Merge pull request #3501 from stsewd/fix-integrations-regex

Fix regex for getting project and user
readthedocs · Jan 15, 2018 · bf23b2b · bf23b2b
2 parents 909cb56 + a7b8a2e
commit bf23b2b
Show file tree

Hide file tree

Showing 4 changed files with 84 additions and 49 deletions.
diff --git a/readthedocs/builds/models.py b/readthedocs/builds/models.py
@@ -21,15 +21,17 @@
 
 from readthedocs.core.utils import broadcast
 from readthedocs.projects.constants import (
-    BITBUCKET_REGEXS, BITBUCKET_URL, GITHUB_REGEXS, GITHUB_URL, GITLAB_REGEXS,
-    GITLAB_URL, PRIVACY_CHOICES, PRIVATE)
+    BITBUCKET_URL, GITHUB_URL, GITLAB_URL, PRIVACY_CHOICES, PRIVATE)
 from readthedocs.projects.models import APIProject, Project
 
 from .constants import (
     BRANCH, BUILD_STATE, BUILD_STATE_FINISHED, BUILD_TYPES, LATEST,
     NON_REPOSITORY_VERSIONS, STABLE, TAG, VERSION_TYPES)
 from .managers import VersionManager
 from .querysets import BuildQuerySet, RelatedBuildQuerySet, VersionQuerySet
+from .utils import (
+    get_bitbucket_username_repo, get_github_username_repo,
+    get_gitlab_username_repo)
 from .version_slug import VersionSlugField
 
 DEFAULT_VERSION_PRIVACY_LEVEL = getattr(
@@ -277,12 +279,8 @@ def get_github_url(
         elif action == 'edit':
             action_string = 'edit'
 
-        for regex in GITHUB_REGEXS:
-            match = regex.search(repo_url)
-            if match:
-                user, repo = match.groups()
-                break
-        else:
+        user, repo = get_github_username_repo(repo_url)
+        if not user and not repo:
             return ''
         repo = repo.rstrip('/')
 
@@ -315,12 +313,8 @@ def get_gitlab_url(
         elif action == 'edit':
             action_string = 'edit'
 
-        for regex in GITLAB_REGEXS:
-            match = regex.search(repo_url)
-            if match:
-                user, repo = match.groups()
-                break
-        else:
+        user, repo = get_gitlab_username_repo(repo_url)
+        if not user and not repo:
             return ''
         repo = repo.rstrip('/')
 
@@ -341,12 +335,8 @@ def get_bitbucket_url(self, docroot, filename, source_suffix='.rst'):
         if not docroot:
             return ''
 
-        for regex in BITBUCKET_REGEXS:
-            match = regex.search(repo_url)
-            if match:
-                user, repo = match.groups()
-                break
-        else:
+        user, repo = get_bitbucket_username_repo(repo_url)
+        if not user and not repo:
             return ''
         repo = repo.rstrip('/')
 

diff --git a/readthedocs/builds/utils.py b/readthedocs/builds/utils.py
@@ -4,32 +4,13 @@
 from __future__ import (
     absolute_import, division, print_function, unicode_literals)
 
-import re
-
-GH_REGEXS = [
-    re.compile('github.com/(.+)/(.+)(?:\.git){1}'),
-    re.compile('github.com/(.+)/(.+)'),
-    re.compile('github.com:(.+)/(.+).git'),
-]
-
-BB_REGEXS = [
-    re.compile('bitbucket.org/(.+)/(.+)/'),
-    re.compile('bitbucket.org/(.+)/(.+)'),
-    re.compile('bitbucket.org:(.+)/(.+)\.git'),
-]
-
-# TODO: I think this can be different than `gitlab.com`
-# self.adapter.provider_base_url
-GL_REGEXS = [
-    re.compile('gitlab.com/(.+)/(.+)(?:\.git){1}'),
-    re.compile('gitlab.com/(.+)/(.+)'),
-    re.compile('gitlab.com:(.+)/(.+)\.git'),
-]
+from readthedocs.projects.constants import (
+    BITBUCKET_REGEXS, GITHUB_REGEXS, GITLAB_REGEXS)
 
 
 def get_github_username_repo(url):
     if 'github' in url:
-        for regex in GH_REGEXS:
+        for regex in GITHUB_REGEXS:
             match = regex.search(url)
             if match:
                 return match.groups()
@@ -38,7 +19,7 @@ def get_github_username_repo(url):
 
 def get_bitbucket_username_repo(url=None):
     if 'bitbucket' in url:
-        for regex in BB_REGEXS:
+        for regex in BITBUCKET_REGEXS:
             match = regex.search(url)
             if match:
                 return match.groups()
@@ -47,7 +28,7 @@ def get_bitbucket_username_repo(url=None):
 
 def get_gitlab_username_repo(url=None):
     if 'gitlab' in url:
-        for regex in GL_REGEXS:
+        for regex in GITLAB_REGEXS:
             match = regex.search(url)
             if match:
                 return match.groups()

diff --git a/readthedocs/projects/constants.py b/readthedocs/projects/constants.py
@@ -292,19 +292,20 @@
 PROJECT_SLUG_REGEX = '(?:[-\w]+)'
 
