clone.sh/update.sh: set fastimport.unpackLimit to 0
[girocco.git] / taskd / clone.sh
blob45fbd071855ed45b41135ed2894220c6f9ebc5b5
1 #!/bin/sh
3 # Invoked from taskd/taskd.pl
5 . @basedir@/shlib.sh
7 set -e
9 umask 002
10 [ "$cfg_permission_control" != "Hooks" ] || umask 000
11 clean_git_env
13 # darcs fast-export | git fast-import with error handling
14 git_darcs_fetch() (
15 set_utf8_locale
16 _err1=
17 _err2=
18 exec 3>&1
19 { read -r _err1 || :; read -r _err2 || :; } <<-EOT
21 exec 4>&3 3>&1 1>&4 4>&-
23 _e1=0
24 "$cfg_basedir"/bin/darcs-fast-export \
25 --export-marks="$(pwd)/dfe-marks" "$1" 3>&- || _e1=$?
26 echo $_e1 >&3
27 } | \
29 _e2=0
30 git fast-import \
31 --export-marks="$(pwd)/gfi-marks" \
32 --export-pack-edges="$(pwd)/gfi-packs" \
33 --force 3>&- || _e2=$?
34 echo $_e2 >&3
37 EOT
38 exec 3>&-
39 [ "$_err1" = 0 -a "$_err2" = 0 ]
40 return $?
43 # bzr fast-export | git fast-import with error handling
44 git_bzr_fetch() (
45 set_utf8_locale
46 BZR_LOG=/dev/null
47 export BZR_LOG
48 _err1=
49 _err2=
50 exec 3>&1
51 { read -r _err1 || :; read -r _err2 || :; } <<-EOT
53 exec 4>&3 3>&1 1>&4 4>&-
55 _e1=0
56 bzr fast-export --plain \
57 --export-marks="$(pwd)/bfe-marks" "$1" 3>&- || _e1=$?
58 echo $_e1 >&3
59 } | \
61 _e2=0
62 git fast-import \
63 --export-marks="$(pwd)/gfi-marks" \
64 --export-pack-edges="$(pwd)/gfi-packs" \
65 --force 3>&- || _e2=$?
66 echo $_e2 >&3
69 EOT
70 exec 3>&-
71 [ "$_err1" = 0 -a "$_err2" = 0 ]
72 return $?
75 send_clone_failed() {
76 trap "" EXIT
77 # We must now close the .clonelog file that is open on stdout and stderr
78 exec >/dev/null 2>&1
79 failaddrs="$(config_get owner || :)"
80 [ -z "$cfg_admincc" -o "$cfg_admincc" = "0" -o -z "$cfg_admin" ] || \
81 if [ -z "$failaddrs" ]; then failaddrs="$cfg_admin"; else failaddrs="$failaddrs,$cfg_admin"; fi
82 [ -z "$failaddrs" ] || \
84 cat <<EOT
85 Condolences. The clone of project $proj just failed.
87 * Source URL: $url
88 * Project settings: $cfg_webadmurl/editproj.cgi?name=$(echo "$proj" | LC_ALL=C sed -e 's/[+]/%2B/g')
90 The project settings link may be used to adjust the settings
91 and restart the clone in order to try the clone again.
