Tim Duesterhus | dbd25c3 | 2021-04-15 21:45:55 +0200 | [diff] [blame] | 1 | /* |
| 2 | * HTTP request URI normalization. |
| 3 | * |
| 4 | * Copyright 2021 Tim Duesterhus <tim@bastelstu.be> |
| 5 | * |
| 6 | * This program is free software; you can redistribute it and/or |
| 7 | * modify it under the terms of the GNU General Public License |
| 8 | * as published by the Free Software Foundation; either version |
| 9 | * 2 of the License, or (at your option) any later version. |
| 10 | * |
| 11 | */ |
| 12 | |
Tim Duesterhus | d371e99 | 2021-04-15 21:45:58 +0200 | [diff] [blame] | 13 | #include <import/ist.h> |
| 14 | |
Tim Duesterhus | dbd25c3 | 2021-04-15 21:45:55 +0200 | [diff] [blame] | 15 | #include <haproxy/api.h> |
| 16 | #include <haproxy/uri_normalizer.h> |
| 17 | |
Tim Duesterhus | 9982fc2 | 2021-04-15 21:45:59 +0200 | [diff] [blame^] | 18 | /* Merges `/../` with preceding path segments. */ |
| 19 | enum uri_normalizer_err uri_normalizer_path_dotdot(const struct ist path, struct ist *dst) |
| 20 | { |
| 21 | enum uri_normalizer_err err; |
| 22 | |
| 23 | const size_t size = istclear(dst); |
| 24 | char * const tail = istptr(*dst) + size; |
| 25 | char *head = tail; |
| 26 | |
| 27 | ssize_t offset = istlen(path) - 1; |
| 28 | |
| 29 | int up = 0; |
| 30 | |
| 31 | /* The path will either be shortened or have the same length. */ |
| 32 | if (size < istlen(path)) { |
| 33 | err = URI_NORMALIZER_ERR_ALLOC; |
| 34 | goto fail; |
| 35 | } |
| 36 | |
| 37 | /* Handle `/..` at the end of the path without a trailing slash. */ |
| 38 | if (offset >= 2 && istmatch(istadv(path, offset - 2), ist("/.."))) { |
| 39 | up++; |
| 40 | offset -= 2; |
| 41 | } |
| 42 | |
| 43 | while (offset >= 0) { |
| 44 | if (offset >= 3 && istmatch(istadv(path, offset - 3), ist("/../"))) { |
| 45 | up++; |
| 46 | offset -= 3; |
| 47 | continue; |
| 48 | } |
| 49 | |
| 50 | if (up > 0) { |
| 51 | /* Skip the slash. */ |
| 52 | offset--; |
| 53 | |
| 54 | /* First check whether we already reached the start of the path, |
| 55 | * before popping the current `/../`. |
| 56 | */ |
| 57 | if (offset >= 0) { |
| 58 | up--; |
| 59 | |
| 60 | /* Skip the current path segment. */ |
| 61 | while (offset >= 0 && istptr(path)[offset] != '/') |
| 62 | offset--; |
| 63 | } |
| 64 | } |
| 65 | else { |
| 66 | /* Prepend the slash. */ |
| 67 | *(--head) = istptr(path)[offset]; |
| 68 | offset--; |
| 69 | |
| 70 | /* Prepend the current path segment. */ |
| 71 | while (offset >= 0 && istptr(path)[offset] != '/') { |
| 72 | *(--head) = istptr(path)[offset]; |
| 73 | offset--; |
| 74 | } |
| 75 | } |
| 76 | } |
| 77 | |
| 78 | if (up > 0) { |
| 79 | /* Prepend a trailing slash. */ |
| 80 | *(--head) = '/'; |
| 81 | |
| 82 | /* Prepend unconsumed `/..`. */ |
| 83 | do { |
| 84 | *(--head) = '.'; |
| 85 | *(--head) = '.'; |
| 86 | *(--head) = '/'; |
| 87 | up--; |
| 88 | } while (up > 0); |
| 89 | } |
| 90 | |
| 91 | *dst = ist2(head, tail - head); |
| 92 | |
| 93 | return URI_NORMALIZER_ERR_NONE; |
| 94 | |
| 95 | fail: |
| 96 | |
| 97 | return err; |
| 98 | } |
| 99 | |
Tim Duesterhus | d371e99 | 2021-04-15 21:45:58 +0200 | [diff] [blame] | 100 | /* Merges adjacent slashes in the given path. */ |
| 101 | enum uri_normalizer_err uri_normalizer_path_merge_slashes(const struct ist path, struct ist *dst) |
| 102 | { |
| 103 | enum uri_normalizer_err err; |
| 104 | |
| 105 | const size_t size = istclear(dst); |
| 106 | struct ist newpath = *dst; |
| 107 | |
| 108 | struct ist scanner = path; |
| 109 | |
| 110 | /* The path will either be shortened or have the same length. */ |
| 111 | if (size < istlen(path)) { |
| 112 | err = URI_NORMALIZER_ERR_ALLOC; |
| 113 | goto fail; |
| 114 | } |
| 115 | |
| 116 | while (istlen(scanner) > 0) { |
| 117 | const char current = istshift(&scanner); |
| 118 | |
| 119 | if (current == '/') { |
| 120 | while (istlen(scanner) > 0 && *istptr(scanner) == '/') |
| 121 | scanner = istnext(scanner); |
| 122 | } |
| 123 | |
| 124 | newpath = __istappend(newpath, current); |
| 125 | } |
| 126 | |
| 127 | *dst = newpath; |
| 128 | |
| 129 | return URI_NORMALIZER_ERR_NONE; |
| 130 | |
| 131 | fail: |
| 132 | |
| 133 | return err; |
| 134 | } |
| 135 | |
| 136 | |
Tim Duesterhus | dbd25c3 | 2021-04-15 21:45:55 +0200 | [diff] [blame] | 137 | /* |
| 138 | * Local variables: |
| 139 | * c-indent-level: 8 |
| 140 | * c-basic-offset: 8 |
| 141 | * End: |
| 142 | */ |