Merge ~cjwatson/launchpad:unsixify-tagfile-parsing into launchpad:master

[Colin Watson]

Remove six from code dealing with parsed tagfiles

`parse_tagfile_content` always returns header values as bytes (now type-annotated), so we can just use `.decode`/`.encode` rather than `six.ensure_*`.

[Jürgen Gmach (jugmac00)]

review: Approve

diff --git a/lib/lp/archiveuploader/changesfile.py b/lib/lp/archiveuploader/changesfile.py
index 61c722c..6f0f81d 100644
--- a/lib/lp/archiveuploader/changesfile.py
+++ b/lib/lp/archiveuploader/changesfile.py
@@ -15,8 +15,6 @@ __all__ = [
 
 import os
 
-import six
-
 from lp.archiveuploader.buildinfofile import BuildInfoFile
 from lp.archiveuploader.dscfile import DSCFile, SignableTagFile
 from lp.archiveuploader.nascentuploadfile import (
@@ -267,7 +265,7 @@ class ChangesFile(SignableTagFile):
             # Urgency is recommended but not mandatory. Default to 'low'
             self._dict["Urgency"] = b"low"
 
-        raw_urgency = six.ensure_text(self._dict["Urgency"]).lower()
+        raw_urgency = self._dict["Urgency"].decode().lower()
         if raw_urgency not in self.urgency_map:
             yield UploadWarning(
                 "Unable to grok urgency %s, overriding with 'low'"
@@ -326,7 +324,7 @@ class ChangesFile(SignableTagFile):
 
         For example, 'hoary' or 'hoary-security'.
         """
-        return six.ensure_text(self._dict["Distribution"])
+        return self._dict["Distribution"].decode()
 
     @property
     def architectures(self):
@@ -335,24 +333,22 @@ class ChangesFile(SignableTagFile):
         For instance ['source', 'all'] or ['source', 'i386', 'amd64']
         or ['source'].
         """
-        return set(six.ensure_text(self._dict["Architecture"]).split())
+        return set(self._dict["Architecture"].decode().split())
 
     @property
     def binaries(self):
         """Return set of binary package names listed."""
-        return set(
-            six.ensure_text(self._dict.get("Binary", "")).strip().split()
-        )
+        return set(self._dict.get("Binary", "").decode().strip().split())
 
     @property
     def converted_urgency(self):
         """Return the appropriate SourcePackageUrgency item."""
-        return self.urgency_map[six.ensure_text(self._dict["Urgency"]).lower()]
+        return self.urgency_map[self._dict["Urgency"].decode().lower()]
 
     @property
     def version(self):
         """Return changesfile claimed version."""
-        return six.ensure_text(self._dict["Version"])
+        return self._dict["Version"].decode()
 
     @classmethod
     def formatChangesComment(cls, comment):
@@ -376,17 +372,17 @@ class ChangesFile(SignableTagFile):
     @property
     def date(self):
         """Return changesfile date."""
-        return six.ensure_text(self._dict["Date"])
+        return self._dict["Date"].decode()
 
     @property
     def source(self):
         """Return changesfile claimed source name."""
-        return six.ensure_text(self._dict["Source"])
+        return self._dict["Source"].decode()
 
     @property
     def architecture_line(self):
         """Return changesfile archicteture line."""
-        return six.ensure_text(self._dict["Architecture"])
+        return self._dict["Architecture"].decode()
 
     @property
     def simulated_changelog(self):
diff --git a/lib/lp/archiveuploader/dscfile.py b/lib/lp/archiveuploader/dscfile.py
index e7516ff..59b5087 100644
--- a/lib/lp/archiveuploader/dscfile.py
+++ b/lib/lp/archiveuploader/dscfile.py
@@ -23,7 +23,6 @@ import shutil
 import tempfile
 import warnings
 
-import six
 from debian.deb822 import Deb822Dict, PkgRelation
 from zope.component import getUtility
 
@@ -220,8 +219,8 @@ class SignableTagFile:
             raise UploadError("Invalid Maintainer.")
 
         if person is None and self.policy.create_people:
-            package = six.ensure_text(self._dict["Source"])
-            version = six.ensure_text(self._dict["Version"])
+            package = self._dict["Source"].decode()
+            version = self._dict["Version"].decode()
             if self.policy.distroseries and self.policy.pocket:
                 policy_suite = "%s/%s" % (
                     self.policy.distroseries.name,
@@ -344,8 +343,7 @@ class DSCFile(SourceUploadFile, SignableTagFile):
 
         if self.format is None:
             raise EarlyReturnUploadError(
-                "Unsupported source format: %s"
-                % six.ensure_str(self._dict["Format"])
+                "Unsupported source format: %s" % self._dict["Format"].decode()
             )
 
     #
@@ -354,19 +352,19 @@ class DSCFile(SourceUploadFile, SignableTagFile):
     @property
     def source(self):
         """Return the DSC source name."""
-        return six.ensure_text(self._dict["Source"])
+        return self._dict["Source"].decode()
 
