New Validator: Git URL

tests/test_git.py

@@ -0,0 +1,163 @@
+# -*- coding: utf-8 -*-
+import pytest
+
+from validators import url, ValidationFailure
+
+
+@pytest.mark.parametrize('address', [
+    u'[email protected]:username/reponame.git',
+    u'https://[email protected]/otherusername/reponame.git',
+    u'https://github.com/username/reponame.git',
+    u'[email protected]:username/reponame.git',
+    u'[email protected]:groupname/reponame.git',
+    u'https://gitlab.com/groupname/reponame.git',
+    u'[email protected]:workspace/reponame.git',
+    u'https://[email protected]/workspace/reponame.git',
+    u'https://bitbucket.org/workspace/reponame.git',
+    u'[email protected]:username/reponame.git',
+    u'https://gitlab.com/username/reponame.git',
+    u'git://github.com/somthing/somthing.git#ff786f9f',
+    u'git://github.com/somthing/somthing.git#gh-pages',
+    u'git://github.com/somthing/somthing.git#master',
+    u'git://github.com/somthing/somthing.git#Quick-Fix',
+    u'git://github.com/somthing/somthing.git#quick_fix',
+    u'git://github.com/somthing/somthing.git#v0.1.0',
+    u'git://host.xz/path/to/repo.git/',
+    u'git://host.xz/~user/path/to/repo.git/',
+    u'[email protected]:user/project.git',
+    u'[email protected]:user/project.git',
+    u'[email protected]:user/some-project.git',
+    u'[email protected]:user/some-project.git',
+    u'[email protected]:user/some_project.git',
+    u'[email protected]:user/some_project.git',
+    u'http://github.com/user/project.git',
+    u'http://host.xz/path/to/repo.git/',
+    u'https://github.com/user/project.git',
+    u'https://host.xz/path/to/repo.git/',
+    u'https://username::;*%$:@github.com/username/repository.git',
+    u'https://username:$fooABC@:@github.com/username/repository.git',
+    u'https://username:[email protected]/username/repository.git',
+    u'ssh://host.xz/path/to/repo.git/',
+    u'ssh://host.xz/path/to/repo.git/',
+    u'ssh://host.xz/~/path/to/repo.git',
+    u'ssh://host.xz/~user/path/to/repo.git/',
+    u'ssh://[email protected]/path/to/repo.git/',
+    u'ssh://[email protected]/path/to/repo.git/',
+    u'ssh://[email protected]/~/path/to/repo.git',
+    u'ssh://[email protected]/~user/path/to/repo.git/',
+    u'ssh://[email protected]:port/path/to/repo.git/',
+    u'https://[email protected]:8080/otherusername/reponame.git',
+    u'ssh://host.xz:port/path/to/repo.git/',
+    u'ssh://host.xz:1234/path/to/repo.git/',
+    u'https://192.168.1.127/user/project.git',
+    u'http://192.168.1.127/user/project.git',
+    u'http://192.168.1.127:port/user/project.git',
+])
+def test_returns_true_on_valid_url(address):
+    assert url(address)
+
+
+@pytest.mark.parametrize('address', [
+    u'[email protected]:username/reponame',
+    u'https://[email protected]/otherusername/reponame',
+    u'https://github.com/username/reponame',
+    u'[email protected]:username/reponame',
+    u'[email protected]:groupname/reponame',
+    u'https://gitlab.com/groupname/reponame',
+    u'[email protected]:workspace/reponame',
+    u'https://[email protected]/workspace/reponame',
+    u'https://bitbucket.org/workspace/reponame',
+    u'[email protected]:username/reponame',
+    u'https://gitlab.com/username/reponame',
+    u'git://host.xz/path/to/repo',
+    u'git://host.xz/~user/path/to/repo.',
+    u'[email protected]:user/project',
+    u'[email protected]:user/project',
+    u'[email protected]:user/some-project',
+    u'[email protected]:user/some-project',
+    u'[email protected]:user/some_project',
+    u'[email protected]:user/some_project',
+    u'http://github.com/user/project',
+    u'http://host.xz/path/to/repo.',
+    u'https://github.com/user/project',
+    u'https://host.xz/path/to/repo',
+    u'https://username::;*%$:@github.com/username/repository',
+    u'https://username:$fooABC@:@github.com/username/repository',
+    u'https://username:[email protected]/username/repository',
+    u'ssh://host.xz/path/to/repo',
+    u'ssh://host.xz/path/to/repo',
+    u'ssh://host.xz/~/path/to/rep',
+    u'ssh://host.xz/~user/path/to/repo',
+    u'ssh://[email protected]/path/to/repo',
+    u'ssh://[email protected]/path/to/repo',
+    u'ssh://[email protected]/~/path/to/repo',
+    u'ssh://[email protected]/~user/path/to/repo',
+    u'ssh://[email protected]:port/path/to/repo',
+    u'https://[email protected]:8080/otherusername/reponame',
+    u'ssh://host.xz:port/path/to/repo',
+    u'ssh://host.xz:1234/path/to/repo',
+    u'https://192.168.1.127/user/project',
+    u'http://192.168.1.127/user/project',
+    u'http://192.168.1.127:port/user/project',
+])
+def test_returns_true_on_valid_loose_url(address):
+    assert url(address, strict=False)
+
+
+@pytest.mark.parametrize('address', 'strict', [
+    u'https://www.github.com',
+    u'[email protected]:username/repo name',
+    u'[email protected]:username/repo name.git',
+    u'https://gitlab.com/username/reponame.git -b branch',
+    u'https://gitlab.com/username/reponame.git --flag',
+    u'https://gitlab.com/username/reponame.git \\-f',
+    u'/path/to/repo.git/',
+    u'file:///path/to/repo.git/',
+    u'file://~/path/to/repo.git/',
+    u'[email protected]:user/some_project.git/foo',
+    u'[email protected]:user/some_project.gitfoo',
+    u'host.xz:/path/to/repo.git/',
+    u'host.xz:path/to/repo.git',
+    u'host.xz:path/to/repo',
+    u'host.xz:~user/path/to/repo.git/',
+    u'path/to/repo.git/',
+    u'rsync://host.xz/path/to/repo.git/',
+    u'[email protected]:/path/to/repo.git/',
+    u'[email protected]:path/to/repo.git',
+    u'[email protected]:~user/path/to/repo.git/',
+    u'~/path/to/repo.git',
+    u'[email protected]:username/reponame.gi',
+])
+def test_returns_failed_loose_validation_on_invalid_url(address):
+    assert isinstance(url(address), ValidationFailure)
+
+
+@pytest.mark.parametrize('address', [
+    u'https://www.github.com',
+    u'[email protected]:username/"repo name".git',
+    u'https://gitlab.com/username/"reponame".git',
+    u'https://gitlab.com/username/\'reponame\'.git',
+    u'[email protected]:username/repo name',
+    u'[email protected]:username/repo name.git',
+    u'https://gitlab.com/username/reponame.git -b branch',
+    u'https://gitlab.com/username/reponame.git --flag',
+    u'https://gitlab.com/username/reponame.git \\-f',
+    u'/path/to/repo.git/',
+    u'file:///path/to/repo.git/',
+    u'file://~/path/to/repo.git/',
+    u'[email protected]:user/some_project.git/foo',
+    u'[email protected]:user/some_project.gitfoo',
+    u'host.xz:/path/to/repo.git/',
+    u'host.xz:path/to/repo.git',
+    u'host.xz:path/to/repo',
+    u'host.xz:~user/path/to/repo.git/',
+    u'path/to/repo.git/',
+    u'rsync://host.xz/path/to/repo.git/',
+    u'[email protected]:/path/to/repo.git/',
+    u'[email protected]:path/to/repo.git',
+    u'[email protected]:~user/path/to/repo.git/',
+    u'~/path/to/repo.git',
+    u'[email protected]:username/reponame.gi',
+])
+def test_returns_failed_strict_validation_on_invalid_url(address):
+    assert isinstance(url(address, strict=True), ValidationFailure)