92 EOT
93 if [ -f .clonelog -a -r .clonelog ]; then
94 echo ""
95 echo "Log follows:"
96 echo ""
97 loglines=$(LC_ALL=C wc -l <.clonelog)
98 if [ $loglines -le 203 ]; then
99 cat .clonelog
100 else
101 head -n 100 .clonelog
102 echo ""
103 echo "[ ... elided $(( $loglines - 200 )) middle lines ... ]"
104 echo ""
105 tail -n 100 .clonelog
108 } | mail -s "[$cfg_name] $proj clone failed" "$failaddrs" || :
111 # removes all leftovers from a previous failed clone attempt
112 cleanup_failed_clone() {
114 # Remove any left-over svn-remote.svn or remote.origin config
115 git config --remove-section svn-remote.svn 2>/dev/null || :
116 git config --remove-section remote.origin 2>/dev/null || :
118 # If there is a remote-template.origin section, pre-seed the
119 # remote.origin section with its contents
120 git config --get-regexp '^remote-template\.origin\..' |
121 while read name value; do
122 if [ -n "$name" -a -n "$value" ]; then
123 git config "remote${name#remote-template}" "$value"
125 done
127 # Any pre-existing FETCH_HEAD from a previous clone failed or not is
128 # now garbage to be removed
129 rm -f FETCH_HEAD
131 # Remove any left-over svn dir from a previous failed attempt
132 rm -rf svn
134 # Remove any left-over .darcs dirs from a previous failed attempt
135 rm -rf *.darcs
137 # Remove any left-over repo.hg dir from a previous failed attempt
138 rm -rf repo.hg
140 # Remove any left-over import/export/temp files from a previous failed attempt
141 rm -f bfe-marks dfe-marks hg2git-heads hg2git-mapping hg2git-marks* hg2git-state \
142 gfi-marks gfi-packs .pkts-temp .refs-temp
144 # We want a gc right after the clone, so re-enable that just in case.
145 # There's a potential race where we could add it and gc.sh could remove
146 # it, but we'll reunset lastgc just before we remove .delaygc at the end.
147 [ -e .delaygc ] || >.delaygc
148 git config --unset gitweb.lastgc 2>/dev/null || :
150 # Remove all pre-existing refs
151 rm -f packed-refs
152 git for-each-ref --format='delete %(refname)' | git_updateref_stdin 2>/dev/null || :
154 # The initial state before a clone starts has HEAD as a symbolic-ref to master
155 git symbolic-ref HEAD refs/heads/master
157 # HEAD is no longer "ok"
158 git config --unset girocco.headok 2>/dev/null || :
160 # We, perhaps, ought to remove any packs/loose objects now, but the next gc
161 # will get rid of any extras. Also, if we're recloning the same thing, any
162 # preexisting packs/loose objects containing what we're recloning will only
163 # speed up the reclone by avoiding some disk writes. So we don't kill them.
166 proj="${1%.git}"
167 cd "$cfg_reporoot/$proj.git"
168 bang_reset
170 ! [ -e .delaygc ] || >.allowgc || :
172 trap "echo '@OVER@'; touch .clone_failed; send_clone_failed" EXIT
173 echo "Project: $proj"
174 echo ""
175 [ -n "$cfg_mirror" ] || { echo "Mirroring is disabled" >&2; exit 1; }
176 url="$(config_get baseurl || :)"
177 case "$url" in *" "*|*" "*|"")
178 echo "Bad mirror URL (\"$url\")"
179 exit 1
180 esac
182 cleanup_failed_clone
184 # Record original mirror type for use by update.sh
185 mirror_type="$(get_url_mirror_type "$url")"
186 git config girocco.mirrortype "$mirror_type"
188 echo "Mirroring from URL \"$url\""
189 echo ""
191 if [ "$cfg_project_owners" = "source" ]; then
192 config_set owner "$(stat -c %U "$url" 2>/dev/null)"
195 mailaddrs="$(config_get owner || :)"
196 [ -z "$cfg_admin" ] || \
197 if [ -z "$mailaddrs" ]; then mailaddrs="$cfg_admin"; else mailaddrs="$mailaddrs,$cfg_admin"; fi
199 # Make sure we don't get any unwanted loose objects
200 # Starting with Git v2.10.0 fast-import can generate loose objects unless we
201 # tweak its configuration to prevent that
202 git_add_config 'fetch.unpackLimit=1'
203 # Note the git config documentation is wrong
204 # transfer.unpackLimit, if set, overrides fetch.unpackLimit
205 git_add_config 'transfer.unpackLimit=1'
206 # But not the Git v2.10.0 and later fastimport.unpackLimit which improperly uses <= instead of <
207 git_add_config 'fastimport.unpackLimit=0'
209 # Initial mirror
210 echo "Initiating mirroring..."
