shlib.sh

   1 #!/bin/sh
   2 #
   3 # This is generic shell library for all the scripts used by Girocco;
   4 # most importantly, it introduces all the $cfg_* shell variables.
   5
   6 # SHA-1 patterns
   7 octet='[0-9a-f][0-9a-f]'
   8 octet4="$octet$octet$octet$octet"
   9 octet19="$octet4$octet4$octet4$octet4$octet$octet$octet"
  10 octet20="$octet4$octet4$octet4$octet4$octet4"
  11 nullsha="0000000000000000000000000000000000000000"
  12 # tab
  13 tab="$(printf '\t')"
  14
  15 # set a sane umask that never excludes any user or group permissions
  16 umask $(printf '0%03o' $(( $(umask) & ~0770 )) )
  17
  18 vcmp() {
  19         # Compare $1 to $2 each of which must match \d+(\.\d+)*
  20         # An empty string ('') for $1 or $2 is treated like 0
  21         # Outputs:
  22         #  -1 if $1 < $2
  23         #   0 if $1 = $2
  24         #   1 if $1 > $2
  25         # Note that `vcmp 1.8 1.8.0.0.0.0` correctly outputs 0.
  26         while
  27                 _a="${1%%.*}"
  28                 _b="${2%%.*}"
  29                 [ -n "$_a" -o -n "$_b" ]
  30         do
  31                 if [ "${_a:-0}" -lt "${_b:-0}" ]; then
  32                         echo -1
  33                         return
  34                 elif [ "${_a:-0}" -gt "${_b:-0}" ]; then
  35                         echo 1
  36                         return
  37                 fi
  38                 _a2="${1#$_a}"
  39                 _b2="${2#$_b}"
  40                 set -- "${_a2#.}" "${_b2#.}"
  41         done
  42         echo 0
  43 }
  44
  45 get_girocco_config_pm_var_list() {
  46         # Export all the variables from Girocco::Config to suitable var= lines
  47         # prefixing them with 'cfg_'. E.g. $cfg_admin is admin's mail address now
  48         # and also setting a 'defined_cfg_' prefix to 1 if they are not undef.
  49         __girocco_conf="$GIROCCO_CONF"
  50         [ -n "$__girocco_conf" ] || __girocco_conf="Girocco::Config"
  51         [ -z "$basedir" ] || __girocco_extrainc="-I$basedir"
  52         perl -I@basedir@ $__girocco_extrainc -M$__girocco_conf -le \
  53                 'foreach (sort {uc($a) cmp uc($b)} keys %Girocco::Config::) {
  54                         my $val = ${$Girocco::Config::{$_}}; defined($val) or $val="";
  55                         $val =~ s/([\\"\$\`])/\\$1/gos;
  56                         $val =~ s/(?:\r\n|\r|\n)$//os;
  57                         print "cfg_$_=\"$val\"";
  58                         print "defined_cfg_$_=",
  59                                 (defined(${$Girocco::Config::{$_}})?"1":"");
  60                 }'
  61 }
  62
  63 get_girocco_config_var_list() (
  64         # Same as get_girocco_config_pm_var_list except that
  65         # the following variables (all starting with var_) are added:
  66         # var_group             cfg_owning_group if defined otherwise `id -gn`
  67         # var_git_ver           The version number part from `git version`
  68         # var_git_exec_path     The result of $cfg_git_bin --exec-dir
  69         # var_have_git_172      Set to 1 if git version >= 1.7.2 otherwise ''
