repozo --verify now correctly handles gzipped backups

src/ZODB/scripts/repozo.py

@@ -287,6 +287,22 @@ def checksum(fp, n):
     return sum.hexdigest()
 
 
+def file_size(fp):
+    # Compute number of bytes that can be read from fp
+    def func(data):
+        pass
+    return dofile(func, fp, None)
+
+
+def checksum_and_size(fp):
+    # Checksum and return it with the size of the file
+    sum = md5()
+    def func(data):
+        sum.update(data)
+    size = dofile(func, fp, None)
+    return sum.hexdigest(), size
+
+
 def copyfile(options, dst, start, n):
     # Copy bytes from file src, to file dst, starting at offset start, for n
     # length of bytes.  For robustness, we first write, flush and fsync
@@ -654,21 +670,39 @@ def do_verify(options):
                                     os.path.basename(fn))
             expected_size = endpos - startpos
             log("Verifying %s", filename)
-            # XXX: if the file is gzipped, we need to unzip it
             try:
+                if filename.endswith('fsz'):
+                    fp = gzip.open(filename, 'rb')
+                    when_uncompressed = ' (when uncompressed)'
+                else:
                     fp = open(filename, 'rb')
+                    when_uncompressed = ''
             except IOError:
                 error("%s is missing", filename)
+                continue
+            try:
+                fp.seek(0, 2)
+            except ValueError:
+                # can't seek in gzipped files
+                if options.quick:
+                    size = file_size(fp)
+                    actual_sum = None
                 else:
-                size = os.fstat(fp.fileno()).st_size
+                    actual_sum, size = checksum_and_size(fp)
+            else:
+                size = fp.tell()
+                if options.quick or size != expected_size:
+                    actual_sum = None
+                else:
+                    fp.seek(0)
+                    actual_sum = checksum(fp, size)
             if size != expected_size:
-                    error("%s is %d bytes, should be %d bytes", filename,
-                          size, expected_size)
+                error("%s is %d bytes%s, should be %d bytes", filename,
+                      size, when_uncompressed, expected_size)
             elif not options.quick:
-                    actual_sum = checksum(fp, size)
                 if actual_sum != sum:
-                        error("%s has checksum %s instead of %s", filename,
-                              actual_sum, sum)
+                    error("%s has checksum %s%s instead of %s", filename,
+                          actual_sum, when_uncompressed, sum)
             fp.close()
 
 

src/ZODB/scripts/tests/test_repozo.py

@@ -14,12 +14,7 @@
 from __future__ import print_function
 import unittest
 import os
-try:
-    # the hashlib package is available from Python 2.5
-    from hashlib import md5
-except ImportError:
-    # the md5 package is deprecated in Python 2.6
-    from md5 import new as md5
+from hashlib import md5
 
 import ZODB.tests.util  # layer used at class scope
 
@@ -812,7 +807,13 @@ class Test_do_verify(OptionsTestBase, unittest.TestCase):
         if text is None:
             text = name
         fqn = os.path.join(self._repository_directory, name)
-        f = _write_file(fqn, text.encode())
+        if ext.endswith('fsz'):
+            _opener = _GzipCloser
+        else:
+            _opener = open
+        with _opener(fqn, 'wb') as f:
+            f.write(text)
+            f.flush()
         return fqn
 
     def test_no_files(self):
@@ -829,6 +830,15 @@ class Test_do_verify(OptionsTestBase, unittest.TestCase):
            '/backup/2010-05-14-04-05-06.deltafs 3 7 f50881ced34c7d9e6bce100bf33dec60\n')
         self.assertEqual(self._callFUT(options), [])
 
+    def test_all_is_fine_gzip(self):
+        options = self._makeOptions(quick=False)
+        self._makeFile(2, 3, 4, '.fsz', 'AAA')
+        self._makeFile(4, 5, 6, '.deltafsz', 'BBBB')
+        self._makeFile(2, 3, 4, '.dat',
+           '/backup/2010-05-14-02-03-04.fsz 0 3 e1faffb3e614e6c2fba74296962386b7\n'
+           '/backup/2010-05-14-04-05-06.deltafsz 3 7 f50881ced34c7d9e6bce100bf33dec60\n')
+        self.assertEqual(self._callFUT(options), [])
+
     def test_missing_file(self):
         options = self._makeOptions(quick=True)
         self._makeFile(2, 3, 4, '.fs', 'AAA')
@@ -839,6 +849,16 @@ class Test_do_verify(OptionsTestBase, unittest.TestCase):
                          [options.repository + os.path.sep +
                           '2010-05-14-04-05-06.deltafs is missing'])
 