211 headref=
212 showheadwarn=
213 warnempty=
214 case "$url" in
215 svn://* | svn+http://* | svn+https://* | svn+file://* | svn+ssh://*)
216 [ -n "$cfg_mirror_svn" ] || { echo "Mirroring svn is disabled" >&2; exit 1; }
217 # We just remove svn+ here, so svn+http://... becomes http://...
218 # We also remove a trailing '/' to match what git-svn will do
219 case "$url" in svn+ssh://*) svnurl="$url";; *) svnurl="${url#svn+}";; esac
220 svnurl="${svnurl%/}"
221 # Use an 'anonsvn' username as is commonly used for anonymous svn
222 # Use an 'anonsvn' password as is commonly used for anonymous svn
223 GIT_ASKPASS_PASSWORD=anonsvn
224 export GIT_ASKPASS_PASSWORD
225 # We require svn info to succeed on the URL otherwise it's
226 # simply not a valid URL and without using -s on the init it
227 # will not otherwise be tested until the fetch
228 svn --non-interactive --username anonsvn --password anonsvn info "$svnurl" > /dev/null
229 # We initially use -s for the init which will possibly shorten
230 # the URL. However, the shortening can fail if a password is
231 # not required for the longer version but is for the shorter,
232 # so try again without -s if the -s version fails.
233 # We must use GIT_DIR=. here or ever so "helpful" git-svn will
234 # create a .git subdirectory!
235 GIT_DIR=. git svn init --username=anonsvn --prefix "" -s "$svnurl" < /dev/null || \
236 GIT_DIR=. git svn init --username=anonsvn --prefix "" "$svnurl" < /dev/null
237 # We need to remember this url so we can detect changes because
238 # ever so "helpful" git-svn may shorten it!
239 config_set svnurl "$svnurl"
240 # At this point, since we asked for a standard layout (-s) git-svn
241 # may have been "helpful" and adjusted our $svnurl to a prefix and
242 # then glued the removed suffix onto the front of any svn-remote.svn.*
243 # config items. We could avoid this by not using the '-s' option
244 # but then we might not get all the history. If, for example, we
245 # are cloning an http://svn.example.com/repos/public repository that
246 # early in its history moved trunk => public/trunk we would miss that
247 # earlier history without allowing the funky shorten+prefix behavior.
248 # So we read back the svn-remote.svn.fetch configuration and compute
249 # the prefix. This way we are sure to get the correct prefix.
250 gitsvnurl="$(git config --get svn-remote.svn.url || :)"
251 gitsvnfetch="$(git config --get-all svn-remote.svn.fetch | tail -1 || :)"
252 gitsvnprefix="${gitsvnfetch%%:*}"
253 gitsvnsuffix="${gitsvnprefix##*/}"
254 gitsvnprefix="${gitsvnprefix%$gitsvnsuffix}"
255 # Ask git-svn to store everything in the normal non-remote
256 # locations being careful to use the correct prefix
257 git config --replace-all svn-remote.svn.fetch "${gitsvnprefix}trunk:refs/heads/master"
258 git config --replace-all svn-remote.svn.branches "${gitsvnprefix}branches/*:refs/heads/*"
259 git config --replace-all svn-remote.svn.tags "${gitsvnprefix}tags/*:refs/tags/*"
260 # look for additional non-standard directories to fetch
261 # check for standard layout at the same time
262 foundstd=
263 foundfile=
264 { svn --non-interactive --username anonsvn --password anonsvn ls "$gitsvnurl/${gitsvnprefix}" 2>/dev/null || :; } | \
265 { while read file; do case $file in
266 # skip the already-handled standard ones and any with a space or tab
267 *' '*|*' '*) :;;
268 trunk/|branches/|tags/) foundstd=1;;
269 # only fetch extra directories from the $svnurl root (not any files)
270 *?/) git config --add svn-remote.svn.fetch \
271 "${gitsvnprefix}${file%/}:refs/heads/${file%/}";;
272 *?) foundfile=1;;
273 esac; done
274 # if files found and no standard directories present use a simpler layout
275 if [ -z "$foundstd" ] && [ -n "$foundfile" ]; then
276 git config --unset svn-remote.svn.branches
277 git config --unset svn-remote.svn.tags
278 git config --replace-all svn-remote.svn.fetch ':refs/heads/master'
279 fi; }
280 # remember the starting time so we can easily combine fetched loose objects
281 # we sleep for 1 second after creating .svnpack to make sure all objects are newer
282 if ! [ -e .svnpack ]; then
283 rm -f .svnpack
284 >.svnpack
285 sleep 1
287 # Again, be careful to use GIT_DIR=. here or else new .git subdirectory!
