diff options
Diffstat (limited to 'bin')
| -rwxr-xr-x | bin/cdxget | 14 | ||||
| -rwxr-xr-x | bin/fmt-metadata | 7 | ||||
| -rwxr-xr-x | bin/gitify | 28 | ||||
| -rwxr-xr-x | bin/poolify | 54 | ||||
| -rwxr-xr-x | bin/urlkey2url | 6 | ||||
| -rwxr-xr-x | bin/wayfore | 4 |
6 files changed, 85 insertions, 28 deletions
@@ -1,7 +1,15 @@ #!/usr/bin/env bash +# Copyright (c) 2017, 2023 Luke Shumaker <lukeshu@lukeshu.com> +# +# This work is free. You can redistribute it and/or modify it under +# the terms of the Do What The Fuck You Want To Public License, +# Version 2, as published by Sam Hocevar. See the COPYING file for +# more details. -url='http://web.archive.org/cdx/search/cdx?' +url='http://web.archive.org/cdx/search/cdx' +s='?' for arg in "$@"; do - url+="$s${arg%%=*}=$(printf '%s' "${arg#*=}"|urlencode)&" + url+="$s${arg%%=*}=$(printf '%s' "${arg#*=}"|urlencode)" + s='&' done -curl -sL "$url" +curl -sfL "$url" diff --git a/bin/fmt-metadata b/bin/fmt-metadata index c92419b..ec82451 100755 --- a/bin/fmt-metadata +++ b/bin/fmt-metadata @@ -1,4 +1,11 @@ #!/usr/bin/env ruby +# Copyright (c) 2017 Luke Shumaker <lukeshu@lukeshu.com> +# +# This work is free. You can redistribute it and/or modify it under +# the terms of the Do What The Fuck You Want To Public License, +# Version 2, as published by Sam Hocevar. See the COPYING file for +# more details. + require 'time' snapshot = ARGV.first.to_i @@ -1,4 +1,10 @@ #!/usr/bin/env bash +# Copyright (c) 2017, 2023 Luke Shumaker <lukeshu@lukeshu.com> +# +# This work is free. You can redistribute it and/or modify it under +# the terms of the Do What The Fuck You Want To Public License, +# Version 2, as published by Sam Hocevar. See the COPYING file for +# more details. empty() { [[ $(stat -c %s "$1") -eq 0 ]] @@ -74,20 +80,36 @@ main() { waurl="http://web.archive.org/web/$(murl2url "${listingdir##*/content-dir/}/")" msg="$waurl" fi - gitdate="$(sed -r 's/(....)(..)(..)(..)(..)(..)/\1-\2-\3T\4:\5:\6/' <<<"$time")" + + HACK_TZ=-0500 + if [[ "$branch" == BETA/CVTUTF-1-4 || "$branch" == ALPHA/CVTUTF-1-5-draft ]]; then + HACK_TZ=-0400 + fi + HACK_NAME='Luke Shumaker' + HACK_EMAIL='lukeshu@lukeshu.com' + + gitdate="$(sed -E 's/(....)(..)(..)(..)(..)(..)/\1-\2-\3T\4:\5:\6 '"$HACK_TZ"'/' <<<"$time")" + git add . + export GIT_AUTHOR_DATE=$gitdate + export GIT_AUTHOR_NAME=$HACK_NAME + export GIT_AUTHOR_EMAIL=$HACK_EMAIL + export GIT_COMMITTER_DATE=$gitdate + export GIT_COMMITTER_NAME=$HACK_NAME + export GIT_COMMITTER_EMAIL=$HACK_EMAIL + git commit --allow-empty -m "$msg" if [[ "$branch" != *.OLD ]]; then lastbranch="$branch" fi if [[ "$branch" == PROGRAMS/CVTUTF ]] && git log -n1 --stat|grep -qF 'ExpectedOutput.txt'; then - git filter-branch -f --parent-filter 'cat; echo " -p BETA/CVTUTF-1-3"' HEAD^..HEAD + FILTER_BRANCH_SQUELCH_WARNING=1 git filter-branch -f --parent-filter 'cat; echo " -p BETA/CVTUTF-1-3"' HEAD^..HEAD git update-ref -d refs/original/refs/heads/"$branch" fi if [[ "$branch" == PROGRAMS/CVTUTF.OLD ]] && git log -n1 --stat|grep -qi '.*\.c\s'; then - git filter-branch -f --parent-filter 'cat; echo " -p PROGRAMS/CVTUTF^"' HEAD^..HEAD + FILTER_BRANCH_SQUELCH_WARNING=1 git filter-branch -f --parent-filter 'cat; echo " -p PROGRAMS/CVTUTF^"' HEAD^..HEAD git update-ref -d refs/original/refs/heads/"$branch" fi fi diff --git a/bin/poolify b/bin/poolify index 6f9a109..b49cbd9 100755 --- a/bin/poolify +++ b/bin/poolify @@ -1,4 +1,10 @@ #!