From 151bceefa12d7f60d722a9c094bee119e80572e8 Mon Sep 17 00:00:00 2001
From: Grzegorz Drozd <1885137+GrzegorzDrozd@users.noreply.github.com>
Date: Wed, 8 Nov 2023 19:43:58 +0100
Subject: [PATCH] Add support for local files with file:// schema. (#22)

Using local file vs content of a file ( for parse method ) allows one to use recursive parsing or queue parsing approach and just replace url with local file. Easier for dev time and testing.
I also added a test that creates local file in temp.

Co-authored-by: Grzegorz Drozd <grzegorz.drozd@gmail.com>
---
 src/SitemapParser.php           | 10 +++++++++
 src/SitemapParser/UrlParser.php | 22 +++++++++++++++++++-
 tests/LocalFileTest.php         | 36 +++++++++++++++++++++++++++++++++
 3 files changed, 67 insertions(+), 1 deletion(-)
 create mode 100644 tests/LocalFileTest.php

diff --git a/src/SitemapParser.php b/src/SitemapParser.php
index bafbf32..f755e1c 100644
--- a/src/SitemapParser.php
+++ b/src/SitemapParser.php
@@ -224,6 +224,16 @@ protected function getContent()
             throw new Exceptions\SitemapParserException('Invalid URL');
         }
         try {
+            if (strpos($this->currentURL, 'file://') === 0) {
+                $path = parse_url($this->currentURL, PHP_URL_PATH);
+                if (!$this->urlValidatePath($path)) {
+                    throw new Exceptions\SitemapParserException('Invalid file path');
+                }
+                if (!file_exists($path) && PHP_OS === 'WINNT') {
+                    return file_get_contents(urldecode($path));
+                }
+                return file_get_contents($path);
+            }
             if (!isset($this->config['guzzle']['headers']['User-Agent'])) {
                 $this->config['guzzle']['headers']['User-Agent'] = $this->userAgent;
             }
diff --git a/src/SitemapParser/UrlParser.php b/src/SitemapParser/UrlParser.php
index 62fb24e..6647a5b 100644
--- a/src/SitemapParser/UrlParser.php
+++ b/src/SitemapParser/UrlParser.php
@@ -54,8 +54,12 @@ protected function urlValidate($url)
         return (
             filter_var($url, FILTER_VALIDATE_URL) &&
             ($parsed = parse_url($url)) !== false &&
-            $this->urlValidateHost($parsed['host']) &&
             $this->urlValidateScheme($parsed['scheme']) &&
+            (
+                (in_array($parsed['scheme'], ['http', 'https'], true) && $this->urlValidateHost($parsed['host']))
+                ||
+                (in_array($parsed['scheme'], ['file'], true) && $this->urlValidatePath($parsed['path']))
+            ) &&
             $this->urlValidateAgainstBlackList($url)
         );
     }
@@ -88,10 +92,26 @@ protected static function urlValidateScheme($scheme)
         return in_array($scheme, [
                 'http',
                 'https',
+                'file'
             ]
         );
     }
 
+    /**
+     * Check if local file exists at given path.
+     *
+     * @param mixed $path
+     * @return bool
+     */
+    public function urlValidatePath(mixed $path) {
+        $result = file_exists($path);
+        if ($result === false && PHP_OS === 'WINNT') {
+            // try to reverse url encoding for windows paths:
+            return file_exists(urldecode($path));
+        }
+        return $result;
+    }
+
     protected function urlValidateAgainstBlackList($url)
     {
         if (empty($this->config['url_black_list'])) {
diff --git a/tests/LocalFileTest.php b/tests/LocalFileTest.php
new file mode 100644
index 0000000..e06eddb
--- /dev/null
+++ b/tests/LocalFileTest.php
@@ -0,0 +1,36 @@
+<?php
+
+namespace vipnytt\SitemapParser\Tests;
+
+use PHPUnit\Framework\TestCase;
+use vipnytt\SitemapParser;
+
+class RecursiveTest extends TestCase {
+
+    public function testLocalFileXMLFile()
+    {
+        $parser = new SitemapParser('SitemapParser');
+        $this->assertInstanceOf('vipnytt\SitemapParser', $parser);
+
+        $tmpfname = tempnam(sys_get_temp_dir(), "sitemap_parser_test_file");
+        $fileContent = <<<XMLSITEMAP
+<?xml version="1.0" encoding="UTF-8"?>
+<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
+  <sitemap>
+    <loc>http://www.example.com/sitemap.xml</loc>
+    <lastmod>2004-10-01T18:23:17+00:00</lastmod>
+  </sitemap>
+</sitemapindex>
+XMLSITEMAP;
+        file_put_contents($tmpfname, $fileContent);
+        $parser->parse('file:///'.$tmpfname);
+        $this->assertEquals([
+            'http://www.example.com/sitemap.xml' => [
+                'loc' => 'http://www.example.com/sitemap.xml',
+                'lastmod' => '2004-10-01T18:23:17+00:00',
+                'namespaces' => [],
+            ],
+        ], $parser->getSitemaps());
+    }
+
+}