From 952f2444763af9100339876b1bfc88c169346967 Mon Sep 17 00:00:00 2001
From: Luke Shumaker <LukeShu@sbcglobal.net>
Date: Thu, 25 Apr 2013 21:06:10 -0400
Subject: improve jh-checksource

Code-wise:
 * add `-m` flag for machine readable
 * use mime-types instead of user-readable strings
 * ignore files less than 3 bytes

Settings-wise:
 * ignore _all_ `inode/*`, `text/*`, `image/*` and `video/*` types
 * ignore PDFs and PostScript
 * ignore .hg files
---
 jh-checksource.sh | 43 +++++++++++++++++++++++++++++++------------
 1 file changed, 31 insertions(+), 12 deletions(-)

diff --git a/jh-checksource.sh b/jh-checksource.sh
index 1fecfa5..9e3a9d6 100644
--- a/jh-checksource.sh
+++ b/jh-checksource.sh
@@ -1,13 +1,15 @@
 #!/bin/bash
 
-sep='<nofileevercontainsthis>'
-resep='@'
+sep='<no-filename-ever-contains-this>'
 
-safe_types_regexp=('text' '(GIF|JPEG|PNG) image data' 'MS Windows icon')
-safe_types_string=('empty')
-safe_files_regexp=('/\.(git|svn)/')
+safe_types_regexp=('^(inode|text|image|video)/')
+safe_types_string=('application/pdf' 'application/postscript')
+safe_files_regexp=('/\.(git|hg|svn)/')
 safe_files_string=()
 
+# don't care about files less than 3 bytes.
+min_size=3
+
 normalize_filename() {
 	local cwd="`pwd`"
 	readlink -m -- "$1"|sed "s|^$cwd/|./|"
@@ -35,18 +37,36 @@ matches_regexp() {
 	return 1
 }
 
+print-human() {
+	libremessages warning "The source directory `pwd` contains binary files:"
+	sed 's/./  -> &/'
+}
+
+print-machine() {
+	cat
+}
+
 main() {
+	format=human
 	# Parse arguments
-	for file in "$@"; do safe_files_string+=("$(normalize_filename "$file")"); done
+	for arg in "$@"; do
+		case "$arg" in
+			-m) format=machine;;
+			*) safe_files_string+=("$(normalize_filename "$arg")");;
+		esac
+	done
 
 	# Init
 	unsafe_files="$(mktemp)"
 
 	# Heavy lifting
-	find . -type f -exec file -F"$sep" {} + | while read line; do
-		file="$(echo "$line"|sed "s${resep}${sep}.*${resep}${resep}")"
-		type="$(echo "$line"|sed "s${resep}.*${sep}\s*${resep}${resep}")"
-
+	find . -type f -printf '%s %h/%f\n' | # find all files
+	while read -r size file; do # filter out files smaller than $min_size
+		[[ $size < $min_size ]] || printf '%s\n' "$file"
+	done |
+	xargs -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
 		file="$(normalize_filename "$file")"
 
 		if \
@@ -61,8 +81,7 @@ main() {
 	done > "$unsafe_files"
 
 	if [[ -n "$(cat "$unsafe_files")" ]]; then
-		echo "==> WARNING: The source directory `pwd` contains binary files:"
-		<"$unsafe_files" sort | sed 's/./  -> &/'
+		<"$unsafe_files" sort | print-$format
 		rm -f "$unsafe_files"
 		exit 1
 	else
-- 
cgit v1.2.3-54-g00ecf