From 7fcfe05f8536a94189281b08f41d712b2421345a Mon Sep 17 00:00:00 2001
From: Filip Valder <fvalder@redhat.com>
Date: Fri, 9 Dec 2016 10:55:25 +0100
Subject: [PATCH] Fix issue #251:

- support for mupltiple SCM URLs (logic was broken when there were more
than one SCM URL in SCMURLS conf. var)

- unsplit-brain handling of SCM URLs between frontend and backend
---
 module_build_service/scm.py   | 27 ++++++++-------------------
 module_build_service/utils.py | 31 +++++++++++++++++++++++++++++++
 module_build_service/views.py |  5 +++--
 3 files changed, 42 insertions(+), 21 deletions(-)

diff --git a/module_build_service/scm.py b/module_build_service/scm.py
index 44570c29..09b8809d 100644
--- a/module_build_service/scm.py
+++ b/module_build_service/scm.py
@@ -45,23 +45,12 @@ class SCM(object):
     "SCM abstraction class"
 
     # Assuming git for HTTP schemas
-    types = {
-                "git": ("git://", "git+http://", "git+https://",
-                    "git+rsync://", "http://", "https://", "file://")
-            }
+    types = module_build_service.utils.scm_url_schemes()
 
     def __init__(self, url, allowed_scm=None, allow_local = False):
         """Initialize the SCM object using the specified scmurl.
 
         If url is not in the list of allowed_scm, an error will be raised.
-        NOTE: only git URLs in the following formats are supported atm:
-            git://
-            git+http://
-            git+https://
-            git+rsync://
-            http://
-            https://
-            file://
 
         :param str url: The unmodified scmurl
         :param list allowed_scm: The list of allowed SCMs, optional
@@ -69,16 +58,15 @@ class SCM(object):
         """
 
         if allowed_scm:
-            for allowed in allowed_scm:
-                if (url.startswith(allowed)
-                        or (allow_local and url.startswith("file://"))):
-                    break
-                else:
-                    raise Unauthorized(
-                        '%s is not in the list of allowed SCMs' % url)
+            if not (url.startswith(tuple(allowed_scm)) or
+                    (allow_local and url.startswith("file://"))):
+                raise Unauthorized(
+                    '%s is not in the list of allowed SCMs' % url)
 
         self.url = url
 
+        # once we have more than one SCM provider, we will need some more
+        # sophisticated lookup logic
         for scmtype, schemes in SCM.types.items():
             if self.url.startswith(schemes):
                 self.scheme = scmtype
@@ -86,6 +74,7 @@ class SCM(object):
         else:
             raise ValidationError('Invalid SCM URL: %s' % url)
 
+        # git is the only one supported SCM provider atm
         if self.scheme == "git":
             match = re.search(r"^(?P<repository>.*/(?P<name>[^?]*))(\?#(?P<commit>.*))?", url)
             self.repository = match.group("repository")
diff --git a/module_build_service/utils.py b/module_build_service/utils.py
index 023e3c59..5a14211f 100644
--- a/module_build_service/utils.py
+++ b/module_build_service/utils.py
@@ -494,3 +494,34 @@ def insert_fake_baseruntime():
     module.state = models.BUILD_STATES['done']
     module.state_reason = 'Artificially created.'
     db.session.commit()
+
+
+def scm_url_schemes(terse=False):
+    """
+    Definition of URL schemes supported by both frontend and scheduler.
+
+    NOTE: only git URLs in the following formats are supported atm:
+        git://
+        git+http://
+        git+https://
+        git+rsync://
+        http://
+        https://
+        file://
+
+    :param terse=False: Whether to return terse list of unique URL schemes
+                        even without the "://".
+    """
+
+    scm_types = {
+                "git": ("git://", "git+http://", "git+https://",
+                        "git+rsync://", "http://", "https://", "file://")
+            }
+
+    if not terse:
+        return scm_types
+    else:
+        scheme_list = []
+        for scm_type, scm_schemes in scm_types.items():
+            scheme_list.extend([scheme[:-3] for scheme in scm_schemes])
+        return list(set(scheme_list))
diff --git a/module_build_service/views.py b/module_build_service/views.py
index bd04e189..57c7af1b 100644
--- a/module_build_service/views.py
+++ b/module_build_service/views.py
@@ -36,7 +36,7 @@ from flask.views import MethodView
 
 from module_build_service import app, conf, log
 from module_build_service import models, db
-from module_build_service.utils import pagination_metadata, filter_module_builds, submit_module_build
+from module_build_service.utils import pagination_metadata, filter_module_builds, submit_module_build, scm_url_schemes
 from module_build_service.errors import (
     ValidationError, Unauthorized, NotFound)
 
@@ -115,8 +115,9 @@ class ModuleBuildAPI(MethodView):
             log.error("The submitted scmurl %r is not allowed" % url)
             raise Unauthorized("The submitted scmurl %s is not allowed" % url)
 
+        schemes_re = '|'.join(map(re.escape, scm_url_schemes(terse=True)))
         scmurl_re = re.compile(
-            r"(?P<giturl>(?:(?P<scheme>git)://(?P<host>[^/]+))?"
+            r"(?P<giturl>(?:(?P<scheme>(" + schemes_re + r"))://(?P<host>[^/]+))?"
             r"(?P<repopath>/[^\?]+))\?(?P<modpath>[^#]*)#(?P<revision>.+)")
         if not scmurl_re.match(url):
             log.error("The submitted scmurl %r is not valid" % url)