Switch to walker for tree print

git-importer.py

@@ -130,6 +130,7 @@ def main():
 
     if args.db:
         importer.import_into_db()
+        importer.export_as_git()
         return
     importer.import_all_revisions(args.gc)
 

lib/importer.py

@@ -259,6 +259,10 @@ class Importer:
             )
             return [DBRevision(row) for row in cur.fetchall()]
 
+    def export_as_git(self):
+        db = DB()
+        TreeBuilder(db).build(self.package).print()
+
     def import_into_db(self):
         db = DB()
         for project, _, api_url in self.projects:
@@ -303,7 +307,6 @@ class Importer:
             self.obs.request(number).import_into_db(db)
 
         db.conn.commit()
-        TreeBuilder(db).build(self.package).print()
 
     def import_all_revisions(self, gc):
         # Fetch all the requests and sort them.  Ideally we should

lib/tree_builder.py

@@ -2,6 +2,19 @@ from lib.db_revision import DBRevision
 from lib.request import Request
 
 
+class AbstractWalker:
+    def call(self, node, is_source):
+        pass
+
+
+class PrintWalker(AbstractWalker):
+    def call(self, node, is_source):
+        if is_source:
+            print("   ", node.revision, node.revision.files_hash)
+        else:
+            print(node.revision, node.revision.files_hash)
+
+
 class TreeNode:
     """
     Nodes in this "tree" have either no parent (root), one parent (in a chain)
@@ -13,20 +26,24 @@ class TreeNode:
         self.merged = None
         self.revision = rev
         self.merged_into = None
+        self.git_commit = None
 
-    def print(self):
+    def walk(self, walker: AbstractWalker):
         node = self
         while node:
-            print(node.revision, node.revision.files_hash)
+            walker.call(node, False)
             if node.merged:
                 source_node = node.merged
                 while source_node:
-                    print("   ", source_node.revision, source_node.revision.files_hash)
+                    walker.call(node, True)
                     source_node = source_node.parent
                     if source_node and source_node.merged_into:
                         break
             node = node.parent
 
+    def print(self):
+        self.walk(PrintWalker())
+
     def as_list(self):
         """Return a list for test cases"""
         node = self
@@ -51,6 +68,8 @@ class TreeBuilder:
         self.db = db
 
     def revisions_chain(self, project, package):
+        """Build a tree without branches (chain) from a project's
+        history ignoring empty and broken revisions"""
         revisions = DBRevision.all_revisions(self.db, project, package)
         revisions.sort()
         prev = None
@@ -69,6 +88,8 @@ class TreeBuilder:
         return tree
 
     def find_merge(self, revision, source_chain):
+        """For a given revision in the target, find the node in the source chain
+        that matches the files"""
         node = source_chain
         while node:
             # exclude reverts happening after the merge
@@ -80,6 +101,8 @@ class TreeBuilder:
             node = node.parent
 
     def add_merge_points(self, factory_revisions):
+        """For all target revisions that accepted a request, look up the merge
+        points in the source chains (ignoring the actual revision submitted for now)"""
         source_revisions = dict()
         factory_node = factory_revisions
         while factory_node:
@@ -120,6 +143,7 @@ class TreeBuilder:
             factory_node = factory_node.parent
 
     def build(self, package):
+        """Create a Factory tree (returning the top)"""
         factory_revisions = self.revisions_chain("openSUSE:Factory", package)
         self.add_merge_points(factory_revisions)