  70         # var_have_git_173      Set to 1 if git version >= 1.7.3 otherwise ''
  71         # var_window_memory     Value to use for repack --window-memory=
  72         # var_log_window_size   Value to use for git-svn --log-window-size=
  73         # var_utf8_locale       Value to use for a UTF-8 locale if available
  74         # var_xargs_r           A "-r" if xargs needs it to behave correctly
  75         # var_du_exclude        Option to exclude PATTERN from du if available
  76         # var_du_follow         Option to follow command line sym links if available
  77         _cfg_vars="$(get_girocco_config_pm_var_list)"
  78         eval "$_cfg_vars"
  79         printf '%s\n' "$_cfg_vars"
  80         printf 'var_group=%s\n' "${cfg_owning_group:-$(id -gn)}"
  81         _gver="$("$cfg_git_bin" version 2>/dev/null | \
  82                 sed -ne 's/^[^0-9]*\([0-9][0-9]*\(\.[0-9][0-9]*\)*\).*$/\1/p')"
  83         printf 'var_git_ver=%s\n' "$_gver"
  84         printf 'var_git_exec_path="%s"\n' "$("$cfg_git_bin" --exec-path 2>/dev/null)"
  85         printf 'var_have_git_172=%s\n' "$([ $(vcmp "$_gver" 1.7.2) -ge 0 ] && echo 1)"
  86         printf 'var_have_git_173=%s\n' "$([ $(vcmp "$_gver" 1.7.3) -ge 0 ] && echo 1)"
  87         __girocco_conf="$GIROCCO_CONF"
  88         [ -n "$__girocco_conf" ] || __girocco_conf="Girocco::Config"
  89         [ -z "$basedir" ] || __girocco_extrainc="-I$basedir"
  90         printf "var_window_memory=%s\n" \
  91                 "$(perl -I@basedir@ $__girocco_extrainc -M$__girocco_conf \
  92                 -MGirocco::Util -e 'print calc_windowmemory')"
  93         printf 'var_log_window_size=%s\n' "${cfg_svn_log_window_size:-250}"
  94         # We parse the output of `locale -a` and select a suitable UTF-8 locale.
  95         _guess_locale="$(locale -a | grep -viE '^(posix|c)(\..*)?$' | \
  96                 grep -iE '\.utf-?8$' | sed -e 's/\.[Uu][Tt][Ff]-*8$//' | \
  97                 sed -e '/en_US/ s/^/0 /; /en_US/ !s/^/1 /' | LC_ALL=C sort | \
  98                 head -n 1 | cut -d ' ' -f 2)"
  99         [ -z "$_guess_locale" ] || printf 'var_utf8_locale=%s.UTF-8\n' "$_guess_locale"
 100         # On some broken platforms running xargs without -r and empty input runs the command
 101         printf 'var_xargs_r=%s\n' "$(: | command xargs echo -r)"
 102         # The disk usage report produces better numbers if du has an exclude option
 103         _x0="$(basename "$0")"
 104         _x0="${_x0%?}?*"
 105         for _duopt in --exclude -I; do
 106                 if _test="$(du $_duopt 's?lib.s*' $_duopt "$_x0" "$0" 2>/dev/null)" && [ -z "$_test" ]; then
 107                         printf 'var_du_exclude=%s\n' "$_duopt"
 108                         break
 109                 fi
 110         done
 111         if _test="$(du -H "$0" 2>/dev/null)" && [ -n "$_test" ]; then
 112                 printf 'var_du_follow=%s\n' "-H"
 113                 break
 114         fi
 115 )
 116
 117 # If basedir has been replaced, and shlib_vars.sh exists, get the config
 118 # definitions from it rather than running Perl.