 GITHUB_REGEXS = [
-    re.compile('github.com/(.+)/(.+)(?:\.git){1}'),
+    re.compile('github.com/(.+)/(.+)(?:\.git){1}$'),
     re.compile('github.com/(.+)/(.+)'),
-    re.compile('github.com:(.+)/(.+).git'),
+    re.compile('github.com:(.+)/(.+)\.git$'),
 ]
 BITBUCKET_REGEXS = [
-    re.compile('bitbucket.org/(.+)/(.+).git'),
+    re.compile('@bitbucket.org/(.+)/(.+)\.git$'),
     re.compile('bitbucket.org/(.+)/(.+)/'),
     re.compile('bitbucket.org/(.+)/(.+)'),
+    re.compile('bitbucket.org:(.+)/(.+)\.git$'),
 ]
 GITLAB_REGEXS = [
-    re.compile('gitlab.com/(.+)/(.+)(?:\.git){1}'),
+    re.compile('gitlab.com/(.+)/(.+)(?:\.git){1}$'),
     re.compile('gitlab.com/(.+)/(.+)'),
-    re.compile('gitlab.com:(.+)/(.+).git'),
+    re.compile('gitlab.com:(.+)/(.+)\.git$'),
 ]
 GITHUB_URL = (
     'https://github.com/{user}/{repo}/'

diff --git a/readthedocs/rtd_tests/tests/test_repo_parsing.py b/readthedocs/rtd_tests/tests/test_repo_parsing.py
@@ -22,26 +22,89 @@ def test_github(self):
         self.pip.repo = 'https://github.com/user/repo/'
         self.assertEqual(self.version.get_github_url(docroot='/docs/', filename='file'), 'https://github.com/user/repo/blob/master/docs/file.rst')
 
+        self.pip.repo = 'https://github.com/user/repo.github.io'
+        self.assertEqual(self.version.get_github_url(docroot='/docs/', filename='file'), 'https://github.com/user/repo.github.io/blob/master/docs/file.rst')
+
+        self.pip.repo = 'https://github.com/user/repo.github.io/'
+        self.assertEqual(self.version.get_github_url(docroot='/docs/', filename='file'), 'https://github.com/user/repo.github.io/blob/master/docs/file.rst')
+
         self.pip.repo = 'https://github.com/user/repo.git'
         self.assertEqual(self.version.get_github_url(docroot='/docs/', filename='file'), 'https://github.com/user/repo/blob/master/docs/file.rst')
 
+        self.pip.repo = 'https://github.com/user/repo.github.io.git'
+        self.assertEqual(self.version.get_github_url(docroot='/docs/', filename='file'), 'https://github.com/user/repo.github.io/blob/master/docs/file.rst')
+
+        self.pip.repo = 'https://github.com/user/repo.git.git'
+        self.assertEqual(self.version.get_github_url(docroot='/docs/', filename='file'), 'https://github.com/user/repo.git/blob/master/docs/file.rst')
+
+    def test_github_ssh(self):
+        self.pip.repo = 'git@github.com:user/repo.git'
+        self.assertEqual(self.version.get_github_url(docroot='/docs/', filename='file'), 'https://github.com/user/repo/blob/master/docs/file.rst')
+
+        self.pip.repo = 'git@github.com:user/repo.github.io.git'
+        self.assertEqual(self.version.get_github_url(docroot='/docs/', filename='file'), 'https://github.com/user/repo.github.io/blob/master/docs/file.rst')
+
     def test_gitlab(self):
         self.pip.repo = 'https://gitlab.com/user/repo'
         self.assertEqual(self.version.get_gitlab_url(docroot='/foo/bar/', filename='file'), 'https://gitlab.com/user/repo/blob/master/foo/bar/file.rst')
 
         self.pip.repo = 'https://gitlab.com/user/repo/'
         self.assertEqual(self.version.get_gitlab_url(docroot='/foo/bar/', filename='file'), 'https://gitlab.com/user/repo/blob/master/foo/bar/file.rst')
 
+        self.pip.repo = 'https://gitlab.com/user/repo.gitlab.io'
+        self.assertEqual(self.version.get_gitlab_url(docroot='/foo/bar/', filename='file'), 'https://gitlab.com/user/repo.gitlab.io/blob/master/foo/bar/file.rst')
+
+        self.pip.repo = 'https://gitlab.com/user/repo.gitlab.io/'
+        self.assertEqual(self.version.get_gitlab_url(docroot='/foo/bar/', filename='file'), 'https://gitlab.com/user/repo.gitlab.io/blob/master/foo/bar/file.rst')
+
         self.pip.repo = 'https://gitlab.com/user/repo.git'
         self.assertEqual(self.version.get_gitlab_url(docroot='/foo/bar/', filename='file'), 'https://gitlab.com/user/repo/blob/master/foo/bar/file.rst')
 
