Add a --help/--version to AsciiDoc convertor

liberty is now self-contained, from opt_handler to manual page.

4 files changed, 466 insertions, 11 deletions

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 9e04875..2e8520e 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -10,7 +10,7 @@ if ("${CMAKE_C_COMPILER_ID}" MATCHES "GNU" OR CMAKE_COMPILER_IS_GNUCC)
 endif ()
 
 # Dependencies
-set (CMAKE_MODULE_PATH ${PROJECT_SOURCE_DIR}/cmake)
+set (CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake")
 include (AddThreads)
 
 find_package (PkgConfig REQUIRED)
@@ -35,7 +35,7 @@ foreach (extra iconv rt)
 endforeach ()
 
 # Build some unit tests
-include_directories (${PROJECT_SOURCE_DIR})
+include_directories ("${PROJECT_SOURCE_DIR}")
 enable_testing ()
 set (tests liberty proto xdg)
 
@@ -57,7 +57,7 @@ endforeach ()
 # --- Tools --------------------------------------------------------------------
 
 # Test the AsciiDoc manual page generator for a successful parse
-set (ASCIIMAN ${PROJECT_SOURCE_DIR}/tools/asciiman.awk)
+set (ASCIIMAN "${PROJECT_SOURCE_DIR}/tools/asciiman.awk")
 add_custom_command (OUTPUT libertyxdr.7
 	COMMAND env LC_ALL=C awk -f ${ASCIIMAN}
 		"${PROJECT_SOURCE_DIR}/libertyxdr.adoc" > libertyxdr.7
@@ -65,10 +65,14 @@ add_custom_command (OUTPUT libertyxdr.7
 	COMMENT "Generating man page for libertyxdr" VERBATIM)
 add_custom_target (docs ALL DEPENDS libertyxdr.7)
 
+# Test the --help/--version to AsciiDoc convertor
+add_test (test-help2adoc
+	env LC_ALL=C "${PROJECT_SOURCE_DIR}/tests/help2adoc.sh")
+
 # Test CMake script parsing
 add_test (test-cmake-parser
-	env LC_ALL=C awk -f ${PROJECT_SOURCE_DIR}/tools/cmake-parser.awk
-	-f ${PROJECT_SOURCE_DIR}/tools/cmake-dump.awk ${CMAKE_CURRENT_LIST_FILE})
+	env LC_ALL=C awk -f "${PROJECT_SOURCE_DIR}/tools/cmake-parser.awk"
+	-f "${PROJECT_SOURCE_DIR}/tools/cmake-dump.awk" ${CMAKE_CURRENT_LIST_FILE})
 
 # Test protocol code generation
 set (lxdrgen_outputs)
@@ -77,15 +81,15 @@ foreach (backend c cpp go mjs swift)
 	list (APPEND lxdrgen_outputs ${lxdrgen_base}.${backend})
 	add_custom_command (OUTPUT ${lxdrgen_base}.${backend}
 		COMMAND env LC_ALL=C awk
-			-f ${PROJECT_SOURCE_DIR}/tools/lxdrgen.awk
-			-f ${PROJECT_SOURCE_DIR}/tools/lxdrgen-${backend}.awk
+			-f "${PROJECT_SOURCE_DIR}/tools/lxdrgen.awk"
+			-f "${PROJECT_SOURCE_DIR}/tools/lxdrgen-${backend}.awk"
 			-v PrefixCamel=ProtoGen
-			${PROJECT_SOURCE_DIR}/tests/lxdrgen.lxdr
+			"${PROJECT_SOURCE_DIR}/tests/lxdrgen.lxdr"
 			> ${lxdrgen_base}.${backend}
 		DEPENDS
-			${PROJECT_SOURCE_DIR}/tools/lxdrgen.awk
-			${PROJECT_SOURCE_DIR}/tools/lxdrgen-${backend}.awk
-			${PROJECT_SOURCE_DIR}/tests/lxdrgen.lxdr
+			"${PROJECT_SOURCE_DIR}/tools/lxdrgen.awk"
+			"${PROJECT_SOURCE_DIR}/tools/lxdrgen-${backend}.awk"
+			"${PROJECT_SOURCE_DIR}/tests/lxdrgen.lxdr"
 		COMMENT "Generating test protocol code (${backend})" VERBATIM)
 endforeach ()
 add_custom_target (test-lxdrgen-outputs ALL DEPENDS ${lxdrgen_outputs})
diff --git a/README.adoc b/README.adoc
index 20a4612..d6f6252 100644
--- a/README.adoc
+++ b/README.adoc
@@ -36,6 +36,9 @@ cmake-dump.awk::
 	This can be used in conjunction with the previous script to dump CMake
 	scripts in a normalized format for further processing.
 
