pyrox.dev
1#! /usr/bin/env bash
2
3set -e -o pipefail
4
5url=
6rev=
7expHash=
8hashType=$NIX_HASH_ALGO
9deepClone=$NIX_PREFETCH_GIT_DEEP_CLONE
10leaveDotGit=$NIX_PREFETCH_GIT_LEAVE_DOT_GIT
11fetchSubmodules=
12fetchLFS=
13builder=
14fetchTags=
15branchName=$NIX_PREFETCH_GIT_BRANCH_NAME
16
17# ENV params
18out=${out:-}
19http_proxy=${http_proxy:-}
20
21# NOTE: use of NIX_GIT_SSL_CAINFO is for backwards compatibility; NIX_SSL_CERT_FILE is preferred
22# as of PR#303307
23GIT_SSL_CAINFO=${NIX_GIT_SSL_CAINFO:-$NIX_SSL_CERT_FILE}
24
25# populated by clone_user_rev()
26fullRev=
27humanReadableRev=
28commitDate=
29commitDateStrict8601=
30
31if test -n "$deepClone"; then
32 deepClone=true
33else
34 deepClone=
35fi
36
37if test "$leaveDotGit" != 1; then
38 leaveDotGit=
39else
40 leaveDotGit=true
41fi
42
43usage(){
44 echo >&2 "syntax: nix-prefetch-git [options] [URL [REVISION [EXPECTED-HASH]]]
45
46Options:
47 --out path Path where the output would be stored.
48 --url url Any url understood by 'git clone'.
49 --rev ref Any sha1 or references (such as refs/heads/master)
50 --hash h Expected hash.
51 --name n Symbolic store path name to use for the result (default: based on URL)
52 --branch-name Branch name to check out into
53 --sparse-checkout Only fetch and checkout part of the repository.
54 --non-cone-mode Use non-cone mode for sparse checkouts.
55 --deepClone Clone the entire repository.
56 --no-deepClone Make a shallow clone of just the required ref.
57 --leave-dotGit Keep the .git directories.
58 --fetch-lfs Fetch git Large File Storage (LFS) files.
59 --fetch-submodules Fetch submodules.
60 --fetch-tags Fetch all tags (useful for git describe).
61 --builder Clone as fetchgit does, but url, rev, and out option are mandatory.
62 --no-add-path Do not actually add the contents of the git repo to the store.
63 --root-dir dir Directory in the repository that will be copied to the output instead of the full repository.
64 --quiet Only print the final json summary.
65"
66 exit 1
67}
68
69# some git commands print to stdout, which would contaminate our JSON output
70clean_git(){
71 git "$@" >&2
72}
73
74argi=0
75argfun=""
76for arg; do
77 if test -z "$argfun"; then
78 case $arg in
79 --out) argfun=set_out;;
80 --url) argfun=set_url;;
81 --rev) argfun=set_rev;;
82 --hash) argfun=set_hashType;;
83 --name) argfun=set_symbolicName;;
84 --branch-name) argfun=set_branchName;;
85 --deepClone) deepClone=true;;
86 --sparse-checkout) argfun=set_sparseCheckout;;
87 --non-cone-mode) nonConeMode=true;;
88 --quiet) QUIET=true;;
89 --no-deepClone) deepClone=;;
90 --leave-dotGit) leaveDotGit=true;;
91 --fetch-lfs) fetchLFS=true;;
92 --fetch-submodules) fetchSubmodules=true;;
93 --fetch-tags) fetchTags=true;;
94 --builder) builder=true;;
95 --no-add-path) noAddPath=true;;
96 --root-dir) argfun=set_rootDir;;
97 -h|--help) usage; exit;;
98 *)
99 : $((++argi))
100 case $argi in
101 1) url=$arg;;
102 2) rev=$arg;;
103 3) expHash=$arg;;
104 *) exit 1;;
105 esac
106 ;;
107 esac
108 else
109 case $argfun in
110 set_*)
111 var=${argfun#set_}
112 eval "$var=$(printf %q "$arg")"
113 ;;
114 esac
115 argfun=""
116 fi
117done
118
119if test -z "$url"; then
120 usage
121fi
122
123
124init_remote(){
125 local url=$1
126 clean_git init --initial-branch=master
127 clean_git remote add origin "$url"
128 if [ -n "$sparseCheckout" ]; then
129 git config remote.origin.partialclonefilter "blob:none"
130 echo "$sparseCheckout" | git sparse-checkout set --stdin ${nonConeMode:+--no-cone}
131 fi
132 ( [ -n "$http_proxy" ] && clean_git config --global http.proxy "$http_proxy" ) || true
133 local proxy_pairs i
134 read -a proxy_pairs <<< "${FETCHGIT_HTTP_PROXIES:-}"
135 for ((i = 1; i < ${#proxy_pairs[@]}; i += 2)); do
136 clean_git config --global "http.${proxy_pairs[$i - 1]}.proxy" "${proxy_pairs[$i]}"
137 done
138}
139
140# Return the reference of an hash if it exists on the remote repository.
