X-Git-Url: https://mattmccutchen.net/utils/utils.git/blobdiff_plain/40ea9b7868f2b7746e7cbabfba6aba982096392a..273c390351c42303171c25215304d1cfd6ca02d4:/lesspipe.sh

diff --git a/lesspipe.sh b/lesspipe.sh
new file mode 100755
index 0000000..372abb5
--- /dev/null
+++ b/lesspipe.sh
@@ -0,0 +1,196 @@
+#!/bin/bash
+# Matt's improved version of Fedora's lesspipe.sh
+#
+# To use this filter with less, define LESSOPEN:
+# export LESSOPEN="|lesspipe.sh %s"
+
+# Updated for "less" that checks our exit code ~ Matt 2010-12-28
+
+# This script is entirely driven by the file content, with type detected by
+# file(1).  For reproducibility and to avoid problems with strange names and
+# even symlinks, we do not use the original name of the file for anything.
+
+set -e
+set -o pipefail
+#set -x
+
+# Attach the file to an FD.  We will reopen /dev/fd/$f to get as many fresh FDs as
+# we need.
+exec {f}<"$1"
+shift
+
+function run_stream_stdin {
+	if [ "$decompressor" == cat ]; then
+		<&$f "$@"
+	else
+		<&$f "$decompressor" | "$@"
+	fi
+}
+
+function stage_to_file_stdin {
+	if [ "$decompressor" != cat ]; then
+		# There is an unavoidable race here, so I am not trying to avoid
+		# it.  If we rewrite the script in Perl, we can use open(undef).
+		# Integration with in2tempfile looks too hard.
+		tmpfile="$(mktemp --tmpdir lesspipe.XXXXXXXXXX)"
+		exec {tf}>"$tmpfile"
+		rm -f "$tmpfile"
+		<&$f "$decompressor" >&$tf # Synchronous
+		# Get a new FD for reading on the tempfile.  This might not be
+		# necessary if the command is just going to reopen it.
+		exec {f}>&- </dev/fd/$tf {tf}>&-
+	else
+		exec <&$f-
+	fi
+}
+
+# Note: the bash file tests (except -h) follow symlinks.
+if [ -f /dev/fd/$f ]; then
+	# Specify the input as -.  This avoids irrelevant "set[ug]id" output for
+	# executables and also follows the symlink without us having to pass -L.
+	# We have to reopen stdin to not consume it.
+	content_type="$(</dev/fd/$f file -b -)"
+
+	case "$content_type" in
+	('gzip compressed data'*)
+		decompressor=gunzip;;
+	('bzip2 compressed data'*)
+		decompressor=bunzip2;;
+	('XZ compressed data'*)
+		decompressor=unxz;;
+	(*)
+		decompressor=cat;;
+	esac
+
+	if [ "$decompressor" != cat ]; then
+		# Get the real content type.  Whatever work the decompressor
+		# does before it is killed with SIGPIPE, we will duplicate later
+		# when we read the file, but there is no good alternative:
+		# - If we do a "tee" with one pipe held for later use, "tee"
+		#   will block once it fills up that pipe.  So if "file" needs
+		#   more data than the pipe buffer to make its decision
+		#   (probably unlikely), we could hang.	
+		# - Going ahead and dumping the decompressed data to a temporary
+		#   file is a waste of work if the content type turns out to be
+		#   one for which we have a streaming viewer, but of course we
+		#   don't know that yet.
+		content_type="$(</dev/fd/$f "$decompressor" | file -b -)" || [ $? == 141 ] # SIGPIPE
+	fi
+
+	case "$content_type" in
+	('troff or preprocessor input'*)
+		## Presuming our parent is "less", set COLUMNS based on the width
+		## of the terminal.
+		#exec {pt}>/proc/$PPID/fd/1
+		#if [ -t $pt ]; then
+		#	# XXX Does this really use stderr rather than the controlling terminal?
+		#	export COLUMNS="$(bash -i -c 'echo $COLUMNS' 2>&$pt)"
+		#fi
+		#exec {pt}>&-
+		# viapty is a more comprehensive solution than the above. ~ 2011-09-05
+		# Hm, by the time we hack in PAGER=cat, maybe it's not any better... ~ 2011-09-06
+		run_stream_stdin viapty man -l -;;
+	(*'tar archive'*)
+		run_stream_stdin tar -tvv;;
+	(*'cpio archive'*)
+		run_stream_stdin cpio --quiet -itv;;
+	(RPM*)
+		run_stream_stdin rpm -qivl --changelog -p -;;
+	(*image*) # FIXME: will catch disk "images" as well as graphical images
+		if type identify &>/dev/null; then
+			identify_cmd=(identify)
+		elif type gm &>/dev/null; then
+			identify_cmd=(gm identify)
+		else
+			### Give them the binary data.
