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 $cfg_* shell variables.
   5
   6 # hash patterns
   7 hexdig='[0-9a-f]'
   8 octet="$hexdig$hexdig"
   9 octet4="$octet$octet$octet$octet"
  10 octet19="$octet4$octet4$octet4$octet4$octet$octet$octet"
  11 octet20="$octet4$octet4$octet4$octet4$octet4"
  12 # tab (single \t between single quotes)
  13 tab='   '
  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" ] || [ -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 unset orig_path
  46 get_girocco_config_pm_var_list() (
  47         # Export all the variables from Girocco::Config to suitable var= lines
  48         # prefixing them with 'cfg_'. E.g. $cfg_admin is admin's mail address now
  49         # and also setting a 'defined_cfg_' prefix to 1 if they are not undef.
  50         __girocco_conf="$GIROCCO_CONF"
  51         [ -n "$__girocco_conf" ] || __girocco_conf="Girocco::Config"
  52         [ -z "$basedir" ] || __girocco_extrainc="-I$basedir"
  53         [ -z "$orig_path" ] || { PATH="$orig_path" && export PATH; }
  54         perl -I@basedir@ $__girocco_extrainc -M$__girocco_conf -le \
  55                 'foreach (sort {uc($a) cmp uc($b)} keys %Girocco::Config::) {
  56                         my $val = ${$Girocco::Config::{$_}}; defined($val) or $val="";
  57                         $val =~ s/([\\"\$\`])/\\$1/gos;
  58                         $val =~ s/(?:\r\n|\r|\n)$//os;
  59                         print "cfg_$_=\"$val\"";
  60                         print "defined_cfg_$_=",
  61                                 (defined(${$Girocco::Config::{$_}})?"1":"");
  62                 }'
  63 )
  64
  65 # Returns full command path for "$1" if it's a valid command otherwise returns "$1"
  66 _fcp() {
  67         if _fp="$(command -v "$1" 2>/dev/null)"; then
  68                 printf '%s\n' "$_fp"
  69         else
  70                 printf '%s\n' "$1"
  71         fi
  72 }
  73
  74 get_girocco_config_var_list() (
  75         # Same as get_girocco_config_pm_var_list except that
  76         # the following variables (all starting with var_) are added:
  77         # var_group             cfg_owning_group if defined otherwise `id -gn`
  78         # var_git_ver           The version number part from `git version`
  79         # var_git_exec_path     The result of $cfg_git_bin --exec-dir
  80         # var_sh_bin            Full path to the posix sh interpreter to use
  81         # var_perl_bin          Full path to the perl interpreter to use
  82         # var_gzip_bin          Full path to the gzip executable to use
  83         # var_nc_openbsd_bin    Full path to the netcat (nc) with -U support
  84         # var_have_git_171      Set to 1 if git version >= 1.7.1 otherwise ''
  85         # var_have_git_172      Set to 1 if git version >= 1.7.2 otherwise ''
  86         # var_have_git_173      Set to 1 if git version >= 1.7.3 otherwise ''
  87         # var_have_git_1710     Set to 1 if git version >= 1.7.10 otherwise ''
  88         # var_have_git_185      Set to 1 if git version >= 1.8.5 otherwise ''
  89         # var_have_git_210      Set to 1 if git version >= 2.1.0 otherwise ''
  90         # var_have_git_260      Set to 1 if git version >= 2.6.0 otherwise ''
  91         # var_have_git_2101     Set to 1 if git version >= 2.10.1 otherwise ''
  92         # var_window_memory     Value to use for repack --window-memory=
  93         # var_big_file_threshold Value to use for core.bigFileThreshold
  94         # var_redelta_threshold Recompute deltas if no more than this many objs
  95         # var_upload_window     If not "", pack.window to use for upload-pack
  96         # var_log_window_size   Value to use for git-svn --log-window-size=
  97         # var_utf8_locale       Value to use for a UTF-8 locale if available
  98         # var_xargs_r           A "-r" if xargs needs it to behave correctly
  99         # var_du_exclude        Option to exclude PATTERN from du if available
 100         # var_du_follow         Option to follow command line sym links if available
 101         _cfg_vars="$(get_girocco_config_pm_var_list)"
 102         eval "$_cfg_vars"
 103         printf '%s\n' "$_cfg_vars"
 104         printf 'var_group=%s\n' "${cfg_owning_group:-$(id -gn)}"
 105         _gver="$("$cfg_git_bin" version 2>/dev/null |
 106                 LC_ALL=C sed -ne 's/^[^0-9]*\([0-9][0-9]*\(\.[0-9][0-9]*\)*\).*$/\1/p')"
 107         printf 'var_git_ver=%s\n' "$_gver"
