[utils/utils.git] / lesspipe.sh

#!/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/$f/
else
	# Device/special file.  Don't print anything; let less complain about it.
	# (For now, we are not interested in content sniffing on pipes.  If less
	# gained support for calling LESSOPEN for stdin, we might reconsider.)
	exit 77
fi
Commit	Line	Data
273c3903 MM	1	#!/bin/bash
	2	# Matt's improved version of Fedora's lesspipe.sh
	3	#
	4	# To use this filter with less, define LESSOPEN:
	5	# export LESSOPEN="\|lesspipe.sh %s"
	6
	7	# Updated for "less" that checks our exit code ~ Matt 2010-12-28
	8
	9	# This script is entirely driven by the file content, with type detected by
	10	# file(1). For reproducibility and to avoid problems with strange names and
	11	# even symlinks, we do not use the original name of the file for anything.
	12
	13	set -e
	14	set -o pipefail
	15	#set -x
	16
	17	# Attach the file to an FD. We will reopen /dev/fd/$f to get as many fresh FDs as
	18	# we need.
	19	exec {f}<"$1"
	20	shift
	21
	22	function run_stream_stdin {
	23	if [ "$decompressor" == cat ]; then
	24	<&$f "$@"
	25	else
	26	<&$f "$decompressor" \| "$@"
	27	fi
	28	}
	29
	30	function stage_to_file_stdin {
	31	if [ "$decompressor" != cat ]; then
	32	# There is an unavoidable race here, so I am not trying to avoid
	33	# it. If we rewrite the script in Perl, we can use open(undef).
	34	# Integration with in2tempfile looks too hard.
	35	tmpfile="$(mktemp --tmpdir lesspipe.XXXXXXXXXX)"
	36	exec {tf}>"$tmpfile"
	37	rm -f "$tmpfile"
	38	<&$f "$decompressor" >&$tf # Synchronous
	39	# Get a new FD for reading on the tempfile. This might not be
	40	# necessary if the command is just going to reopen it.
	41	exec {f}>&- </dev/fd/$tf {tf}>&-
	42	else
	43	exec <&$f-
	44	fi
	45	}
	46
	47	# Note: the bash file tests (except -h) follow symlinks.
	48	if [ -f /dev/fd/$f ]; then
	49	# Specify the input as -. This avoids irrelevant "set[ug]id" output for
	50	# executables and also follows the symlink without us having to pass -L.
	51	# We have to reopen stdin to not consume it.
	52	content_type="$(</dev/fd/$f file -b -)"
	53
	54	case "$content_type" in
	55	('gzip compressed data'*)
	56	decompressor=gunzip;;
	57	('bzip2 compressed data'*)
	58	decompressor=bunzip2;;
	59	('XZ compressed data'*)
	60	decompressor=unxz;;
	61	(*)
	62	decompressor=cat;;
	63	esac
	64
65	if [ "$decompressor" != cat ]; then
66	# Get the real content type. Whatever work the decompressor
67	# does before it is killed with SIGPIPE, we will duplicate later
68	# when we read the file, but there is no good alternative:
69	# - If we do a "tee" with one pipe held for later use, "tee"
70	# will block once it fills up that pipe. So if "file" needs
71	# more data than the pipe buffer to make its decision
72	# (probably unlikely), we could hang.
73	# - Going ahead and dumping the decompressed data to a temporary
74	# file is a waste of work if the content type turns out to be
75	# one for which we have a streaming viewer, but of course we
76	# don't know that yet.
77	content_type="$(</dev/fd/$f "$decompressor" \| file -b -)" \|\| [ $? == 141 ] # SIGPIPE
78	fi
79
80	case "$content_type" in
81	('troff or preprocessor input'*)
82	## Presuming our parent is "less", set COLUMNS based on the width
83	## of the terminal.
84	#exec {pt}>/proc/$PPID/fd/1
85	#if [ -t $pt ]; then
86	# # XXX Does this really use stderr rather than the controlling terminal?
87	# export COLUMNS="$(bash -i -c 'echo $COLUMNS' 2>&$pt)"
88	#fi
89	#exec {pt}>&-
90	# viapty is a more comprehensive solution than the above. ~ 2011-09-05
91	# Hm, by the time we hack in PAGER=cat, maybe it's not any better... ~ 2011-09-06
92	run_stream_stdin viapty man -l -;;
93	('tar archive')
94	run_stream_stdin tar -tvv;;
95	('cpio archive')
96	run_stream_stdin cpio --quiet -itv;;
97	(RPM*)
98	run_stream_stdin rpm -qivl --changelog -p -;;
99	(image) # FIXME: will catch disk "images" as well as graphical images
100	if type identify &>/dev/null; then
101	identify_cmd=(identify)
102	elif type gm &>/dev/null; then
103	identify_cmd=(gm identify)
104	else
105	### Give them the binary data.
