fork of https://github.com/sourcegraph/zoekt
zoekt-git-index: ignore UTF-8 BOM in .gitmodules
This is causing most of our persistent indexing errors, and fixing it
here is a better solution than explicitly blocking the strange
repositories.
author
Ryan Hitchman
committer
Ryan Hitchman
date
(Jun 7, 2021, 7:54 AM -0600)
commit
b491c4af
b491c4afee121202244d5851a3c92d1d96a77e6c
parent
800377be
800377bed22a37ea80c1d7840e6e8bdfff5f99e9
+49
-18
+13
-1
gitindex/submodule.go
+13
-1
gitindex/submodule.go
···
29
29
30
30
// ParseGitModules parses the contents of a .gitmodules file.
31
31
func ParseGitModules(content []byte) (map[string]*SubmoduleEntry, error) {
32
-
dec := config.NewDecoder(bytes.NewBuffer(content))
32
+
buf := bytes.NewBuffer(content)
33
+
34
+
// Handle the possibility that .gitmodules has a UTF-8 BOM, which would
35
+
// otherwise break the scanner.
36
+
// https://stackoverflow.com/a/21375405
37
+
r, _, err := buf.ReadRune()
38
+
if err != nil {
39
+
return nil, err
40
+
}
41
+
if r != '\uFEFF' {
42
+
buf.UnreadRune()
43
+
}
44
+
dec := config.NewDecoder(buf)
33
45
cfg := &config.Config{}
34
46
35
47
if err := dec.Decode(cfg); err != nil {
+36
-17
gitindex/submodule_test.go
+36
-17
gitindex/submodule_test.go
···
20
20
)
21
21
22
22
func TestParseGitModules(t *testing.T) {
23
-
testData := `[submodule "plugins/abc"]
24
-
path = plugins/abc
25
-
url = ../plugins/abc
26
-
branch = .`
27
-
28
-
got, err := ParseGitModules([]byte(testData))
29
-
if err != nil {
30
-
t.Fatalf("ParseGitModules: %T", err)
23
+
cases := []struct {
24
+
data string
25
+
want map[string]*SubmoduleEntry
26
+
}{{
27
+
`[submodule "plugins/abc"]
28
+
path = plugins/abc
29
+
url = ../plugins/abc
30
+
branch = .`,
31
+
map[string]*SubmoduleEntry{
32
+
"plugins/abc": {
33
+
Path: "plugins/abc",
34
+
URL: "../plugins/abc",
35
+
Branch: ".",
36
+
},
37
+
}},
38
+
{
39
+
"\uFEFF" + `[submodule "plugins/abc"]
40
+
path = plugins/abc
41
+
url = ../plugins/abc
42
+
branch = .`,
43
+
map[string]*SubmoduleEntry{
44
+
"plugins/abc": {
45
+
Path: "plugins/abc",
46
+
URL: "../plugins/abc",
47
+
Branch: ".",
48
+
},
49
+
}},
31
50
}
32
51
33
-
want := map[string]*SubmoduleEntry{
34
-
"plugins/abc": &SubmoduleEntry{
35
-
Path: "plugins/abc",
36
-
URL: "../plugins/abc",
37
-
Branch: ".",
38
-
},
39
-
}
40
-
if !reflect.DeepEqual(got, want) {
41
-
t.Fatalf("got %v, want %v", got, want)
52
+
for _, tc := range cases {
53
+
got, err := ParseGitModules([]byte(tc.data))
54
+
if err != nil {
55
+
t.Fatalf("ParseGitModules: %T", err)
56
+
}
57
+
58
+
if !reflect.DeepEqual(got, tc.want) {
59
+
t.Fatalf("got %v, want %v", got, tc.want)
60
+
}
42
61
}
43
62
}