 119 if [ "@basedir@" = '@'basedir'@' ] || ! [ -r "@basedir@/shlib_vars.sh" ]; then
 120         # Import all the variables from Girocco::Config to the local environment,
 121         eval "$(get_girocco_config_var_list)"
 122 else
 123         # Import the variables from shlib_vars.sh which avoids needlessly
 124         # running another copy of Perl
 125         . "@basedir@/shlib_vars.sh"
 126 fi
 127
 128 # Make sure we have a reproducible environment by using a controlled HOME dir
 129 XDG_CONFIG_HOME="$cfg_chroot/var/empty"
 130 HOME="$cfg_chroot/etc/girocco"
 131 GIT_CONFIG_NOSYSTEM=1
 132 GIT_ATTR_NOSYSTEM=1
 133 GIT_NO_REPLACE_OBJECTS=1
 134 GIT_TERMINAL_PROMPT=0
 135 GIT_ASKPASS="$cfg_basedir/bin/git-askpass-password"
 136 export XDG_CONFIG_HOME
 137 export HOME
 138 export GIT_CONFIG_NOSYSTEM
 139 export GIT_ATTR_NOSYSTEM
 140 export GIT_NO_REPLACE_OBJECTS
 141 export GIT_TERMINAL_PROMPT
 142 export GIT_ASKPASS
 143 unset GIT_USER_AGENT
 144 unset GIT_HTTP_USER_AGENT
 145 if [ -n "$defined_cfg_git_client_ua" ]; then
 146         GIT_USER_AGENT="$cfg_git_client_ua"
 147         export GIT_USER_AGENT
 148         GIT_HTTP_USER_AGENT="$cfg_git_client_ua"
 149         export GIT_HTTP_USER_AGENT
 150 fi
 151
 152 # We cannot use a git() {} or nc_openbsd() {} function to redirect git
 153 # and nc_openbsd to the desired executables because when using
 154 # "ENV_VAR=xxx func" the various /bin/sh implementations behave in various
 155 # different and unexpected ways:
 156 #   a) treat "ENV_VAR=xxx" like a separate, preceding "export ENV_VAR=xxx"
 157 #   b) treat "ENV_VAR=xxx" like a separate, prededing "ENV_VAR=xxx"
 158 #   c) treat "ENV_VAR=xxx" like a temporary setting only while running func
 159 # None of these are good.  We want a temporary "export ENV_VAR=xxx"
 160 # setting only while running func which none of the /bin/sh's do.
 161 #
 162 # Instead we'd like to use an alias that provides the desired behavior without
 163 # any of the bad (a), (b) or (c) effects.
 164 #
 165 # However, unfortunately, some of the crazy /bin/sh implementations do not
 166 # recognize alias expansions when preceded by variable assignments!
 167 #
 168 # So we are left with git() {} and nc_openbsd() {} functions and in the
 169 # case of git() {} we can compensate for (b) and (c) failing to export
 170 # but not (a) and (b) persisting the values so the caller will simply
 171 # have to beware and explicitly unset any variables that should not persist
 172 # beyond the function call itself.
 173
 174 git() (
 175         [ "${GIT_DIR+set}" = "set" ] && export GIT_DIR
 176         [ "${GIT_SSL_NO_VERIFY+set}" = "set" ] && export GIT_SSL_NO_VERIFY
 177         [ "${GIT_TRACE_PACKET+set}" = "set" ] && export GIT_TRACE_PACKET
 178         [ "${GIT_USER_AGENT+set}" = "set" ] && export GIT_USER_AGENT
 179         [ "${GIT_HTTP_USER_AGENT+set}" = "set" ] && export GIT_HTTP_USER_AGENT
 180         exec "$cfg_git_bin" "$@"
 181 )
 182
 183 nc_openbsd() { "$cfg_nc_openbsd_bin" "$@"; }
 184
 185 # Some platforms' broken xargs runs the command always at least once even if
 186 # there's no input unless given a special option.  Automatically supply the
 187 # option on those platforms by providing an xargs function.
