build git repo

2 files changed, 74 insertions, 1 deletions

diff --git a/Makefile b/Makefile
index 5cb9fdc..12c5bb6 100644
--- a/Makefile
+++ b/Makefile
@@ -49,6 +49,9 @@ dat/content-file/%:
 content-file = $(foreach u,$(filter-out %/,$(index)),dat/content-file/$(call url2murl,$(u)))
 download: $(content-file)
 
-.PHONY: all fix download
+git: download
+	gitthing dat/git < dat/index.txt
+
+.PHONY: all fix download git
 .DELETE_ON_ERROR:
 .SECONDARY:
diff --git a/bin/gitthing b/bin/gitthing
new file mode 100755
index 0000000..a54c017
--- /dev/null
+++ b/bin/gitthing
@@ -0,0 +1,70 @@
+#!/usr/bin/env bash
+
+empty() {
+	[[ $(stat -c %s "$1") -eq 0 ]]
+}
+
+url2murl() {
+	local x
+	x=$1
+	x=${x//'^'/'^5E'}
+	x=${x//':'/'^3A'}
+	x=${x//'%'/'^25'}
+	printf '%s' "$x"
+}
+
+murl2url() {
+	local x
+	x=$1
+	x=${x//'^25'/'%'}
+	x=${x//'^3A'/':'}
+	x=${x//'^5E'/'^'}
+	printf '%s' "$x"
+}
+
+main() {
+	set -euE -o pipefail
+	top=$PWD
+
+	mkdir -p "$1"
+	cd "$1"
+	git init
+	echo 'ref: refs/heads/PROGRAMS/CVTUTF' > .git/HEAD
+	git commit --allow-empty -m 'initial commit'
+
+	while read -r time url; do
+		suffix="${url##*/Public/}"
+		dirpart="${suffix%/*}"
+		filepart="${suffix##*/}"
+
+		branch=$dirpart
+
+		git checkout PROGRAMS/CVTUTF
+		git checkout -b "$branch" || true
+		git checkout "$branch"
+
+		waurl="http://web.archive.org/web/$time/$url"
+		if [[ -n "$filepart" ]]; then
+			file="$top/dat/content-file/$(url2murl "${waurl#http://}")"
+			cp "$file" .
+		else
+			dir="$top/dat/content-dir/$(url2murl "${waurl#http://}")"
+			comm -23 \
+			    <(git ls-files) \
+			    <(< "$dir/metadata.txt" awk '{print $1}') \
+			    | xargs -r0 rm -f --
+			if ! empty "$dir/readme.txt"; then
+			   cp "$dir/readme.txt" .
+			fi
+			cp "$dir/metadata.txt" .metadata.txt
+		fi
+
+		if [[ -n "$(git status -s .)" ]]; then
+			gitdate="$(sed -r 's/(....)(..)(..)(..)(..)(..)/\1-\2-\3T\4:\5:\6/' <<<"$time")"
+			git add .
+			GIT_AUTHOR_DATE=$gitdate GIT_AUTHOR_DATE=$gitdate git commit -m "$time $url"
+		fi
+	done
+}
+
+main "$@"