From 9dca5e16376ab8a1115feffec843f06451f9b372 Mon Sep 17 00:00:00 2001
From: Martin Bornhold <martin@rhodecode.com>
Date: Thu, 22 Sep 2016 15:33:11 +0200
Subject: [PATCH] shadow: Use only a single regular expression to generate and
 match repo/group slugs.

---
 rhodecode/lib/middleware/simplevcs.py | 13 +++++-----
 rhodecode/lib/utils.py                | 37 +++++++++------------------
 2 files changed, 18 insertions(+), 32 deletions(-)

diff --git a/rhodecode/lib/middleware/simplevcs.py b/rhodecode/lib/middleware/simplevcs.py
index b0e3f6b0..1a438ca0 100644
--- a/rhodecode/lib/middleware/simplevcs.py
+++ b/rhodecode/lib/middleware/simplevcs.py
@@ -112,14 +112,13 @@ class SimpleVCS(object):
         vcs_repo_name and is_shadow_repo on the current instance.
         """
         # TODO: martinb: Move to class or module scope.
-        # TODO: martinb: Check if we have to use re.UNICODE.
-        # TODO: martinb: Check which chars are allowed for repo/group names.
-        #       These chars are excluded: '`?=[]\;\'"<>,/~!@#$%^&*()+{}|: '
-        #       Code from: rhodecode/lib/utils.py:repo_name_slug()
+        from rhodecode.lib.utils import SLUG_RE
         pr_regex = re.compile(
-            '(?P<base_name>(?:[\w-]+)(?:/[\w-]+)*)/'    # repo groups
-            '(?P<repo_name>[\w-]+)'                     # target repo name
-            '/pull-request/(?P<pr_id>\d+)/repository')  # pr suffix
+            '(?P<groups>(?:{slug_pat})(?:/{slug_pat})*)'  # repo groups
+            '/(?P<target>{slug_pat})'                     # target repo
+            '/pull-request/(?P<pr_id>\d+)'                # pull request
+            '/repository$'                                # shadow repo
+            .format(slug_pat=SLUG_RE.pattern))
 
         # Get url repo name from environment.
         self.url_repo_name = self._get_repository_name(environ)
diff --git a/rhodecode/lib/utils.py b/rhodecode/lib/utils.py
index 54dae19e..48e47f23 100644
--- a/rhodecode/lib/utils.py
+++ b/rhodecode/lib/utils.py
@@ -58,42 +58,29 @@ log = logging.getLogger(__name__)
 
 REMOVED_REPO_PAT = re.compile(r'rm__\d{8}_\d{6}_\d{6}__.*')
 
+# String of characters which are not allowed in repo/group slugs.
+SLUG_BAD_CHARS = re.escape('`?=[]\;\'"<>,/~!@#$%^&*()+{}|:')
+# Regex that matches forbidden characters in repo/group slugs.
+SLUG_BAD_CHAR_RE = re.compile('[{}]'.format(SLUG_BAD_CHARS))
+# Regex that matches allowed characters in repo/group slugs.
+SLUG_GOOD_CHAR_RE = re.compile('[^{}]'.format(SLUG_BAD_CHARS))
+# Regex that matches whole repo/group slugs.
+SLUG_RE = re.compile('[^{}]+'.format(SLUG_BAD_CHARS))
+
 _license_cache = None
 
 
-def recursive_replace(str_, replace=' '):
-    """
-    Recursive replace of given sign to just one instance
-
-    :param str_: given string
-    :param replace: char to find and replace multiple instances
-
-    Examples::
-    >>> recursive_replace("Mighty---Mighty-Bo--sstones",'-')
-    'Mighty-Mighty-Bo-sstones'
-    """
-
-    if str_.find(replace * 2) == -1:
-        return str_
-    else:
-        str_ = str_.replace(replace * 2, replace)
-        return recursive_replace(str_, replace)
-
-
 def repo_name_slug(value):
     """
     Return slug of name of repository
     This function is called on each creation/modification
     of repository to prevent bad names in repo
     """
+    replacement_char = '-'
 
     slug = remove_formatting(value)
-    slug = strip_tags(slug)
-
-    for c in """`?=[]\;'"<>,/~!@#$%^&*()+{}|: """:
-        slug = slug.replace(c, '-')
-    slug = recursive_replace(slug, '-')
-    slug = collapse(slug, '-')
+    slug = SLUG_BAD_CHAR_RE.sub(replacement_char, slug)
+    slug = collapse(slug, replacement_char)
     return slug