@@ -139,6 +139,17 @@ def test_revisionrange_parse(revision_range, expect):
139139 assert result == expect
140140
141141
142+ def git_call (cmd , encoding = None ):
143+ """Returns a mocked call to git"""
144+ return call (
145+ cmd .split (),
146+ cwd = str (Path ("/path" )),
147+ encoding = encoding ,
148+ stderr = PIPE ,
149+ env = {"LC_ALL" : "C" , "PATH" : os .environ ["PATH" ]},
150+ )
151+
152+
142153@pytest .mark .kwparametrize (
143154 dict (
144155 revision = ":WORKTREE:" ,
@@ -147,31 +158,31 @@ def test_revisionrange_parse(revision_range, expect):
147158 dict (
148159 revision = "HEAD" ,
149160 expect_git_calls = [
150- "git show HEAD:./my.txt" ,
151- "git log -1 --format=%ct HEAD -- my.txt" ,
161+ git_call ( "git show HEAD:./my.txt" ) ,
162+ git_call ( "git log -1 --format=%ct HEAD -- my.txt" , encoding = "utf-8" ) ,
152163 ],
153164 expect_textdocument_calls = [
154- call .from_lines ([ b"1627107028" ] , mtime = "2021-07-24 06:10:28.000000 +0000" )
165+ call .from_bytes ( b"1627107028" , mtime = "2021-07-24 06:10:28.000000 +0000" )
155166 ],
156167 ),
157168 dict (
158169 revision = "HEAD^" ,
159170 expect_git_calls = [
160- "git show HEAD^:./my.txt" ,
161- "git log -1 --format=%ct HEAD^ -- my.txt" ,
171+ git_call ( "git show HEAD^:./my.txt" ) ,
172+ git_call ( "git log -1 --format=%ct HEAD^ -- my.txt" , encoding = "utf-8" ) ,
162173 ],
163174 expect_textdocument_calls = [
164- call .from_lines ([ b"1627107028" ] , mtime = "2021-07-24 06:10:28.000000 +0000" )
175+ call .from_bytes ( b"1627107028" , mtime = "2021-07-24 06:10:28.000000 +0000" )
165176 ],
166177 ),
167178 dict (
168179 revision = "master" ,
169180 expect_git_calls = [
170- "git show master:./my.txt" ,
171- "git log -1 --format=%ct master -- my.txt" ,
181+ git_call ( "git show master:./my.txt" ) ,
182+ git_call ( "git log -1 --format=%ct master -- my.txt" , encoding = "utf-8" ) ,
172183 ],
173184 expect_textdocument_calls = [
174- call .from_lines ([ b"1627107028" ] , mtime = "2021-07-24 06:10:28.000000 +0000" )
185+ call .from_bytes ( b"1627107028" , mtime = "2021-07-24 06:10:28.000000 +0000" )
175186 ],
176187 ),
177188 expect_git_calls = [],
@@ -189,17 +200,7 @@ def test_git_get_content_at_revision_obtain_file_content(
189200
190201 git .git_get_content_at_revision (Path ("my.txt" ), revision , Path ("/path" ))
191202
192- expected_calls = [
193- call (
194- expected_call .split (),
195- cwd = str (Path ("/path" )),
196- encoding = "utf-8" ,
197- stderr = PIPE ,
198- env = {"LC_ALL" : "C" , "PATH" : os .environ ["PATH" ]},
199- )
200- for expected_call in expect_git_calls
201- ]
202- assert check_output .call_args_list == expected_calls
203+ assert check_output .call_args_list == expect_git_calls
203204 assert text_document_class .method_calls == expect_textdocument_calls
204205
205206
@@ -403,6 +404,58 @@ def test_git_get_content_at_revision_stderr(git_repo, capfd, caplog):
403404 assert caplog .text == ""
404405
405406
407+ @pytest .fixture (scope = "module" )
408+ def encodings_repo (tmp_path_factory ):
409+ """Create an example Git repository using various encodings for the same file"""
410+ tmpdir = tmp_path_factory .mktemp ("branched_repo" )
411+ git_repo = GitRepoFixture .create_repository (tmpdir )
412+ # Commit without an encoding cookie, defaults to utf-8
413+ git_repo .add ({"file.py" : "darker = 'plus foncé'\n " }, commit = "Default encoding" )
414+ git_repo .create_tag ("default" )
415+ # Commit without an encoding cookie but with a utf-8 signature
416+ content = "darker = 'plus foncé'\n " .encode ("utf-8-sig" )
417+ git_repo .add ({"file.py" : content }, commit = "utf-8-sig" )
418+ git_repo .create_tag ("utf-8-sig" )
419+ # Commit with an iso-8859-1 encoding cookie
420+ content = "# coding: iso-8859-1\n darker = 'plus foncé'\n " .encode ("iso-8859-1" )
421+ git_repo .add ({"file.py" : content }, commit = "iso-8859-1" )
422+ git_repo .create_tag ("iso-8859-1" )
423+ # Commit with a utf-8 encoding cookie
424+ content = "# coding: utf-8\n python = 'パイソン'\n " .encode ("utf-8" )
425+ git_repo .add ({"file.py" : content }, commit = "utf-8" )
426+ git_repo .create_tag ("utf-8" )
427+ # Current worktree content (not committed) with a shitfjs encoding cookie
428+ content = "# coding: shiftjis\n python = 'パイソン'\n " .encode ("shiftjis" )
429+ git_repo .add ({"file.py" : content })
430+ return git_repo
431+
432+
433+ @pytest .mark .kwparametrize (
434+ dict (commit = "default" , encoding = "utf-8" , lines = ("darker = 'plus foncé'" ,)),
435+ dict (commit = "utf-8-sig" , encoding = "utf-8-sig" , lines = ("darker = 'plus foncé'" ,)),
436+ dict (
437+ commit = "iso-8859-1" ,
438+ encoding = "iso-8859-1" ,
439+ lines = ("# coding: iso-8859-1" , "darker = 'plus foncé'" ),
440+ ),
441+ dict (
442+ commit = "utf-8" , encoding = "utf-8" , lines = ("# coding: utf-8" , "python = 'パイソン'" )
443+ ),
444+ dict (
445+ commit = ":WORKTREE:" ,
446+ encoding = "shiftjis" ,
447+ lines = ("# coding: shiftjis" , "python = 'パイソン'" ),
448+ ),
449+ )
450+ def test_git_get_content_at_revision_encoding (encodings_repo , commit , encoding , lines ):
451+ """Git file is loaded using its historical encoding"""
452+ result = git .git_get_content_at_revision (
453+ Path ("file.py" ), commit , encodings_repo .root
454+ )
455+ assert result .encoding == encoding
456+ assert result .lines == lines
457+
458+
406459@pytest .mark .kwparametrize (
407460 dict (retval = 0 , expect = True ),
408461 dict (retval = 1 , expect = False ),
0 commit comments