From d17e60a6087a9b140599d59aa747139b03c7207c Mon Sep 17 00:00:00 2001
From: Stephan Kulow <coolo@suse.de>
Date: Mon, 24 Oct 2022 12:01:28 +0200
Subject: [PATCH] Add another table to store linked_revs

We need to create fake revisions when packages were touched that are
linked themselves
---
 lib/db.py           | 11 ++++++++++-
 lib/db_revision.py  |  6 +++++-
 lib/importer.py     | 41 +++++++++++++++++++++++++++++++++++++++--
 lib/tree_builder.py |  5 +++--
 4 files changed, 57 insertions(+), 6 deletions(-)

diff --git a/lib/db.py b/lib/db.py
index 7958c50..87acbe8 100644
--- a/lib/db.py
+++ b/lib/db.py
@@ -163,8 +163,17 @@ class DB:
             )
             """,
             "UPDATE scheme SET version=13",
+        ),
+        schemes[14] = (
+            """
+            CREATE TABLE linked_revs (
+                id              SERIAL PRIMARY KEY,
+                revision_id     INTEGER NOT NULL,
+                linked_id       INTEGER NOT NULL
+            )
+            """,
+            "UPDATE scheme SET version=14",
         )
-
         schema_version = self.schema_version()
         if (schema_version + 1) not in schemes:
             return
diff --git a/lib/db_revision.py b/lib/db_revision.py
index 11dfecb..c8065e8 100644
--- a/lib/db_revision.py
+++ b/lib/db_revision.py
@@ -25,7 +25,7 @@ class DBRevision:
         self._hash = None
 
     def __str__(self):
-        return f"Rev {self.project}/{self.rev} Md5 {self.unexpanded_srcmd5} {self.commit_time} {self.userid} {self.request_number}"
+        return f"Rev {self.project}/{self.package}/{self.rev} Md5 {self.unexpanded_srcmd5} {self.commit_time} {self.userid} {self.request_number}"
 
     def __repr__(self):
         return f"[{self.__str__()}]"
@@ -119,6 +119,8 @@ class DBRevision:
         return ret
 
     def linked_rev(self, db):
+        if self.broken:
+            return None
         with db.cursor() as cur:
             cur.execute(
                 "SELECT project,package FROM links where revision_id=%s", (self.dbid,)
@@ -134,6 +136,8 @@ class DBRevision:
             revisions = [DBRevision(row) for row in cur.fetchall()]
         if revisions:
             return revisions[0]
+        else:
+            self.set_broken(db)
         return None
 
     def set_broken(self, db):
diff --git a/lib/importer.py b/lib/importer.py
index 8bd08bf..011da01 100644
--- a/lib/importer.py
+++ b/lib/importer.py
@@ -1,5 +1,6 @@
 import functools
 import logging
+import xml.etree.ElementTree as ET
 
 from lib.binary import is_binary_or_large
 from lib.db import DB
@@ -145,6 +146,36 @@ class Importer:
                     dbrev.links_to(db, tprj, tpkg)
         db.conn.commit()
 
+    def find_linked_revs(self, db):
+        cur = db.cursor()
+        cur.execute(
+            """SELECT * from revisions WHERE id in (SELECT l.revision_id FROM links l
+                        LEFT JOIN linked_revs lrevs ON lrevs.revision_id=l.revision_id
+                        WHERE lrevs.id IS NULL) and broken is FALSE;"""
+        )
+        for row in cur.fetchall():
+            rev = DBRevision(row)
+            linked_rev = rev.linked_rev(db)
+            if not linked_rev:
+                logging.debug("No link", rev)
+                continue
+            cur.execute(
+                """INSERT INTO linked_revs (revision_id, linked_id)
+                            VALUES (%s,%s)""",
+                (rev.dbid, linked_rev.dbid),
+            )
+
+    def fetch_all_linked_packages(self, db, project, package):
+        cur = db.cursor()
+        cur.execute(
+            """SELECT DISTINCT l.project, l.package from links l JOIN revisions r
+                on r.id=l.revision_id WHERE r.project=%s AND r.package=%s""",
+            (project, package),
+        )
+        for row in cur.fetchall():
+            (lproject, lpackage) = row
+            self.update_db_package(db, lproject, lpackage)
+
     def import_into_db(self):
         db = DB()
         for project, _, api_url in self.projects:
@@ -169,9 +200,15 @@ class Importer:
                 # TODO move into SELECT
                 if rev.broken or rev.expanded_srcmd5:
                     continue
-                linked_rev = rev.linked_rev(db)
+                with db.cursor() as cur:
+                    cur.execute(
+                        """SELECT unexpanded_srcmd5 from revisions WHERE
+                                id=(SELECT linked_id FROM linked_revs WHERE revision_id=%s""",
+                        (rev.dbid,),
+                    )
+                    linked_rev = cur.fetchone()
                 if linked_rev:
-                    linked_rev = linked_rev.unexpanded_srcmd5
+                    linked_rev = linked_rev[0]
                 list = self.obs.list(
                     project, self.package, rev.unexpanded_srcmd5, linked_rev
                 )
diff --git a/lib/tree_builder.py b/lib/tree_builder.py
index 69d260d..af74331 100644
--- a/lib/tree_builder.py
+++ b/lib/tree_builder.py
@@ -12,14 +12,15 @@ class TreeBuilder:
         ret = []
         prev = None
         for rev in revisions:
-            if rev.broken: continue
+            print(rev, rev.files_hash(self.db))
+            if rev.broken:
+                continue
             if prev and prev.files_hash(self.db) == rev.files_hash(self.db):
                 continue
             ret.append(rev)
             prev = rev
         return ret
 
-
     def build(self, package):
         factory_revisions = self.filtered_revisions("openSUSE:Factory", package)
         source_revisions = dict()