' class='card-link'>
@@ -48,6 +49,7 @@
                        <div class='text'>
                            <h2>$title</h2>
                            <h3>$frist_p</h3>
+                            <div class='source'>$source</div>
                        </div>
                </div>
            </a>";
--- a/news_logic.php
+++ b/news_logic.php
@@ -1,5 +1,7 @@
 <?php

+use JetBrains\PhpStorm\Internal\ReturnTypeContract;
+
 error_reporting(~0);
 ini_set('display_errors', 1);

@@ -26,7 +28,11 @@ function know_page($news_unique) {
        case $news_unique === './news/genbeta.html' || $news_unique === './news/Genbeta.html':
            return 'Genbeta';
            break;
-            
+
+        case $news_unique === './news/infobae.html' || $news_unique === './news/Infobae.html':
+            return 'Infobae';
+            break;
+
        default:
            return 0;
            break;
@@ -49,6 +55,10 @@ function know_author($page, $news_unique) {
            return str_replace('>', '', $second_cut);
            break;

+        case $page === 'Infobae':
+            return $author = get_string_between($content, ',t.authors="', '",');
+            break;
+
        default:
            return 0;
            break;
@@ -87,12 +97,24 @@ function create_entry_in_DB($news_unique, $page, $author) {
            $inner_HTML = get_string_between($content, '<div class="article-content">', '<div class="article-content-outer">') . '<script id="script-estructurator" src="./scripts/genbeta.js"></script>';
            $frist_p = strip_tags(get_string_between($inner_HTML, '<p>', '</p>'));
            break;
+
+        case $page === 'Infobae':
+            $title = get_string_between($content, '<html lang="es"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><title>', '- Infobae</title>');
+            $pre_icon = get_string_between($content, '<div class="visual__image ">', '</div>');
+            $icon = get_string_between($pre_icon, '992w,', ' 1200w" ');
+            $inner_HTML = get_string_between($content, '<section class="article-section page-container">', '</section>') . '<script id="script-estructurador" src="./scripts/infobae.js"></script>';
+            $frist_p = get_string_between($content, '<meta property="og:description" content="', '">');
+            break;
        
        default:
            echo 0;
            return 0;
            break;
    }
+    $title;
+    $icon;
+    $inner_HTML;
+    $frist_p;

    $insert_news = $mySQLconnect -> prepare('insert into noticias (title, content, icon_route, page_source, author, frist_paragraph) values (?, ?, ?, ?, ?, ?)');

@@ -119,7 +141,7 @@ function bring_the_news_back_home($actual_page, $news_per_page) {
    $frist_calc = $actual_page * $news_per_page;
    $second_calc = ($actual_page * $news_per_page) - $news_per_page;

-    $prepared_query = $mySQLconnect -> prepare("select id, title, frist_paragraph, icon_route from noticias where id < ? and id > ?");
+    $prepared_query = $mySQLconnect -> prepare("select id, title, frist_paragraph, icon_route, page_source from noticias where id < ? and id > ?");
    $prepared_query -> bindParam(1, $frist_calc, PDO::PARAM_INT);
    $prepared_query -> bindParam(2, $second_calc, PDO::PARAM_INT);
    //$prepared_query -> execute(array($actual_page * 10, $actual_page * 10 - 10));
--- a/scripts/genbeta.js
+++ b/scripts/genbeta.js
--- a/scripts/infobae.js
+++ b/scripts/infobae.js
@@ -0,0 +1,87 @@
+//todo - Informacion de infobae
+
+title = document.getElementsByClassName('article-headline')[0];
+console.log(title);
+
+sub_title = document.getElementsByClassName('article-subheadline')[0];
+console.log(sub_title);
+
+content = document.getElementsByClassName('nd-body-article')[0];
+console.log(content);
+
+trash = [document.getElementsByClassName('infogram-embed')[0]];
+console.log(trash);
+
+paragraphs = document.querySelectorAll('.paragraph, .visual__image');
+console.log(paragraphs);
+
+infobae = document.getElementsByTagName('article')[0];
+
+
+//todo - Estructura del sitio web cliente
+
+all = document.getElementsByClassName('all')[0];
+console.log(all);
+
+//todo - Purificacion de la informacion
+
+for (let i = 0; i < paragraphs.length; i++) {
+    if (paragraphs[i].innerHTML == "<b>SEGUIR LEYENDO:</b>" || paragraphs[i].innerHTML == "<b>SEGUIR LEYENDO</b>" || paragraphs[i].innerHTML == "<br>") {
+        console.log("AAAAA");
+        trash.push(paragraphs[i])
+
+    }
+    
+}
+
+title.innerHTML = ''
+
+for (let i = 0; i < trash.length; i++) {
+    if (trash[i] !== undefined){
+        trash[i].innerHTML = ''
+        trash[i].remove();
+    }
+}
+
+//todo - Copiar e insertar informacion en el sitio web cliente
+
+//! TITULO
+all.appendChild(title);
+
+frist_p = document.createElement("P")
+frist_p.innerText = sub_title.innerText
+
+//! PARRAFOS
+paragraphsFragment = document.createDocumentFragment();
+
+paragraphsFragment.appendChild(frist_p)
+
+for (let i = 0; i < paragraphs.length; i++) {
+
+    if (paragraphs[i].className == "visual__image ") {
+        console.log("AAAA");
+
+        paragraphs[i].children[0].style.height = "auto"
+        paragraphs[i].children[0].style.width = "100%"
+    }
+
+    console.log(paragraphs[i]);
+    paragraphsFragment.appendChild(paragraphs[i])
+    
+}
+
+article = document.createElement('article')
+
+article.appendChild(paragraphsFragment)
+
+all.appendChild(article)
+
+//todo - Eliminar informacion de infobae
+
+infobae.remove()
+
+//todo - Eliminar el script
+
+script = document.getElementById('script-estructurator');
+
+script.remove()
--- a/style.css
+++ b/style.css
@@ -80,6 +80,8 @@ header {
    margin-right: 2vw;
    width: calc(74vw - 150px - 6vw);
    margin-bottom: 2vh;
+    display: flex;
+    flex-direction: column;
 }

 .card h2 {
@@ -113,3 +115,7 @@ header {
    color: #797979;
 }

+.card .source {
+    align-self: flex-end;
+    flex-grow: 1;
+}
--- a/test-page.php
+++ b/test-page.php
@@ -0,0 +1,31 @@
+<link rel="stylesheet" href="news.css">
+
+<body>
+
+<header>
+    <a href="index.php" class="name">
+        <img class="logo" src="./img/logo.svg">
+        <h1>AllNews.com</h1>
+    </a>
+</header>
+
+<div class="all">
+    <article>
+            
+            <?php
+
+        require 'news_logic.php';
+
+        check_news();
+
+        ?>
+
+    </article>
+</div>
+
+<div class="source">
+    Autor: <b><?php echo $author;?></b> | Pagina fuente: <b><?php echo $page_source;?></b>
+</div>
+
+
+</body>