106	#echo "No identify available"
107	#echo "Install ImageMagick or GraphicsMagick to browse images"
108	identify_cmd=(false)
109	fi
110	if [ "${identify_cmd[0]}" != false ]; then
111	# Make a file so identify won't stage it again with a
112	# nonreproducible name. (Note that "identify -" will
113	# stage again even if /dev/stdin is a file!)
114	stage_to_file_stdin
115	"${identify_cmd[@]}" /dev/stdin
116	fi;;
117	('OpenDocument Text'\|_,'OpenDocument Text Template')
118	stage_to_file_stdin
119	### CUSTOM SCRIPT -- This part cannot be upstreamed unless odt2text is.
120	odt2text /dev/stdin # based on unzip, needs a file
121	;;
122	# This has to come after the special case for odt2text.
123	('Zip archive data'\|_,'OpenDocument'\|_,'OpenOffice 1.x'*)
124	stage_to_file_stdin
125	zipinfo /dev/stdin # needs a file
126	;;
127	('PDF document'*)
128	run_stream_stdin pdftotext - -;;
129	('PostScript document'*)
130	# If you wanted to see the PostScript source code, sorry...
131	run_stream_stdin ps2ascii;;
132	(ELF'dynamically linked')
133	stage_to_file_stdin # read twice
134	objdump -xRTdgl /dev/stdin
135	# The above does not seem to show data sections. Run a separate command
136	# to show them. (Of course I would prefer a single command.)
137	objdump -s -j .rodata -j .data /dev/stdin
138	;;
139	(ELF*)
140	stage_to_file_stdin # read twice
141	objdump -xdgl /dev/stdin
142	objdump -s -j .rodata -j .data /dev/stdin
143	;;
144	('SQLite 3.x database'*)
145	# TODO: Reimplement copy to a temporary directory to
146	# handle databases that use write-ahead mode, etc. I had
147	# implemented this in main VM and I forgot to copy it to
148	# main-template before shutting down. :( ~ Noted 2015-12-28
149	stage_to_file_stdin
150	# I believe that this will honor a write lock held on an
151	# uncompressed database by another instance of SQLite. I
152	# further believe that is what we want.
153	sqlite3 /dev/stdin .dump # needs a file
154	;;
155	('compiled Java class data'*)
156	# Agghhh... why couldn't javap just take a filename? It's clear
157	# that all it is doing is translating our argument to a file and
158	# opening that, without checking the correctness of the package
159	# or class name.
160	# XXX: Proper cleanup on failure
161	dir="$(mktemp -d --tmpdir lesspipe-javap.XXXXXXXXXX)"
162	ln -s /dev/stdin "$dir/Stdin.class"
163	run_stream_stdin javap -classpath "$dir" -private -s -c -verbose Stdin
164	rm -rf "$dir"
165	;;
166	# Disabled for now.
167	#(*.grl)
168	# stage_to_file_stdin
169	# head -n 1 /dev/stdin \| fold -w 28
170	# tail -n +2;;
171	(*) # Unknown content type.
172	if [ "$decompressor" != cat ]; then
173	# Just decompress the file.
174	<&$f "$decompressor"
175	else
176	# We cannot add value in this case.
177	# Let "less" show the original file.
178	exit 77
179	fi;;
180	esac
181
182	elif [ -d /dev/fd/$f ]; then
183	# nip blank line at the end
184	# XXX: how to make sure "less" has -R enabled?
185	#ls -H -al --color=always -- /dev/fd/$f/ \| head --bytes=-3
186	### Between coreutils-7.6-9.matt1.fc12 and coreutils-8.5-7.fc14, the
187	### output has changed to put the \n at the end, so the special hack is
188	### no longer necessary (less ignores a final \n) and in fact corrupts
189	### the colors. ~ Matt 2011-06-19
190	ls -H -al --color=always -- /dev/fd/$f/
191	else
192	# Device/special file. Don't print anything; let less complain about it.
193	# (For now, we are not interested in content sniffing on pipes. If less
194	# gained support for calling LESSOPEN for stdin, we might reconsider.)
195	exit 77
196	fi