     @property
     def dsc_version(self):
         """Return the DSC source version."""
-        return six.ensure_text(self._dict["Version"])
+        return self._dict["Version"].decode()
 
     @property
     def format(self):
         """Return the DSC format."""
         try:
             return SourcePackageFormat.getTermByToken(
-                six.ensure_text(self._dict["Format"])
+                self._dict["Format"].decode()
             ).value
         except LookupError:
             return None
@@ -374,12 +372,12 @@ class DSCFile(SourceUploadFile, SignableTagFile):
     @property
     def architecture(self):
         """Return the DSC source architecture."""
-        return six.ensure_text(self._dict["Architecture"])
+        return self._dict["Architecture"].decode()
 
     @property
     def binary(self):
         """Return the DSC claimed binary line."""
-        return six.ensure_text(self._dict["Binary"])
+        return self._dict["Binary"].decode()
 
     #
     # DSC file checks.
@@ -449,7 +447,7 @@ class DSCFile(SourceUploadFile, SignableTagFile):
         for field_name in ["Build-Depends", "Build-Depends-Indep"]:
             field = self._dict.get(field_name, None)
             if field is not None:
-                field = six.ensure_text(field)
+                field = field.decode()
                 if field.startswith("ARRAY"):
                     yield UploadError(
                         "%s: invalid %s field produced by a broken version "
diff --git a/lib/lp/archiveuploader/tagfiles.py b/lib/lp/archiveuploader/tagfiles.py
index 7470f27..7815394 100644
--- a/lib/lp/archiveuploader/tagfiles.py
+++ b/lib/lp/archiveuploader/tagfiles.py
@@ -7,6 +7,7 @@ __all__ = ["TagFileParseError", "parse_tagfile", "parse_tagfile_content"]
 
 
 import tempfile
+from typing import Dict, Optional
 
 import apt_pkg
 
@@ -19,7 +20,9 @@ class TagFileParseError(Exception):
     pass
 
 
-def parse_tagfile_content(content, filename=None):
+def parse_tagfile_content(
+    content: bytes, filename: Optional[str] = None
+) -> Dict[str, bytes]:
     """Parses a tag file and returns a dictionary where each field is a key.
 
     The mandatory first argument is the contents of the tag file as a
@@ -56,7 +59,7 @@ def parse_tagfile_content(content, filename=None):
     return trimmed_dict
 
 
-def parse_tagfile(filename):
+def parse_tagfile(filename: str) -> Dict[str, bytes]:
     """Parses a tag file and returns a dictionary where each field is a key.
 
     The mandatory first argument is the filename of the tag file, and
diff --git a/lib/lp/archiveuploader/tests/nascentupload-closing-bugs.rst b/lib/lp/archiveuploader/tests/nascentupload-closing-bugs.rst
index 68eff30..4a47593 100644
--- a/lib/lp/archiveuploader/tests/nascentupload-closing-bugs.rst
+++ b/lib/lp/archiveuploader/tests/nascentupload-closing-bugs.rst
@@ -82,7 +82,7 @@ This new version fixes bug #6 according its changesfiles:
     >>> print(bar2_src.changes.changed_by["person"].name)
     kinnison
 
-    >>> print(six.ensure_str(bar2_src.changes._dict["Launchpad-bugs-fixed"]))
+    >>> print(bar2_src.changes._dict["Launchpad-bugs-fixed"].decode())
     6
 
     >>> print(bar2_src.changes.changes_comment)
diff --git a/lib/lp/archiveuploader/tests/nascentuploadfile.rst b/lib/lp/archiveuploader/tests/nascentuploadfile.rst
index 357205c..78bfed5 100644
--- a/lib/lp/archiveuploader/tests/nascentuploadfile.rst
+++ b/lib/lp/archiveuploader/tests/nascentuploadfile.rst
@@ -274,8 +274,8 @@ Some fields extracted from the tag_file are required, they are always
 present in ChangesFile and DSCFile:
 
     >>> sig_file._dict = {}
-    >>> sig_file._dict["Source"] = "some-source"
-    >>> sig_file._dict["Version"] = "6.6.6"
+    >>> sig_file._dict["Source"] = b"some-source"
+    >>> sig_file._dict["Version"] = b"6.6.6"
 
