|
1
|
|
|
package util |
|
2
|
|
|
|
|
3
|
|
|
import ( |
|
4
|
|
|
"strings" |
|
5
|
|
|
"unicode" |
|
6
|
|
|
) |
|
7
|
|
|
|
|
8
|
|
|
var ( |
|
9
|
|
|
separators = &unicode.RangeTable{ |
|
10
|
|
|
R16: []unicode.Range16{ |
|
11
|
|
|
{0x002c, 0x002c, 1}, //comma |
|
12
|
|
|
{0x003b, 0x003b, 1}, //semicolon |
|
13
|
|
|
}, |
|
14
|
|
|
} |
|
15
|
|
|
quotationMarks = &unicode.RangeTable{ |
|
16
|
|
|
R16: []unicode.Range16{ |
|
17
|
|
|
{0x0022, 0x0022, 1}, //quot |
|
18
|
|
|
}, |
|
19
|
|
|
} |
|
20
|
|
|
trimRunes = `,; "` |
|
21
|
|
|
) |
|
22
|
|
|
|
|
23
|
|
|
type tokenizer struct { |
|
24
|
|
|
keyword string |
|
25
|
|
|
keywords []string |
|
26
|
|
|
lastIndex int |
|
27
|
|
|
quotFound bool |
|
28
|
|
|
} |
|
29
|
|
|
|
|
30
|
|
|
var t tokenizer |
|
31
|
|
|
|
|
32
|
|
|
func (t *tokenizer) addKeyword(input string, index int) { |
|
33
|
|
|
keyword := fetchKeyword(input, t.lastIndex, index) |
|
34
|
|
|
t.lastIndex = index |
|
35
|
|
|
|
|
36
|
|
|
if len(keyword) != 0 { |
|
37
|
|
|
t.append(keyword) |
|
38
|
|
|
} |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
func (t *tokenizer) append(keyword string) { |
|
42
|
|
|
t.keywords = append(t.keywords, keyword) |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
func (t *tokenizer) toggleQuot() { |
|
46
|
|
|
t.quotFound = !t.quotFound |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
func SplitKeywords(input string) []string { |
|
50
|
|
|
t = tokenizer{} |
|
51
|
|
|
|
|
52
|
|
|
for index, r := range []rune(input) { |
|
53
|
|
|
if unicode.In(r, separators) && !t.quotFound { |
|
54
|
|
|
t.addKeyword(input, index) |
|
55
|
|
|
|
|
56
|
|
|
continue |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
if index == len(input)-1 { |
|
60
|
|
|
t.addKeyword(input, len(input)) |
|
61
|
|
|
|
|
62
|
|
|
continue |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
if unicode.In(r, quotationMarks) { |
|
66
|
|
|
t.toggleQuot() |
|
67
|
|
|
|
|
68
|
|
|
continue |
|
69
|
|
|
} |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
return t.keywords |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
func fetchKeyword(input string, start, end int) string { |
|
76
|
|
|
return trim(cut(input, start, end)) |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
func trim(input string) string { |
|
80
|
|
|
return strings.Trim(input, trimRunes) |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
func cut(input string, start, end int) string { |
|
84
|
|
|
r := []rune(input) |
|
85
|
|
|
if end > len(r) { |
|
86
|
|
|
end = len(r) |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
return string(r[start:end]) |
|
90
|
|
|
} |
|
91
|
|
|
|
vvval /
go-metadata-scanner
