diff-pr.sh 5.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235
  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. # "09/401f736622f2c9258d14388ebd47900bbab126"
  32. 'usr/lib/.build-id/'
  33. )
  34. # prints "$2$1$3$1...$N"
  35. join() {
  36. local sep="$1"; shift
  37. local out; printf -v out "${sep//%/%%}%s" "$@"
  38. echo "${out#$sep}"
  39. }
  40. uninterestingTarballGrep="^([.]?/)?($(join '|' "${uninterestingTarballContent[@]}"))"
  41. if [ "$#" -eq 0 ]; then
  42. usage >&2
  43. exit 1
  44. fi
  45. pull="$1" # PR number
  46. shift
  47. #dir="$(dirname "$(readlink -f "$BASH_SOURCE")")"
  48. tempDir="$(mktemp -d)"
  49. trap "rm -rf '$tempDir'" EXIT
  50. cd "$tempDir"
  51. git clone --quiet \
  52. https://github.com/docker-library/official-images.git \
  53. oi
  54. git -C oi fetch --quiet \
  55. origin "pull/$pull/merge":pull
  56. images=( "$@" )
  57. if [ "${#images[@]}" -eq 0 ]; then
  58. images=( $(git -C oi/library diff --name-only master...pull -- . | xargs -n1 basename) )
  59. fi
  60. export BASHBREW_CACHE="${BASHBREW_CACHE:-${XDG_CACHE_HOME:-$HOME/.cache}/bashbrew}"
  61. export BASHBREW_LIBRARY="$PWD/oi/library"
  62. : "${BASHBREW_ARCH:=amd64}" # TODO something smarter with arches
  63. export BASHBREW_ARCH
  64. # "bashbrew cat" template for duplicating something like "bashbrew list --uniq" but with architectures too
  65. archesListTemplate='
  66. {{- range $e := $.Entries -}}
  67. {{- range .Architectures -}}
  68. {{- $.RepoName -}}:{{- $e.Tags | last -}}
  69. {{- " @ " -}}
  70. {{- . -}}
  71. {{- "\n" -}}
  72. {{- end -}}
  73. {{- end -}}
  74. '
  75. # ... and SharedTags
  76. sharedTagsListTemplate='
  77. {{- range $group := .Manifest.GetSharedTagGroups -}}
  78. {{- range $tag := $group.SharedTags -}}
  79. {{- join ":" $.RepoName $tag -}}
  80. {{- " -- " -}}
  81. {{- range $i, $e := $group.Entries -}}
  82. {{- if gt $i 0 -}}
  83. {{- ", " -}}
  84. {{- end -}}
  85. {{- join ":" $.RepoName ($e.Tags | last) -}}
  86. {{- end -}}
  87. {{- "\n" -}}
  88. {{- end -}}
  89. {{- end -}}
  90. '
  91. # TODO something less hacky than "git archive" hackery, like a "bashbrew archive" or "bashbrew context" or something
  92. template='
  93. {{- range $.Entries -}}
  94. {{- $arch := .HasArchitecture arch | ternary arch (.Architectures | first) -}}
  95. {{- $from := $.ArchDockerFrom $arch . -}}
  96. git -C "$BASHBREW_CACHE/git" archive --format=tar
  97. {{- " " -}}
  98. {{- "--prefix=" -}}
  99. {{- $.RepoName -}}
  100. _
  101. {{- .Tags | last -}}
  102. {{- "/" -}}
  103. {{- " " -}}
  104. {{- .ArchGitCommit $arch -}}
  105. {{- ":" -}}
  106. {{- $dir := .ArchDirectory $arch -}}
  107. {{- (eq $dir ".") | ternary "" $dir -}}
  108. {{- "\n" -}}
  109. {{- end -}}
  110. '
  111. copy-tar() {
  112. local src="$1"; shift
  113. local dst="$1"; shift
  114. if [ "$allFiles" ]; then
  115. mkdir -p "$dst"
  116. cp -al "$src"/*/ "$dst/"
  117. return
  118. fi
  119. # "Dockerfile*" at the end here ensures we capture "Dockerfile.builder" style repos in a useful way too (busybox, hello-world)