141ref_from_hash(){
142 local hash=$1
143 git ls-remote origin | sed -n "\,$hash\t, { s,\(.*\)\t\(.*\),\2,; p; q}"
144}
145
146# Return the hash of a reference if it exists on the remote repository.
147hash_from_ref(){
148 local ref=$1
149 git ls-remote origin | sed -n "\,\t$ref, { s,\(.*\)\t\(.*\),\1,; p; q}"
150}
151
152# Returns a name based on the url and reference
153#
154# This function needs to be in sync with nix's fetchgit implementation
155# of urlToName() to re-use the same nix store paths.
156url_to_name(){
157 local url=$1
158 local ref=$2
159 local base
160 base=$(basename "$url" .git | cut -d: -f2)
161
162 if test -n "$rootDir"; then
163 # Sanitize by removing leading dots and replacing all invalid character sequences with dashes.
164 # See sanitizeDerivationName in ../../../lib/strings.nix for reference.
165 echo "$base-$(sed -E 's/^\.+//;s/[^[:alnum:]+._?=-]+/-/g' <<< $rootDir)"
166 else
167 if [[ $ref =~ ^[a-z0-9]+$ ]]; then
168 echo "$base-${ref:0:7}"
169 else
170 echo "$base"
171 fi
172 fi
173}
174
175# Fetch and checkout the right sha1
176checkout_hash(){
177 local hash="$1"
178 local ref="$2"
179
180 if test -z "$hash"; then
181 hash=$(hash_from_ref "$ref")
182 fi
183
184 [[ -z "$deepClone" ]] && \
185 clean_git fetch ${builder:+--progress} --depth=1 origin "$hash" || \
186 clean_git fetch -t ${builder:+--progress} origin || return 1
187
188 local object_type=$(git cat-file -t "$hash")
189 if [[ "$object_type" == "commit" || "$object_type" == "tag" ]]; then
190 clean_git checkout -b "$branchName" "$hash" || return 1
191 elif [[ "$object_type" == "tree" ]]; then
192 clean_git config user.email "nix-prefetch-git@localhost"
193 clean_git config user.name "nix-prefetch-git"
194 local commit_id=$(git commit-tree "$hash" -m "Commit created from tree hash $hash")
195 clean_git checkout -b "$branchName" "$commit_id" || return 1
196 else
197 echo "Unrecognized git object type: $object_type"
198 return 1
199 fi
200}
201
202# Fetch only a branch/tag and checkout it.
203checkout_ref(){
204 local hash="$1"
205 local ref="$2"
206
207 if [[ -n "$deepClone" ]]; then
208 # The caller explicitly asked for a deep clone. Deep clones
209 # allow "git describe" and similar tools to work. See
210 # https://marc.info/?l=nix-dev&m=139641582514772
211 # for a discussion.
212 return 1
213 fi
214
215 if test -z "$ref"; then
216 ref=$(ref_from_hash "$hash")
217 fi
218
219 if test -n "$ref"; then
220 # --depth option is ignored on http repository.