288 GIT_DIR=. git svn fetch --log-window-size=$var_log_window_size --username=anonsvn --quiet < /dev/null
289 # git svn does not preserve group permissions in the svn subdirectory
290 chmod -R ug+rw,o+r svn
291 # git svn also leaves behind ref turds that end with @nnn
292 # We get rid of them now
293 git for-each-ref --format='%(refname)' |
294 LC_ALL=C sed '/^..*@[1-9][0-9]*$/!d; s/^/delete /' |
295 git_updateref_stdin
296 unset GIT_ASKPASS_PASSWORD
298 darcs://*)
299 [ -n "$cfg_mirror_darcs" ] || { echo "Mirroring darcs is disabled" >&2; exit 1; }
300 httpurl="http://${url#darcs://}"
301 git_darcs_fetch "$httpurl"
303 bzr://*)
304 [ -n "$cfg_mirror_bzr" ] || { echo "Mirroring bzr is disabled" >&2; exit 1; }
305 # we just remove bzr:// here, a typical bzr url is just
306 # "lp:foo"
307 bzrurl="${url#bzr://}"
308 git_bzr_fetch "$bzrurl"
310 hg+http://* | hg+https://* | hg+file://* | hg+ssh://*)
311 [ -n "$cfg_mirror_hg" ] || { echo "Mirroring hg is disabled" >&2; exit 1; }
312 # We just remove hg+ here, so hg+http://... becomes http://...
313 hgurl="${url#hg+}"
314 # Perform the initial hg clone
315 hg clone -U "$hgurl" "$(pwd)/repo.hg"
316 # Do the fast-export | fast-import
317 git_hg_fetch
320 # We manually add remote.origin.url and remote.origin.fetch
321 # to simulate a `git remote add --mirror=fetch` since that's
322 # not available until Git 1.7.5 and this way we guarantee we
323 # always get exactly the intended configuration and nothing else.
324 git config remote.origin.url "$url"
325 if ! is_gfi_mirror_url "$url" && [ "$(git config --bool girocco.cleanmirror 2>/dev/null || :)" = "true" ]; then
326 git config --replace-all remote.origin.fetch "+refs/heads/*:refs/heads/*"
327 git config --add remote.origin.fetch "+refs/tags/*:refs/tags/*"
328 git config --add remote.origin.fetch "+refs/notes/*:refs/notes/*"
329 git config --add remote.origin.fetch "+refs/top-bases/*:refs/top-bases/*"
330 git config --bool girocco.lastupdateclean true
331 else
332 git config --replace-all remote.origin.fetch "+refs/*:refs/*"
333 git config --bool girocco.lastupdateclean false
335 # Set the correct HEAD symref by using ls-remote first
336 GIT_SSL_NO_VERIFY=1 GIT_TRACE_PACKET=1 git ls-remote origin >.refs-temp 2>.pkts-temp || \
338 # Since everything was redirected, on failure there'd be no output,
339 # so let's make some failure output
340 cat .pkts-temp
341 echo ""
342 echo "git ls-remote \"$url\" failed"
343 exit 1
345 # Compensate for git() {} side effects
346 unset GIT_TRACE_PACKET
347 # If the server is running at least Git 1.8.4.3 then it will send us the actual
348 # symref for HEAD. If we are running at least Git 1.7.5 then we can snarf that
349 # out of the packet trace data.