/usr/bin/env bash +# Copyright (c) 2017, 2023 Luke Shumaker <lukeshu@lukeshu.com> +# +# This work is free. You can redistribute it and/or modify it under +# the terms of the Do What The Fuck You Want To Public License, +# Version 2, as published by Sam Hocevar. See the COPYING file for +# more details. url2murl() { local x @@ -13,14 +19,27 @@ main() { set -euE -o pipefail shopt -s nullglob - echo '# Pass 1' - declare -A rewrite - rewrite[200109261739]=200303310700 - while read -r snap name date time size; do + arg_metadata_txt=$1 + arg_index_txt=$2 + + # Overrides ############################################################ + + declare -A override_datetime + override_datetime[200109261739]=200303310700 + + override_synthetic_listings=( + #YYYYMMDDHHMM branch_name newfiles + '200307291500 ALPHA/CVTUTF-1-1 ExpectedOutput.txt readme.txt' + ) + + # Main ################################################################# + + echo '# Pass 1 (initialize snapshots from $arg_metadata_txt)' + while read -r snap name date time; do dirpart="${name%/*}" filepart="${name##*/}" datetime="${date//-/}${time//:/}" - datetime="${rewrite[$datetime]:-$datetime}" + datetime="${override_datetime[$datetime]:-$datetime}" filedir=dat/pools/files/"${datetime}-${name//\//_}" snapdir=dat/pools/snaps/"${snap}-${dirpart//\//_}" if [[ -d "${filedir/.OLD/}" ]]; then @@ -28,9 +47,9 @@ main() { fi mkdir -p -- "$filedir" "$snapdir" ln -sr "$filedir/$filepart" "$snapdir" - done < "$1" + done < "$arg_metadata_txt" - echo '# Pass 1.5' + echo '# Pass 1.5 (initialize synthetic snapshots)' # Looking at the data, there are 3 revisions that we DON'T # have directory listings for. So we need to synthesize # those. @@ -39,7 +58,8 @@ main() { # synthesizing anything, then looking for files ending in # ".1". They are created during pass 2 if we have a file with # no matching listing. - while read -r datetime dirpart newfiles; do + for line in "${override_synthetic_listings[@]}"; do + read -r datetime dirpart newfiles <<<"$line" # We need to figure out which files to put in the # directory listing. We're going to do that by # mimicking the previous listing with that dirpart. @@ -65,20 +85,18 @@ main() { rm -- "$snapdir/$filepart" ln -sr "$filedir/$filepart" "$snapdir" done - done < <(printf '%s\n' \ - '200307291500 ALPHA/CVTUTF-1-1 ExpectedOutput.txt readme.txt' \ - ) + done - echo '# Pass 2' + echo '# Pass 2 (resolve files)' while read -r time url; do + if [[ "$url" == */ ]]; then + # Skip directories + continue + fi name="${url##*/Public/}" dirpart="${name%/*}" filepart="${name##*/}" - if [[ -z "$filepart" ]]; then - continue - fi - pools=(dat/pools/files/*-"${name//\//_}") if [[ "$name" = *.OLD* ]]; then pname="${name//\//_}" @@ -115,9 +133,9 @@ main() { fi i+=1 done - done < "$2" + done < "$arg_index_txt" - echo '# Pass 3' + echo '# Pass 3 (resolve missing files)' while read -r missing; do if [[ -f "${missing/.OLD}/${missing##*_}" ]]; then ln -sr "${missing/.OLD}/${missing##*_}" "$missing" diff --git a/bin/urlkey2url b/bin/urlkey2url index 5d0ec3d..813e66f 100755 --- a/bin/urlkey2url +++ b/bin/urlkey2url @@ -1,4 +1,10 @@ #!/usr/bin/env bash +# Copyright (c) 2017 Luke Shumaker <lukeshu@lukeshu.com> +# +# This work is free. You can redistribute it and/or modify it under +# the terms of the Do What The Fuck You Want To Public License, +# Version 2, as published by Sam Hocevar. See the COPYING file for +# more details. for arg in "$@"; do keydomain="${arg%%)*}" diff --git a/bin/wayfore b/bin/wayfore deleted file mode 100755 index b0bde8a..0000000 --- a/bin/wayfore +++ /dev/null @@ -1,4 +0,0 @@ -#!/usr/bin/sed -zrf -# The opposite of 'wayback' -s/(<[hH][eE][aA][dD]>).*<!-- End Wayback Rewrite JS Include -->/\1/ -s/<!-- BEGIN WAYBACK TOOLBAR INSERT -->.*<!-- END WAYBACK TOOLBAR INSERT -->// |