From 71df5bcdccf298d84be49262c01ddf2bace85bb3 Mon Sep 17 00:00:00 2001
From: loviuz <loviuz@gmail.com>
Date: Sat, 8 Jan 2022 14:28:07 +0100
Subject: [PATCH] Completamento classe scraping

---
 .gitignore               |  3 ++-
 scrape.php               |  3 ++-
 src/Scraping/Scraper.php | 29 +++++++++++++++++++----------
 3 files changed, 23 insertions(+), 12 deletions(-)

diff --git a/.gitignore b/.gitignore
index 1577a3c..48f15df 100644
--- a/.gitignore
+++ b/.gitignore
@@ -2,4 +2,5 @@ vendor/
 composer.phar
 composer.lock
 pdf/*
-config.php
\ No newline at end of file
+config.php
+launch.json
\ No newline at end of file
diff --git a/scrape.php b/scrape.php
index db9d0a3..cef5aae 100644
--- a/scrape.php
+++ b/scrape.php
@@ -1,11 +1,12 @@
 <?php
 
 require __DIR__.'/vendor/autoload.php';
+require __DIR__.'/config.php';
 
 use Scraping\Scraper;
 
 $scraper = new Scraper();
 
 foreach ($filters as $i => $filter) {
-    $scraper->scrape('GET', $start_url, $filters, $i);
+    $scraper->scrape('GET', $start_url, $filters, $i, $download_dir);
 }
diff --git a/src/Scraping/Scraper.php b/src/Scraping/Scraper.php
index 73b098d..7f9e175 100644
--- a/src/Scraping/Scraper.php
+++ b/src/Scraping/Scraper.php
@@ -8,23 +8,32 @@ class Scraper
 {
     public $links = [];
 
-    public static function scrape($method, $url, $filters, $level){
+    public static function scrape($method, $url, $filters, $level, $download_dir){
         $client = new Client();
 
         $crawler = $client->request($method, $url);
 
-        print str_pad('['.$level.']', $level*2, ' ', STR_PAD_LEFT).' '.$url."\n";
+        if (strstr($client->getResponse()->getHeaders()['content-type'][0], 'text/html')) {
+            print str_pad('['.$level.']', $level*2, ' ', STR_PAD_LEFT).' '.$url."\n";
 
-        // 1) Lista atti
-        $crawler->filterXPath($filters[$level])->each(function ($node) use ($method, $filters, $level) {
-            $new_url = $node->link()->getUri();
-            $new_level = $level+1;
+            $crawler->filterXPath($filters[$level])->each(function ($node) use ($method, $filters, $level) {
+                $new_url = $node->link()->getUri();
+                $new_level = $level+1;
 
-            //print str_pad('['.$new_level.']', $new_level*2, ' ', STR_PAD_RIGHT).' '.$new_url."\n";
+                if ($new_level < count($filters)) {
+                    self::scrape($method, $new_url, $filters, $new_level, $download_dir);
+                }
+            });
+        } else {
+            $content_disposition = $client->getResponse()->getHeaders()['content-disposition'][0];
 
-            if ($new_level < count($filters)) {
-                self::scrape($method, $new_url, $filters, $new_level);
+            $filename = time();
+            
+            if (preg_match('/filename="([^"]+)"/', $content_disposition, $m)) {
+                $filename = $m[1];
             }
-        });
+
+            file_put_contents( $download_dir.'/'.$filename, $client->getResponse()->getContent() );
+        }
     }
 }
\ No newline at end of file