 After initialising sig_file we can parse addresses and look them up in
 Launchpad:
diff --git a/lib/lp/archiveuploader/tests/test_dscfile.py b/lib/lp/archiveuploader/tests/test_dscfile.py
index b3365a2..23a15ca 100644
--- a/lib/lp/archiveuploader/tests/test_dscfile.py
+++ b/lib/lp/archiveuploader/tests/test_dscfile.py
@@ -168,8 +168,8 @@ class TestSignableTagFile(TestCaseWithFactory):
         tagfile.logger = DevNullLogger()
         tagfile.policy = FakePolicy(None, None, create_people=True)
         tagfile._dict = {
-            "Source": "arbitrary-source-package-name",
-            "Version": "1.0",
+            "Source": b"arbitrary-source-package-name",
+            "Version": b"1.0",
         }
         return tagfile
 
diff --git a/lib/lp/soyuz/doc/soyuz-upload.rst b/lib/lp/soyuz/doc/soyuz-upload.rst
index f32f5f7..b456a95 100644
--- a/lib/lp/soyuz/doc/soyuz-upload.rst
+++ b/lib/lp/soyuz/doc/soyuz-upload.rst
@@ -60,14 +60,14 @@ been uploaded over FTP.
     ...         tf = {}
     ...
     ...     if "Source" in tf:
-    ...         package_names.append(six.ensure_text(tf["Source"]))
+    ...         package_names.append(tf["Source"].decode())
     ...
     ...     send_filepaths = [changes_filepath]
     ...     if "Files" in tf:
     ...         send_filepaths.extend(
     ...             [
     ...                 os.path.join(test_files_dir, line.split()[-1])
-    ...                 for line in six.ensure_text(tf["Files"]).splitlines()
+    ...                 for line in tf["Files"].decode().splitlines()
     ...                 if line
     ...             ]
     ...         )
diff --git a/lib/lp/soyuz/scripts/gina/archive.py b/lib/lp/soyuz/scripts/gina/archive.py
index f283190..ad06de5 100644
--- a/lib/lp/soyuz/scripts/gina/archive.py
+++ b/lib/lp/soyuz/scripts/gina/archive.py
@@ -20,7 +20,6 @@ import tempfile
 from collections import defaultdict
 
 import apt_pkg
-import six
 
 from lp.services.scripts import log
 from lp.soyuz.scripts.gina import call
@@ -236,10 +235,8 @@ class PackagesMap:
                 for section in sources:
                     try:
                         src_tmp = dict(section)
-                        src_tmp["Component"] = six.ensure_binary(
-                            info_set.component
-                        )
-                        src_name = six.ensure_text(src_tmp["Package"])
+                        src_tmp["Component"] = info_set.component.encode()
+                        src_name = src_tmp["Package"].decode()
                     except KeyError:
                         log.exception(
                             "Invalid Sources stanza in %s",
@@ -267,10 +264,8 @@ class PackagesMap:
                 try:
                     bin_tmp = dict(section)
                     # The component isn't listed in the tagfile.
-                    bin_tmp["Component"] = six.ensure_binary(
-                        info_set.component
-                    )
-                    bin_name = six.ensure_text(bin_tmp["Package"])
+                    bin_tmp["Component"] = info_set.component.encode()
+                    bin_name = bin_tmp["Package"].decode()
                 except KeyError:
                     log.exception(
                         "Invalid Releases stanza in %s",
@@ -284,10 +279,8 @@ class PackagesMap:
             for section in dibinaries:
                 try:
                     dibin_tmp = dict(section)
-                    dibin_tmp["Component"] = six.ensure_binary(
-                        info_set.component
-                    )
-                    dibin_name = six.ensure_text(dibin_tmp["Package"])
+                    dibin_tmp["Component"] = info_set.component.encode()
+                    dibin_name = dibin_tmp["Package"].decode()
                 except KeyError:
                     log.exception(
                         "Invalid D-I Releases stanza in %s" % info_set.difile
diff --git a/lib/lp/soyuz/scripts/tests/test_gina.py b/lib/lp/soyuz/scripts/tests/test_gina.py
index 3c37297..edff14e 100644
--- a/lib/lp/soyuz/scripts/tests/test_gina.py
+++ b/lib/lp/soyuz/scripts/tests/test_gina.py
@@ -11,7 +11,6 @@ from textwrap import dedent
 from unittest import TestLoader
 
 import apt_pkg
-import six
 import transaction
 from fixtures import EnvironmentVariableFixture
 from testtools.matchers import MatchesSetwise, MatchesStructure
@@ -338,7 +337,7 @@ class TestSourcePackageData(TestCaseWithFactory):
             )
 
         dsc_contents = parse_tagfile(dsc_path)
-        dsc_contents["Directory"] = six.ensure_binary(pool_dir)
+        dsc_contents["Directory"] = pool_dir.encode()
         dsc_contents["Package"] = b"foo"
         dsc_contents["Component"] = b"main"
         dsc_contents["Section"] = b"misc"
@@ -409,7 +408,7 @@ class TestSourcePackageData(TestCaseWithFactory):
             )
 
         dsc_contents = parse_tagfile(dsc_path)
-        dsc_contents["Directory"] = six.ensure_binary(pool_dir)
+        dsc_contents["Directory"] = pool_dir.encode()
         dsc_contents["Package"] = b"foo"
         dsc_contents["Component"] = b"main"
         dsc_contents["Section"] = b"misc"