Another take on parsing pull request references

Parse only references that start with 'PR:'.
Parse both owner/repo#number and gitea_url/owner/repo/pulls/number

osc/commands_git/pr_dump.py

@@ -309,47 +309,18 @@ class PullRequestDumpCommand(osc.commandline_git.GitObsCommand):
             with open(os.path.join(metadata_dir, "submodules-diff.json"), "w", encoding="utf-8") as f:
                 json.dump(submodule_diff, f, indent=4, sort_keys=True)
 
-            linked_prs = {}
-
-            # body may contain references with both https:// or without, which look indetical in UI. so we must handle both cases:
-            for url in re.findall(r"https?://[^\s]+/pulls/\d+", pr_obj.body):
-                if not self.gitea_conn.host in url:
-                    print(f"ignoring PR {url}")
-                    linked_prs[url] = None
-                    continue
-
-                print(f"Linking PR {url}...")
-                _, _, linked_id = url.partition(self.gitea_conn.host + "/")
-
-                try:
-                    linked_owner, linked_repo, linked_number = gitea_api.PullRequest.split_id(linked_id)
-                    linked_pr_obj = gitea_api.PullRequest.get(self.gitea_conn, linked_owner, linked_repo, linked_number)
-                    if linked_pr_obj is None:
-                        linked_prs[url] = None
-                    else:
-                        linked_prs[url] = linked_pr_obj.to_light_dict()
-                except:
-                    linked_prs[url] = None
-
-            for m in re.findall(r"([^\s\/]+)\/([^\s\/]+)\#(\d+)", pr_obj.body):
-                uri = f"{m[0]}/{m[1]}#{m[2]}"
-                print(f"Linking PR {uri}...")
-
-                try:
-                    linked_pr_obj = gitea_api.PullRequest.get(self.gitea_conn, m[0], m[1], m[2])
-                    if linked_pr_obj is None:
-                        linked_prs[uri] = None
-                    else:
-                        linked_prs[uri] = linked_pr_obj.to_light_dict()
-                except:
-                    linked_prs[uri] = None
+            referenced_pull_requests = {}
+            for ref_owner, ref_repo, ref_number in pr_obj.parse_pr_references():
+                ref_id = f"{ref_owner}/{ref_repo}#{ref_number}"
+                referenced_pr_obj = gitea_api.PullRequest.get(self.gitea_conn, ref_owner, ref_repo, ref_number)
+                referenced_pull_requests[ref_id] = referenced_pr_obj.dict()
 
             with open(
                 os.path.join(metadata_dir, "referenced-pull-requests.json"),
                 "w",
                 encoding="utf-8",
             ) as f:
-                json.dump(linked_prs, f, indent=4, sort_keys=True)
+                json.dump(referenced_pull_requests, f, indent=4, sort_keys=True)
 
             if warnings:
                 return 38

osc/gitea_api/pr.py

@@ -10,11 +10,7 @@ from .connection import GiteaHTTPResponse
 from .user import User
 
 
-class PullRequestReview:
-    def __init__(self, data: dict, *, response: Optional[GiteaHTTPResponse] = None):
-        self._data = data
-        self._response = response
-
+class PullRequestReview(GiteaModel):
     @property
     def state(self) -> str:
         return self._data["state"]
@@ -104,9 +100,34 @@ class PullRequest(GiteaModel):
         parsed_url = urllib.parse.urlparse(url)
         path = parsed_url.path
         owner, repo, pulls, number = path.strip("/").split("/")
-        assert pulls in ("pulls", "issues")
+        if pulls not in ("pulls", "issues"):
+            raise ValueError(f"URL doesn't point to a pull request or an issue: {url}")
         return owner, repo, int(number)
 
+    def parse_pr_references(self) -> List[Tuple[str, str, int]]:
+        refs = re.findall(r"^PR: *(.*)$", self.body, re.M)
+        result = []
+
+        for ref in refs:
+            # try owner/repo#number first
+            try:
+                result.append(PullRequest.split_id(ref))
+                continue
+            except ValueError:
+                pass
+
+            # parse owner, repo, number from a pull request url
+            if ref.startswith(f"{self._conn.login.url.rstrip('/')}/"):
+                try:
+                    result.append(PullRequest.get_owner_repo_number(ref))
+                    continue
+                except ValueError:
+                    pass
+
+            raise ValueError(f"Unable to parse pull request reference: {ref}")
+
+        return result
+
     @property
     def is_pull_request(self):
         # determine if we're working with a proper pull request or an issue without pull request details
@@ -337,7 +358,7 @@ class PullRequest(GiteaModel):
             "body": description,
         }
         response = conn.request("POST", url, json_data=data)
-        obj = cls(response.json(), response=response)
+        obj = cls(response.json(), response=response, conn=conn)
         return obj
 
     @classmethod
@@ -358,7 +379,7 @@ class PullRequest(GiteaModel):
         """
         url = conn.makeurl("repos", owner, repo, "pulls", str(number))
         response = conn.request("GET", url)
-        obj = cls(response.json(), response=response)
+        obj = cls(response.json(), response=response, conn=conn)
         return obj
 
     @classmethod
@@ -391,7 +412,7 @@ class PullRequest(GiteaModel):
         }
         url = conn.makeurl("repos", owner, repo, "pulls", str(number))
         response = conn.request("PATCH", url, json_data=json_data)
-        obj = cls(response.json(), response=response)
+        obj = cls(response.json(), response=response, conn=conn)
         return obj
 
     @classmethod
@@ -420,7 +441,7 @@ class PullRequest(GiteaModel):
         }
         url = conn.makeurl("repos", owner, repo, "pulls", query=q)
         response = conn.request("GET", url)
-        obj_list = [cls(i, response=response) for i in response.json()]
+        obj_list = [cls(i, response=response, conn=conn) for i in response.json()]
         return obj_list
 
     @classmethod
@@ -464,7 +485,7 @@ class PullRequest(GiteaModel):
         url = conn.makeurl("repos", "issues", "search", query=q)
         obj_list = []
         for response in conn.request_all_pages("GET", url):
-            obj_list.extend([cls(i, response=response) for i in response.json()])
+            obj_list.extend([cls(i, response=response, conn=conn) for i in response.json()])
         return obj_list
 
     @classmethod