From 9db2ec22d90078b52182fcd36a1bc7239b1dd93c Mon Sep 17 00:00:00 2001
From: loviuz <loviuz@gmail.com>
Date: Thu, 6 Jan 2022 17:20:40 +0100
Subject: [PATCH] Primo commit

---
 composer.json            | 10 ++++++++++
 scrape.php               | 20 ++++++++++++++++++++
 src/Scraping/Scraper.php | 30 ++++++++++++++++++++++++++++++
 3 files changed, 60 insertions(+)
 create mode 100644 composer.json
 create mode 100644 scrape.php
 create mode 100644 src/Scraping/Scraper.php

diff --git a/composer.json b/composer.json
new file mode 100644
index 0000000..2b67faf
--- /dev/null
+++ b/composer.json
@@ -0,0 +1,10 @@
+{
+    "require": {
+        "fabpot/goutte": "^4.0"
+    },
+    "autoload": {
+        "psr-4": {
+            "": "src/"
+        }
+    }
+}
\ No newline at end of file
diff --git a/scrape.php b/scrape.php
new file mode 100644
index 0000000..e90cb20
--- /dev/null
+++ b/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);
+}
diff --git a/src/Scraping/Scraper.php b/src/Scraping/Scraper.php
new file mode 100644
index 0000000..73b098d
--- /dev/null
+++ b/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);
+            }
+        });
+    }
+}
\ No newline at end of file