 108         printf 'var_git_exec_path="%s"\n' "$("$cfg_git_bin" --exec-path 2>/dev/null)"
 109         printf 'var_sh_bin="%s"\n' "$(_fcp "${cfg_posix_sh_bin:-/bin/sh}")"
 110         printf 'var_perl_bin="%s"\n' "$(_fcp "${cfg_perl_bin:-$(unset -f perl; command -v perl)}")"
 111         printf 'var_gzip_bin="%s"\n' "$(_fcp "${cfg_gzip_bin:-$(unset -f gzip; command -v gzip)}")"
 112         printf 'var_nc_openbsd_bin="%s"\n' "$(_fcp "${cfg_nc_openbsd_bin:-$(unset -f nc; command -v nc)}")"
 113         printf 'var_have_git_171=%s\n' "$([ $(vcmp "$_gver" 1.7.1) -ge 0 ] && echo 1)"
 114         printf 'var_have_git_172=%s\n' "$([ $(vcmp "$_gver" 1.7.2) -ge 0 ] && echo 1)"
 115         printf 'var_have_git_173=%s\n' "$([ $(vcmp "$_gver" 1.7.3) -ge 0 ] && echo 1)"
 116         printf 'var_have_git_1710=%s\n' "$([ $(vcmp "$_gver" 1.7.10) -ge 0 ] && echo 1)"
 117         printf 'var_have_git_185=%s\n' "$([ $(vcmp "$_gver" 1.8.5) -ge 0 ] && echo 1)"
 118         printf 'var_have_git_210=%s\n' "$([ $(vcmp "$_gver" 2.1.0) -ge 0 ] && echo 1)"
 119         printf 'var_have_git_260=%s\n' "$([ $(vcmp "$_gver" 2.6.0) -ge 0 ] && echo 1)"
 120         printf 'var_have_git_2101=%s\n' "$([ $(vcmp "$_gver" 2.10.1) -ge 0 ] && echo 1)"
 121         __girocco_conf="$GIROCCO_CONF"
 122         [ -n "$__girocco_conf" ] || __girocco_conf="Girocco::Config"
 123         [ -z "$basedir" ] || __girocco_extrainc="-I$basedir"
 124         printf "var_window_memory=%s\n" \
 125                 "$(perl -I@basedir@ $__girocco_extrainc -M$__girocco_conf \
 126                 -MGirocco::Util -e 'print calc_windowmemory')"
 127         printf "var_big_file_threshold=%s\n" \
 128                 "$(perl -I@basedir@ $__girocco_extrainc -M$__girocco_conf \
 129                 -MGirocco::Util -e 'print calc_bigfilethreshold')"
 130         printf "var_redelta_threshold=%s\n" \
 131                 "$(perl -I@basedir@ $__girocco_extrainc -M$__girocco_conf \
 132                 -MGirocco::Util -e 'print calc_redeltathreshold')"
 133         if [ -n "$cfg_upload_pack_window" ] && [ "$cfg_upload_pack_window" -ge 2 ] &&
 134            [ "$cfg_upload_pack_window" -le 50 ]; then
 135                 printf "var_upload_window=%s\n" "$cfg_upload_pack_window"
 136         else
 137                 printf "var_upload_window=%s\n" ""
 138         fi
 139         printf 'var_log_window_size=%s\n' "${cfg_svn_log_window_size:-250}"
 140         # We parse the output of `locale -a` and select a suitable UTF-8 locale.
 141         _guess_locale="$(locale -a | LC_ALL=C grep -viE '^(posix|c)(\..*)?$' |
 142                 LC_ALL=C grep -iE '\.utf-?8$' | LC_ALL=C sed -e 's/\.[Uu][Tt][Ff]-*8$//' |
 143                 LC_ALL=C sed -e '/en_US/ s/^/0 /; /en_US/ !s/^/1 /' | LC_ALL=C sort |
 144                 head -n 1 | LC_ALL=C cut -d ' ' -f 2)"
 145         [ -z "$_guess_locale" ] || printf 'var_utf8_locale=%s.UTF-8\n' "$_guess_locale"
 146         # On some broken platforms running xargs without -r and empty input runs the command
 147         printf 'var_xargs_r=%s\n' "$(</dev/null command xargs printf %s -r)"
 148         # The disk usage report produces better numbers if du has an exclude option
 149         _x0="$(basename "$0")"
 150         _x0="${_x0%?}?*"
 151         for _duopt in --exclude -I; do
 152                 if _test="$(du $_duopt 's?lib.s*' $_duopt "$_x0" "$0" 2>/dev/null)" && [ -z "$_test" ]; then
 153                         printf 'var_du_exclude=%s\n' "$_duopt"
 154                         break
 155                 fi
 156         done
 157         if _test="$(du -H "$0" 2>/dev/null)" && [ -n "$_test" ]; then
 158                 printf 'var_du_follow=%s\n' "-H"
 159                 break
 160         fi
 161 )
 162
 163 # If basedir has been replaced, and shlib_vars.sh exists, get the config
 164 # definitions from it rather than running Perl.
 165 if [ "@basedir@" = '@'basedir'@' ] || ! [ -r "@basedir@/shlib_vars.sh" ]; then
 166         # Import all the variables from Girocco::Config to the local environment,
 167         eval "$(get_girocco_config_var_list)"
 168 else
 169         # Import the variables from shlib_vars.sh which avoids needlessly
 170         # running another copy of Perl
 171         . "@basedir@/shlib_vars.sh"
 172 fi
 173
 174 # git_add_config "some.var=value"
 175 # every ' in value must be replaced with the 4-character sequence '\'' before
 176 # calling this function or Git will barf.  Will not be effective unless running
 177 # Git version 1.7.3 or later.