221 clean_git fetch ${builder:+--progress} --depth 1 origin +"$ref" || return 1
222 clean_git checkout -b "$branchName" FETCH_HEAD || return 1
223 else
224 return 1
225 fi
226}
227
228# Update submodules
229init_submodules(){
230 # shallow with leaveDotGit will change hashes
231 [[ -z "$deepClone" ]] && [[ -z "$leaveDotGit" ]] && \
232 clean_git submodule update --init --recursive --checkout -j ${NIX_BUILD_CORES:-1} --progress --depth 1 || \
233 clean_git submodule update --init --recursive --checkout -j ${NIX_BUILD_CORES:-1} --progress
234}
235
236clone(){
237 local top=$PWD
238 local dir="$1"
239 local url="$2"
240 local hash="$3"
241 local ref="$4"
242
243 cd "$dir"
244
245 # Initialize the repository.
246 init_remote "$url"
247
248 # Download data from the repository.
249 checkout_ref "$hash" "$ref" ||
250 checkout_hash "$hash" "$ref" || (
251 echo 1>&2 "Unable to checkout $hash$ref from $url."
252 exit 1
253 )
254
255 # Fetch all tags if requested
256 if test -n "$fetchTags"; then
257 echo "fetching all tags..." >&2
258 clean_git fetch origin 'refs/tags/*:refs/tags/*' || echo "warning: failed to fetch some tags" >&2
259 fi
260
261 # Checkout linked sources.
262 if test -n "$fetchSubmodules"; then
263 init_submodules
264 fi
265
266 if [ -z "$builder" ] && [ -f .topdeps ]; then
267 if tg help &>/dev/null; then
268 echo "populating TopGit branches..."
269 tg remote --populate origin
270 else
271 echo "WARNING: would populate TopGit branches but TopGit is not available" >&2
272 echo "WARNING: install TopGit to fix the problem" >&2
273 fi
274 fi
275
276 cd "$top"
277}
278
279# Remove all remote branches, remove tags not reachable from HEAD, do a full
280# repack and then garbage collect unreferenced objects.
281make_deterministic_repo(){
282 local repo="$1"
283
284 # run in sub-shell to not touch current working directory
285 (
286 cd "$repo"
287 # Remove files that contain timestamps or otherwise have non-deterministic
288 # properties.
289 if [ -f .git ]; then
290 local dotgit_content=$(<.git)
291 local dotgit_dir="${dotgit_content#gitdir: }"
292 else
293 local dotgit_dir=".git"
294 fi
295 pushd "$dotgit_dir" >/dev/null
296 rm -rf logs/ hooks/ index FETCH_HEAD ORIG_HEAD refs/remotes/origin/HEAD config
297 popd >/dev/null
298 # Remove all remote branches.
299 git branch -r | while read -r branch; do
300 clean_git branch -rD "$branch"
301 done
302
303 # Remove tags not reachable from HEAD. If we're exactly on a tag, don't
304 # delete it.
305 maybe_tag=$(git tag --points-at HEAD)
306 git tag --contains HEAD | while read -r tag; do
307 if [ "$tag" != "$maybe_tag" ]; then
308 clean_git tag -d "$tag"
309 fi
310 done
311
312 # Do a full repack. Must run single-threaded, or else we lose determinism.
313 clean_git config pack.threads 1
314 clean_git repack -A -d -f
315 rm -f "$dotgit_dir/config"
316
317 # Garbage collect unreferenced objects.
318 # Note: --keep-largest-pack prevents non-deterministic ordering of packs
319 # listed in .git/objects/info/packs by only using a single pack
320 clean_git gc --prune=all --keep-largest-pack
321 )
322}
323
324
325clone_user_rev() {
326 local dir="$1"
327 local url="$2"
328 local rev="${3:-HEAD}"
329
330 if [ -n "$fetchLFS" ]; then
331 clean_git lfs install
332 fi
333
334 # Perform the checkout.
