fix: git URLs working with bootstrap-parallel

The bootstrap-parallel command failed when top-level requirements were
specified as git URLs. During the bootstrap phase, git URLs were
correctly stored in the dependency graph, but the build-parallel phase
reconstructed requirements as name==version, losing the URL
information. This caused the system to treat local directory paths as
HTTP URLs, resulting in invalid URL errors.

Changes:
- Add get_top_level_requirement() method to DependencyGraph to retrieve
  original requirement specifications from ROOT node edges
- Modify build_parallel() to use original top-level requirements when
  available, preserving git URLs during parallel builds
- Add URL parsing logic to download_git_source() to extract @ref from
  git URLs like repo.git@5.2.0

This fix enables nightly build pipelines that rely on git URLs with
the bootstrap-parallel command for faster builds.

Now #762 can be revisited.

Fixes: #761

Co-authored-by: Claude <claude@anthropic.com>

Signed-off-by: Ioannis Angelakopoulos <iangelak@redhat.com>

Author: iangelak

src/fromager/commands/build.py

@@ -696,7 +696,19 @@ def update_progressbar_cb(future: concurrent.futures.Future) -> None:
                     "starting to build: %s", sorted(n.key for n in buildable_nodes)
                 )
                 for node in buildable_nodes:
-                    req = Requirement(f"{node.canonicalized_name}=={node.version}")
+                    # Try to get the original top-level requirement if it includes a URL.
+                    # We only use the top-level requirement if:
+                    # 1. The node is a direct child of ROOT (it's actually top-level)
+                    # 2. The requirement includes a URL (git, etc.)
+                    # Otherwise, we use the already-resolved version from the node to avoid:
+                    # - Building wrong versions when same package appears multiple times
+                    # - Re-resolving unpinned version specifiers
+                    top_level_req = graph.get_top_level_requirement(node)
+                    if top_level_req and top_level_req.url:
+                        req = top_level_req
+                        logger.debug("using top-level requirement with URL: %s", req)
+                    else:
+                        req = Requirement(f"{node.canonicalized_name}=={node.version}")
                     reqs.append(req)
                     future = executor.submit(
                         _build_parallel,

src/fromager/dependency_graph.py

@@ -352,6 +352,19 @@ def get_nodes_by_name(self, req_name: str | None) -> list[DependencyNode]:
     def get_root_node(self) -> DependencyNode:
         return self.nodes[ROOT]
 
+    def get_top_level_requirement(self, node: DependencyNode) -> Requirement | None:
+        """Get the top-level requirement specification for a node.
+
+        For packages that were specified as top-level requirements (e.g., with git URLs),
+        this returns the original requirement specification. Returns None if the node
+        is not a direct child of ROOT.
+        """
+        root = self.get_root_node()
+        for edge in root.children:
+            if edge.destination_node.key == node.key:
+                return edge.req
+        return None
+
     def get_all_nodes(self) -> typing.Iterable[DependencyNode]:
         return self.nodes.values()
 

src/fromager/sources.py

@@ -204,6 +204,17 @@ def download_git_source(
     if url_to_clone.startswith("git+"):
         url_to_clone = url_to_clone[len("git+") :]
 
+    # Parse ref from URL if it's in the form repo@ref
+    from urllib.parse import urlparse
+
+    if ref is None:
+        parsed_url = urlparse(url_to_clone)
+        if "@" in parsed_url.path:
+            # Extract the ref from the path
+            new_path, _, git_ref = parsed_url.path.rpartition("@")
+            url_to_clone = parsed_url._replace(path=new_path).geturl()
+            ref = git_ref
+
     logger.info(f"cloning source from {url_to_clone}@{ref} to {destination_dir}")
     # Get git options from package settings
     pbi = ctx.package_build_info(req)