diff-pr.sh 5.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229
  1. #!/bin/bash
  2. set -eo pipefail
  3. shopt -s dotglob
  4. # make sure we can GTFO
  5. trap 'echo >&2 Ctrl+C captured, exiting; exit 1' SIGINT
  6. usage() {
  7. cat <<-EOUSAGE
  8. usage: $0 [PR number] [repo[:tag]]
  9. ie: $0 1024
  10. $0 9001 debian php django
  11. EOUSAGE
  12. }
  13. # TODO flags parsing
  14. allFiles=
  15. listTarballContents=1
  16. findCopies='20%'
  17. uninterestingTarballContent=(
  18. # "config_diff_2017_01_07.log"
  19. 'var/log/YaST2/'
  20. # "ks-script-mqmz_080.log"
  21. # "ks-script-ycfq606i.log"
  22. 'var/log/anaconda/'
  23. # "2016-12-20/"
  24. 'var/lib/yum/history/'
  25. 'var/lib/dnf/history/'
  26. # "a/f8c032d2be757e1a70f00336b55c434219fee230-acl-2.2.51-12.el7-x86_64/var_uuid"
  27. 'var/lib/yum/yumdb/'
  28. 'var/lib/dnf/yumdb/'
  29. # "b42ff584.0"
  30. 'etc/pki/tls/rootcerts/'
  31. )
  32. # prints "$2$1$3$1...$N"
  33. join() {
  34. local sep="$1"; shift
  35. local out; printf -v out "${sep//%/%%}%s" "$@"
  36. echo "${out#$sep}"
  37. }
  38. uninterestingTarballGrep="^([.]?/)?($(join '|' "${uninterestingTarballContent[@]}"))"
  39. if [ "$#" -eq 0 ]; then
  40. usage >&2
  41. exit 1
  42. fi
  43. pull="$1" # PR number
  44. shift
  45. #dir="$(dirname "$(readlink -f "$BASH_SOURCE")")"
  46. tempDir="$(mktemp -d)"
  47. trap "rm -rf '$tempDir'" EXIT
  48. cd "$tempDir"
  49. git clone --quiet \
  50. https://github.com/docker-library/official-images.git \
  51. oi
  52. git -C oi fetch --quiet \
  53. origin "pull/$pull/merge":pull
  54. images=( "$@" )
  55. if [ "${#images[@]}" -eq 0 ]; then
  56. images=( $(git -C oi/library diff --name-only master...pull -- . | xargs -n1 basename) )
  57. fi
  58. export BASHBREW_CACHE="${BASHBREW_CACHE:-${XDG_CACHE_HOME:-$HOME/.cache}/bashbrew}"
  59. export BASHBREW_LIBRARY="$PWD/oi/library"
  60. export BASHBREW_ARCH='amd64' # TODO something smarter with arches
  61. # "bashbrew cat" template for duplicating something like "bashbrew list --uniq" but with architectures too
  62. archesListTemplate='
  63. {{- range $e := $.Entries -}}
  64. {{- range .Architectures -}}
  65. {{- $.RepoName -}}:{{- $e.Tags | last -}}
  66. {{- " @ " -}}
  67. {{- . -}}
  68. {{- "\n" -}}
  69. {{- end -}}
  70. {{- end -}}
  71. '
  72. # ... and SharedTags
  73. sharedTagsListTemplate='
  74. {{- range $group := .Manifest.GetSharedTagGroups -}}
  75. {{- range $tag := $group.SharedTags -}}
  76. {{- join ":" $.RepoName $tag -}}
  77. {{- " -- " -}}
  78. {{- range $i, $e := $group.Entries -}}
  79. {{- if gt $i 0 -}}
  80. {{- ", " -}}
  81. {{- end -}}
  82. {{- join ":" $.RepoName ($e.Tags | last) -}}
  83. {{- end -}}
  84. {{- "\n" -}}
  85. {{- end -}}
  86. {{- end -}}
  87. '
  88. # TODO something less hacky than "git archive" hackery, like a "bashbrew archive" or "bashbrew context" or something
  89. template='
  90. {{- range $.Entries -}}
  91. {{- $arch := .Architectures | first -}}
  92. {{- $from := $.ArchDockerFrom $arch . -}}
  93. git -C "$BASHBREW_CACHE/git" archive --format=tar
  94. {{- " " -}}
  95. {{- "--prefix=" -}}
  96. {{- $.RepoName -}}
  97. _
  98. {{- .Tags | last -}}
  99. {{- "/" -}}
  100. {{- " " -}}
  101. {{- .ArchGitCommit $arch -}}
  102. {{- ":" -}}
  103. {{- $dir := .ArchDirectory $arch -}}
  104. {{- (eq $dir ".") | ternary "" $dir -}}
  105. {{- "\n" -}}