 178 git_add_config() {
 179         GIT_CONFIG_PARAMETERS="${GIT_CONFIG_PARAMETERS:+$GIT_CONFIG_PARAMETERS }'$1'"
 180         export GIT_CONFIG_PARAMETERS
 181 }
 182
 183 # Make sure we have a reproducible environment by using a controlled HOME dir
 184 XDG_CONFIG_HOME="$cfg_chroot/var/empty"
 185 HOME="$cfg_chroot/etc/girocco"
 186 TMPDIR="/tmp"
 187 GIT_CONFIG_NOSYSTEM=1
 188 GIT_ATTR_NOSYSTEM=1
 189 GIT_NO_REPLACE_OBJECTS=1
 190 GIT_TERMINAL_PROMPT=0
 191 GIT_PAGER="cat"
 192 PAGER="cat"
 193 GIT_ASKPASS="$cfg_basedir/bin/git-askpass-password"
 194 GIT_SVN_NOTTY=1
 195 GIROCCO_SUPPRESS_AUTO_GC_UPDATE=1
 196 export XDG_CONFIG_HOME
 197 export HOME
 198 export TMPDIR
 199 export GIT_CONFIG_NOSYSTEM
 200 export GIT_ATTR_NOSYSTEM
 201 export GIT_NO_REPLACE_OBJECTS
 202 export GIT_TERMINAL_PROMPT
 203 export GIT_PAGER
 204 export PAGER
 205 export GIT_ASKPASS
 206 export GIT_SVN_NOTTY
 207 export GIROCCO_SUPPRESS_AUTO_GC_UPDATE
 208 unset GIT_USER_AGENT
 209 unset GIT_HTTP_USER_AGENT
 210 if [ -n "$defined_cfg_git_client_ua" ]; then
 211         GIT_USER_AGENT="$cfg_git_client_ua"
 212         export GIT_USER_AGENT
 213         GIT_HTTP_USER_AGENT="$cfg_git_client_ua"
 214         export GIT_HTTP_USER_AGENT
 215 fi
 216 unset GIT_CONFIG_PARAMETERS
 217 git_add_config "core.ignoreCase=false"
 218 git_add_config "core.pager=cat"
 219 if [ -n "$cfg_git_no_mmap" ]; then
 220         # Just like compiling with NO_MMAP
 221         git_add_config "core.packedGitWindowSize=1m"
 222 else
 223         # Always use the 32-bit default (32m) even on 64-bit to avoid memory blowout
 224         git_add_config "core.packedGitWindowSize=32m"
 225 fi
 226 [ -z "$var_big_file_threshold" ] ||
 227         git_add_config "core.bigFileThreshold=$var_big_file_threshold"
 228 git_add_config "gc.auto=0"
 229
 230 # Make sure any sendmail.pl config is always available
 231 unset SENDMAIL_PL_HOST
 232 unset SENDMAIL_PL_PORT
 233 unset SENDMAIL_PL_NCBIN
 234 unset SENDMAIL_PL_NCOPT
 235 [ -z "$cfg_sendmail_pl_host"  ] || { SENDMAIL_PL_HOST="$cfg_sendmail_pl_host"   && export SENDMAIL_PL_HOST; }
 236 [ -z "$cfg_sendmail_pl_port"  ] || { SENDMAIL_PL_PORT="$cfg_sendmail_pl_port"   && export SENDMAIL_PL_PORT; }
 237 [ -z "$cfg_sendmail_pl_ncbin" ] || { SENDMAIL_PL_NCBIN="$cfg_sendmail_pl_ncbin" && export SENDMAIL_PL_NCBIN; }
 238 [ -z "$cfg_sendmail_pl_ncopt" ] || { SENDMAIL_PL_NCOPT="$cfg_sendmail_pl_ncopt" && export SENDMAIL_PL_NCOPT; }
 239
 240 # Set PATH and PYTHON to the values set by Config.pm, if any
 241 unset PYTHON
 242 [ -z "$cfg_python" ] || { PYTHON="$cfg_python" && export PYTHON; }
 243 [ -z "$cfg_path" ] || { orig_path="$PATH" && PATH="$cfg_path" && export PATH; }
 244
 245 # Extra GIT variables that generally ought to be cleared, but whose clearing
 246 # could potentially interfere with the correct operation of hook scripts so
 247 # they are segregated into a separate function for use as appropriate
 248 clean_git_env() {
 249         unset GIT_ALTERNATE_OBJECT_DIRECTORIES
 250         unset GIT_CONFIG
 251         unset GIT_DIR
 252         unset GIT_GRAFT_FILE
 253         unset GIT_INDEX_FILE
 254         unset GIT_OBJECT_DIRECTORY
 255         unset GIT_NAMESPACE
 256 }
 257
 258 # We cannot use a git() {} or nc_openbsd() {} function to redirect git
 259 # and nc_openbsd to the desired executables because when using
 260 # "ENV_VAR=xxx func" the various /bin/sh implementations behave in various
 261 # different and unexpected ways:
 262 #   a) treat "ENV_VAR=xxx" like a separate, preceding "export ENV_VAR=xxx"
 263 #   b) treat "ENV_VAR=xxx" like a separate, prededing "ENV_VAR=xxx"
 264 #   c) treat "ENV_VAR=xxx" like a temporary setting only while running func
 265 # None of these are good.  We want a temporary "export ENV_VAR=xxx"
 266 # setting only while running func which none of the /bin/sh's do.
 267 #
 268 # Instead we'd like to use an alias that provides the desired behavior without
 269 # any of the bad (a), (b) or (c) effects.
 270 #
 271 # However, unfortunately, some of the crazy /bin/sh implementations do not
 272 # recognize alias expansions when preceded by variable assignments!
 273 #
 274 # So we are left with git() {} and nc_openbsd() {} functions and in the
 275 # case of git() {} we can compensate for (b) and (c) failing to export
 276 # but not (a) and (b) persisting the values so the caller will simply
 277 # have to beware and explicitly unset any variables that should not persist
 278 # beyond the function call itself.
 279
 280 git() (
 281         [ z"${GIT_DIR+set}" != z"set" ] || export GIT_DIR
 282         [ z"${GIT_SSL_NO_VERIFY+set}" != z"set" ] || export GIT_SSL_NO_VERIFY
 283         [ z"${GIT_TRACE_PACKET+set}" != z"set" ] || export GIT_TRACE_PACKET
 284         [ z"${GIT_USER_AGENT+set}" != z"set" ] || export GIT_USER_AGENT
 285         [ z"${GIT_HTTP_USER_AGENT+set}" != z"set" ] || export GIT_HTTP_USER_AGENT
 286         exec "$cfg_git_bin" "$@"
 287 )
 288
 289 # Since we do not yet require at least Git 1.8.5 this is a compatibility function
 290 # that allows us to use git update-ref --stdin where supported and the slow shell
 291 # script where not, but only the "delete" operation is currently supported.
