Create fake revisions for every commit in the base of a linked package

This is a rather complex operation, but whenever a package changes in
Factory, the inherited package gets a shadow revision consisting of the
3-way merge. If this happens due to a request being accepted, this is
actually in most cases also commited by 'buildservice-autocommit',
so we're making sure this is always happening (and are actually
duplicating revisions in cases that we filter out later as empty
commits).

To differenciate the fake revisions from the real revisions, I add
a fraction part

README.md

@@ -1,3 +1,5 @@
 sudo zypper in python3-psycopg2
 sudo su - postgres
 # `createdb -O <LOCAL_USER> imported_git`
+
+To reset the database, drop table scheme

lib/db.py

@@ -1,4 +1,7 @@
+import logging
+
 import psycopg2
+from psycopg2.extras import LoggingConnection
 
 from lib.config import config
 
@@ -14,7 +17,9 @@ class DB:
             # read the connection parameters
             params = config(section=self.config_section)
             # connect to the PostgreSQL server
-            self.conn = psycopg2.connect(**params)
+            self.conn = psycopg2.connect(connection_factory=LoggingConnection, **params)
+            logger = logging.getLogger(__name__)
+            self.conn.initialize(logger)
 
         except (Exception, psycopg2.DatabaseError) as error:
             print(error)
@@ -96,6 +101,7 @@ class DB:
             "UPDATE scheme SET version=4",
         )
         schemes[5] = (
+            """DROP TABLE IF EXISTS files""",
             """
             CREATE TABLE files (
                 id              SERIAL PRIMARY KEY,
@@ -109,6 +115,7 @@ class DB:
             "UPDATE scheme SET version=5",
         )
         schemes[6] = (
+            """DROP TABLE IF EXISTS requests""",
             """
             CREATE TABLE requests (
                 id              SERIAL PRIMARY KEY,
@@ -154,6 +161,7 @@ class DB:
             "UPDATE scheme SET version=12",
         )
         schemes[13] = (
+            """DROP TABLE IF EXISTS linked_revs""",
             """
             CREATE TABLE users (
                 id              SERIAL PRIMARY KEY,
@@ -174,6 +182,30 @@ class DB:
             """,
             "UPDATE scheme SET version=14",
         )
+        schemes[14] = (
+            "ALTER TABLE revisions ALTER COLUMN rev TYPE real USING rev::real",
+            "UPDATE scheme SET version=14",
+        )
+        schemes[15] = (
+            """DROP TABLE IF EXISTS fake_revs""",
+            """
+            CREATE TABLE fake_revs (
+                id              SERIAL PRIMARY KEY,
+                revision_id     INTEGER NOT NULL,
+                linked_id       INTEGER NOT NULL
+            )
+            """,
+            "create index revs_linked on fake_revs (revision_id,linked_id)",
+            "UPDATE scheme SET version=15",
+        )
+        schemes[16] = (
+            "ALTER TABLE revisions ADD COLUMN files_hash VARCHAR(40)",
+            "UPDATE scheme SET version=16",
+        )
+        schemes[17] = (
+            "ALTER TABLE linked_revs ADD COLUMN considered BOOLEAN DEFAULT FALSE",
+            "UPDATE scheme SET version=17",
+        )
         schema_version = self.schema_version()
         if (schema_version + 1) not in schemes:
             return

lib/db_revision.py

@@ -19,10 +19,10 @@ class DBRevision:
             self.expanded_srcmd5,
             self.request_number,
             self.request_id,
+            self.files_hash,
         ) = row
-        self.rev = int(self.rev)
+        self.rev = float(self.rev)
         self._files = None
-        self._hash = None
 
     def __str__(self):
         return f"Rev {self.project}/{self.package}/{self.rev} Md5 {self.unexpanded_srcmd5} {self.commit_time} {self.userid} {self.request_number}"
@@ -52,6 +52,7 @@ class DBRevision:
             "comment": self.comment,
             "broken": self.broken,
             "expanded_srcmd5": self.expanded_srcmd5,
+            "files_hash": self.files_hash,
             "files": self.files_list(db),
         }
         if self.request_id:
@@ -92,6 +93,7 @@ class DBRevision:
                 (project, package, str(rev)),
             )
             row = cur.fetchone()
+        if row:
             return DBRevision(row)
 
     @staticmethod
@@ -103,7 +105,7 @@ class DBRevision:
             )
             max = cur.fetchone()[0]
         if max:
-            return DBRevision.fetch_revision(db, project, package, int(max))
+            return DBRevision.fetch_revision(db, project, package, max)
         return None
 
     @staticmethod
@@ -144,7 +146,6 @@ class DBRevision:
         with db.cursor() as cur:
             cur.execute("UPDATE revisions SET broken=TRUE where id=%s", (self.dbid,))
 
