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