335 case "$rev" in
336 HEAD|refs/*)
337 clone "$dir" "$url" "" "$rev" 1>&2;;
338 *)
339 if test -z "$(echo "$rev" | tr -d 0123456789abcdef)"; then
340 clone "$dir" "$url" "$rev" "" 1>&2
341 else
342 # if revision is not hexadecimal it might be a tag
343 clone "$dir" "$url" "" "refs/tags/$rev" 1>&2
344 fi;;
345 esac
346
347 pushd "$dir" >/dev/null
348 fullRev=$( (git rev-parse "$rev" 2>/dev/null || git rev-parse "refs/heads/$branchName") | tail -n1)
349 humanReadableRev=$(git describe "$fullRev" 2> /dev/null || git describe --tags "$fullRev" 2> /dev/null || echo -- none --)
350 commitDate=$(git show -1 --no-patch --pretty=%ci "$fullRev")
351 commitDateStrict8601=$(git show -1 --no-patch --pretty=%cI "$fullRev")
352 popd >/dev/null
353
354 # Allow doing additional processing before .git removal
355 eval "$NIX_PREFETCH_GIT_CHECKOUT_HOOK"
356 if test -z "$leaveDotGit"; then
357 echo "removing \`.git'..." >&2
358 find "$dir" -name .git -print0 | xargs -0 rm -rf
359 else
360 find "$dir" -name .git | while read -r gitdir; do
361 make_deterministic_repo "$(readlink -f "$(dirname "$gitdir")")"
362 done
363 fi
364}
365
366clone_user_rev_to_tmpfile(){
367 local url="$1"
368 local rev="${2:-HEAD}"
369
370 # nix>=2.20 rejects adding symlinked paths to the store, so use realpath
371 # to resolve to a physical path. https://github.com/NixOS/nix/issues/11941
372 tmpPath="$(realpath "$(mktemp -d --tmpdir git-checkout-tmp-XXXXXXXX)")"
373 exit_handlers+=(remove_tmpPath)
374
375 tmpOut="$tmpPath/out/$storePathName"
376 tmpClone="$tmpPath/clone"
377 mkdir -p "$tmpPath/out" "$tmpClone"
378
379 # Perform the checkout.
380 clone_user_rev "$tmpClone" "$url" "$rev"
381}
382
383exit_handlers=()
384
385run_exit_handlers() {
386 exit_status=$?
387 for handler in "${exit_handlers[@]}"; do
388 eval "$handler $exit_status"
389 done
390}
391
392trap run_exit_handlers EXIT
393
394quiet_exit_handler() {
395 exec 2>&3 3>&-
396 if [ $1 -ne 0 ]; then
397 cat "$errfile" >&2
398 fi
399 rm -f "$errfile"
400}
401
402quiet_mode() {
403 errfile="$(mktemp "${TMPDIR:-/tmp}/git-checkout-err-XXXXXXXX")"
404 exit_handlers+=(quiet_exit_handler)
405 exec 3>&2 2>"$errfile"
406}
407
408json_escape() {
409 local s="$1"
410 s="${s//\\/\\\\}" # \
411 s="${s//\"/\\\"}" # "
412 s="${s//^H/\\\b}" # \b (backspace)
413 s="${s//^L/\\\f}" # \f (form feed)
414 s="${s//
415/\\\n}" # \n (newline)
416 s="${s//^M/\\\r}" # \r (carriage return)
417 s="${s// /\\t}" # \t (tab)
418 echo "$s"
419}
420
421print_results() {
422 hash="$1"
423 if ! test -n "$QUIET"; then
424 echo "" >&2
425 echo "git revision is $fullRev" >&2
426 if test -n "$finalPath"; then
427 echo "path is $finalPath" >&2
428 fi
429 echo "git human-readable version is $humanReadableRev" >&2
430 echo "Commit date is $commitDate" >&2
431 if test -n "$hash"; then
432 echo "hash is $hash" >&2
433 fi
434 fi
435 if test -n "$hash"; then
436 cat <<EOF
437{
438 "url": "$(json_escape "$url")",
439 "rev": "$(json_escape "$fullRev")",
440 "date": "$(json_escape "$commitDateStrict8601")",