 292 git_updateref_stdin() {
 293         if [ -n "$var_have_git_185" ]; then
 294                 git update-ref --stdin
 295         else
 296                 while read -r _op _ref; do
 297                         case "$_op" in
 298                                 delete)
 299                                         git update-ref -d "$_ref"
 300                                         ;;
 301                                 *)
 302                                         echo "bad git_updateref_stdin op: $_op" >&2
 303                                         exit 1
 304                                         ;;
 305                         esac
 306                 done
 307         fi
 308 }
 309
 310 # see comments for git() -- callers must explicitly export all variables
 311 # intended for the commands these functions run before calling them
 312 perl() { command "${var_perl_bin:-perl}" "$@"; }
 313 gzip() { command "${var_gzip_bin:-gzip}" "$@"; }
 314
 315 nc_openbsd() { command "$var_nc_openbsd_bin" "$@"; }
 316
 317 list_packs() { command "$cfg_basedir/bin/list_packs" "$@"; }
 318
 319 readlink() { command "$cfg_basedir/bin/readlink" "$@"; }
 320
 321 strftime() { command "$cfg_basedir/bin/strftime" "$@"; }
 322
 323 # Some platforms' broken xargs runs the command always at least once even if
 324 # there's no input unless given a special option.  Automatically supply the
 325 # option on those platforms by providing an xargs function.
 326 xargs() { command xargs $var_xargs_r "$@"; }
 327
 328 _addrlist() {
 329         _list=
 330         for _addr in "$@"; do
 331                 [ -z "$_list" ] || _list="$_list, "
 332                 _list="$_list$_addr"
 333         done
 334         echo "$_list"
 335 }
 336
 337 _sendmail() {
 338         _mailer="${cfg_sendmail_bin:-/usr/sbin/sendmail}"
 339         if [ -n "$cfg_sender" ]; then
 340                 "$_mailer" -i -f "$cfg_sender" "$@"
 341         else
 342                 "$_mailer" -i "$@"
 343         fi
 344 }
 345
 346 # First argument is an id WITHOUT surrounding '<' and '>' to use in a
 347 # "References:" header.  It may be "" to suppress the "References" header.
 348 # Following arguments are just like mail function
 349 mailref() {
 350         _references=
 351         if [ $# -ge 1 ]; then
 352                 _references="$1"
 353                 shift
 354         fi
 355         _subject=
 356         if [ "$1" = "-s" ]; then
 357                 shift
 358                 _subject="$1"
 359                 shift
 360         fi
 361         {
 362                 echo "From: \"$cfg_name\" ($cfg_title) <$cfg_admin>"
 363                 echo "To: $(_addrlist "$@")"
 364                 [ -z "$_subject" ] || echo "Subject: $_subject"
 365                 echo "MIME-Version: 1.0"
 366                 echo "Content-Type: text/plain; charset=UTF-8"
 367                 echo "Content-Transfer-Encoding: 8bit"
 368                 [ -z "$_references" ] || echo "References: <$_references>"
 369                 [ -n "$cfg_suppress_x_girocco" ] || echo "X-Girocco: $cfg_gitweburl"
 370                 echo "Auto-Submitted: auto-generated"
 371                 echo ""
 372                 cat
 373         } | _sendmail "$@"
 374 }
 375
 376 # Usage: mail [-s <subject>] <addr> [<addr>...]
 377 mail() {
 378         mailref "" "$@"
 379 }
 380
 381 # bang CMD... will execute the command with well-defined failure mode;
 382 # set bang_action to string of the failed action ('clone', 'update', ...);
 383 # re-define the bang_trap() function to do custom cleanup before bailing out
 384 bang() {
 385         bang_active=1
 386         bang_cmd="$*"
 387         bang_errcode=0
 388         if [ "${show_progress:-0}" != "0" ]; then
 389                 exec 3>&1
 390                 read -r bang_errcode <<-EOT || :
 391                 $(
 392                         exec 4>&3 3>&1 1>&4 4>&-
 393                         { "$@" 3>&- || echo $? >&3; } 2>&1 | tee -i -a "$bang_log"
 394                 )
 395                 EOT
 396                 exec 3>&-
 397                 if [ -z "$bang_errcode" ] || [ "$bang_errcode" = "0" ]; then
 398                         # All right. Cool.
 399                         bang_active=
 400                         bang_cmd=
 401                         return;
 402                 fi
 403         else
 404                 if "$@" >>"$bang_log" 2>&1; then
 405                         # All right. Cool.
 406                         bang_active=
 407                         bang_cmd=
 408                         return;
 409                 else
 410                         bang_errcode="$?"
