Primo commit

composer.json

@@ -0,0 +1,10 @@
+{
+    "require": {
+        "fabpot/goutte": "^4.0"
+    },
+    "autoload": {
+        "psr-4": {
+            "": "src/"
+        }
+    }
+}

scrape.php

@@ -0,0 +1,20 @@
+<?php
+
+require __DIR__.'/vendor/autoload.php';
+
+use Scraping\Scraper;
+
+$start_url = 'https://este.trasparenza-valutazione-merito.it/web/amministrazionetrasparente/papca-ap/-/papca/igrid/1452834656/1452816457';
+
+$filters = [
+    '//td[@class="actions"]/a[@title="Visualizza Allegato"]',
+    '//table[@class="allegati"]/a[@title="Download versione non firmata"]'
+];
+
+
+$scraper = new Scraper();
+$scraper->levels = count($filters);
+
+foreach ($filters as $i => $filter) {
+    $scraper->scrape('GET', $start_url, $filters, $i);
+}

src/Scraping/Scraper.php

@@ -0,0 +1,30 @@
+<?php
+
+namespace Scraping;
+
+use \Goutte\Client;
+
+class Scraper
+{
+    public $links = [];
+
+    public static function scrape($method, $url, $filters, $level){
+        $client = new Client();
+
+        $crawler = $client->request($method, $url);
+
+        print str_pad('['.$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 = $level+1;
+
+            //print str_pad('['.$new_level.']', $new_level*2, ' ', STR_PAD_RIGHT).' '.$new_url."\n";
+
+            if ($new_level < count($filters)) {
+                self::scrape($method, $new_url, $filters, $new_level);
+            }
+        });
+    }
+}