validators/git.py

@@ -0,0 +1,70 @@
+import re
+
+from .utils import validator
+
+"""
+sources:
+https://github.com/git/git/blob/master/url.c
+https://git-scm.com/docs/git-clone
+https://stackoverflow.com/questions/23976019/how-to-verify-valid-format-of-url-as-a-git-repo/60000569#comment111902087_60000569
+https://github.com/jonschlinkert/is-git-url
+"""
+url_regex = re.compile(
+    # protocol
+    # either prot:// (protocol captured as group 1) or git@serv (server captured as group 2)
+    r"(?:(git|ssh|https?)|git@([A-Za-z0-9][A-Za-z0-9\+\.\-_]+)):(?:\/\/)?"
+    # username and maybe password, if either provided (both are captured as group 3)
+    r"(?:(\S*)@)?"
+    # the actual targeted URL as per STD66 (RFC3986) [A-Za-z][A-Za-z0-9+.-]* 
+    # with some added leniency for IPs and allowed URI special chars (_~/:) in order to capture full path+ports
+    # (entire URI captured as group 4)
+    r"([A-Za-z0-9][A-Za-z0-9\+\.\-_\/\~\:]*?)"
+    # .git matched if provided
+    r"(?:\.git)"
+    # ending '/' OR branch name/commit id (captured as group 5)
+    r"(\/?|\#[-\d\w._]+?)",
+    re.UNICODE | re.IGNORECASE
+)
+
+
+@validator
+def git(value: str, strict: bool = False):
+    """
+    Return whether a given value is a valid git URL.
+
+    If the value is valid git URL this function returns ``True``, otherwise
+    :class:`~validators.utils.ValidationFailure`.
+
+    Examples::
+
+        >>> git('[email protected]:username/reponame.git')
+        True
+
+        >>> git('[email protected]:username/reponame')
+        True
+
+        >>> git('http://192.168.1.127/user/project.git')
+        True
+
+        >>> git('git://bitbucket.org/org/repo.git#ff786f9f')
+        True
+
+        >>> git('https://www.github.com')
+        ValidationFailure(func=git, ...)
+
+        >>> git('[email protected]:username/reponame', strict=True)
+        ValidationFailure(func=git, ...)
+
+    :param value: URL address string to validate
+    :param strict: (default=False) enfocre URL end with .git (if false, upon validation failure '.git; is added
+    to the end if the procided value and validation retried)
+    """
+    result = url_regex.match(value)
+
+    if result or strict:
+        return result
+    else: # not strict and failed to match
+        # try adding ".git" at end of value and retry
+        value = value[:-1] if value.endswith('/') else value
+        result = url_regex.match(value + ".git")
+        return result