+        self.pip.repo = 'https://gitlab.com/user/repo.gitlab.io.git'
+        self.assertEqual(self.version.get_gitlab_url(docroot='/foo/bar/', filename='file'), 'https://gitlab.com/user/repo.gitlab.io/blob/master/foo/bar/file.rst')
+
+        self.pip.repo = 'https://gitlab.com/user/repo.git.git'
+        self.assertEqual(self.version.get_gitlab_url(docroot='/foo/bar/', filename='file'), 'https://gitlab.com/user/repo.git/blob/master/foo/bar/file.rst')
+
+    def test_gitlab_ssh(self):
+        self.pip.repo = 'git@gitlab.com:user/repo.git'
+        self.assertEqual(self.version.get_gitlab_url(docroot='/foo/bar/', filename='file'), 'https://gitlab.com/user/repo/blob/master/foo/bar/file.rst')
+
+        self.pip.repo = 'git@gitlab.com:user/repo.gitlab.io.git'
+        self.assertEqual(self.version.get_gitlab_url(docroot='/foo/bar/', filename='file'), 'https://gitlab.com/user/repo.gitlab.io/blob/master/foo/bar/file.rst')
+
     def test_bitbucket(self):
         self.pip.repo = 'https://bitbucket.org/user/repo'
         self.assertEqual(self.version.get_bitbucket_url(docroot='/foo/bar/', filename='file'), 'https://bitbucket.org/user/repo/src/master/foo/bar/file.rst')
 
         self.pip.repo = 'https://bitbucket.org/user/repo/'
         self.assertEqual(self.version.get_bitbucket_url(docroot='/foo/bar/', filename='file'), 'https://bitbucket.org/user/repo/src/master/foo/bar/file.rst')
 
+        self.pip.repo = 'https://bitbucket.org/user/repo.gitbucket.io'
+        self.assertEqual(self.version.get_bitbucket_url(docroot='/foo/bar/', filename='file'), 'https://bitbucket.org/user/repo.gitbucket.io/src/master/foo/bar/file.rst')
+
+        self.pip.repo = 'https://bitbucket.org/user/repo.gitbucket.io/'
+        self.assertEqual(self.version.get_bitbucket_url(docroot='/foo/bar/', filename='file'), 'https://bitbucket.org/user/repo.gitbucket.io/src/master/foo/bar/file.rst')
+
         self.pip.repo = 'https://bitbucket.org/user/repo.git'
+        self.assertEqual(self.version.get_bitbucket_url(docroot='/foo/bar/', filename='file'), 'https://bitbucket.org/user/repo.git/src/master/foo/bar/file.rst')
+
+        self.pip.repo = 'https://bitbucket.org/user/repo.gitbucket.io.git'
+        self.assertEqual(self.version.get_bitbucket_url(docroot='/foo/bar/', filename='file'), 'https://bitbucket.org/user/repo.gitbucket.io.git/src/master/foo/bar/file.rst')
+
+        self.pip.repo = 'https://bitbucket.org/user/repo.git.git'
+        self.assertEqual(self.version.get_bitbucket_url(docroot='/foo/bar/', filename='file'), 'https://bitbucket.org/user/repo.git.git/src/master/foo/bar/file.rst')
+
+    def test_bitbucket_https(self):
+        self.pip.repo = 'https://user@bitbucket.org/user/repo.git'
+        self.assertEqual(self.version.get_bitbucket_url(docroot='/foo/bar/', filename='file'), 'https://bitbucket.org/user/repo/src/master/foo/bar/file.rst')
+
+        self.pip.repo = 'https://user@bitbucket.org/user/repo.gitbucket.io.git'
+        self.assertEqual(self.version.get_bitbucket_url(docroot='/foo/bar/', filename='file'), 'https://bitbucket.org/user/repo.gitbucket.io/src/master/foo/bar/file.rst')
+
+    def test_bitbucket_ssh(self):
+        self.pip.repo = 'git@bitbucket.org:user/repo.git'
         self.assertEqual(self.version.get_bitbucket_url(docroot='/foo/bar/', filename='file'), 'https://bitbucket.org/user/repo/src/master/foo/bar/file.rst')
 
+        self.pip.repo = 'git@bitbucket.org:user/repo.gitbucket.io.git'
+        self.assertEqual(self.version.get_bitbucket_url(docroot='/foo/bar/', filename='file'), 'https://bitbucket.org/user/repo.gitbucket.io/src/master/foo/bar/file.rst')