From 2914a74b751779327081b508ee104c3ed2919b58 Mon Sep 17 00:00:00 2001
From: Fabio Manganiello <info@fabiomanganiello.com>
Date: Sat, 27 Nov 2021 01:19:55 +0100
Subject: [PATCH] Replace relative links in converted markdown

---
 platypush/plugins/http/webpage/__init__.py | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/platypush/plugins/http/webpage/__init__.py b/platypush/plugins/http/webpage/__init__.py
index 02e54d343..23008890a 100644
--- a/platypush/plugins/http/webpage/__init__.py
+++ b/platypush/plugins/http/webpage/__init__.py
@@ -1,8 +1,10 @@
 import datetime
 import json
 import os
+import re
 import subprocess
 import tempfile
+from urllib.parse import urlparse
 
 from platypush.plugins import action
 from platypush.plugins.http.request import Plugin
@@ -29,6 +31,12 @@ class HttpWebpagePlugin(Plugin):
         with subprocess.Popen(proc, stdout=subprocess.PIPE, stderr=None) as parser:
             return parser.communicate()[0].decode()
 
+    @staticmethod
+    def _fix_relative_links(markdown: str, url: str) -> str:
+        url = urlparse(url)
+        base_url = f'{url.scheme}://{url.netloc}'
+        return re.sub(r'(!?\[.+?])\((/.+?)\)', f'\1({base_url}\2)', markdown)
+
     # noinspection PyShadowingBuiltins
     @action
     def simplify(self, url, type='html', html=None, outfile=None):
@@ -101,6 +109,9 @@ class HttpWebpagePlugin(Plugin):
         except Exception as e:
             raise RuntimeError('Could not parse JSON: {}. Response: {}'.format(str(e), response))
 
+        if type == 'markdown':
+            self._fix_relative_links(response['content'], url)
+
         self.logger.debug('Got response from Mercury API: {}'.format(response))
         title = response.get('title', '{} on {}'.format(
             'Published' if response.get('date_published') else 'Generated',