  120. for d in "$src"/*/Dockerfile*; do
  121. dDir="$(dirname "$d")"
  122. dDirName="$(basename "$dDir")"
  123. IFS=$'\n'
  124. files=(
  125. "$(basename "$d")"
  126. $(awk '
  127. toupper($1) == "COPY" || toupper($1) == "ADD" {
  128. for (i = 2; i < NF; i++) {
  129. print $i
  130. }
  131. }
  132. ' "$d")
  133. # some extra files which are likely interesting if they exist, but no big loss if they do not
  134. ' *.manifest' # debian/ubuntu "package versions" list
  135. ' *.ks' # fedora "kickstart" (rootfs build script)
  136. ' build*.txt' # ubuntu "build-info.txt", debian "build-command.txt"
  137. # usefulness yet to be proven:
  138. #' *.log'
  139. #' {MD5,SHA1,SHA256}SUMS'
  140. #' *.{md5,sha1,sha256}'
  141. # (the space prefix is removed below and is used to ignore non-matching globs so that bad "Dockerfile" entries appropriately lead to failure)
  142. )
  143. unset IFS
  144. mkdir -p "$dst/$dDirName"
  145. for origF in "${files[@]}"; do
  146. f="${origF# }" # trim off leading space (indicates we don't care about failure)
  147. [ "$f" = "$origF" ] && failureMatters=1 || failureMatters=
  148. globbed=( $(cd "$dDir" && eval "echo $f") )
  149. for g in "${globbed[@]}"; do
  150. if [ -z "$failureMatters" ] && [ ! -e "$dDir/$g" ]; then
  151. continue
  152. fi
  153. mkdir -p "$(dirname "$dst/$dDirName/$g")"
  154. cp -alT "$dDir/$g" "$dst/$dDirName/$g"
  155. if [ "$listTarballContents" ]; then
  156. case "$g" in
  157. *.tar.*|*.tgz)
  158. tar -tf "$dst/$dDirName/$g" \
  159. | grep -vE "$uninterestingTarballGrep" \
  160. | sed -e 's!^./!!' \
  161. | sort \
  162. > "$dst/$dDirName/$g 'tar -t'"
  163. ;;
  164. esac
  165. fi
  166. done
  167. done
  168. done
  169. }
  170. mkdir temp
  171. git -C temp init --quiet
  172. bashbrew list "${images[@]}" | sort -uV > temp/_bashbrew-list || :
  173. bashbrew cat --format "$archesListTemplate" "${images[@]}" | sort -V > temp/_bashbrew-arches || :
  174. bashbrew cat --format "$sharedTagsListTemplate" "${images[@]}" | grep -vE '^$' | sort -V > temp/_bashbrew-shared-tags || :
  175. for image in "${images[@]}"; do
  176. if script="$(bashbrew cat -f "$template" "$image")"; then
  177. mkdir tar
  178. ( eval "$script" | tar -xiC tar )
  179. copy-tar tar temp
  180. rm -rf tar
  181. fi
  182. done
  183. git -C temp add . || :
  184. git -C temp commit --quiet --allow-empty -m 'initial' || :
  185. git -C oi checkout --quiet pull
  186. git -C temp rm --quiet -rf . || :
  187. bashbrew list "${images[@]}" | sort -uV > temp/_bashbrew-list || :
  188. bashbrew cat --format "$archesListTemplate" "${images[@]}" | sort -V > temp/_bashbrew-arches || :
  189. bashbrew cat --format "$sharedTagsListTemplate" "${images[@]}" | grep -vE '^$' | sort -V > temp/_bashbrew-shared-tags || :
  190. script="$(bashbrew cat -f "$template" "${images[@]}")"
  191. mkdir tar
  192. ( eval "$script" | tar -xiC tar )
  193. copy-tar tar temp
  194. rm -rf tar
  195. git -C temp add .
  196. git -C temp diff --minimal --ignore-all-space --find-copies="$findCopies" --find-copies-harder --irreversible-delete --staged