khenaidoo | ffe076b | 2019-01-15 16:08:08 -0500 | [diff] [blame^] | 1 | #!/usr/bin/env bash |
| 2 | |
| 3 | # Purpose: plain text tar format |
| 4 | # Limitations: - only suitable for text files, directories, and symlinks |
| 5 | # - stores only filename, content, and mode |
| 6 | # - not designed for untrusted input |
| 7 | # |
| 8 | # Note: must work with bash version 3.2 (macOS) |
| 9 | |
| 10 | # Copyright 2017 Roger Luethi |
| 11 | # |
| 12 | # Licensed under the Apache License, Version 2.0 (the "License"); |
| 13 | # you may not use this file except in compliance with the License. |
| 14 | # You may obtain a copy of the License at |
| 15 | # |
| 16 | # http://www.apache.org/licenses/LICENSE-2.0 |
| 17 | # |
| 18 | # Unless required by applicable law or agreed to in writing, software |
| 19 | # distributed under the License is distributed on an "AS IS" BASIS, |
| 20 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 21 | # See the License for the specific language governing permissions and |
| 22 | # limitations under the License. |
| 23 | |
| 24 | set -o errexit -o nounset |
| 25 | |
| 26 | # Sanitize environment (for instance, standard sorting of glob matches) |
| 27 | export LC_ALL=C |
| 28 | |
| 29 | path="" |
| 30 | CMD="" |
| 31 | ARG_STRING="$*" |
| 32 | |
| 33 | #------------------------------------------------------------------------------ |
| 34 | # Not all sed implementations can work on null bytes. In order to make ttar |
| 35 | # work out of the box on macOS, use Python as a stream editor. |
| 36 | |
| 37 | USE_PYTHON=0 |
| 38 | |
| 39 | PYTHON_CREATE_FILTER=$(cat << 'PCF' |
| 40 | #!/usr/bin/env python |
| 41 | |
| 42 | import re |
| 43 | import sys |
| 44 | |
| 45 | for line in sys.stdin: |
| 46 | line = re.sub(r'EOF', r'\EOF', line) |
| 47 | line = re.sub(r'NULLBYTE', r'\NULLBYTE', line) |
| 48 | line = re.sub('\x00', r'NULLBYTE', line) |
| 49 | sys.stdout.write(line) |
| 50 | PCF |
| 51 | ) |
| 52 | |
| 53 | PYTHON_EXTRACT_FILTER=$(cat << 'PEF' |
| 54 | #!/usr/bin/env python |
| 55 | |
| 56 | import re |
| 57 | import sys |
| 58 | |
| 59 | for line in sys.stdin: |
| 60 | line = re.sub(r'(?<!\\)NULLBYTE', '\x00', line) |
| 61 | line = re.sub(r'\\NULLBYTE', 'NULLBYTE', line) |
| 62 | line = re.sub(r'([^\\])EOF', r'\1', line) |
| 63 | line = re.sub(r'\\EOF', 'EOF', line) |
| 64 | sys.stdout.write(line) |
| 65 | PEF |
| 66 | ) |
| 67 | |
| 68 | function test_environment { |
| 69 | if [[ "$(echo "a" | sed 's/a/\x0/' | wc -c)" -ne 2 ]]; then |
| 70 | echo "WARNING sed unable to handle null bytes, using Python (slow)." |
| 71 | if ! which python >/dev/null; then |
| 72 | echo "ERROR Python not found. Aborting." |
| 73 | exit 2 |
| 74 | fi |
| 75 | USE_PYTHON=1 |
| 76 | fi |
| 77 | } |
| 78 | |
| 79 | #------------------------------------------------------------------------------ |
| 80 | |
| 81 | function usage { |
| 82 | bname=$(basename "$0") |
| 83 | cat << USAGE |
| 84 | Usage: $bname [-C <DIR>] -c -f <ARCHIVE> <FILE...> (create archive) |
