zodbanalyze: tests: Verify CSV mode via golden output generated during gen_testdata

Instead of putting expected output into test code.

The reason for this change is that soon, with many kinds of generated
ZODB databases, analyze output will be different for each kind, and the
only practical way to make sure that analyze works correctly is to have
good expected output for each kind.

We already do this way for zodbdump. Start doing the same for
zodbanalyze as a preparatory step as well.

zodbtools/test/gen_testdata.py

@@ -18,7 +18,10 @@
 #
 # See COPYING file for full licensing terms.
 # See https://www.nexedi.com/licensing for rationale and options.
-"""generate reference database and index for tests"""
+"""generate reference database and index for tests
+
+Golden zodbdump & zodbanalyze outputs are also generated besides database itself.
+"""
 
 # NOTE result of this script must be saved in version control and should not be
 # generated at the time when tests are run. This is because even though we make
@@ -306,6 +309,7 @@ def _gen_testdb(outfs_path, zext):
 # ----------------------------------------
 
 from zodbtools.zodbdump import zodbdump
+from zodbtools import zodbanalyze
 from zodbtools.test.testutil import zext_supported
 
 def main():
@@ -321,10 +325,23 @@ def main():
         for f in glob.glob(dbname + '.*'):
             os.remove(f)
         gen_testdb("%s.fs" % dbname, zext=zext)
+
+        # prepare zdump.ok for generated database
         stor = FileStorage("%s.fs" % dbname, read_only=True)
         for pretty in ('raw', 'zpickledis'):
             with open("%s.zdump.%s.ok" % (dbname, pretty), "wb") as f:
                 zodbdump(stor, None, None, pretty=pretty, out=f)
 
+        # prepare zanalyze.csv.ok
+        sys_stdout = sys.stdout
+        sys.stdout = open("%s.zanalyze.csv.ok" % dbname, "w")
+        zodbanalyze.report(
+            zodbanalyze.analyze("%s.fs" % dbname, use_dbm=False, delta_fs=False, tidmin=None, tidmax=None),
+            csv=True,
+        )
+        sys.stdout.close()
+        sys.stdout = sys_stdout
+
+
 if __name__ == '__main__':
     main()

zodbtools/test/test_analyze.py

@@ -25,8 +25,9 @@ from golang import b
 
 
 def test_zodbanalyze(tmpdir, capsys):
+    testdata = os.path.join(os.path.dirname(__file__), "testdata")
     tfs1 = fs1_testdata_py23(tmpdir,
-                    os.path.join(os.path.dirname(__file__), "testdata", "1.fs"))
+                    os.path.join(testdata, "1.fs"))
 
     for use_dbm in (False, True):
         report(
@@ -55,13 +56,11 @@ def test_zodbanalyze(tmpdir, capsys):
         csv=True,
     )
     captured = capsys.readouterr()
-    assert (
-        """Class Name,T.Count,T.Bytes,Pct,AvgSize,C.Count,C.Bytes,O.Count,O.Bytes
-persistent.mapping.PersistentMapping,3,639,22.468354%,213.000000,1,213,2,426
-__main__.Object,65,2205,77.531646%,33.923077,9,310,56,1895
-"""
-        == captured.out
-    )
+
+    with open('%s/1.zanalyze.csv.ok' % testdata, 'r') as f:
+        zanalyze_csv_ok = f.read()
+
+    assert captured.out == zanalyze_csv_ok
     assert captured.err == ""
 
     # empty range

zodbtools/test/testdata/1.zanalyze.csv.ok

+Class Name,T.Count,T.Bytes,Pct,AvgSize,C.Count,C.Bytes,O.Count,O.Bytes
+persistent.mapping.PersistentMapping,3,639,22.523793%,213.000000,1,213,2,426
+__main__.Object,65,2198,77.476207%,33.815385,9,317,56,1881

zodbtools/test/testdata/1_!zext.zanalyze.csv.ok

+Class Name,T.Count,T.Bytes,Pct,AvgSize,C.Count,C.Bytes,O.Count,O.Bytes
+persistent.mapping.PersistentMapping,3,639,22.523793%,213.000000,1,213,2,426
+__main__.Object,65,2198,77.476207%,33.815385,9,317,56,1881