 188 xargs() { command xargs $var_xargs_r "$@"; }
 189
 190 _addrlist() {
 191         _list=
 192         for _addr in "$@"; do
 193                 [ -z "$_list" ] || _list="$_list, "
 194                 _list="$_list$_addr"
 195         done
 196         echo "$_list"
 197 }
 198
 199 _sendmail() {
 200         _mailer="${cfg_sendmail_bin:-/usr/sbin/sendmail}"
 201         if [ -n "$cfg_sender" ]; then
 202                 "$_mailer" -i -f "$cfg_sender" "$@"
 203         else
 204                 "$_mailer" -i "$@"
 205         fi
 206 }
 207
 208 mail() {
 209         _subject=
 210         if [ "$1" = "-s" ]; then
 211                 shift
 212                 _subject="$1"
 213                 shift
 214         fi
 215         {
 216                 echo "From: \"$cfg_name\" ($cfg_title) <$cfg_admin>"
 217                 echo "To: $(_addrlist "$@")"
 218                 [ -z "$_subject" ] || echo "Subject: $_subject"
 219                 echo "MIME-Version: 1.0"
 220                 echo "Content-Type: text/plain; charset=utf-8"
 221                 echo "Content-Transfer-Encoding: 8bit"
 222                 echo "Auto-Submitted: auto-generated"
 223                 echo ""
 224                 cat
 225         } | _sendmail "$@"
 226 }
 227
 228 # bang CMD... will execute the command with well-defined failure mode;
 229 # set bang_action to string of the failed action ('clone', 'update', ...);
 230 # pre-set bang_once=1 to make sure jobs banging on a repo repeatedly will
 231 # not spam the owner; re-define the bang_trap() function to do custom
 232 # cleanup before bailing out
 233 bang() {
 234         if [ -n "$show_progress" ]; then
 235                 exec 3>&1
 236                 errcode=
 237                 read -r errcode <<-EOT || :
 238                 $(
 239                         exec 4>&3 3>&1 1>&4 4>&-
 240                         { "$@" 3>&- || echo $? >&3; } 2>&1 | tee -a "$bang_log"
 241                 )
 242                 EOT
 243                 exec 3>&-
 244                 if [ -z "$errcode" ]; then
 245                         # All right. Cool.
 246                         return;
 247                 fi
 248         else
 249                 if "$@" >>"$bang_log" 2>&1; then
 250                         # All right. Cool.
 251                         return;
 252                 else
 253                         errcode="$?"
 254                 fi
 255         fi
 256         if ! [ -e .banged ] || [ -e .bangagain ]; then
 257                 rm -f .bangagain
 258                 bangmailok=true
 259                 ! [ -f HEAD -a -f config -a -d objects ] ||
 260                 bangmailok="$(GIT_DIR=. git config --bool gitweb.statusupdates 2>/dev/null || echo true)"
 261                 bangaddrs=''
 262                 [ "$bangmailok" = "false" -o -z "$mail" ] || bangaddrs="$mail"
 263                 [ -z "$cfg_admincc" -o "$cfg_admincc" = "0" -o -z "$cfg_admin" ] ||
 264                 if [ -z "$bangaddrs" ]; then bangaddrs="$cfg_admin"; else bangaddrs="$bangaddrs,$cfg_admin"; fi
 265                 [ -z "$bangaddrs" ] ||
 266                 {
 267                         echo "$* failed with error code $errcode"
 268                         echo ""
 269                         [ ! -n "$bang_once" ] || echo "you will not receive any more notifications until recovery"
 270                         echo "this status message may be disabled on the project admin page"
 271                         echo ""
 272                         echo "Log follows:"
 273                         echo ""
 274                         cat "$bang_log"
 275                 } | mail -s "[$cfg_name] $proj $bang_action failed" "$bangaddrs"
 276         fi
 277         touch .banged
 278         cat "$bang_log" > .banglog
 279         bang_trap
 280         exit 1
 281 }
 282
 283 # bang_eval CMD... will evaluate the command with well-defined failure mode;
 284 # Identical to bang CMD... except the command is eval'd instead of executed.
 285 bang_eval() {
 286         bang eval "$*"
 287 }
 288
 289 # Default bang settings:
 290 bang_setup() {
 291         bang_action="lame_programmer"
 292         bang_once=
 293         bang_trap() { :; }
 294         bang_log="$(mktemp -t repomgr-XXXXXX)"
 295         trap 'rm -f "$bang_log"' EXIT
 296         trap 'exit 130' INT
 297         trap 'exit 143' TERM
 298 }
 299
 300 # Remove banged status
 301 bang_reset() {
 302         rm -f .banged .bangagain .banglog
 303 }
 304
 305 # Check to see if banged status
 306 is_banged() {
 307         [ -e .banged ]
 308 }
 309
 310
 311 # Progress report - if show_progress is set, shows the given message.