| 85 | $bname -t -f <ARCHIVE> (list archive contents) |
| 86 | $bname [-C <DIR>] -x -f <ARCHIVE> (extract archive) |
| 87 | |
| 88 | Options: |
| 89 | -C <DIR> (change directory) |
| 90 | -v (verbose) |
| 91 | |
| 92 | Example: Change to sysfs directory, create ttar file from fixtures directory |
| 93 | $bname -C sysfs -c -f sysfs/fixtures.ttar fixtures/ |
| 94 | USAGE |
| 95 | exit "$1" |
| 96 | } |
| 97 | |
| 98 | function vecho { |
| 99 | if [ "${VERBOSE:-}" == "yes" ]; then |
| 100 | echo >&7 "$@" |
| 101 | fi |
| 102 | } |
| 103 | |
| 104 | function set_cmd { |
| 105 | if [ -n "$CMD" ]; then |
| 106 | echo "ERROR: more than one command given" |
| 107 | echo |
| 108 | usage 2 |
| 109 | fi |
| 110 | CMD=$1 |
| 111 | } |
| 112 | |
| 113 | unset VERBOSE |
| 114 | |
| 115 | while getopts :cf:htxvC: opt; do |
| 116 | case $opt in |
| 117 | c) |
| 118 | set_cmd "create" |
| 119 | ;; |
| 120 | f) |
| 121 | ARCHIVE=$OPTARG |
| 122 | ;; |
| 123 | h) |
| 124 | usage 0 |
| 125 | ;; |
| 126 | t) |
| 127 | set_cmd "list" |
| 128 | ;; |
| 129 | x) |
| 130 | set_cmd "extract" |
| 131 | ;; |
| 132 | v) |
| 133 | VERBOSE=yes |
| 134 | exec 7>&1 |
| 135 | ;; |
| 136 | C) |
| 137 | CDIR=$OPTARG |
| 138 | ;; |
| 139 | *) |
| 140 | echo >&2 "ERROR: invalid option -$OPTARG" |
| 141 | echo |
| 142 | usage 1 |
| 143 | ;; |
| 144 | esac |
| 145 | done |
| 146 | |
| 147 | # Remove processed options from arguments |
| 148 | shift $(( OPTIND - 1 )); |
| 149 | |
| 150 | if [ "${CMD:-}" == "" ]; then |
| 151 | echo >&2 "ERROR: no command given" |
| 152 | echo |
| 153 | usage 1 |
| 154 | elif [ "${ARCHIVE:-}" == "" ]; then |
| 155 | echo >&2 "ERROR: no archive name given" |
| 156 | echo |
| 157 | usage 1 |
| 158 | fi |
| 159 | |
| 160 | function list { |
| 161 | local path="" |
| 162 | local size=0 |
| 163 | local line_no=0 |
| 164 | local ttar_file=$1 |
| 165 | if [ -n "${2:-}" ]; then |
| 166 | echo >&2 "ERROR: too many arguments." |
| 167 | echo |
| 168 | usage 1 |
| 169 | fi |
| 170 | if [ ! -e "$ttar_file" ]; then |
| 171 | echo >&2 "ERROR: file not found ($ttar_file)" |
| 172 | echo |
| 173 | usage 1 |
| 174 | fi |
| 175 | while read -r line; do |
| 176 | line_no=$(( line_no + 1 )) |
| 177 | if [ $size -gt 0 ]; then |
| 178 | size=$(( size - 1 )) |
| 179 | continue |
| 180 | fi |
| 181 | if [[ $line =~ ^Path:\ (.*)$ ]]; then |
| 182 | path=${BASH_REMATCH[1]} |
| 183 | elif [[ $line =~ ^Lines:\ (.*)$ ]]; then |
| 184 | size=${BASH_REMATCH[1]} |
| 185 | echo "$path" |
| 186 | elif [[ $line =~ ^Directory:\ (.*)$ ]]; then |
| 187 | path=${BASH_REMATCH[1]} |
| 188 | echo "$path/" |
| 189 | elif [[ $line =~ ^SymlinkTo:\ (.*)$ ]]; then |
| 190 | echo "$path -> ${BASH_REMATCH[1]}" |
| 191 | fi |
| 192 | done < "$ttar_file" |
| 193 | } |
| 194 | |
| 195 | function extract { |
| 196 | local path="" |
| 197 | local size=0 |
| 198 | local line_no=0 |
| 199 | local ttar_file=$1 |
| 200 | if [ -n "${2:-}" ]; then |
| 201 | echo >&2 "ERROR: too many arguments." |