-
     def import_dir_list(self, db, xml):
         with db.cursor() as cur:
             cur.execute(
@@ -164,9 +165,13 @@ class DBRevision:
                     ),
                 )
 
-    def files_hash(self, db):
+    def previous_commit(self, db):
-        if self._hash:
+        return self.fetch_revision(db, self.project, self.package, int(self.rev) - 1)
-            return self._hash
+
+    def next_commit(self, db):
+        return self.fetch_revision(db, self.project, self.package, int(self.rev) + 1)
+
+    def calculate_files_hash(self, db):
         m = md5()
         for file_dict in self.files_list(db):
             m.update(
@@ -178,8 +183,7 @@ class DBRevision:
                     + str(file_dict["size"])
                 ).encode("utf-8")
             )
-        self._hash = m.hexdigest()
+        return m.hexdigest()
-        return self._hash
 
     def files_list(self, db):
         if self._files:
@@ -194,7 +198,7 @@ class DBRevision:
         return self._files
 
     @staticmethod
-    def requests_to_fetch(self, db, project, package):
+    def requests_to_fetch(db, project, package):
         with db.cursor() as cur:
             cur.execute(
                 """SELECT request_number FROM revisions revs LEFT JOIN requests
@@ -209,8 +213,9 @@ class DBRevision:
         """Used in test cases to read a revision from fixtures into the test database"""
         with db.cursor() as cur:
             cur.execute(
-                """INSERT INTO revisions (project, package, rev, unexpanded_srcmd5, expanded_srcmd5, commit_time, userid, comment, broken)  
+                """INSERT INTO revisions (project, package, rev, unexpanded_srcmd5, expanded_srcmd5, 
-                        VALUES(%s, %s, %s, %s, %s, %s, %s, %s, %s) RETURNING id""",
+                                        commit_time, userid, comment, broken, files_hash)  
+                        VALUES(%s, %s, %s, %s, %s, %s, %s, %s, %s, %s) RETURNING id""",
                 (
                     rev_dict["project"],
                     rev_dict["package"],
@@ -221,6 +226,7 @@ class DBRevision:
                     rev_dict["userid"],
                     rev_dict["comment"],
                     rev_dict["broken"],
+                    rev_dict["files_hash"],
                 ),
             )
             rev_id = cur.fetchone()[0]

lib/exporter.py

@@ -13,7 +13,10 @@ class Exporter:
     def run(self):
         db = DB()
         cur = db.cursor()
-        cur.execute("SELECT * from revisions where package=%s", (self.package,))
+        cur.execute(
+            "SELECT * from revisions where package=%s ORDER BY project,rev",
+            (self.package,),
+        )
         data = {"revisions": []}
         for row in cur.fetchall():
             data["revisions"].append(DBRevision(row).as_dict(db))

lib/importer.py

@@ -2,6 +2,8 @@ import functools
 import logging
 import xml.etree.ElementTree as ET
 
+import psycopg2
+
 from lib.binary import is_binary_or_large
 from lib.db import DB
 from lib.db_revision import DBRevision
@@ -165,6 +167,18 @@ class Importer:
                 (rev.dbid, linked_rev.dbid),
             )
 
+    def calculate_file_hashes(self, db):
+        cur = db.cursor()
+        cur.execute(
+            "SELECT * from revisions where files_hash IS NULL AND broken is FALSE"
+        )
+        for row in cur.fetchall():
+            rev = DBRevision(row)
+            md5 = rev.calculate_files_hash(db)
+            cur.execute(
+                "UPDATE revisions SET files_hash=%s WHERE id=%s", (md5, rev.dbid)
+            )
+
     def fetch_all_linked_packages(self, db, project, package):
         cur = db.cursor()
         cur.execute(
@@ -176,19 +190,86 @@ class Importer:
             (lproject, lpackage) = row
             self.update_db_package(db, lproject, lpackage)
 