 312 progress() {
 313         [ ! -n "$show_progress" ] || echo "$@"
 314 }
 315
 316
 317 # Project config accessors; must be run in project directory
 318 config_get() {
 319         git config "gitweb.$1"
 320 }
 321
 322 config_set() {
 323         git config "gitweb.$1" "$2" && chgrp $var_group config && chmod g+w config
 324 }
 325
 326 config_set_raw() {
 327         git config "$1" "$2"        && chgrp $var_group config && chmod g+w config
 328 }
 329
 330 config_get_date_seconds() {
 331         _dt="$(config_get "$1" || :)"
 332         [ -n "$_dt" ] || return 1
 333         _ds="$(perl -I@basedir@ -MGirocco::Util -e "print parse_rfc2822_date('$_dt')")"
 334         [ -n "$_ds" ] || return 1
 335         echo "$_ds"
 336 }
 337
 338 # Tool for checking whether given number of seconds has not passed yet
 339 check_interval() {
 340         os="$(config_get_date_seconds "$1")" || return 1
 341         ns="$(date +%s)"
 342         [ $ns -lt $(($os+$2)) ]
 343 }
 344
 345 # Check if we are running with effective root permissions
 346 is_root() {
 347         [ "$(id -u 2>/dev/null)" = "0" ]
 348 }
 349
 350 # Check to see if the single argument is a Git directory
 351 is_git_dir() {
 352         # Just like Git's test except we ignore GIT_OBJECT_DIRECTORY
 353         # And we are slightly more picky (must be refs/.+ not refs/.*)
 354         [ -d "$1/objects" -a -x "$1/objects" ] || return 1
 355         [ -d "$1/refs" -a -x "$1/refs" ] || return 1
 356         if [ -L "$1/HEAD" ]; then
 357                 _hr="$(readlink "$1/HEAD")"
 358                 case "$_hr" in "refs/"?*) :;; *) return 1;; esac
 359         fi
 360         [ -f "$1/HEAD" -a -r "$1/HEAD" ] || return 1
 361         read -r _hr <"$1/HEAD" || return 1
 362         case "$_hr" in
 363                 $octet20)
 364                         return 0;;
 365                 ref:*)
 366                         _hr="${_hr##ref:*[ $tab]}"
 367                         case "$_hr" in "refs/"?*) return 0;; esac
 368         esac
 369         return 1
 370 }
 371
 372 # List all Git repositories, with given prefix if specified, one-per-line
 373 # All project names starting with _ are always excluded from the result
 374 get_repo_list() {
 375         if [ -n "$1" ]; then
 376                 cut -d : -f 1,3 "$cfg_chroot"/etc/group | grep "^$1"
 377         else
 378                 cut -d : -f 1,3 "$cfg_chroot"/etc/group
 379         fi | while IFS=: read name id; do
 380                 [ $id -lt 65536 ] || case "$name" in _*) :;; ?*) echo "$name"; esac
 381         done
 382 }
 383
 384 # returns empty string for non-mirrors, otherwise one of:
 385 #   m  =>  normal Git mirror
 386 #   s  =>  mirror from svn source
 387 #   d  =>  mirror from darcs source
 388 #   b  =>  mirror from bzr source
 389 #   h  =>  mirror from hg source
 390 # the optional passed in git dir defaults to "."
 391 # will fail if the directory does not have .nofetch and gitweb.baseurl
 392 # comes back empty
 393 get_mirror_type() {
 394         _gitdir="${1:-.}"
 395         # always return empty for non-mirrors
 396         [ ! -e "$_gitdir/.nofetch" ] || return 0
 397         _url="$(GIT_DIR="$_gitdir" config_get baseurl 2>/dev/null || :)"
 398         _url="${_url##* }"
 399         case "$_url" in
 400                 "")
 401                         return 1
 402                         ;;
 403                 svn://* | svn+http://* | svn+https://*)
 404                         echo 's'
 405                         ;;
 406                 darcs://*)
 407                         echo 'd'
 408                         ;;
 409                 bzr://*)
 410                         echo 'b'
 411                         ;;
 412                 hg+http://* | hg+https://*)
 413                         echo 'h'
 414                         ;;
 415                 *)
 416                         echo 'm'
 417                         ;;
 418         esac
 419         return 0
 420 }
 421
 422 # returns true if the passed in git dir (defaults to ".") is a mirror using git fast-import
 423 is_gfi_mirror() {
 424         case "$(get_mirror_type "$1" 2>/dev/null || :)" in
 425                 d|b|h)
 426                         # darcs, bzr and hg mirrors use git fast-import
 427                         return 0
 428                         ;;
 429                 *)
 430                         # Don't think git-svn currently uses git fast-import
 431                         # And Git mirrors certainly do not
 432                         return 1
 433                         ;;
 434         esac
 435         # assume it does not use git fast-import
 436         return 1
 437 }
 438
 439 # returns true if the passed in git dir (defaults to ".") is a mirror using git-svn
 440 is_svn_mirror() {
 441         [ "$(get_mirror_type "$1" 2>/dev/null || :)" = "s" ]
 442 }
 443
 444 # A well-known UTF-8 locale is required for some of the fast-import providers
 445 # in order to avoid mangling characters.  Ideally we could use "POSIX.UTF-8"
 446 # but that is not reliably UTF-8 but rather usually US-ASCII.