| 202 | echo |
| 203 | usage 1 |
| 204 | fi |
| 205 | if [ ! -e "$ttar_file" ]; then |
| 206 | echo >&2 "ERROR: file not found ($ttar_file)" |
| 207 | echo |
| 208 | usage 1 |
| 209 | fi |
| 210 | while IFS= read -r line; do |
| 211 | line_no=$(( line_no + 1 )) |
| 212 | local eof_without_newline |
| 213 | if [ "$size" -gt 0 ]; then |
| 214 | if [[ "$line" =~ [^\\]EOF ]]; then |
| 215 | # An EOF not preceeded by a backslash indicates that the line |
| 216 | # does not end with a newline |
| 217 | eof_without_newline=1 |
| 218 | else |
| 219 | eof_without_newline=0 |
| 220 | fi |
| 221 | # Replace NULLBYTE with null byte if at beginning of line |
| 222 | # Replace NULLBYTE with null byte unless preceeded by backslash |
| 223 | # Remove one backslash in front of NULLBYTE (if any) |
| 224 | # Remove EOF unless preceeded by backslash |
| 225 | # Remove one backslash in front of EOF |
| 226 | if [ $USE_PYTHON -eq 1 ]; then |
| 227 | echo -n "$line" | python -c "$PYTHON_EXTRACT_FILTER" >> "$path" |
| 228 | else |
| 229 | # The repeated pattern makes up for sed's lack of negative |
| 230 | # lookbehind assertions (for consecutive null bytes). |
| 231 | echo -n "$line" | \ |
| 232 | sed -e 's/^NULLBYTE/\x0/g; |
| 233 | s/\([^\\]\)NULLBYTE/\1\x0/g; |
| 234 | s/\([^\\]\)NULLBYTE/\1\x0/g; |
| 235 | s/\\NULLBYTE/NULLBYTE/g; |
| 236 | s/\([^\\]\)EOF/\1/g; |
| 237 | s/\\EOF/EOF/g; |
| 238 | ' >> "$path" |
| 239 | fi |
| 240 | if [[ "$eof_without_newline" -eq 0 ]]; then |
| 241 | echo >> "$path" |
| 242 | fi |
| 243 | size=$(( size - 1 )) |
| 244 | continue |
| 245 | fi |
| 246 | if [[ $line =~ ^Path:\ (.*)$ ]]; then |
| 247 | path=${BASH_REMATCH[1]} |
| 248 | if [ -e "$path" ] || [ -L "$path" ]; then |
| 249 | rm "$path" |
| 250 | fi |
| 251 | elif [[ $line =~ ^Lines:\ (.*)$ ]]; then |
| 252 | size=${BASH_REMATCH[1]} |
| 253 | # Create file even if it is zero-length. |
| 254 | touch "$path" |
| 255 | vecho " $path" |
| 256 | elif [[ $line =~ ^Mode:\ (.*)$ ]]; then |
| 257 | mode=${BASH_REMATCH[1]} |
| 258 | chmod "$mode" "$path" |
| 259 | vecho "$mode" |
| 260 | elif [[ $line =~ ^Directory:\ (.*)$ ]]; then |
| 261 | path=${BASH_REMATCH[1]} |
| 262 | mkdir -p "$path" |
| 263 | vecho " $path/" |
| 264 | elif [[ $line =~ ^SymlinkTo:\ (.*)$ ]]; then |
| 265 | ln -s "${BASH_REMATCH[1]}" "$path" |
| 266 | vecho " $path -> ${BASH_REMATCH[1]}" |
| 267 | elif [[ $line =~ ^# ]]; then |
| 268 | # Ignore comments between files |
| 269 | continue |
| 270 | else |
| 271 | echo >&2 "ERROR: Unknown keyword on line $line_no: $line" |
| 272 | exit 1 |
| 273 | fi |
| 274 | done < "$ttar_file" |
| 275 | } |
| 276 | |
| 277 | function div { |
| 278 | echo "# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -" \ |
| 279 | "- - - - - -" |
| 280 | } |
| 281 | |
| 282 | function get_mode { |
| 283 | local mfile=$1 |
| 284 | if [ -z "${STAT_OPTION:-}" ]; then |
| 285 | if stat -c '%a' "$mfile" >/dev/null 2>&1; then |
| 286 | # GNU stat |
| 287 | STAT_OPTION='-c' |
| 288 | STAT_FORMAT='%a' |