+    def find_fake_revisions(self, db):
+        cur = db.cursor()
+        cur.execute(
+            """SELECT * from revisions WHERE
+                        id in (SELECT revision_id from linked_revs WHERE considered=FALSE) AND
+                        id not in (SELECT revision_id FROM fake_revs) ORDER by project,package,rev"""
+        )
+        for row in cur.fetchall():
+            rev = DBRevision(row)
+            prev = rev.previous_commit(db)
+            if not prev:
+                cur.execute(
+                    "UPDATE linked_revs SET considered=TRUE where revision_id=%s",
+                    (rev.dbid,),
+                )
+                continue
+            cur2 = db.cursor()
+            cur2.execute(
+                """SELECT * from revisions where id in 
+                            (SELECT revision_id from linked_revs WHERE linked_id=%s) 
+                            AND commit_time <= %s ORDER BY commit_time""",
+                (prev.dbid, rev.commit_time),
+            )
+            last_linked = None
+            for linked in cur2.fetchall():
+                linked = DBRevision(linked)
+                nextrev = linked.next_commit(db)
+                if nextrev and nextrev.commit_time < rev.commit_time:
+                    continue
+                last_linked = linked
+            cur.execute(
+                "UPDATE linked_revs SET considered=TRUE where revision_id=%s",
+                (rev.dbid,),
+            )
+            if last_linked:
+                linked = last_linked
+                cur2.execute(
+                    "SELECT 1 FROM fake_revs where revision_id=%s AND linked_id=%s",
+                    (rev.dbid, linked.dbid),
+                )
+                if cur2.fetchone():
+                    cur.execute(
+                        "UPDATE linked_revs SET considered=TRUE where revision_id=%s",
+                        (rev.dbid,),
+                    )
+                    continue
+                fake_rev = linked.rev + rev.rev / 1000.0
+                comment = f"Updating link to change in {rev.project}/{rev.package} revision {rev.rev}"
+                cur2.execute(
+                    """INSERT INTO revisions (project,package,rev,unexpanded_srcmd5, 
+                                commit_time, userid, comment) VALUES(%s,%s,%s,%s,%s,%s,%s) RETURNING id""",
+                    (
+                        linked.project,
+                        linked.package,
+                        fake_rev,
+                        linked.unexpanded_srcmd5,
+                        rev.commit_time,
+                        "buildservice-autocommit",
+                        comment,
+                    ),
+                )
+                new_id = cur2.fetchone()[0]
+                cur2.execute(
+                    """INSERT INTO linked_revs (revision_id, linked_id) VALUES (%s,%s)""",
+                    (new_id, rev.dbid),
+                )
+                cur2.execute(
+                    """INSERT INTO fake_revs (revision_id, linked_id) VALUES (%s,%s)""",
+                    (rev.dbid, linked.dbid),
+                )
+
     def import_into_db(self):
         db = DB()
         for project, _, api_url in self.projects:
             self.obs.change_url(api_url)
             self.update_db_package(db, project, self.package)
-            with db.cursor() as cur:
+            self.fetch_all_linked_packages(db, project, self.package)
-                cur.execute(
+            # all remaining, no filtering here
-                    "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",
+            self.find_linked_revs(db)
-                    (project, self.package),
+            self.find_fake_revisions(db)
-                )
-                for row in cur.fetchall():
-                    (lproject, lpackage) = row
-                    self.update_db_package(db, lproject, lpackage)
 
             missing_users = User.missing_users(db)
             for userid in missing_users:
@@ -203,7 +284,7 @@ class Importer:
                 with db.cursor() as cur:
                     cur.execute(
                         """SELECT unexpanded_srcmd5 from revisions WHERE
-                                id=(SELECT linked_id FROM linked_revs WHERE revision_id=%s""",
+                                id=(SELECT linked_id FROM linked_revs WHERE revision_id=%s)""",
                         (rev.dbid,),
                     )
                     linked_rev = cur.fetchone()

lib/obs_revision.py

@@ -56,7 +56,7 @@ class OBSRevision:
         return self
 
     def __str__(self):
-        return f"Rev {self.project}/{self.rev} Md5 {self.srcmd5} {self.time} {self.userid} {self.request_number}"
+        return f"Rev {self.project}/{self.package}/{self.rev}.0 Md5 {self.srcmd5} {self.time} {self.userid} {self.request_number}"
 
     def __repr__(self):
         return f"[{self.__str__()}]"

lib/tree_builder.py

@@ -12,10 +12,10 @@ class TreeBuilder:
         ret = []
         prev = None
         for rev in revisions:
-            print(rev, rev.files_hash(self.db))
+            print(rev, rev.files_hash)
             if rev.broken:
                 continue
-            if prev and prev.files_hash(self.db) == rev.files_hash(self.db):
+            if prev and prev.files_hash == rev.files_hash:
                 continue
             ret.append(rev)
             prev = rev
@@ -25,7 +25,7 @@ class TreeBuilder:
         factory_revisions = self.filtered_revisions("openSUSE:Factory", package)
         source_revisions = dict()
         for rev in factory_revisions:
-            print(rev, rev.files_hash(self.db))
+            print(rev, rev.files_hash)
             if rev.request_id:
                 req = Request.find(self.db, rev.request_id)
                 print("  ", req)
@@ -35,5 +35,5 @@ class TreeBuilder:
                         req.source_project, req.source_package
                     )
                 for rev2 in source_revisions.get(key):
-                    if rev2.files_hash(self.db) == rev.files_hash(self.db):
+                    if rev2.files_hash == rev.files_hash:
                         print("    ", rev2)