 447 # We parse the output of `locale -a` and select a suitable UTF-8 locale at
 448 # install time and store that in $var_utf8_locale if one is found.
 449 # If we cannot find one in the `locale -a` output then we just use a well-known
 450 # UTF-8 locale and hope for the best.  We set LC_ALL to our choice and export
 451 # it.  We only set this temporarily when running the fast-import providers.
 452 set_utf8_locale() {
 453         LC_ALL="${var_utf8_locale:-en_US.UTF-8}"
 454         export LC_ALL
 455 }
 456
 457 # hg-fast-export | git fast-import with error handling in current directory GIT_DIR
 458 git_hg_fetch() (
 459         set_utf8_locale
 460         _python="${PYTHON:-python}"
 461         _err1=
 462         _err2=
 463         exec 3>&1
 464         { read -r _err1 || :; read -r _err2 || :; } <<-EOT
 465         $(
 466                 exec 4>&3 3>&1 1>&4 4>&-
 467                 {
 468                         _e1=0
 469                         [ -f hg2git-marks ] || touch hg2git-marks
 470                         _af="$(git config hg.authorsfile || :)"
 471                         _cmd='GIT_DIR="$(pwd)" "$_python" "$cfg_basedir/bin/hg-fast-export.py" \
 472                                 --repo    "$(pwd)/repo.hg" \
 473                                 --marks   "$(pwd)/hg2git-marks" \
 474                                 --mapping "$(pwd)/hg2git-mapping" \
 475                                 --heads   "$(pwd)/hg2git-heads" \
 476                                 --status  "$(pwd)/hg2git-state" \
 477                                 -U unknown --force --flatten --hg-hash'
 478                         [ -z "$_af" ] || _cmd="$_cmd"' --authors "$_af"'
 479                         eval "$_cmd" 3>&- || _e1=$?
 480                         echo $_e1 >&3
 481                 } | \
 482                 {
 483                         _e2=0
 484                         rm -f hg2git-marks.new
 485                         git fast-import \
 486                                 --export-marks="$(pwd)/hg2git-marks.new" \
 487                                 --export-pack-edges="$(pwd)/gfi-packs" \
 488                                 --force 3>&- || _e2=$?
 489                         echo $_e2 >&3
 490                 }
 491         )
 492         EOT
 493         exec 3>&-
 494         [ "$_err1" = 0 -a "$_err2" = 0 ] || return 1
 495         if [ -f hg2git-marks ]; then
 496                 rm -f hg2git-marks.old
 497                 mv hg2git-marks hg2git-marks.old
 498         else
 499                 touch hg2git-marks.old
 500         fi
 501         cat hg2git-marks.old hg2git-marks.new | \
 502                 LC_ALL=C sort -t : -k2,2n -u | \
 503                 sed -ne "/^:[1-9][0-9]* $octet20\$/p" > hg2git-marks
 504         rm hg2git-marks.old hg2git-marks.new
 505         rm -f hg2git-heads
 506         git branch --no-color | \
 507         while IFS= read -r _head; do
 508                 echo ":${_head#??} $(git rev-parse "refs/heads/${_head#??}")"
 509         done > hg2git-heads
 510 )