350 if [ -s .refs-temp ]; then
351 # Nothing to do unless the remote repository has at least 1 ref
352 # See if we got a HEAD ref
353 head="$(LC_ALL=C grep -E "^$octet20[ $tab]+HEAD\$" <.refs-temp | LC_ALL=C awk '{print $1}')"
354 # If the remote has HEAD set to a symbolic ref that does not exist
355 # then we will not receive a HEAD ref in the ls-remote output
356 headref=
357 showheadwarn=
358 symrefcap=
359 if [ -n "$head" ]; then
360 symrefcap="$(LC_ALL=C sed -ne <.pkts-temp \
361 "/packet:.*git<.*[ $tab]symref="'HEAD:refs\/heads\/'"[^ $tab]/\
362 {s/^.*[ $tab]symref="'HEAD:\(refs\/heads\/'"[^ $tab][^ $tab]*"'\).*$/\1/;p;}')"
363 # prefer $symrefcap (refs/heads/master if no $symrefcap) if it
364 # matches HEAD otherwise take the first refs/heads/... match
365 matchcnt=0
366 while read ref; do
367 [ -n "$ref" ] || continue
368 matchcnt=$(( $matchcnt + 1 ))
369 if [ -z "$headref" ] || [ "$ref" = "${symrefcap:-refs/heads/master}" ]; then
370 headref="$ref"
372 [ "$headref" = "${symrefcap:-refs/heads/master}" -a $matchcnt -gt 1 ] && break
373 done <<-EOT
374 $(LC_ALL=C grep -E "^$head[ $tab]+refs/heads/[^ $tab]+\$" <.refs-temp | \
375 LC_ALL=C awk '{print $2}')
377 # Warn if there was more than one match and $symrefcap is empty
378 # or $symrefcap is not the same as $headref since our choice might
379 # differ from the source repository's HEAD
380 if [ $matchcnt -ge 1 -a "$symrefcap" != "$headref" ] && \
381 [ -n "$symrefcap" -o $matchcnt -gt 1 ]; then
382 showheadwarn=1
385 if [ -z "$headref" ]; then
386 # If we still don't have a HEAD ref then prefer refs/heads/master
387 # if it exists otherwise take the first refs/heads/...
388 # We do not support having a detached HEAD.
389 # We always warn now because we will be setting HEAD differently
390 # than the source repository had HEAD set
391 showheadwarn=1
392 while read ref; do
393 [ -n "$ref" ] || continue
394 if [ -z "$headref" ] || [ "$ref" = "refs/heads/master" ]; then
395 headref="$ref"
397 [ "$headref" = "refs/heads/master" ] && break
398 done <<-EOT
399 $(LC_ALL=C grep -E "^$octet20[ $tab]+refs/heads/[^ $tab]+\$" <.refs-temp | \
400 LC_ALL=C awk '{print $2}')
403 # If we STILL do not have a HEAD ref (perhaps the source repository
404 # contains only tags) then use refs/heads/master. It will be invalid
405 # but is no worse than we used to do by default and we'll warn about
406 # it. We do not support a HEAD symref to anything other than refs/heads/...
