Implement a fileshash function for revision

lib/db_revision.py

@@ -1,3 +1,5 @@
+from hashlib import md5
+
 from lib.request import Request
 
 
@@ -19,6 +21,7 @@ class DBRevision:
             self.request_id,
         ) = row
         self.rev = int(self.rev)
+        self._files = None
 
     def __str__(self):
         return f"Rev {self.project}/{self.rev} Md5 {self.unexpanded_srcmd5} {self.commit_time} {self.userid} {self.request_number}"
@@ -156,14 +159,31 @@ class DBRevision:
                     ),
                 )
 
+    def fileshash(self, db):
+        m = md5()
+        for file_dict in self.files_list(db):
+            m.update(
+                (
+                    file_dict["name"]
+                    + "/"
+                    + file_dict["md5"]
+                    + "/"
+                    + str(file_dict["size"])
+                ).encode("utf-8")
+            )
+        return m.hexdigest()
+
     def files_list(self, db):
+        if self._files:
+            return self._files
         cur = db.cursor()
         cur.execute("SELECT * from files where revision_id=%s", (self.dbid,))
-        ret = []
+        self._files = []
         for row in cur.fetchall():
             (_, _, name, md5, size, mtime) = row
-            ret.append({"md5": md5, "size": size, "mtime": mtime, "name": name})
+            self._files.append({"md5": md5, "size": size, "mtime": mtime, "name": name})
-        return ret
+        self._files.sort(key=lambda x: x["name"])
+        return self._files
 
     @staticmethod
     def requests_to_fetch(self, db, project, package):

lib/request.py

@@ -3,6 +3,7 @@ class Request:
         self.number = int(xml.get("id"))
         self.creator = xml.get("creator")
         self.state = xml.find("state").get("name")
+        # not used in the database
         self.target = xml.find("action/target").get("project")
 
         self.type_ = xml.find("action").get("type")
@@ -28,7 +29,7 @@ class Request:
         return self.type_
 
     def __str__(self):
-        return f"Req {self.number} {self.creator} {self.type_} {self.source_project}->{self.target} {self.state}"
+        return f"Req {self.number} {self.creator} {self.type_} {self.source_project} {self.state}"
 
     def __repr__(self):
         return f"[{self.__str__()}]"

tests/tree_test.py

@@ -1,11 +1,11 @@
 import os
 import unittest
-from pprint import pprint
 
 import yaml
 
 from lib.db import DB
 from lib.db_revision import DBRevision
+from lib.request import Request
 
 
 class TestTreeMethods(unittest.TestCase):
@@ -19,7 +19,21 @@ class TestTreeMethods(unittest.TestCase):
 
     def test_create_tree(self):
         factory_revisions = DBRevision.all_revisions(self.db, "openSUSE:Factory", "zsh")
-        pprint(sorted(factory_revisions))
+        source_revisions = dict()
+        for rev in sorted(factory_revisions):
+            print(rev, rev.fileshash(self.db))
+            if rev.request_id:
+                req = Request.find(self.db, rev.request_id)
+                print("  ", req)
+                key = f"{req.source_project}/{req.source_package}"
+                if key not in source_revisions:
+                    source_revisions[key] = sorted(
+                        DBRevision.all_revisions(
+                            self.db, req.source_project, req.source_package
+                        )
+                    )
+                for rev2 in source_revisions.get(key):
+                    print(rev2, rev2.fileshash(self.db))
 
 
 if __name__ == "__main__":