#1160 Fix the way how KojiContentGenerator computes filesize.
Merged a month ago by jkaluza. Opened a month ago by jkaluza.
jkaluza/fm-orchestrator utf  into  master

@@ -407,11 +407,12 @@ 

          mmd_path = os.path.join(output_path, mmd_filename)

          try:

              with open(mmd_path, 'rb') as mmd_f:

-                 data = to_text_type(mmd_f.read())

+                 raw_data = mmd_f.read()

+                 data = to_text_type(raw_data)

                  mmd = Modulemd.Module().new_from_string(data)

                  ret['filename'] = mmd_filename

-                 ret['filesize'] = len(data)

-                 ret['checksum'] = hashlib.md5(data.encode('utf-8')).hexdigest()

+                 ret['filesize'] = len(raw_data)

+                 ret['checksum'] = hashlib.md5(raw_data).hexdigest()

          except IOError:

              if arch == "src":

                  # This might happen in case the Module is submitted directly

@@ -276,7 +276,7 @@ 

      @patch("module_build_service.builder.KojiContentGenerator.open", create=True)

      def test_get_arch_mmd_output(self, patched_open):

          patched_open.return_value = mock_open(

-             read_data=self.cg.mmd).return_value

+             read_data=self.cg.mmd.encode("utf-8")).return_value

          ret = self.cg._get_arch_mmd_output("./fake-dir", "x86_64")

          assert ret == {

              'arch': 'x86_64',

@@ -286,7 +286,7 @@ 

              'components': [],

              'extra': {'typeinfo': {'module': {}}},

              'filename': 'modulemd.x86_64.txt',

-             'filesize': 1136,

+             'filesize': 1138,

              'type': 'file'

          }

  

@@ -296,7 +296,7 @@ 

          rpm_artifacts = mmd.get_rpm_artifacts()

          rpm_artifacts.add("dhcp-libs-12:4.3.5-5.module_2118aef6.x86_64")

          mmd.set_rpm_artifacts(rpm_artifacts)

-         mmd_data = to_text_type(mmd.dumps())

+         mmd_data = bytes(mmd.dumps())

  

          patched_open.return_value = mock_open(

              read_data=mmd_data).return_value

@@ -336,7 +336,7 @@ 

                              u'version': '4.3.5'}],

              'extra': {'typeinfo': {'module': {}}},

              'filename': 'modulemd.x86_64.txt',

-             'filesize': 317,

+             'filesize': 319,

              'type': 'file'

          }

  

@@ -625,7 +625,7 @@ 

                  }

              ],

              "arch": "noarch",

-             "filesize": 1136,

+             "filesize": 1138,

              "checksum": "96b7739ffa3918e6ac3e3bd422b064ea",

              "checksum_type": "md5",

              "type": "file",

@@ -625,7 +625,7 @@ 

                  }

              ],

              "arch": "noarch",

-             "filesize": 1136,

+             "filesize": 1138,

              "checksum": "96b7739ffa3918e6ac3e3bd422b064ea",

              "checksum_type": "md5",

              "type": "file",

The current code reads the data, converts them to unicode string and
then uses the len() of that string as filesize. This is wrong,
because Koji expects filesize to be really number of bytes, not number
of characters.

Therefore, in this commit, the filesize is computed from raw data (bytes).

ret['checksum'] = hashlib.md5(raw_data).hexdigest()

Ah, I was wondering why the size is different (I would expect only ASCII characters in test data). It's because there is stray character in tests/staged_data/platform.yaml.

rebased onto 0408204

a month ago

@lholecek, updated based on your comment. Yes, we have added unicode character causing problems to test data in the past.

Commit b38feb5 fixes this pull-request

Pull-Request has been merged by jkaluza

a month ago

Pull-Request has been merged by jkaluza

a month ago