|
| 1 | +// Copyright 2025 The Gitea Authors. All rights reserved. |
| 2 | +// SPDX-License-Identifier: MIT |
| 3 | + |
| 4 | +package glob |
| 5 | + |
| 6 | +import ( |
| 7 | + "errors" |
| 8 | + "fmt" |
| 9 | + "regexp" |
| 10 | + |
| 11 | + "code.gitea.io/gitea/modules/util" |
| 12 | +) |
| 13 | + |
| 14 | +// Reference: https://github.com/gobwas/glob/blob/master/glob.go |
| 15 | + |
| 16 | +type Glob interface { |
| 17 | + Match(string) bool |
| 18 | +} |
| 19 | + |
| 20 | +type globCompiler struct { |
| 21 | + nonSeparatorChars string |
| 22 | + globPattern []rune |
| 23 | + regexpPattern string |
| 24 | + regexp *regexp.Regexp |
| 25 | + pos int |
| 26 | +} |
| 27 | + |
| 28 | +// compileChars compiles character class patterns like [abc] or [!abc] |
| 29 | +func (g *globCompiler) compileChars() (string, error) { |
| 30 | + result := "" |
| 31 | + if g.pos < len(g.globPattern) && g.globPattern[g.pos] == '!' { |
| 32 | + g.pos++ |
| 33 | + result += "^" |
| 34 | + } |
| 35 | + |
| 36 | + for g.pos < len(g.globPattern) { |
| 37 | + c := g.globPattern[g.pos] |
| 38 | + g.pos++ |
| 39 | + |
| 40 | + if c == ']' { |
| 41 | + return "[" + result + "]", nil |
| 42 | + } |
| 43 | + |
| 44 | + if c == '\\' { |
| 45 | + if g.pos >= len(g.globPattern) { |
| 46 | + return "", errors.New("unterminated character class escape") |
| 47 | + } |
| 48 | + result += "\\" + string(g.globPattern[g.pos]) |
| 49 | + g.pos++ |
| 50 | + } else { |
| 51 | + result += string(c) |
| 52 | + } |
| 53 | + } |
| 54 | + |
| 55 | + return "", errors.New("unterminated character class") |
| 56 | +} |
| 57 | + |
| 58 | +// compile compiles the glob pattern into a regular expression |
| 59 | +func (g *globCompiler) compile(subPattern bool) (string, error) { |
| 60 | + result := "" |
| 61 | + |
| 62 | + for g.pos < len(g.globPattern) { |
| 63 | + c := g.globPattern[g.pos] |
| 64 | + g.pos++ |
| 65 | + |
| 66 | + if subPattern && c == '}' { |
| 67 | + return "(" + result + ")", nil |
| 68 | + } |
| 69 | + |
| 70 | + switch c { |
| 71 | + case '*': |
| 72 | + if g.pos < len(g.globPattern) && g.globPattern[g.pos] == '*' { |
| 73 | + g.pos++ |
| 74 | + result += ".*" // match any sequence of characters |
| 75 | + } else { |
| 76 | + result += g.nonSeparatorChars + "*" // match any sequence of non-separator characters |
| 77 | + } |
| 78 | + case '?': |
| 79 | + result += g.nonSeparatorChars // match any single non-separator character |
| 80 | + case '[': |
| 81 | + chars, err := g.compileChars() |
| 82 | + if err != nil { |
| 83 | + return "", err |
| 84 | + } |
| 85 | + result += chars |
| 86 | + case '{': |
| 87 | + subResult, err := g.compile(true) |
| 88 | + if err != nil { |
| 89 | + return "", err |
| 90 | + } |
| 91 | + result += subResult |
| 92 | + case ',': |
| 93 | + if subPattern { |
| 94 | + result += "|" |
| 95 | + } else { |
| 96 | + result += "," |
| 97 | + } |
| 98 | + case '\\': |
| 99 | + if g.pos >= len(g.globPattern) { |
| 100 | + return "", errors.New("no character to escape") |
| 101 | + } |
| 102 | + result += "\\" + string(g.globPattern[g.pos]) |
| 103 | + g.pos++ |
| 104 | + case '.', '+', '^', '$', '(', ')', '|': |
| 105 | + result += "\\" + string(c) // escape regexp special characters |
| 106 | + default: |
| 107 | + result += string(c) |
| 108 | + } |
| 109 | + } |
| 110 | + |
| 111 | + return result, nil |
| 112 | +} |
| 113 | + |
| 114 | +func newGlobCompiler(pattern string, separators ...rune) (Glob, error) { |
| 115 | + g := &globCompiler{globPattern: []rune(pattern)} |
| 116 | + |
| 117 | + // Escape separators for use in character class |
| 118 | + escapedSeparators := regexp.QuoteMeta(string(separators)) |
| 119 | + if escapedSeparators != "" { |
| 120 | + g.nonSeparatorChars = "[^" + escapedSeparators + "]" |
| 121 | + } else { |
| 122 | + g.nonSeparatorChars = "." |
| 123 | + } |
| 124 | + |
| 125 | + compiled, err := g.compile(false) |
| 126 | + if err != nil { |
| 127 | + return nil, err |
| 128 | + } |
| 129 | + |
| 130 | + g.regexpPattern = "^" + compiled + "$" |
| 131 | + |
| 132 | + regex, err := regexp.Compile(g.regexpPattern) |
| 133 | + if err != nil { |
| 134 | + return nil, fmt.Errorf("failed to compile regexp: %w", err) |
| 135 | + } |
| 136 | + |
| 137 | + g.regexp = regex |
| 138 | + return g, nil |
| 139 | +} |
| 140 | + |
| 141 | +func (g *globCompiler) Match(s string) bool { |
| 142 | + return g.regexp.MatchString(s) |
| 143 | +} |
| 144 | + |
| 145 | +func Compile(pattern string, separators ...rune) (Glob, error) { |
| 146 | + return newGlobCompiler(pattern, separators...) |
| 147 | +} |
| 148 | + |
| 149 | +func MustCompile(pattern string, separators ...rune) Glob { |
| 150 | + g, err := Compile(pattern, separators...) |
| 151 | + if err != nil { |
| 152 | + panic(err) |
| 153 | + } |
| 154 | + return g |
| 155 | +} |
| 156 | + |
| 157 | +// QuoteMeta returns a string that quotes all glob pattern meta characters |
| 158 | +// inside the argument text; For example, QuoteMeta(`{foo*}`) returns `\[foo\*\]`. |
| 159 | +// Reference: https://github.com/gobwas/glob/blob/master/glob.go |
| 160 | +func QuoteMeta(s string) string { |
| 161 | + b := make([]byte, 2*len(s)) |
| 162 | + j := 0 |
| 163 | + for i := 0; i < len(s); i++ { |
| 164 | + c := s[i] |
| 165 | + isSpecial := c == '*' || c == '?' || c == '\\' || c == '[' || c == ']' || c == '{' || c == '}' |
| 166 | + if isSpecial { |
| 167 | + b[j] = '\\' |
| 168 | + j++ |
| 169 | + } |
| 170 | + b[j] = s[i] |
| 171 | + j++ |
| 172 | + } |
| 173 | + return util.UnsafeBytesToString(b[0:j]) |
| 174 | +} |
0 commit comments