| 289 | else |
| 290 | # BSD stat |
| 291 | STAT_OPTION='-f' |
| 292 | # Octal output, user/group/other (omit file type, sticky bit) |
| 293 | STAT_FORMAT='%OLp' |
| 294 | fi |
| 295 | fi |
| 296 | stat "${STAT_OPTION}" "${STAT_FORMAT}" "$mfile" |
| 297 | } |
| 298 | |
| 299 | function _create { |
| 300 | shopt -s nullglob |
| 301 | local mode |
| 302 | local eof_without_newline |
| 303 | while (( "$#" )); do |
| 304 | file=$1 |
| 305 | if [ -L "$file" ]; then |
| 306 | echo "Path: $file" |
| 307 | symlinkTo=$(readlink "$file") |
| 308 | echo "SymlinkTo: $symlinkTo" |
| 309 | vecho " $file -> $symlinkTo" |
| 310 | div |
| 311 | elif [ -d "$file" ]; then |
| 312 | # Strip trailing slash (if there is one) |
| 313 | file=${file%/} |
| 314 | echo "Directory: $file" |
| 315 | mode=$(get_mode "$file") |
| 316 | echo "Mode: $mode" |
| 317 | vecho "$mode $file/" |
| 318 | div |
| 319 | # Find all files and dirs, including hidden/dot files |
| 320 | for x in "$file/"{*,.[^.]*}; do |
| 321 | _create "$x" |
| 322 | done |
| 323 | elif [ -f "$file" ]; then |
| 324 | echo "Path: $file" |
| 325 | lines=$(wc -l "$file"|awk '{print $1}') |
| 326 | eof_without_newline=0 |
| 327 | if [[ "$(wc -c "$file"|awk '{print $1}')" -gt 0 ]] && \ |
| 328 | [[ "$(tail -c 1 "$file" | wc -l)" -eq 0 ]]; then |
| 329 | eof_without_newline=1 |
| 330 | lines=$((lines+1)) |
| 331 | fi |
| 332 | echo "Lines: $lines" |
| 333 | # Add backslash in front of EOF |
| 334 | # Add backslash in front of NULLBYTE |
| 335 | # Replace null byte with NULLBYTE |
| 336 | if [ $USE_PYTHON -eq 1 ]; then |
| 337 | < "$file" python -c "$PYTHON_CREATE_FILTER" |
| 338 | else |
| 339 | < "$file" \ |
| 340 | sed 's/EOF/\\EOF/g; |
| 341 | s/NULLBYTE/\\NULLBYTE/g; |
| 342 | s/\x0/NULLBYTE/g; |
| 343 | ' |
| 344 | fi |
| 345 | if [[ "$eof_without_newline" -eq 1 ]]; then |
| 346 | # Finish line with EOF to indicate that the original line did |
| 347 | # not end with a linefeed |
| 348 | echo "EOF" |
| 349 | fi |
| 350 | mode=$(get_mode "$file") |
| 351 | echo "Mode: $mode" |
| 352 | vecho "$mode $file" |
| 353 | div |
| 354 | else |
| 355 | echo >&2 "ERROR: file not found ($file in $(pwd))" |
| 356 | exit 2 |
| 357 | fi |
| 358 | shift |
| 359 | done |
| 360 | } |
| 361 | |
| 362 | function create { |
| 363 | ttar_file=$1 |
| 364 | shift |
| 365 | if [ -z "${1:-}" ]; then |
| 366 | echo >&2 "ERROR: missing arguments." |
| 367 | echo |
| 368 | usage 1 |
| 369 | fi |
| 370 | if [ -e "$ttar_file" ]; then |
| 371 | rm "$ttar_file" |
| 372 | fi |
| 373 | exec > "$ttar_file" |
| 374 | echo "# Archive created by ttar $ARG_STRING" |
| 375 | _create "$@" |
| 376 | } |
| 377 | |
| 378 | test_environment |
| 379 | |
| 380 | if [ -n "${CDIR:-}" ]; then |
| 381 | if [[ "$ARCHIVE" != /* ]]; then |
| 382 | # Relative path: preserve the archive's location before changing |
| 383 | # directory |
| 384 | ARCHIVE="$(pwd)/$ARCHIVE" |
| 385 | fi |
| 386 | cd "$CDIR" |
| 387 | fi |
| 388 | |
| 389 | "$CMD" "$ARCHIVE" "$@" |