jh-checksource: optimize further

1 files changed, 20 insertions, 52 deletions

diff --git a/jh-checksource.sh b/jh-checksource.sh
index 9de91c0..5296226 100644
--- a/jh-checksource.sh
+++ b/jh-checksource.sh
@@ -4,16 +4,19 @@
 # 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.
 
-sep='<no-filename-ever-contains-this>'
+# Regular expressions are POSIX EREs, and must match the entirety of the string
 
-safe_types_regexp=('^(inode|text|image|video|audio)/')
-safe_types_string=('application/pdf' 'application/postscript' 'application/xml' 'application/ogg' 'message/rfc822')
 safe_dirs_glob=(.{git,hg,svn} '*.git')
-safe_files_regexp=('/po/[^/]*.gmo$' '\.flw$' '\.odg$' '\.ppt$')
+safe_types_regexp=('(inode|text|image|video|audio)/.*' 'application/(pdf|postscript|xml|ogg)' message/rfc822)
+safe_files_regexp=('.*/po/[^/]*.gmo' '.*\.(flw|odg|ppt)')
 safe_files_string=()
 
-# don't care about files less than 3 bytes.
-min_size=3
+# don't care about files less than 3 bytes ('c' is for characters)
+min_size=3c
+
+sep='<no-filename-ever-contains-this>'
+
+################################################################################
 
 cwd="$(readlink -m -- "$PWD")"
 
@@ -23,28 +26,6 @@ normalize_filenames() {
 	done | sed "s|^$cwd/|./|"
 }
 
-matches_string() {
-	local needle=$1
-	shift
-	for straw in "$@"; do
-		if [[ "$needle" = "$straw" ]]; then
-			return 0
-		fi
-	done
-	return 1
-}
-
-matches_regexp() {
-	local needle=$1
-	shift
-	for straw in "$@"; do
-		if [[ "$needle" =~ $straw ]]; then
-			return 0
-		fi
-	done
-	return 1
-}
-
 print-human() {
 	libremessages warning "The source directory %s contains binary files:" "$PWD"
 	sed 's/^/  -> /'
@@ -69,36 +50,23 @@ main() {
 	local unsafe_files="$(mktemp --tmpdir "${0##*/}.XXXXXXXXXX")"
 	trap "$(printf 'rm -f -- %q' "$unsafe_files")" EXIT
 
-	# Heavy lifting
-	local filter_dirs=()
+	# Turn the variables up top into a bunch of `find(1)` filters
+	local filters=()
 	local glob
 	for glob in "${safe_dirs_glob[@]}"; do
-		filter_dirs+=(-type d -name "$glob" -prune -o)
+		filters+=(-type d -name "$glob" -prune -o)
 	done
-	find . "${filter_dirs[@]}" -type f -printf '%s %p\n' | # find all files
-	while read -r size file; do # filter out files smaller than $min_size
-		[[ $size -lt $min_size ]] || printf '%s\n' "$file"
-	done |
+	filters+=(-type f -size +"${min_size}")
+
+	# Heavy lifting
+	find . "${filters[@]}" -print | # find all files
 	normalize_filenames |
+	grep -Fxvf <(printf '%s\n' "${safe_files_string[@]}") |
+	grep -Exvf <(printf '%s\n' "${safe_files_regexp[@]}") |
 	xargs -r -d'\n' file --mime-type -r -F "$sep" | # identify the filetypes
 	sed -r "s@(.*)${sep}\s*(.*)@\2:\1@" |  # reformat the output to be easier to parse
-	while IFS=: read -r type file; do
-		declare -A cached_types
-		if ! { matches_string "$file" "${safe_files_string[@]}" || \
-		       matches_regexp "$file" "${safe_files_regexp[@]}" ;}; then
-			if [[ -z ${cached_types[$type]} ]]; then
-				if matches_string "$type" "${safe_types_string[@]}" || \
-				   matches_regexp "$type" "${safe_types_regexp[@]}" ; then
-					cached_types[$type]=false
-				else
-					cached_types[$type]=true
-				fi
-			fi
-			if "${cached_types[$type]}"; then
-				printf "%s\n" "$file"
-			fi
-		fi
-	done > "$unsafe_files"
+	grep -Exvf <(printf '%s:.*\n' "${safe_types_regexp[@]}") |
+	cut -d: -f2- > "$unsafe_files"
 
 	if [[ "$(stat -c '%s' -- "$unsafe_files")" -gt 0 ]]; then
 		<"$unsafe_files" sort | print-$format