 411                 fi
 412         fi
 413         bang_failed
 414 }
 415
 416 bang_failed() {
 417         bang_active=
 418         unset GIT_DIR
 419         >.banged
 420         cat "$bang_log" >.banglog
 421         echo "" >>.banglog
 422         echo "$bang_cmd failed with error code $bang_errcode" >>.banglog
 423         if [ "${show_progress:-0}" != "0" ]; then
 424                 echo ""
 425                 echo "$bang_cmd failed with error code $bang_errcode"
 426         fi
 427         if [ -e .bangagain ]; then
 428                 git config --remove-section girocco.bang 2>/dev/null || :
 429                 rm -f .bangagain
 430         fi
 431         bangcount="$(git config --int girocco.bang.count 2>/dev/null)" || :
 432         bangcount=$(( ${bangcount:-0} + 1 ))
 433         git config --int girocco.bang.count $bangcount
 434         if [ $bangcount -eq 1 ]; then
 435                 git config girocco.bang.firstfail "$(TZ=UTC date "+%Y-%m-%d %T UTC")"
 436         fi
 437         if [ $bangcount -ge $cfg_min_mirror_failure_message_count ] &&
 438            [ "$(git config --bool girocco.bang.messagesent 2>/dev/null || :)" != "true" ] &&
 439            ! check_interval "girocco.bang.firstfail" $cfg_min_mirror_failure_message_interval; then
 440                 bangmailok="$(git config --bool gitweb.statusupdates 2>/dev/null || echo true)"
 441                 bangaddrs=
 442                 [ "$bangmailok" = "false" ] || [ -z "$mail" ] || bangaddrs="$mail"
 443                 [ -z "$cfg_admincc" ] || [ "$cfg_admincc" = "0" ] || [ -z "$cfg_admin" ] ||
 444                 if [ -z "$bangaddrs" ]; then bangaddrs="$cfg_admin"; else bangaddrs="$bangaddrs,$cfg_admin"; fi
 445                 rsubj=
 446                 [ $bangcount -le 1 ] || rsubj=" repeatedly"
 447                 [ -z "$bangaddrs" ] ||
 448                 {
 449                         echo "$bang_cmd failed with error code $bang_errcode"
 450                         echo ""
 451                         rsubj=
 452                         if [ $bangcount -gt 1 ]; then
 453                                 echo "$bangcount consecutive update failures have occurred since $(config_get girocco.bang.firstfail)"
 454                                 echo ""
 455                         fi
 456                         echo "you will not receive any more notifications until recovery"
 457                         echo "this status message may be disabled on the project admin page"
 458                         echo ""
 459                         echo "Log follows:"
 460                         echo ""
 461                         cat "$bang_log"
 462                 } | mailref "update@$cfg_gitweburl/$proj.git" -s "[$cfg_name] $proj $bang_action failed$rsubj" "$bangaddrs"
 463                 git config --bool girocco.bang.messagesent true
 464         fi
 465         bangthrottle=
 466         [ $bangcount -lt 15 ] ||
 467                 check_interval "girocco.bang.firstfail" $(( $cfg_min_mirror_interval * 3 / 2 )) ||
 468                 bangthrottle=1
 469         bang_trap $bangthrottle
 470         [ -n "$bang_errcode" ] && [ "$bang_errcode" != "0" ] || bang_errcode=1
 471         exit $bang_errcode
 472 }
 473
 474 # bang_eval CMD... will evaluate the command with well-defined failure mode;
 475 # Identical to bang CMD... except the command is eval'd instead of executed.
 476 bang_eval() {
 477         bang eval "$*"
 478 }
 479
 480 # Default bang settings:
 481 bang_setup() {
 482         bang_active=
 483         bang_action="lame_programmer"
 484         bang_trap() { :; }
 485         bang_tmpdir="${TMPDIR:-/tmp}"
 486         bang_tmpdir="${bang_tmpdir%/}"
 487         bang_log="$(mktemp "${bang_tmpdir:-/tmp}/repomgr-XXXXXX")"
 488         is_git_dir . || {
 489                 echo "bang_setup called with current directory not a git directory" >&2
 490                 exit 1
 491         }
 492         trap 'rm -f "$bang_log"' EXIT
 493         trap '[ -z "$bang_active" ] || { bang_errcode=130; bang_failed; }; exit 130' INT
 494         trap '[ -z "$bang_active" ] || { bang_errcode=143; bang_failed; }; exit 143' TERM
 495 }
 496
 497 # Remove banged status
 498 bang_reset() {
 499         rm -f .banged .bangagain .banglog
 500         git config --remove-section girocco.bang 2>/dev/null || :
 501 }
 502
 503 # Check to see if banged status
 504 is_banged() {
 505         [ -e .banged ]
 506 }
 507
 508 # Check to see if banged message was sent
 509 was_banged_message_sent() {
 510         [ "$(git config --bool girocco.bang.messagesent 2>/dev/null || :)" = "true" ]
 511 }
 512
 513 # Progress report - if show_progress is set, shows the given message.
