|
| 1 | +// Copyright 2018 The Go Authors. All rights reserved. |
| 2 | +// Use of this source code is governed by a BSD-style |
| 3 | +// license that can be found in the LICENSE file. |
| 4 | + |
| 5 | +package get |
| 6 | + |
| 7 | +import ( |
| 8 | + "fmt" |
| 9 | + "strings" |
| 10 | + "unicode" |
| 11 | + "unicode/utf8" |
| 12 | +) |
| 13 | + |
| 14 | +// The following functions are copied verbatim from cmd/go/internal/module/module.go, |
| 15 | +// with a change to additionally reject Windows short-names, |
| 16 | +// and one to accept arbitrary letters (golang.org/issue/29101). |
| 17 | +// |
| 18 | +// TODO(bcmills): After the call site for this function is backported, |
| 19 | +// consolidate this back down to a single copy. |
| 20 | +// |
| 21 | +// NOTE: DO NOT MERGE THESE UNTIL WE DECIDE ABOUT ARBITRARY LETTERS IN MODULE MODE. |
| 22 | + |
| 23 | +// CheckImportPath checks that an import path is valid. |
| 24 | +func CheckImportPath(path string) error { |
| 25 | + if err := checkPath(path, false); err != nil { |
| 26 | + return fmt.Errorf("malformed import path %q: %v", path, err) |
| 27 | + } |
| 28 | + return nil |
| 29 | +} |
| 30 | + |
| 31 | +// checkPath checks that a general path is valid. |
| 32 | +// It returns an error describing why but not mentioning path. |
| 33 | +// Because these checks apply to both module paths and import paths, |
| 34 | +// the caller is expected to add the "malformed ___ path %q: " prefix. |
| 35 | +// fileName indicates whether the final element of the path is a file name |
| 36 | +// (as opposed to a directory name). |
| 37 | +func checkPath(path string, fileName bool) error { |
| 38 | + if !utf8.ValidString(path) { |
| 39 | + return fmt.Errorf("invalid UTF-8") |
| 40 | + } |
| 41 | + if path == "" { |
| 42 | + return fmt.Errorf("empty string") |
| 43 | + } |
| 44 | + if strings.Contains(path, "..") { |
| 45 | + return fmt.Errorf("double dot") |
| 46 | + } |
| 47 | + if strings.Contains(path, "//") { |
| 48 | + return fmt.Errorf("double slash") |
| 49 | + } |
| 50 | + if path[len(path)-1] == '/' { |
| 51 | + return fmt.Errorf("trailing slash") |
| 52 | + } |
| 53 | + elemStart := 0 |
| 54 | + for i, r := range path { |
| 55 | + if r == '/' { |
| 56 | + if err := checkElem(path[elemStart:i], fileName); err != nil { |
| 57 | + return err |
| 58 | + } |
| 59 | + elemStart = i + 1 |
| 60 | + } |
| 61 | + } |
| 62 | + if err := checkElem(path[elemStart:], fileName); err != nil { |
| 63 | + return err |
| 64 | + } |
| 65 | + return nil |
| 66 | +} |
| 67 | + |
| 68 | +// checkElem checks whether an individual path element is valid. |
| 69 | +// fileName indicates whether the element is a file name (not a directory name). |
| 70 | +func checkElem(elem string, fileName bool) error { |
| 71 | + if elem == "" { |
| 72 | + return fmt.Errorf("empty path element") |
| 73 | + } |
| 74 | + if strings.Count(elem, ".") == len(elem) { |
| 75 | + return fmt.Errorf("invalid path element %q", elem) |
| 76 | + } |
| 77 | + if elem[0] == '.' && !fileName { |
| 78 | + return fmt.Errorf("leading dot in path element") |
| 79 | + } |
| 80 | + if elem[len(elem)-1] == '.' { |
| 81 | + return fmt.Errorf("trailing dot in path element") |
| 82 | + } |
| 83 | + |
| 84 | + charOK := pathOK |
| 85 | + if fileName { |
| 86 | + charOK = fileNameOK |
| 87 | + } |
| 88 | + for _, r := range elem { |
| 89 | + if !charOK(r) { |
| 90 | + return fmt.Errorf("invalid char %q", r) |
| 91 | + } |
| 92 | + } |
| 93 | + |
| 94 | + // Windows disallows a bunch of path elements, sadly. |
| 95 | + // See https://docs.microsoft.com/en-us/windows/desktop/fileio/naming-a-file |
| 96 | + short := elem |
| 97 | + if i := strings.Index(short, "."); i >= 0 { |
