From 9afec1a034185dcbfbe25f47198ae6067fae3d40 Mon Sep 17 00:00:00 2001
From: Bernard Cafarelli <bernard.cafarelli@gmail.com>
Date: Thu, 5 Dec 2019 18:22:38 +0100
Subject: [PATCH] More python3 fixes

---
 pym/euscan/handlers/__init__.py | 2 +-
 pym/euscan/handlers/freecode.py | 6 +++---
 pym/euscan/handlers/generic.py  | 4 ++--
 3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/pym/euscan/handlers/__init__.py b/pym/euscan/handlers/__init__.py
index bdae6f8..8ee02b3 100644
--- a/pym/euscan/handlers/__init__.py
+++ b/pym/euscan/handlers/__init__.py
@@ -158,7 +158,7 @@ def scan_url(pkg, urls, options, on_progress=None):
             except Exception as e:
                 output.ewarn(
                     "Handler failed: [%s] %s" %
-                    (e.__class__.__name__, e.message)
+                    (e.__class__.__name__, str(e))
                 )
 
             if versions and CONFIG['oneshot']:
diff --git a/pym/euscan/handlers/freecode.py b/pym/euscan/handlers/freecode.py
index b6f5a70..94ec649 100644
--- a/pym/euscan/handlers/freecode.py
+++ b/pym/euscan/handlers/freecode.py
@@ -22,7 +22,7 @@ def scan_pkg(pkg, options):
     output.einfo("Using FreeCode handler: " + package)
 
     fp = urllib.request.urlopen("http://freecode.com/projects/%s/releases" % package)
-    content = fp.read()
+    content = str(fp.read())
 
     result = re.findall(
         r'<a href="/projects/%s/releases/(\d+)">([^<]+)</a>' % package,
@@ -36,10 +36,10 @@ def scan_pkg(pkg, options):
             continue
         fp = urllib.request.urlopen("http://freecode.com/projects/%s/releases/%s" %
                             (package, release_id))
-        content = fp.read()
+        content = str(fp.read())
         download_page = re.findall(r'<a href="(/urls/[^"]+)"', content)[0]
         fp = urllib.request.urlopen("http://freecode.com%s" % download_page)
-        content = fp.read()
+        content = str(fp.read())
         url = re.findall(
             r'In case it doesn\'t, click here: <a href="([^"]+)"',
             content
diff --git a/pym/euscan/handlers/generic.py b/pym/euscan/handlers/generic.py
index afa17fb..6d6f49e 100644
--- a/pym/euscan/handlers/generic.py
+++ b/pym/euscan/handlers/generic.py
@@ -52,7 +52,7 @@ def confidence_score(found, original, minimum=CONFIDENCE):
 
 
 def scan_html(data, url, pattern):
-    soup = BeautifulSoup(data)
+    soup = BeautifulSoup(data, features="lxml")
     results = []
 
     for link in soup.findAll('a'):
@@ -114,7 +114,7 @@ def scan_directory_recursive(cp, ver, rev, url, steps, orig_url, options):
 
     results = []
 
-    if re.search("<\s*a\s+[^>]*href", data, re.I):
+    if re.search(b"<\s*a\s+[^>]*href", data, re.I):
         results.extend(scan_html(data, url, pattern))
     elif url.startswith('ftp://'):
         results.extend(scan_ftp(data, url, pattern))