euscan: generic handler fixes

* urljoin was broken with dirs ("http://site.com/lol" joined with "wat"
  produces http://site.com/wat)
* fixed _v regex, don't match what is not needed

Signed-off-by: volpino <fox91@anche.no>

2 files changed, 12 insertions, 5 deletions

diff --git a/pym/euscan/handlers/generic.py b/pym/euscan/handlers/generic.py
index 831dfce..12c4a99 100644
--- a/pym/euscan/handlers/generic.py
+++ b/pym/euscan/handlers/generic.py
@@ -35,7 +35,10 @@ def scan_html(data, url, pattern):
 
         match = re.match(pattern, href, re.I)
         if match:
-            results.append((".".join(match.groups()), match.group(0)))
+            results.append(
+                (".".join([x for x in match.groups() if x is not None]),
+                 match.group(0))
+            )
     return results
 
 
@@ -47,8 +50,10 @@ def scan_ftp(data, url, pattern):
         line = line.replace("\n", "").replace("\r", "")
         match = re.search(pattern, line, re.I)
         if match:
-            results.append((".".join(match.groups()), match.group(0)))
-
+            results.append(
+                (".".join([x for x in match.groups() if x is not None]),
+                 match.group(0))
+            )
     return results
 
 
@@ -88,6 +93,8 @@ def scan_directory_recursive(cp, ver, rev, url, steps, orig_url):
         pv = helpers.gentoo_mangle_version(up_pv)
         if helpers.version_filtered(cp, ver, pv):
             continue
+        if not url.endswith("/"):
+            url = url + "/"
         path = urljoin(url, path)
 
         if not steps and path not in orig_url:
diff --git a/pym/euscan/helpers.py b/pym/euscan/helpers.py
index 6582393..ec721b7 100644
--- a/pym/euscan/helpers.py
+++ b/pym/euscan/helpers.py
@@ -30,8 +30,8 @@ VERSION_CMP_PACKAGE_QUIRKS = {
     'sys-process/htop': htop_vercmp
 }
 
-_v_end = '((-|_)(pre|p|beta|b|alpha|a|rc|r)\d*)'
-_v = r'((\d+)((\.\d+)*)([a-zA-Z]*?)(' + _v_end + '*))'
+_v_end = r'(?:(?:-|_)(?:pre|p|beta|b|alpha|a|rc|r)\d*)'
+_v = r'((?:\d+)(?:(?:\.\d+)*)(?:[a-zA-Z]*?)(?:' + _v_end + '*))'
 
 
 # Stolen from g-pypi