From e4816bc2eeff675d54aa84047a4a988d697949c9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Fran=C3=A7ois=20Poulain?= <fpoulain@metrodore.fr>
Date: Sat, 8 Aug 2020 18:24:27 +0200
Subject: [PATCH] feat(convert): ajoute les liens auto dans le filtrage html

---
 drupal2spip_lal/base/convert.py | 42 +++++++++++++++++++++++++++++++++
 1 file changed, 42 insertions(+)

diff --git a/drupal2spip_lal/base/convert.py b/drupal2spip_lal/base/convert.py
index b53c046..76aaa96 100644
--- a/drupal2spip_lal/base/convert.py
+++ b/drupal2spip_lal/base/convert.py
@@ -118,7 +118,49 @@ def filter_html(html):
         except Exception as e:
             raise ValueError("Echec de auto_p: {}".format(e))
 
+    def auto_a(html):
+        soup = bs(html, 'html.parser')
+
+        email_pattern = re.compile(r'(\b[\w\.\+_-]+@(\w+\.)+\w+\b)')
+        for line in soup.find_all(string=email_pattern):
+            if line.parent.name == 'a':
+                continue
+            a_string = email_pattern.sub(
+                r'<a class= "auto-a" href="mailto:\1">\1</a>', line.string
+            )
+            a_soup = bs(a_string, 'html.parser')
+            line.replace_with(a_soup)
+
+        protocols = [
+            'http',
+            'https',
+            'ftp',
+            'news',
+            'nntp',
+            'tel',
+            'telnet',
+            'mailto',
+            'irc',
+            'ssh',
+            'sftp',
+            'webcal',
+            'rtsp',
+        ]
+        link_pattern = re.compile(
+            r'((\b({})s?)?://(\w+\.)+\w+/?[^\s]*)'.format('|'.join(protocols))
+        )
+        for line in soup.find_all(string=link_pattern):
+            if line.parent.name == 'a':
+                continue
+            a_string = link_pattern.sub(
+                r'<a class="auto-a" href="\1">\1</a>', line.string
+            )
+            a_soup = bs(a_string, 'html.parser')
+            line.replace_with(a_soup)
+        return str(soup)
+
     html = auto_p(html)
+    html = auto_a(html)
     return html