git-importer/lib/tree_builder.py

from lib.db_revision import DBRevision
from lib.request import Request


class TreeNode:
    """
    Nodes in this "tree" have either no parent (root), one parent (in a chain)
    or two parents (in this case the merged revision wins in conflicts).
    """

    def __init__(self, rev):
        self.parent = None
        self.merged = None
        self.revision = rev
        self.merged_into = None

    def print(self):
        node = self
        while node:
            print(node.revision, node.revision.files_hash)
            if node.merged:
                source_node = node.merged
                while source_node:
                    print("   ", source_node.revision, source_node.revision.files_hash)
                    source_node = source_node.parent
                    if source_node and source_node.merged_into:
                        break
            node = node.parent

    def as_list(self):
        """Return a list for test cases"""
        node = self
        ret = []
        while node:
            repr = {"commit": node.revision.short_string()}
            if node.merged:
                source_node = node.merged
                repr["merged"] = []
                while source_node:
                    repr["merged"].append(source_node.revision.short_string())
                    source_node = source_node.parent
                    if source_node and source_node.merged_into:
                        break
            node = node.parent
            ret.append(repr)
        return ret


class TreeBuilder:
    def __init__(self, db):
        self.db = db

    def revisions_chain(self, project, package):
        revisions = DBRevision.all_revisions(self.db, project, package)
        revisions.sort()
        prev = None
        tree = None
        for rev in revisions:
            if rev.broken:
                continue
            if prev and prev.files_hash == rev.files_hash:
                continue
            prev = rev
            new_tree = TreeNode(rev)
            if tree:
                new_tree.parent = tree
            tree = new_tree

        return tree

    def find_merge(self, revision, source_chain):
        node = source_chain
        while node:
            # exclude reverts happening after the merge
            if (
                node.revision.commit_time <= revision.commit_time
                and node.revision.files_hash == revision.files_hash
            ):
                return node
            node = node.parent

    def add_merge_points(self, factory_revisions):
        source_revisions = dict()
        factory_node = factory_revisions
        while factory_node:
            if factory_node.revision.request_id:
                req = Request.find(self.db, factory_node.revision.request_id)
                key = f"{req.source_project}/{req.source_package}"
                if key not in source_revisions:
                    source_revisions[key] = self.revisions_chain(
                        req.source_project, req.source_package
                    )
                factory_node.merged = self.find_merge(
                    factory_node.revision, source_revisions[key]
                )
                # add a reverse lookup
                if factory_node.merged:
                    factory_node.merged.merged_into = factory_node

            factory_node = factory_node.parent

    def prune_loose_end(self, factory_node):
        """Look for source revisions that end in a new root and prune them"""
        last_merge = None
        while factory_node:
            if factory_node.merged:
                source_node = factory_node.merged
                ended_without_merge = False
                while source_node:
                    source_node = source_node.parent
                    if source_node and source_node.merged_into:
                        ended_without_merge = True
                        break
                if not ended_without_merge:
                    factory_node.merged = None
                    if last_merge:
                        last_merge.parent = None
                else:
                    last_merge = factory_node.merged
            factory_node = factory_node.parent

    def build(self, package):
        factory_revisions = self.revisions_chain("openSUSE:Factory", package)
        self.add_merge_points(factory_revisions)

        self.prune_loose_end(factory_revisions)
        return factory_revisions