| 98 | + short = short[:i] |
| 99 | + } |
| 100 | + for _, bad := range badWindowsNames { |
| 101 | + if strings.EqualFold(bad, short) { |
| 102 | + return fmt.Errorf("disallowed path element %q", elem) |
| 103 | + } |
| 104 | + } |
| 105 | + |
| 106 | + // Reject path components that look like Windows short-names. |
| 107 | + // Those usually end in a tilde followed by one or more ASCII digits. |
| 108 | + if tilde := strings.LastIndexByte(short, '~'); tilde >= 0 && tilde < len(short)-1 { |
| 109 | + suffix := short[tilde+1:] |
| 110 | + suffixIsDigits := true |
| 111 | + for _, r := range suffix { |
| 112 | + if r < '0' || r > '9' { |
| 113 | + suffixIsDigits = false |
| 114 | + break |
| 115 | + } |
| 116 | + } |
| 117 | + if suffixIsDigits { |
| 118 | + return fmt.Errorf("trailing tilde and digits in path element") |
| 119 | + } |
| 120 | + } |
| 121 | + |
| 122 | + return nil |
| 123 | +} |
| 124 | + |
| 125 | +// pathOK reports whether r can appear in an import path element. |
| 126 | +// |
| 127 | +// NOTE: This function DIVERGES from module mode pathOK by accepting Unicode letters. |
| 128 | +func pathOK(r rune) bool { |
| 129 | + if r < utf8.RuneSelf { |
| 130 | + return r == '+' || r == '-' || r == '.' || r == '_' || r == '~' || |
| 131 | + '0' <= r && r <= '9' || |
| 132 | + 'A' <= r && r <= 'Z' || |
| 133 | + 'a' <= r && r <= 'z' |
| 134 | + } |
| 135 | + return unicode.IsLetter(r) |
| 136 | +} |
| 137 | + |
| 138 | +// fileNameOK reports whether r can appear in a file name. |
| 139 | +// For now we allow all Unicode letters but otherwise limit to pathOK plus a few more punctuation characters. |
| 140 | +// If we expand the set of allowed characters here, we have to |
| 141 | +// work harder at detecting potential case-folding and normalization collisions. |
| 142 | +// See note about "safe encoding" below. |
| 143 | +func fileNameOK(r rune) bool { |
| 144 | + if r < utf8.RuneSelf { |
| 145 | + // Entire set of ASCII punctuation, from which we remove characters: |
| 146 | + // ! " # $ % & ' ( ) * + , - . / : ; < = > ? @ [ \ ] ^ _ ` { | } ~ |
| 147 | + // We disallow some shell special characters: " ' * < > ? ` | |
| 148 | + // (Note that some of those are disallowed by the Windows file system as well.) |
| 149 | + // We also disallow path separators / : and \ (fileNameOK is only called on path element characters). |
| 150 | + // We allow spaces (U+0020) in file names. |
| 151 | + const allowed = "!#$%&()+,-.=@[]^_{}~ " |
| 152 | + if '0' <= r && r <= '9' || 'A' <= r && r <= 'Z' || 'a' <= r && r <= 'z' { |
| 153 | + return true |
| 154 | + } |
| 155 | + for i := 0; i < len(allowed); i++ { |
| 156 | + if rune(allowed[i]) == r { |
| 157 | + return true |
| 158 | + } |
| 159 | + } |
| 160 | + return false |
| 161 | + } |
| 162 | + // It may be OK to add more ASCII punctuation here, but only carefully. |
| 163 | + // For example Windows disallows < > \, and macOS disallows :, so we must not allow those. |
| 164 | + return unicode.IsLetter(r) |
| 165 | +} |
| 166 | + |
| 167 | +// badWindowsNames are the reserved file path elements on Windows. |
| 168 | +// See https://docs.microsoft.com/en-us/windows/desktop/fileio/naming-a-file |
| 169 | +var badWindowsNames = []string{ |
| 170 | + "CON", |
| 171 | + "PRN", |
| 172 | + "AUX", |
| 173 | + "NUL", |
| 174 | + "COM1", |
| 175 | + "COM2", |
| 176 | + "COM3", |
| 177 | + "COM4", |
| 178 | + "COM5", |
| 179 | + "COM6", |
| 180 | + "COM7", |
| 181 | + "COM8", |
| 182 | + "COM9", |
| 183 | + "LPT1", |
| 184 | + "LPT2", |
| 185 | + "LPT3", |
| 186 | + "LPT4", |
| 187 | + "LPT5", |
| 188 | + "LPT6", |
| 189 | + "LPT7", |
| 190 | + "LPT8", |
| 191 | + "LPT9", |
| 192 | +} |
0 commit comments