Build: better support for empty default branch

readthedocs/api/v2/views/integrations.py

@@ -20,6 +20,7 @@
 
 from readthedocs.builds.constants import BRANCH
 from readthedocs.builds.constants import LATEST
+from readthedocs.builds.constants import LATEST_VERBOSE_NAME
 from readthedocs.builds.constants import TAG
 from readthedocs.core.signals import webhook_bitbucket
 from readthedocs.core.signals import webhook_github
@@ -313,7 +314,7 @@ def get_closed_external_version_response(self, project):
 
     def update_default_branch(self, default_branch):
         """
-        Update the `Version.identifer` for `latest` with the VCS's `default_branch`.
+        Update the `Version.identifier` for `latest` with the VCS's `default_branch`.
 
         The VCS's `default_branch` is the branch cloned when there is no specific branch specified
         (e.g. `git clone <URL>`).
@@ -335,7 +336,9 @@ def update_default_branch(self, default_branch):
             # Always check for the machine attribute, since latest can be user created.
             # RTD doesn't manage those.
             self.project.versions.filter(slug=LATEST, machine=True).update(
-                identifier=default_branch
+                identifier=default_branch,
+                verbose_name=LATEST_VERBOSE_NAME,
+                type=BRANCH,
             )
 
 

readthedocs/builds/models.py

@@ -222,6 +222,10 @@ def is_public(self):
     def is_external(self):
         return self.type == EXTERNAL
 
+    @property
+    def is_machine_latest(self):
+        return self.machine and self.slug == LATEST
+
     @property
     def explicit_name(self):
         """

readthedocs/doc_builder/director.py

@@ -223,7 +223,12 @@ def checkout(self):
                     dismissable=True,
                 )
 
-        identifier = self.data.build_commit or self.data.version.identifier
+        # We can't skip the checkout step.
+        # If Feature.DONT_CLEAN_BUILD is enabled, we need to explicitly call checkout
+        # with the default branch, otherwise we could end up in the wrong branch.
+        identifier = (
+            self.data.build_commit or self.data.default_branch or self.data.version.identifier
+        )
         log.info("Checking out.", identifier=identifier)
         self.vcs_repository.checkout(identifier)
 

readthedocs/projects/models.py

@@ -1177,15 +1177,21 @@ def get_original_latest_version(self):
         When latest is machine created, it's basically an alias
         for the default branch/tag (like main/master),
 
-        Returns None if the current default version doesn't point to a valid version.
+        Returns None if latest doesn't point to a valid version,
+        or if isn't managed by RTD (machine=False).
         """
-        default_version_name = self.get_default_branch()
+        # For latest, the identifier is the name of the branch/tag.
+        latest_version_identifier = (
+            self.versions.filter(slug=LATEST, machine=True)
+            .values_list("identifier", flat=True)
+            .first()
+        )
+        if not latest_version_identifier:
+            return None
         return (
             self.versions(manager=INTERNAL)
             .exclude(slug=LATEST)
-            .filter(
-                verbose_name=default_version_name,
-            )
+            .filter(verbose_name=latest_version_identifier)
             .first()
         )
 
@@ -1203,8 +1209,22 @@ def update_latest_version(self):
             return
 
         # default_branch can be a tag or a branch name!
-        default_version_name = self.get_default_branch()
-        original_latest = self.get_original_latest_version()
+        default_version_name = self.get_default_branch(fallback_to_vcs=False)
+        # If the default_branch is not set, it means that the user
+        # wants to use the default branch of the respository, but
+        # we don't know what that is here, `latest` will be updated
+        # on the next build.
+        if not default_version_name:
+            return
+
+        # Search for a branch or tag with the name of the default branch,
+        # so we can sync latest with it.
+        original_latest = (
+            self.versions(manager=INTERNAL)
+            .exclude(slug=LATEST)
+            .filter(verbose_name=default_version_name)
+            .first()
+        )
         latest.verbose_name = LATEST_VERBOSE_NAME
         latest.type = original_latest.type if original_latest else BRANCH
         # For latest, the identifier is the name of the branch/tag.
@@ -1304,14 +1324,22 @@ def get_default_version(self):
             return self.default_version
         return LATEST
 