441 "path": "$(json_escape "$finalPath")",
442 "$(json_escape "$hashType")": "$(json_escape "$hash")",
443 "hash": "$(nix-hash --to-sri --type $hashType $hash)",
444 "fetchLFS": $([[ -n "$fetchLFS" ]] && echo true || echo false),
445 "fetchSubmodules": $([[ -n "$fetchSubmodules" ]] && echo true || echo false),
446 "deepClone": $([[ -n "$deepClone" ]] && echo true || echo false),
447 "fetchTags": $([[ -n "$fetchTags" ]] && echo true || echo false),
448 "leaveDotGit": $([[ -n "$leaveDotGit" ]] && echo true || echo false),
449 "rootDir": "$(json_escape "$rootDir")"
450}
451EOF
452 fi
453}
454
455remove_tmpPath() {
456 rm -rf "$tmpPath"
457}
458
459remove_tmpHomePath() {
460 chmod -R u+w "$tmpHomePath"
461 rm -rf "$tmpHomePath"
462}
463
464if test -n "$QUIET"; then
465 quiet_mode
466fi
467
468if test -z "$branchName"; then
469 branchName=fetchgit
470fi
471
472if [ -v symbolicName ]; then
473 storePathName="$symbolicName"
474else
475 storePathName="$(url_to_name "$url" "$rev")"
476fi
477
478tmpHomePath="$(mktemp -d "${TMPDIR:-/tmp}/nix-prefetch-git-tmp-home-XXXXXXXXXX")"
479exit_handlers+=(remove_tmpHomePath)
480ln -s "${NETRC:-$HOME/.netrc}" "$tmpHomePath/.netrc"
481HOME="$tmpHomePath"
482unset XDG_CONFIG_HOME
483export GIT_CONFIG_NOSYSTEM=1
484
485if test -n "$builder"; then
486 test -n "$out" -a -n "$url" -a -n "$rev" || usage
487 if test -n "$rootDir"; then
488 clone_user_rev_to_tmpfile "$url" "$rev"
489 mv "$tmpClone/$rootDir" "$out"
490 else
491 mkdir -p "$out"
492 clone_user_rev "$out" "$url" "$rev"
493 fi
494else
495 if test -z "$hashType"; then
496 hashType=sha256
497 fi
498
499 # If the hash was given, a file with that hash may already be in the
500 # store.
501 if test -n "$expHash"; then
502 finalPath=$(nix-store --print-fixed-path --recursive "$hashType" "$expHash" "$storePathName")
503 if ! nix-store --check-validity "$finalPath" 2> /dev/null; then
504 finalPath=
505 fi
506 hash=$expHash
507 fi
508
509 # If we don't know the hash or a path with that hash doesn't exist,
510 # download the file and add it to the store.
511 if test -z "$finalPath"; then
512 clone_user_rev_to_tmpfile "$url" "$rev"
513
514 if test -z "$rootDir"; then
515 mv "$tmpClone" "$tmpOut"
516 else
517 mv "$tmpClone/$rootDir" "$tmpOut"
518 fi
519
520 # Compute the hash.
521 hash=$(nix-hash --type $hashType --base32 "$tmpOut")
522
523 # Add the downloaded file to the Nix store.
524
525 if test -z "$noAddPath"; then
526 finalPath=$(nix-store --add-fixed --recursive "$hashType" "$tmpOut") \
527 || { printf "maybe try again with \`nix-prefetch-git --no-add-path <...>' ?\n" >&2; exit 1; }
528 else
529 printf "the path for \`%s' has NOT been added to the store\n" "$url" >&2
530 finalPath=$(nix-store --print-fixed-path --recursive "$hashType" "$hash" "$storePathName")
531 fi
532
533 if test -n "$expHash" -a "$expHash" != "$hash"; then
534 echo "hash mismatch for URL \`$url'. Got \`$hash'; expected \`$expHash'." >&2
535 exit 1
536 fi
537 fi
538
539 print_results "$hash"
540
541 if test -n "$PRINT_PATH"; then
542 echo "$finalPath"
543 fi
544fi