PR#17: Lookaside: encoding repo name to UTF-8 - rpkg

rpkg

#17 Lookaside: encoding repo name to UTF-8

Closed 8 years ago by lsedlar. Opened 8 years ago by araszka.

araszka/rpkg bz1241059 into master

Lookaside: encoding repo name to UTF-8

Ales Raszka • 8 years ago

67964c5

src/pyrpkg/__init__.py

file modified

+5 -2

		`@@ -387,7 +387,6 @@`

		`def load_push_url(self):`
		`"""Find the pushurl or url of remote of branch we're on."""`
		`-`
		`try:`
		`url = self.repo.git.config('--get', 'remote.%s.pushurl'`
		`% self.branch_remote)`
		`@@ -397,7 +396,11 @@`
		`% self.branch_remote)`
		`except git.GitCommandError as e:`
		`raise rpkgError('Unable to find remote push url: %s' % e)`
		`- self._push_url = url`
		`+ if isinstance(url, unicode):`
		`+ # GitPython >= 1.0 return unicode. It must be encoded to string.`
		`+ self._push_url = url.encode('utf-8')`
		`+ else:`
		`+ self._push_url = url`

		`@property`
		`def commithash(self):`

test/commands/test_package_name.py

file added

+24

		`@@ -0,0 +1,24 @@`
		`+ import os`
		`+`
		`+ from . import CommandTestCase`
		`+`
		`+`
		`+ class CommandPackageNameTestCase(CommandTestCase):`
		`+ def test_name_is_not_unicode(self):`
		`+ self.make_new_git(self.module)`
		`+`
		`+ import pyrpkg`
		`+ cmd = pyrpkg.Commands(self.path, self.lookaside, self.lookasidehash,`
		`+ self.lookaside_cgi, self.gitbaseurl,`
		`+ self.anongiturl, self.branchre, self.kojiconfig,`
		`+ self.build_client, self.user, self.dist,`
		`+ self.target, self.quiet)`
		`+ cmd.clone(self.module, anon=True)`
		`+`
		`+ moduledir = os.path.join(self.path, self.module)`
		`+ cmd.path = moduledir`
		`+`
		`+ # pycurl can't handle unicode variable`
		`+ # module_name needs to be string`
		`+ self.assertNotEquals(type(cmd.module_name), unicode)`
		`+ self.assertEquals(type(cmd.module_name), str)`

no initial comment

pbabinca commented 8 years ago

I don't like the proposed change:
- Find where the Unicode string firstly appeared and convert it there.
- Sources have unit tests so write a test to cover this issue

lsedlar commented on line 15 of src/pyrpkg/__init__.py 8 years ago

I would much prefer to only encode if url is actually a unicode instance. As is, the code would crash if someone has older GitPython (that returns strs), and the url actually contains non-ascii characters.

>>> "ščř".encode('utf-8')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
UnicodeDecodeError: 'ascii' codec can't decode byte 0xc5 in position 0: ordinal not in range(128)

Edited 8 years ago by lsedlar

lsedlar commented 8 years ago

I have rebase the patch on master and merged.