-    def get_default_branch(self):
-        """Get the version representing 'latest'."""
+    def get_default_branch(self, fallback_to_vcs=True):
+        """
+        Get the name of the branch or tag that the user wants to use as 'latest'.
+
+        In case the user explicitly set a default branch, we use that,
+        otherwise we try to get it from the remote repository.
+        """
         if self.default_branch:
             return self.default_branch
 
         if self.remote_repository and self.remote_repository.default_branch:
             return self.remote_repository.default_branch
 
+        if not fallback_to_vcs:
+            return None
+
         vcs_class = self.vcs_class()
         if vcs_class:
             return vcs_class.fallback_branch

readthedocs/projects/tasks/builds.py

@@ -25,6 +25,7 @@
 from readthedocs.builds import tasks as build_tasks
 from readthedocs.builds.constants import ARTIFACT_TYPES
 from readthedocs.builds.constants import ARTIFACT_TYPES_WITHOUT_MULTIPLE_FILES_SUPPORT
+from readthedocs.builds.constants import BRANCH
 from readthedocs.builds.constants import BUILD_FINAL_STATES
 from readthedocs.builds.constants import BUILD_STATE_BUILDING
 from readthedocs.builds.constants import BUILD_STATE_CANCELLED
@@ -115,6 +116,10 @@ class TaskData:
     config: BuildConfigV2 = None
     project: APIProject = None
     version: APIVersion = None
+    # Default branch for the repository.
+    # Only set when building the latest version, and the project
+    # doesn't have an explicit default branch.
+    default_branch: str | None = None
 
     # Dictionary returned from the API.
     build: dict = field(default_factory=dict)
@@ -644,18 +649,22 @@ def on_success(self, retval, task_id, args, kwargs):
         # NOTE: we are updating the db version instance *only* when
         # TODO: remove this condition and *always* update the DB Version instance
         if "html" in valid_artifacts:
+            data = {
+                "built": True,
+                "documentation_type": self.data.version.documentation_type,
+                "has_pdf": "pdf" in valid_artifacts,
+                "has_epub": "epub" in valid_artifacts,
+                "has_htmlzip": "htmlzip" in valid_artifacts,
+                "build_data": self.data.version.build_data,
+                "addons": self.data.version.addons,
+            }
+            # Update the latest version to point to the current VCS default branch
+            # if the project doesn't have an explicit default branch set.
+            if self.data.default_branch:
+                data["identifier"] = self.data.default_branch
+                data["type"] = BRANCH
             try:
-                self.data.api_client.version(self.data.version.pk).patch(
-                    {
-                        "built": True,
-                        "documentation_type": self.data.version.documentation_type,
-                        "has_pdf": "pdf" in valid_artifacts,
-                        "has_epub": "epub" in valid_artifacts,
-                        "has_htmlzip": "htmlzip" in valid_artifacts,
-                        "build_data": self.data.version.build_data,
-                        "addons": self.data.version.addons,
-                    }
-                )
+                self.data.api_client.version(self.data.version.pk).patch(data)
             except HttpClientError:
                 # NOTE: I think we should fail the build if we cannot update
                 # the version at this point. Otherwise, we will have inconsistent data
@@ -803,6 +812,15 @@ def execute(self):
         with self.data.build_director.vcs_environment:
             self.data.build_director.setup_vcs()
 
+            # Get the default branch of the repository if the project doesn't
+            # have an explicit default branch set and we are building latest.
+            # The identifier from latest will be updated with this value
+            # if the build succeeds.
+            if self.data.version.is_machine_latest and not self.data.project.default_branch:
+                self.data.default_branch = (
+                    self.data.build_director.vcs_repository.get_default_branch()
+                )
+
             # Sync tags/branches from VCS repository into Read the Docs'
             # `Version` objects in the database. This method runs commands
             # (e.g. "hg tags") inside the VCS environment, so it requires to be