Merge branch 'lukeshu/tidy'

4 files changed, 38 insertions, 28 deletions

diff --git a/bin/cdxget b/bin/cdxget
index 46d56c4..6844fa6 100755
--- a/bin/cdxget
+++ b/bin/cdxget
@@ -1,7 +1,9 @@
 #!/usr/bin/env bash
 
-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/gitify b/bin/gitify
index 1e5d43d..b245f3e 100755
--- a/bin/gitify
+++ b/bin/gitify
@@ -82,7 +82,7 @@ main() {
 			HACK_NAME='Luke Shumaker'
 			HACK_EMAIL='lukeshu@lukeshu.com'
 
-			gitdate="$(sed -r 's/(....)(..)(..)(..)(..)(..)/\1-\2-\3T\4:\5:\6 '"$HACK_TZ"'/' <<<"$time")"
+			gitdate="$(sed -E 's/(....)(..)(..)(..)(..)(..)/\1-\2-\3T\4:\5:\6 '"$HACK_TZ"'/' <<<"$time")"
 
 			git add .
 
@@ -99,11 +99,11 @@ main() {
 				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 e256157..649b71b 100755
--- a/bin/poolify
+++ b/bin/poolify
@@ -13,14 +13,27 @@ 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 +41,9 @@ 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 +52,8 @@ 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.
@@ -65,12 +79,14 @@ 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 [[ "$url" == */ ]]; then
+			# Skip directories
+			continue
+		fi
 		if [[ "$url" == */2.0-Update/* ]]; then
 			# Gross hack
 			continue
@@ -79,10 +95,6 @@ main() {
 		dirpart="${name%/*}"
 		filepart="${name##*/}"
 
-		if [[ -z "$filepart" ]]; then
-			continue
-		fi
-
 		pools=(dat/pools/files/*-"${name//\//_}")
 		if [[ "$name" = *.OLD* ]]; then
 			pname="${name//\//_}"
@@ -119,9 +131,9 @@ 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/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 -->//