Fix gitlab regexp to match more samples (#52)

* Add parsing variable for user to gitlab parser * Fix gitlab selector
nephila · Apr 25, 2023 · fc819d9 · fc819d9
1 parent 8987fc8
commit fc819d9
Show file tree

Hide file tree

Showing 5 changed files with 102 additions and 5 deletions.
diff --git a/changes/47.feature b/changes/47.feature
@@ -0,0 +1 @@
+Add parsing variable for user to gitlab parser
diff --git a/giturlparse/parser.py b/giturlparse/parser.py
@@ -35,7 +35,7 @@ def parse(url, check_domain=True):
 
             # Skip if not matched
             if not match:
-                # print("[%s] URL: %s dit not match %s" % (name, url, regex.pattern))
+                print("[{}] URL: {} dit not match {}".format(name, url, regex.pattern))
                 continue
 
             # Skip if domain is bad

diff --git a/giturlparse/platforms/base.py b/giturlparse/platforms/base.py
@@ -35,6 +35,7 @@ def __init__(self):
 
     @staticmethod
     def clean_data(data):
+        print(data)
         data["path"] = ""
         data["branch"] = ""
         data["protocols"] = list(filter(lambda x: x, data.get("protocols", "").split("+")))

diff --git a/giturlparse/platforms/gitlab.py b/giturlparse/platforms/gitlab.py
@@ -4,20 +4,20 @@
 class GitLabPlatform(BasePlatform):
     PATTERNS = {
         "https": (
-            r"(?P<protocols>(git\+)?(?P<protocol>https))://(?P<domain>.+?)(?P<port>:[0-9]+)?"
+            r"(?P<protocols>(git\+)?(?P<protocol>https))://(?P<domain>[^:/]+)(?P<port>:[0-9]+)?"
             r"(?P<pathname>/(?P<owner>[^/]+?)/"
             r"(?P<groups_path>.*?)?(?(groups_path)/)?(?P<repo>[^/]+?)(?:(\.git)?(/)?)"
             r"(?P<path_raw>(/blob/|/-/tree/).+)?)$"
         ),
         "ssh": (
-            r"(?P<protocols>(git\+)?(?P<protocol>ssh))?(://)?git@(?P<domain>.+?):(?P<port>[0-9]+)?(?(port))?"
+            r"(?P<protocols>(git\+)?(?P<protocol>ssh))?(://)?(?P<_user>.+?)@(?P<domain>[^:/]+)(:)?(?P<port>[0-9]+)?(?(port))?"
             r"(?P<pathname>/?(?P<owner>[^/]+)/"
             r"(?P<groups_path>.*?)?(?(groups_path)/)?(?P<repo>[^/]+?)(?:(\.git)?(/)?)"
             r"(?P<path_raw>(/blob/|/-/tree/).+)?)$"
         ),
         "git": (
-            r"(?P<protocols>(?P<protocol>git))://(?P<domain>.+?):(?P<port>[0-9]+)?(?(port))?"
-            r"(?P<pathname>/?(?P<owner>[^/]+)/"
+            r"(?P<protocols>(?P<protocol>git))://(?P<domain>[^:/]+):?(?P<port>[0-9]+)?(?(port))?"
+            r"(?P<pathname>/(?P<owner>[^/]+?)/"
             r"(?P<groups_path>.*?)?(?(groups_path)/)?(?P<repo>[^/]+?)(?:(\.git)?(/)?)"
             r"(?P<path_raw>(/blob/|/-/tree/).+)?)$"
         ),

diff --git a/giturlparse/tests/test_parse.py b/giturlparse/tests/test_parse.py
@@ -500,6 +500,29 @@
             },
         ),
     ),
+    (
+        "GIT",
+        (
+            "git://host.org/Org/Group/subGroup/Repo.git/blob/master/giturlparse/github.py",
+            {
+                "host": "host.org",
+                "resource": "host.org",
+                "user": "git",
+                "port": "",
+                "owner": "Org",
+                "repo": "Repo",
+                "name": "Repo",
+                "groups": ["Group", "subGroup"],
+                "path": "master/giturlparse/github.py",
+                "path_raw": "/blob/master/giturlparse/github.py",
+                "pathname": "/Org/Group/subGroup/Repo.git/blob/master/giturlparse/github.py",
+                "branch": "",
+                "protocol": "git",
+                "protocols": ["git"],
+                "platform": "gitlab",
+            },
+        ),
+    ),
     (
         "GIT",
         (
@@ -546,6 +569,78 @@
             },
         ),
     ),
+    (
+        "GIT",
+        (
+            "joe@github.com-work:nephila/giturlparse.git",
+            {
+                "host": "github.com-work",
+                "resource": "github.com-work",
+                "user": "joe",
+                "port": "",
+                "owner": "nephila",
+                "repo": "giturlparse",
+                "name": "giturlparse",
+                "groups": [],
+                "path": "",
+                "path_raw": "",
+                "pathname": "nephila/giturlparse.git",
+                "branch": "",
+                "protocol": "ssh",
+                "protocols": [],
+                "github": False,
+                "platform": "gitlab",
+            },
+        ),
+    ),
+    (
+        "SSH",
+        (
+            "git@gitlab.example.com/groupA/projectB.git",
+            {
+                "host": "gitlab.example.com",
+                "resource": "gitlab.example.com",
+                "user": "git",
+                "port": "",
+                "owner": "groupA",
+                "repo": "projectB",
+                "name": "projectB",
+                "groups": [],
+                "path": "",
+                "path_raw": "",
+                "pathname": "/groupA/projectB.git",
+                "branch": "",
+                "protocol": "ssh",
+                "protocols": [],
+                "github": False,
+                "platform": "gitlab",
+            },
+        ),
+    ),
+    (
+        "SSH",
+        (
+            "ssh://git@gitlab.example.com/groupA/projectB.git",
+            {
+                "host": "gitlab.example.com",
+                "resource": "gitlab.example.com",
+                "user": "git",
+                "port": "",
+                "owner": "groupA",
+                "repo": "projectB",
+                "name": "projectB",
+                "groups": [],
+                "path": "",
+                "path_raw": "",
+                "pathname": "/groupA/projectB.git",
+                "branch": "",
+                "protocol": "ssh",
+                "protocols": ["ssh"],
+                "github": False,
+                "platform": "gitlab",
+            },
+        ),
+    ),
 )
 
 INVALID_PARSE_URLS = (