+    def test_missing_file_gzip(self):
+        options = self._makeOptions(quick=True)
+        self._makeFile(2, 3, 4, '.fsz', 'AAA')
+        self._makeFile(2, 3, 4, '.dat',
+           '/backup/2010-05-14-02-03-04.fsz 0 3 e1faffb3e614e6c2fba74296962386b7\n'
+           '/backup/2010-05-14-04-05-06.deltafsz 3 7 f50881ced34c7d9e6bce100bf33dec60\n')
+        self.assertEqual(self._callFUT(options),
+                         [options.repository + os.path.sep +
+                          '2010-05-14-04-05-06.deltafsz is missing'])
+
     def test_bad_size(self):
         options = self._makeOptions(quick=False)
         self._makeFile(2, 3, 4, '.fs', 'AAA')
@@ -851,6 +871,18 @@ class Test_do_verify(OptionsTestBase, unittest.TestCase):
                           '2010-05-14-04-05-06.deltafs is 3 bytes,'
                           ' should be 4 bytes'])
 
+    def test_bad_size_gzip(self):
+        options = self._makeOptions(quick=False)
+        self._makeFile(2, 3, 4, '.fsz', 'AAA')
+        self._makeFile(4, 5, 6, '.deltafsz', 'BBB')
+        self._makeFile(2, 3, 4, '.dat',
+           '/backup/2010-05-14-02-03-04.fsz 0 3 e1faffb3e614e6c2fba74296962386b7\n'
+           '/backup/2010-05-14-04-05-06.deltafsz 3 7 f50881ced34c7d9e6bce100bf33dec60\n')
+        self.assertEqual(self._callFUT(options),
+                         [options.repository + os.path.sep +
+                          '2010-05-14-04-05-06.deltafsz is 3 bytes (when uncompressed),'
+                          ' should be 4 bytes'])
+
     def test_bad_checksum(self):
         options = self._makeOptions(quick=False)
         self._makeFile(2, 3, 4, '.fs', 'AAA')
@@ -864,6 +896,19 @@ class Test_do_verify(OptionsTestBase, unittest.TestCase):
                           ' 36486440db255f0ee6ab109d5d231406 instead of'
                           ' f50881ced34c7d9e6bce100bf33dec60'])
 
+    def test_bad_checksum_gzip(self):
+        options = self._makeOptions(quick=False)
+        self._makeFile(2, 3, 4, '.fsz', 'AAA')
+        self._makeFile(4, 5, 6, '.deltafsz', 'BbBB')
+        self._makeFile(2, 3, 4, '.dat',
+           '/backup/2010-05-14-02-03-04.fsz 0 3 e1faffb3e614e6c2fba74296962386b7\n'
+           '/backup/2010-05-14-04-05-06.deltafsz 3 7 f50881ced34c7d9e6bce100bf33dec60\n')
+        self.assertEqual(self._callFUT(options),
+                         [options.repository + os.path.sep +
+                          '2010-05-14-04-05-06.deltafsz has checksum'
+                          ' 36486440db255f0ee6ab109d5d231406 (when uncompressed) instead of'
+                          ' f50881ced34c7d9e6bce100bf33dec60'])
+
     def test_quick_ignores_checksums(self):
         options = self._makeOptions(quick=True)
         self._makeFile(2, 3, 4, '.fs', 'AAA')
@@ -873,6 +918,15 @@ class Test_do_verify(OptionsTestBase, unittest.TestCase):
            '/backup/2010-05-14-04-05-06.deltafs 3 7 bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\n')
         self.assertEqual(self._callFUT(options), [])
 
+    def test_quick_ignores_checksums_gzip(self):
+        options = self._makeOptions(quick=True)
+        self._makeFile(2, 3, 4, '.fsz', 'AAA')
+        self._makeFile(4, 5, 6, '.deltafsz', 'BBBB')
+        self._makeFile(2, 3, 4, '.dat',
+           '/backup/2010-05-14-02-03-04.fsz 0 3 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\n'
+           '/backup/2010-05-14-04-05-06.deltafsz 3 7 bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\n')
+        self.assertEqual(self._callFUT(options), [])
+
 
 class MonteCarloTests(unittest.TestCase):