  106. {{- end -}}
  107. '
  108. copy-tar() {
  109. local src="$1"; shift
  110. local dst="$1"; shift
  111. if [ "$allFiles" ]; then
  112. mkdir -p "$dst"
  113. cp -al "$src"/*/ "$dst/"
  114. return
  115. fi
  116. # "Dockerfile*" at the end here ensures we capture "Dockerfile.builder" style repos in a useful way too (busybox, hello-world)
  117. for d in "$src"/*/Dockerfile*; do
  118. dDir="$(dirname "$d")"
  119. dDirName="$(basename "$dDir")"
  120. IFS=$'\n'
  121. files=(
  122. "$(basename "$d")"
  123. $(awk '
  124. toupper($1) == "COPY" || toupper($1) == "ADD" {
  125. for (i = 2; i < NF; i++) {
  126. print $i
  127. }
  128. }
  129. ' "$d")
  130. # some extra files which are likely interesting if they exist, but no big loss if they do not
  131. ' *.manifest' # debian/ubuntu "package versions" list
  132. ' *.ks' # fedora "kickstart" (rootfs build script)
  133. ' build*.txt' # ubuntu "build-info.txt", debian "build-command.txt"
  134. # usefulness yet to be proven:
  135. #' *.log'
  136. #' {MD5,SHA1,SHA256}SUMS'
  137. #' *.{md5,sha1,sha256}'
  138. # (the space prefix is removed below and is used to ignore non-matching globs so that bad "Dockerfile" entries appropriately lead to failure)
  139. )
  140. unset IFS
  141. mkdir -p "$dst/$dDirName"
  142. for origF in "${files[@]}"; do
  143. f="${origF# }" # trim off leading space (indicates we don't care about failure)
  144. [ "$f" = "$origF" ] && failureMatters=1 || failureMatters=
  145. globbed=( $(cd "$dDir" && eval "echo $f") )
  146. for g in "${globbed[@]}"; do
  147. if [ -z "$failureMatters" ] && [ ! -e "$dDir/$g" ]; then
  148. continue
  149. fi
  150. mkdir -p "$(dirname "$dst/$dDirName/$g")"
  151. cp -alT "$dDir/$g" "$dst/$dDirName/$g"
  152. if [ "$listTarballContents" ]; then
  153. case "$g" in
  154. *.tar.*|*.tgz)
  155. tar -tf "$dst/$dDirName/$g" \
  156. | grep -vE "$uninterestingTarballGrep" \
  157. | sort \
  158. > "$dst/$dDirName/$g 'tar -t'"
  159. ;;
  160. esac
  161. fi
  162. done
  163. done
  164. done
  165. }
  166. mkdir temp
  167. git -C temp init --quiet
  168. bashbrew list "${images[@]}" | sort -V > temp/_bashbrew-list || :
  169. bashbrew cat --format "$archesListTemplate" "${images[@]}" | sort -V > temp/_bashbrew-arches || :
  170. bashbrew cat --format "$sharedTagsListTemplate" "${images[@]}" | grep -vE '^$' | sort -V > temp/_bashbrew-shared-tags || :
  171. for image in "${images[@]}"; do
  172. if script="$(bashbrew cat -f "$template" "$image")"; then
  173. mkdir tar
  174. ( eval "$script" | tar -xiC tar )
  175. copy-tar tar temp
  176. rm -rf tar
  177. fi
  178. done
  179. git -C temp add . || :
  180. git -C temp commit --quiet --allow-empty -m 'initial' || :
  181. git -C oi checkout --quiet pull
  182. git -C temp rm --quiet -rf . || :
  183. bashbrew list "${images[@]}" | sort -V > temp/_bashbrew-list || :
  184. bashbrew cat --format "$archesListTemplate" "${images[@]}" | sort -V > temp/_bashbrew-arches || :
  185. bashbrew cat --format "$sharedTagsListTemplate" "${images[@]}" | grep -vE '^$' | sort -V > temp/_bashbrew-shared-tags || :
  186. script="$(bashbrew cat -f "$template" "${images[@]}")"
  187. mkdir tar
  188. ( eval "$script" | tar -xiC tar )
  189. copy-tar tar temp
  190. rm -rf tar
  191. git -C temp add .
  192. git -C temp diff --minimal --find-copies="$findCopies" --find-copies-harder --irreversible-delete --staged