 514 progress() {
 515         [ "${show_progress:-0}" = "0" ] || echo "$*"
 516 }
 517
 518 # Project config accessors; must be run in project directory
 519 config_get() {
 520         case "$1" in
 521         *.*)
 522                 git config "$1";;
 523         *)
 524                 git config "gitweb.$1";;
 525         esac
 526 }
 527
 528 config_set() {
 529         git config "gitweb.$1" "$2" && chgrp $var_group config && chmod g+w config
 530 }
 531
 532 config_set_raw() {
 533         git config "$1" "$2"        && chgrp $var_group config && chmod g+w config
 534 }
 535
 536 config_get_date_seconds() {
 537         _dt="$(config_get "$1")" || :
 538         [ -n "$_dt" ] || return 1
 539         _ds="$(perl -I@basedir@ -MGirocco::Util -e "print parse_any_date('$_dt')")"
 540         [ -n "$_ds" ] || return 1
 541         echo "$_ds"
 542 }
 543
 544 # Tool for checking whether given number of seconds has not passed yet
 545 check_interval() {
 546         os="$(config_get_date_seconds "$1")" || return 1
 547         ns="$(date +%s)"
 548         [ $ns -lt $(($os+$2)) ]
 549 }
 550
 551 # Check if we are running with effective root permissions
 552 is_root() {
 553         [ "$(id -u 2>/dev/null)" = "0" ]
 554 }
 555
 556 # Check to see if the single argument is a Git directory
 557 is_git_dir() {
 558         # Just like Git's test except we ignore GIT_OBJECT_DIRECTORY
 559         # And we are slightly more picky (must be refs/.+ not refs/.*)
 560         [ -d "$1/objects" ] && [ -x "$1/objects" ] || return 1
 561         [ -d "$1/refs" ] && [ -x "$1/refs" ] || return 1
 562         if [ -L "$1/HEAD" ]; then
 563                 _hr="$(readlink "$1/HEAD")"
 564                 case "$_hr" in "refs/"?*) :;; *) return 1;; esac
 565         fi
 566         [ -f "$1/HEAD" ] && [ -r "$1/HEAD" ] || return 1
 567         read -r _hr <"$1/HEAD" || return 1
 568         case "$_hr" in
 569                 $octet20*)
 570                         [ "${_hr#*[!0-9a-f]}" = "$_hr" ] || return 1
 571                         return 0;;
 572                 ref:refs/?*)
 573                         return 0;;
 574                 ref:*)
 575                         _hr="${_hr##ref:*[ $tab]}"
 576                         case "$_hr" in "refs/"?*) return 0;; esac
 577         esac
 578         return 1
 579 }
 580
 581 # List all Git repositories, with given prefix if specified, one-per-line
 582 # All project names starting with _ are always excluded from the result
 583 get_repo_list() {
 584         if [ -n "$1" ]; then
 585                 LC_ALL=C cut -d : -f 1,3 "$cfg_chroot"/etc/group | LC_ALL=C grep "^$1"
 586         else
 587                 LC_ALL=C cut -d : -f 1,3 "$cfg_chroot"/etc/group
 588         fi | while IFS=: read name id; do
 589                 [ $id -lt 65536 ] || case "$name" in _*) :;; ?*) echo "$name"; esac
 590         done
 591 }
 592
 593 # set the variable named by the first argument to the project part (i.e. WITH
 594 # the trailing ".git" but WITHOUT the leading $cfg_reporoot) of the directory
 595 # specified by the second argument.
 596 # This function cannot be fooled by symbolic links.
 597 # If the second argument is omitted (or empty) use $(pwd -P) instead.
 598 # The directory specified by the second argument must exist.
 599 v_get_proj_from_dir() {
 600         [ -n "$2" ] || set -- "$1" "$(pwd -P)"
 601         [ -d "$2" ] || return 1
 602         case "$2" in
 603         "$cfg_reporoot/"?*)
 604                 # Simple case that does not need any fancy footwork
 605                 _projpart="${2#$cfg_reporoot/}"
 606                 ;;
 607         *)
 608                 _absrr="$(cd "$cfg_reporoot" && pwd -P)"
 609                 _abspd="$(cd "$2" && pwd -P)"
 610                 case "$_abspd" in
 611                 "$_absrr/"?*)
 612                         # The normal case
 613                         _projpart="${_abspd#$_absrr/}"
 614                         ;;
 615                 *)
 616                         # Must have been reached via a symbolic link, but
 617                         # we have no way to know the source, so use a
 618                         # generic "_external" leader combined with just the
 619                         # trailing directory name
 620                         _abspd="${_abspd%/}"
 621                         _abspd="${_abspd%/.git}"
 622                         _projpart="_external/${_abspd##*/}"
 623                         ;;
 624                 esac
 625         esac
 626         eval "$1="'"$_projpart"'
 627 }
 628
 629 # Return success if the given project name has at least one immediate child fork
 630 # that has a non-zero length alternates file
 631 has_forks_with_alternates() {
 632         _prj="${1%.git}"
 633         [ -n "$_prj" ] || return 1
 634         [ -d "$cfg_reporoot/$_prj" ] || return 1
 635         is_git_dir "$cfg_reporoot/$_prj.git" || return 1
 636         if
 637                 get_repo_list "$_prj/[^/:][^/:]*:" |
 638                 while read -r _prjname && [ -n "$_prjname" ]; do
 639                         ! [ -s "$cfg_reporoot/$_prjname.git/objects/info/alternates" ] ||
 640                         exit 1 # will only exit implicit subshell created by '|'
 641                 done
 642         then
 643                 return 1
 644         fi
 645         return 0
 646 }
 647
 648 # returns empty string and error for empty string otherwise one of
 649 #   m  =>  normal Git mirror
 650 #   s  =>  mirror from svn source
 651 #   d  =>  mirror from darcs source
 652 #   b  =>  mirror from bzr source
 653 #   h  =>  mirror from hg source
 654 #   w  =>  mirror from mediawiki source
 655 #   f  =>  mirror from other fast-import source
 656 # note that if the string is non-empty and none of s, d, b or h match the
 657 # return will always be type m regardless of whether it's a valid Git URL
 658 get_url_mirror_type() {
 659         case "$1" in
 660                 "")
 661                         return 1
 662                         ;;
 663                 svn://* | svn+http://* | svn+https://* | svn+file://* | svn+ssh://*)
 664                         echo 's'
 665                         ;;
 666                 darcs://*)
 667                         echo 'd'
 668                         ;;
 669                 bzr://*)
 670                         echo 'b'
 671                         ;;
 672                 hg+http://* | hg+https://* | hg+file://* | hg+ssh://* | hg::*)
 673                         echo 'h'
 674                         ;;
 675                 mediawiki::*)
 676                         echo 'w'
 677                         ;;
 678                 *)
 679                         echo 'm'
 680                         ;;
 681         esac
 682         return 0
 683 }
 684
 685 # returns false for empty string