+			#echo "No identify available"
+			#echo "Install ImageMagick or GraphicsMagick to browse images"
+			identify_cmd=(false)
+		fi
+		if [ "${identify_cmd[0]}" != false ]; then
+			# Make a file so identify won't stage it again with a
+			# nonreproducible name.  (Note that "identify -" will
+			# stage again even if /dev/stdin is a file!)
+			stage_to_file_stdin
+			"${identify_cmd[@]}" /dev/stdin
+		fi;;
+	('OpenDocument Text'|_,'OpenDocument Text Template')
+		stage_to_file_stdin
+		### CUSTOM SCRIPT -- This part cannot be upstreamed unless odt2text is.
+		odt2text /dev/stdin # based on unzip, needs a file
+		;;
+	# This has to come after the special case for odt2text.
+	('Zip archive data'*|_,'OpenDocument'*|_,'OpenOffice 1.x'*)
+		stage_to_file_stdin
+		zipinfo /dev/stdin # needs a file
+		;;
+	('PDF document'*)
+		run_stream_stdin pdftotext - -;;
+	('PostScript document'*)
+		# If you wanted to see the PostScript source code, sorry...
+		run_stream_stdin ps2ascii;;
+	(ELF*'dynamically linked'*)
+		stage_to_file_stdin # read twice
+		objdump -xRTdgl /dev/stdin
+		# The above does not seem to show data sections.  Run a separate command
+		# to show them.  (Of course I would prefer a single command.)
+		objdump -s -j .rodata -j .data /dev/stdin
+		;;
+	(ELF*)
+		stage_to_file_stdin # read twice
+		objdump -xdgl /dev/stdin
+		objdump -s -j .rodata -j .data /dev/stdin
+		;;
+	('SQLite 3.x database'*)
+		# TODO: Reimplement copy to a temporary directory to
+		# handle databases that use write-ahead mode, etc.  I had
+		# implemented this in main VM and I forgot to copy it to
+		# main-template before shutting down. :( ~ Noted 2015-12-28
+		stage_to_file_stdin
+		# I believe that this will honor a write lock held on an
+		# uncompressed database by another instance of SQLite.  I
+		# further believe that is what we want.
+		sqlite3 /dev/stdin .dump # needs a file
+		;;
+	('compiled Java class data'*)
+		# Agghhh... why couldn't javap just take a filename?  It's clear
+		# that all it is doing is translating our argument to a file and
+		# opening that, without checking the correctness of the package
+		# or class name.
+		# XXX: Proper cleanup on failure
+		dir="$(mktemp -d --tmpdir lesspipe-javap.XXXXXXXXXX)"
+		ln -s /dev/stdin "$dir/Stdin.class"
+		run_stream_stdin javap -classpath "$dir" -private -s -c -verbose Stdin
+		rm -rf "$dir"
+		;;
+	# Disabled for now.
+	#(*.grl)
+	#	stage_to_file_stdin
+	#	head -n 1 /dev/stdin | fold -w 28
+	#	tail -n +2;;
+	(*) # Unknown content type.
+		if [ "$decompressor" != cat ]; then
+			# Just decompress the file.
+			<&$f "$decompressor"
+		else
+			# We cannot add value in this case.
+			# Let "less" show the original file.
+			exit 77
+		fi;;
+	esac
+
+elif [ -d /dev/fd/$f ]; then
+	# nip blank line at the end
+	# XXX: how to make sure "less" has -R enabled?
+	#ls -H -al --color=always -- /dev/fd/$f/ | head --bytes=-3
+	### Between coreutils-7.6-9.matt1.fc12 and coreutils-8.5-7.fc14, the
+	### output has changed to put the \n at the end, so the special hack is
+	### no longer necessary (less ignores a final \n) and in fact corrupts
+	### the colors. ~ Matt 2011-06-19
+	ls -H -al --color=always -- /dev/fd/$f/
+else
+	# Device/special file.  Don't print anything; let less complain about it.
+	# (For now, we are not interested in content sniffing on pipes.  If less
+	# gained support for calling LESSOPEN for stdin, we might reconsider.)
+	exit 77
+fi