From e91e0bf09c0e17c754833358482dc703054361ac Mon Sep 17 00:00:00 2001
From: "Antoine R. Dumont (@ardumont)" <ardumont@softwareheritage.org>
Date: Mon, 22 May 2023 16:51:51 +0200
Subject: [PATCH] cgit: Allow url to be optional

Some cgit instances are at a domain's root path so we can build their url directly from
their 'instance' parameter.

This unifies further the cli to register a lister and the cli to schedule the listed
origins from a forge.

[1]
```
https://git.kernel.org
https://source.codeaurora.org
https://git.trueelena.org
https://dev.sanctum.geek.nz
https://git.trueelena.org
https://git.dpkg.org
https://anongit.mindrot.org
https://git.aurel32.net
https://gitweb.gentoo.org
https://git.joeyh.name
https://git.adrian.geek.nz
```

Refs. swh/devel/swh-lister#4693
---
 swh/lister/cgit/lister.py            |  9 +++++----
 swh/lister/cgit/tests/test_lister.py | 21 ++++++++++++++++++++-
 swh/lister/pattern.py                |  8 +++++---
 3 files changed, 30 insertions(+), 8 deletions(-)

diff --git a/swh/lister/cgit/lister.py b/swh/lister/cgit/lister.py
index 4a9aeab..1ab17cb 100644
--- a/swh/lister/cgit/lister.py
+++ b/swh/lister/cgit/lister.py
@@ -1,4 +1,4 @@
-# Copyright (C) 2019-2022 The Software Heritage developers
+# Copyright (C) 2019-2023 The Software Heritage developers
 # License: GNU General Public License version 3, or any later version
 # See top-level LICENSE file for more information
 
@@ -46,7 +46,7 @@ class CGitLister(StatelessLister[Repositories]):
     def __init__(
         self,
         scheduler: SchedulerInterface,
-        url: str,
+        url: Optional[str] = None,
         instance: Optional[str] = None,
         credentials: Optional[CredentialsType] = None,
         base_git_url: Optional[str] = None,
@@ -57,8 +57,9 @@ class CGitLister(StatelessLister[Repositories]):
         """Lister class for CGit repositories.
 
         Args:
-            url: main URL of the CGit instance, i.e. url of the index
-                of published git repositories on this instance.
+            url: (Optional) Root URL of the CGit instance, i.e. url of the index of
+                published git repositories on this instance. Defaults to
+                :file:`https://{instance}` if unset.
             instance: Name of cgit instance. Defaults to url's network location
                 if unset.
             base_git_url: Optional base git url which allows the origin url
diff --git a/swh/lister/cgit/tests/test_lister.py b/swh/lister/cgit/tests/test_lister.py
index c6ffcf2..bcf0e8d 100644
--- a/swh/lister/cgit/tests/test_lister.py
+++ b/swh/lister/cgit/tests/test_lister.py
@@ -1,4 +1,4 @@
-# Copyright (C) 2019-2022 The Software Heritage developers
+# Copyright (C) 2019-2023 The Software Heritage developers
 # License: GNU General Public License version 3, or any later version
 # See top-level LICENSE file for more information
 
@@ -155,6 +155,25 @@ requests_mock_datadir_missing_url = requests_mock_datadir_factory(
 )
 
 
+def test_lister_cgit_instantiate(swh_scheduler):
+    """Build a lister with either an url or an instance is fine"""
+    url = "https://source.codeaurora.org"
+    lister = CGitLister(swh_scheduler, url=url)
+    assert lister is not None
+    assert lister.url == url
+
+    assert CGitLister(swh_scheduler, instance="source.codeaurora.org") is not None
+    assert lister is not None
+    assert lister.url == url
+
+
+def test_lister_cgit_fail_to_instantiate(swh_scheduler):
+    """Build a lister without its url nor its instance should raise"""
+    # ... It will raise without any of those
+    with pytest.raises(ValueError, match="'url' or 'instance'"):
+        CGitLister(swh_scheduler)
+
+
 def test_lister_cgit_get_origin_from_repo_failing(
     requests_mock_datadir_missing_url, swh_scheduler
 ):
diff --git a/swh/lister/pattern.py b/swh/lister/pattern.py
index 7fb8680..dd3295c 100644
--- a/swh/lister/pattern.py
+++ b/swh/lister/pattern.py
@@ -116,12 +116,14 @@ class Lister(Generic[StateType, PageType]):
             raise ValueError("Must set the LISTER_NAME attribute on Lister classes")
 
         self.url: str
+        # lister can be instantiated using directly their 'url' (the default behavior)
+        # or derive their url through an 'instance' (their domain's root path) parameter
         if url is not None:
-            # Retro-compability with lister already instantiated out of a provided url
+            # direct url instantiation
             self.url = url
         elif url is None and instance is not None:
-            # Allow lister to be instantiated simply with their type and instance
-            # (as in their domain like "gitlab.com", "git.garbaye.fr", ...)
+            # Allow instantiation through their instance parameter (domain's root path)
+            # (e.g. "gitlab.com", "git.garbaye.fr", ...)
             self.url = self.build_url(instance)
         else:
             raise ValueError(