 686 # returns true if the passed in url is a mirror using git fast-import
 687 is_gfi_mirror_url() {
 688         [ -n "$1" ] || return 1
 689         case "$(get_url_mirror_type "$1" 2>/dev/null || :)" in
 690                 d|b|h|w|f)
 691                         # darcs, bzr, hg and mediawiki mirrors use git fast-import
 692                         # and so do generic "f" fast-import mirrors
 693                         return 0
 694                         ;;
 695                 *)
 696                         # Don't think git-svn currently uses git fast-import
 697                         # And Git mirrors certainly do not
 698                         return 1
 699                         ;;
 700         esac
 701         # assume it does not use git fast-import
 702         return 1
 703 }
 704
 705 # returns false for empty string
 706 # returns true if the passed in url is a mirror using git-svn
 707 is_svn_mirror_url() {
 708         [ -n "$1" ] || return 1
 709         [ "$(get_url_mirror_type "$1" 2>/dev/null || :)" = "s" ]
 710 }
 711
 712 # returns mirror url for gitweb.baseurl of git directory
 713 # (GIT_DIR) passed in as the argument (which defaults to "." if omitted)
 714 # will fail if the directory does not have .nofetch and gitweb.baseurl
 715 # comes back empty -- otherwise .nofetch directories succeed with a "" return
 716 # automatically strips any leading "disabled " prefix before returning result
 717 get_mirror_url() {
 718         _gitdir="${1:-.}"
 719         # always return empty for non-mirrors
 720         ! [ -e "$_gitdir/.nofetch" ] || return 0
 721         _url="$(GIT_DIR="$_gitdir" config_get baseurl 2>/dev/null)" || :
 722         _url="${_url##* }"
 723         [ -n "$_url" ] || return 1
 724         printf '%s\n' "$_url"
 725         return 0
 726 }
 727
 728 # returns get_url_mirror_type for gitweb.baseurl of git directory
 729 # (GIT_DIR) passed in as the argument (which defaults to "." if omitted)
 730 # will fail if the directory does not have .nofetch and gitweb.baseurl
 731 # comes back empty -- otherwise .nofetch directories succeed with a "" return
 732 # automatically strips any leading "disabled " prefix before testing
 733 get_mirror_type() {
 734         _url="$(get_mirror_url "$@")" || return 1
 735         [ -n "$_url" ] || return 0
 736         get_url_mirror_type "$_url"
 737 }
 738
 739 # returns true if the passed in git dir (defaults to ".") is a mirror using git fast-import
 740 is_gfi_mirror() {
 741         _url="$(get_mirror_url "$@")" || return 1
 742         is_gfi_mirror_url "$_url"
 743 }
 744
 745 # returns true if the passed in git dir (defaults to ".") is a mirror using git-svn
 746 is_svn_mirror() {
 747         _url="$(get_mirror_url "$@")" || return 1
 748         is_svn_mirror_url "$_url"
 749 }
 750
 751 # current directory must already be set to Git repository
 752 # if girocco.headok is already true succeeds without doing anything
 753 # if rev-parse --verify HEAD succeeds sets headok=true and succeeds
 754 # otherwise tries to set HEAD to a symbolic ref to refs/heads/master
 755 # then refs/heads/trunk and finally the first top-level head from
 756 # refs/heads/* (i.e. only two slashes in the name) and finally any
 757 # existing refs/heads.  The first one to succeed wins and sets headok=true
 758 # and then a successful exit.  Otherwise headok is left unset with a failure exit
 759 # We use the girocco.headok flag to make sure we only force a valid HEAD symref
 760 # when the repository is being set up -- if the HEAD is later deleted (through
 761 # a push or fetch --prune) that's no longer our responsibility to fix
 762 check_and_set_head() {
 763         [ "$(git config --bool girocco.headok 2>/dev/null || :)" != "true" ] || return 0
 764         if git rev-parse --verify --quiet HEAD >/dev/null; then
 765                 git config --bool girocco.headok true
 766                 return 0
 767         fi
 768         for _hr in refs/heads/master refs/heads/trunk; do
 769                 if git rev-parse --verify --quiet "$_hr" >/dev/null; then
 770                         _update_head_symref "$_hr"
 771                         return 0
 772                 fi
 773         done
 774         git for-each-ref --format="%(refname)" refs/heads 2>/dev/null |
 775         while read -r _hr; do
 776                 case "${_hr#refs/heads/}" in */*) :;; *)
 777                         _update_head_symref "$_hr"
 778                         exit 1 # exit subshell created by "|"
 779                 esac
 780         done || return 0
 781         _hr="$(git for-each-ref --format="%(refname)" refs/heads 2>/dev/null | head -n 1)" || :
 782         if [ -n "$_hr" ]; then
 783                 _update_head_symref "$_hr"
 784                 return 0
 785         fi
 786         return 1
 787 }
 788 _update_head_symref() {
 789         git symbolic-ref HEAD "$1"
 790         git config --bool girocco.headok true
 791         ! [ -d htmlcache ] || { >htmlcache/changed; } 2>/dev/null || :
 792 }
 793
 794 # current directory must already be set to Git repository
 795 # if the directory needs to have gc run and .needsgc is not already set
 796 # then .needsgc will be set triggering a "mini" gc at the next opportunity
 797 # Girocco shouldn't generate any loose objects but we check for that anyway
 798 check_and_set_needsgc() {
 799         ! [ -e .needsgc ] || return 0
 800         _packs=
 801         { _packs="$(list_packs --quiet --count --exclude-no-idx --exclude-keep objects/pack || :)" || :; } 2>/dev/null
 802         if [ "${_packs:-0}" -ge 20 ]; then
 803                 >.needsgc
 804                 return 0
 805         fi
 806         _logfiles=
 807         { _logfiles="$(($(find -L reflogs -maxdepth 1 -type f -print | wc -l || :)+0))" || :; } 2>/dev/null
 808         if [ "${_logfiles:-0}" -ge 50 ]; then
 809                 >.needsgc
 810                 return 0
 811         fi
 812         # Truly git gc only checks the number of objects in the objects/17 directory
 813         # We check for -ge 10 which should make the probability of having more than
 814         # 5120 (20*256) loose objects present when there are less than 10 in
 815         # objects/17 vanishingly small (20 is the threshold we use for pack files)
 816         _objfiles=
 817         ! [ -d objects/17 ] ||
 818         { _objfiles="$(($(find -L objects/17 -type f -name "$octet19*" -print | wc -l || :)+0))" || :; } 2>/dev/null
 819         if [ "${_objfiles:-0}" -ge 10 ]; then
 820                 >.needsgc
 821                 return 0
 822         fi
 823 }
 824
 825 # A well-known UTF-8 locale is required for some of the fast-import providers
 826 # in order to avoid mangling characters.  Ideally we could use "POSIX.UTF-8"
 827 # but that is not reliably UTF-8 but rather usually US-ASCII.