407 [ -n "$headref" ] || headref="refs/heads/master"
408 git symbolic-ref HEAD "$headref"
409 pruneopt=--prune
410 [ "$(git config --bool fetch.prune 2>/dev/null || :)" != "false" ] || pruneopt=
411 # remember the starting time so we can easily detect new packs for fast-import mirrors
412 # we sleep for 1 second after creating .gfipack to make sure all packs are newer
413 if is_gfi_mirror_url "$url" && [ ! -e .gfipack ]; then
414 rm -f .gfipack
415 >.gfipack
416 sleep 1
418 GIT_SSL_NO_VERIFY=1 git remote update $pruneopt
419 if [ -e .gfipack ] && is_gfi_mirror_url "$url"; then
420 find objects/pack -type f -newer .gfipack -name "pack-$octet20.pack" -print >>gfi-packs
421 rm -f .gfipack
423 else
424 warnempty=1
425 git symbolic-ref HEAD "refs/heads/master"
427 rm -f .refs-temp .pkts-temp
429 esac
431 # The objects subdirectories permissions must be updated now.
432 # In the case of a dumb http clone, the permissions will not be correct
433 # (missing group write) despite the core.sharedrepository=1 setting!
434 # The objects themselves seem to have the correct permissions.
435 # This problem appears to have been fixed in the most recent git versions.
436 perms=g+w
437 [ "$cfg_permission_control" != "Hooks" ] || perms=go+w
438 chmod $perms $(find objects -maxdepth 1 -type d) 2>/dev/null || :
440 # We may have just cloned a lot of refs and they will all be
441 # individual files at this point. Let's pack them now so we
442 # can have better performance right from the start.
443 git pack-refs --all
445 # Initialize gitweb.lastreceive, gitweb.lastchange and info/lastactivity
446 git config gitweb.lastreceive "$(date '+%a, %d %b %Y %T %z')"
447 git config gitweb.lastchange "$(date '+%a, %d %b %Y %T %z')"
448 git for-each-ref --sort=-committerdate --format='%(committerdate:iso8601)' \
449 --count=1 refs/heads > info/lastactivity || :
450 ! [ -d htmlcache ] || { >htmlcache/changed; } 2>/dev/null || :
452 # Don't leave a multi-megabyte useless FETCH_HEAD behind
453 rm -f FETCH_HEAD
455 # Last ditch attempt to get a valid HEAD for a non-git source
456 check_and_set_head || :
458 # The rest
459 echo "Final touches..."
460 git update-server-info
461 trap "" EXIT
463 # run gc now unless the clone is empty
464 if [ -z "$warnempty" ]; then
465 git config --unset gitweb.lastgc 2>/dev/null || :
466 rm -f .delaygc .allowgc
469 emptynote=
470 [ -z "$warnempty" ] ||
471 emptynote="
472 WARNING: You have mirrored an empty repository.
474 headnote=
475 [ -n "$showheadwarn" -a -n "$headref" ] &&
476 headnote="
477 NOTE: HEAD has been set to a symbolic ref to \"$headref\".
478 Use the \"Project settings\" link to choose a different HEAD symref.
480 sizenote=
481 ! is_gfi_mirror ||
482 sizenote="
483 NOTE: Since this is a mirror of a non-Git source, the initial repository
484 size may be somewhat larger than necessary. This will be corrected
485 shortly. If you intend to clone this repository you may want to
486 wait up to 1 hour before doing so in order to receive the more
487 compact final size.
489 [ -z "$mailaddrs" ] ||
490 mail -s "[$cfg_name] $proj clone completed" "$mailaddrs" <<EOT || :
491 Congratulations! The clone of project $proj just completed.
493 * Source URL: $url
494 * GitWeb interface: $cfg_gitweburl/$proj.git
495 * Project settings: $cfg_webadmurl/editproj.cgi?name=$(echo "$proj" | LC_ALL=C sed -e 's/[+]/%2B/g')
496 $emptynote$headnote$sizenote
497 Have a lot of fun.
500 echo "Mirroring finished successfuly!"
501 # In case this is a re-mirror, lastgc could have been set already so clear it now
502 git config --unset gitweb.lastgc || :
503 rm .clone_in_progress
504 echo "$sizenote@OVER@"