 828 # We parse the output of `locale -a` and select a suitable UTF-8 locale at
 829 # install time and store that in $var_utf8_locale if one is found.
 830 # If we cannot find one in the `locale -a` output then we just use a well-known
 831 # UTF-8 locale and hope for the best.  We set LC_ALL to our choice and export
 832 # it.  We only set this temporarily when running the fast-import providers.
 833 set_utf8_locale() {
 834         LC_ALL="${var_utf8_locale:-en_US.UTF-8}"
 835         export LC_ALL
 836 }
 837
 838 # hg-fast-export | git fast-import with error handling in current directory GIT_DIR
 839 git_hg_fetch() (
 840         set_utf8_locale
 841         _python="${PYTHON:-python}"
 842         rm -f hg2git-marks.old hg2git-marks.new
 843         if [ -f hg2git-marks ] && [ -s hg2git-marks ]; then
 844                 LC_ALL=C sed 's/^:\([^ ][^ ]*\) \([^ ][^ ]*\)$/\2 \1/' <hg2git-marks | {
 845                         if [ -n "$var_have_git_185" ]; then
 846                                 git cat-file --batch-check=':%(rest) %(objectname)'
 847                         else
 848                                 LC_ALL=C sed 's/^\([^ ][^ ]*\) \([^ ][^ ]*\)$/:\2 \1/'
 849                         fi
 850                 } | LC_ALL=C sed '/ missing$/d' >hg2git-marks.old
 851                 if [ -n "$var_have_git_171" ] &&
 852                    git rev-parse --quiet --verify refs/notes/hg >/dev/null; then
 853                         if [ -z "$var_have_git_185" ] ||
 854                            ! LC_ALL=C cmp -s hg2git-marks hg2git-marks.old; then
 855                                 _nm='hg-fast-export'
 856                                 GIT_AUTHOR_NAME="$_nm"
 857                                 GIT_COMMITTER_NAME="$_nm"
 858                                 GIT_AUTHOR_EMAIL="$_nm"
 859                                 GIT_COMMITTER_EMAIL="$_nm"
 860                                 export GIT_AUTHOR_NAME
 861                                 export GIT_COMMITTER_NAME
 862                                 export GIT_AUTHOR_EMAIL
 863                                 export GIT_COMMITTER_EMAIL
 864                                 git notes --ref=refs/notes/hg prune
 865                                 unset GIT_AUTHOR_NAME
 866                                 unset GIT_COMMITTER_NAME
 867                                 unset GIT_AUTHOR_EMAIL
 868                                 unset GIT_COMMITTER_EMAIL
 869                         fi
 870                 fi
 871         else
 872                 >hg2git-marks.old
 873         fi
 874         _err1=
 875         _err2=
 876         exec 3>&1
 877         { read -r _err1 || :; read -r _err2 || :; } <<-EOT
 878         $(
 879                 exec 4>&3 3>&1 1>&4 4>&-
 880                 {
 881                         _e1=0
 882                         _af="$(git config hg.authorsfile)" || :
 883                         _cmd='GIT_DIR="$(pwd)" "$_python" "$cfg_basedir/bin/hg-fast-export.py" \
 884                                 --repo    "$(pwd)/repo.hg" \
 885                                 --marks   "$(pwd)/hg2git-marks.old" \
 886                                 --mapping "$(pwd)/hg2git-mapping" \
 887                                 --heads   "$(pwd)/hg2git-heads" \
 888                                 --status  "$(pwd)/hg2git-state" \
 889                                 -U unknown --force --flatten --hg-hash'
 890                         [ -z "$_af" ] || _cmd="$_cmd"' --authors "$_af"'
 891                         eval "$_cmd" 3>&- || _e1=$?
 892                         echo $_e1 >&3
 893                 } |
 894                 {
 895                         _e2=0
 896                         git fast-import \
 897                                 --import-marks="$(pwd)/hg2git-marks.old" \
 898                                 --export-marks="$(pwd)/hg2git-marks.new" \
 899                                 --export-pack-edges="$(pwd)/gfi-packs" \
 900                                 --force 3>&- || _e2=$?
 901                         echo $_e2 >&3
 902                 }
 903         )
 904         EOT
 905         exec 3>&-
 906         [ "$_err1" = 0 ] && [ "$_err2" = 0 ] || return 1
 907         mv -f hg2git-marks.new hg2git-marks
 908         rm -f hg2git-marks.old
 909         git for-each-ref --format='%(refname) %(objectname)' refs/heads |
 910         LC_ALL=C sed -e 's,^refs/heads/,:,' >hg2git-heads
 911 )