+help2adoc.awk::
+	Produces AsciiDoc manual pages from --version/--help output.
+
 lxdrgen.awk::
 	Protocol code generator for a variant of XDR,
 	which is link:libertyxdr.adoc[documented separately].
diff --git a/tests/help2adoc.sh b/tests/help2adoc.sh
new file mode 100755
index 0000000..393c6a9
--- /dev/null
+++ b/tests/help2adoc.sh
@@ -0,0 +1,214 @@
+#!/bin/sh -e
+# This test very exactly matches the output,
+# but help2adoc is more or less feature-complete already.
+self=$(realpath "$0")
+help2adoc=$(realpath "$(dirname "$0")/../tools/help2adoc.awk")
+
+#- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
+
+test_oneline_help() {
+cat <<END
+Usage: $self [--brightness [+-]BRIGHTNESS] [--input NAME] [--restart]
+END
+}
+
+test_oneline_version() {
+cat <<'END'
+eizoctl 1.0
+END
+}
+
+test_oneline_out() {
+cat <<'END'
+eizoctl(1)
+==========
+:doctype: manpage
+:manmanual: eizoctl Manual
+:mansource: eizoctl 1.0
+
+Name
+----
+eizoctl - manual page for eizoctl 1.0
+
+Synopsis
+--------
+*eizoctl* [**--brightness** [+-]__BRIGHTNESS__] [**--input** __NAME__] [**--restart**]
+END
+}
+
+#- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
+
+test_simple_help() {
+cat <<'END'
+Usage: elksmart-comm [OPTION]... [COMMAND...]
+Usage: elksmart-comm
+Transmit or receive infrared commands.
+
+  -d, --debug                 elksmart-comm will run in debug mode
+ -f, --frequency HZ          frequency (38000 Hz by default)
+   -n, --nec                   use the NEC transmission format
+  -h, --help                  display this help and exit
+ -V, --version               output version information and exit
+END
+}
+
+test_simple_version() {
+cat <<'END'
+elksmart-comm (usb-drivers) dev
+END
+}
+
+test_simple_out() {
+cat <<'END'
+elksmart-comm(1)
+================
+:doctype: manpage
+:manmanual: elksmart-comm Manual
+:mansource: usb-drivers dev
+
+Name
+----
+elksmart-comm - manual page for elksmart-comm dev
+
+Synopsis
+--------
+*elksmart-comm* [__OPTION__]... [__COMMAND__...]
+
+*elksmart-comm*
+
+
+Description
+-----------
+Transmit or receive infrared commands.
+
+
+*-d*, **--debug**::
+	**elksmart-comm** will run in debug mode
+
+*-f*, **--frequency** __HZ__::
+	frequency (38000 Hz by default)
+
+*-n*, **--nec**::
+	use the NEC transmission format
+
+*-h*, **--help**::
+	display this help and exit
+
+*-V*, **--version**::
+	output version information and exit
+END
+}
+
+#- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
+
+test_wild_help() {
+cat <<'END'
+Usage:
+ wild [option]… <command>… — Be wild
+What's happening?
+  -f, --frequency hz-2-foo  frequency to --foo at
+      --foo=bar
+  Foobar.
+  Boo far.
+
+ Subsection:
+  --help
+  --version
+  Oh my.
+
+Major section:
+And now for something completely different.
+  Very wild
+END
+}
+
+test_wild_version() {
+cat <<'END'
+wild 1
+Copies left and right.
+END
+}
+
+test_wild_out() {
+cat <<'END'
+wild(1)
+=======
+:doctype: manpage
+:manmanual: wild Manual
+:mansource: wild 1
+
+Name
+----
+wild - manual page for wild 1
+
+Synopsis
+--------
+*wild* [__option__]... <__command__>...
+
+
+Description
+-----------
+Be **wild**
+
+What's happening?
+
+*-f*, **--frequency** __hz-2-foo__::
+	frequency to **--foo** at
+
+*--foo*=__bar__::
+	Foobar.
+	Boo far.
+
+
+Subsection
+~~~~~~~~~~
+
+*--help*::
+
+*--version*::
+	Oh my.
+
+
+Major section
+-------------
+And now for something completely different.
+  Very wild
+END
+}
+
+#- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
+
+run() {
+	echo "-- help2adoc/$1"
+	local selfquoted=$(echo "$self" | sed 's/\\/&&/g')
+	local output=$(TEST=$1 awk -f "$help2adoc" -v Target="$selfquoted")
+	local expect="$($1_out)"
+	if [ "$output" = "$expect" ]
+	then return
+	fi
+
+	echo "== Expected"
+	sed 's/^/   /' <<-END
+	$expect
+	END
+	echo "== Received"
+	sed 's/^/   /' <<-END
+	$output
+	END
+	exit 1
+}
+
+if [ -z "$TEST" ]
+then
+	run test_oneline
+	run test_simple
+	run test_wild
+	echo "-- OK"
+elif [ "$1" = "--help" ]
+then ${TEST}_help
+elif [ "$1" = "--version" ]
+then ${TEST}_version
+else
+	echo "Wrong usage"
+	exit 1
+fi
diff --git a/tools/help2adoc.awk b/tools/help2adoc.awk
new file mode 100644
index 0000000..b36753f
--- /dev/null
+++ b/tools/help2adoc.awk
@@ -0,0 +1,234 @@
+# help2adoc.awk: convert --version/--help to AsciiDoc manual pages
+#
+# Copyright (c) 2024, Přemysl Eric Janouch <p@janouch.name>
+# SPDX-License-Identifier: 0BSD
+#
+# Usage: awk -f help2adoc.awk -v Target=cat
+#
+# This is not intended to produce great output, merely useful output,
+# if only because there is no real standard of what the input should look like.
+#
+# The only target that needs to work is liberty's own opt_handler.
+# The expected input format is roughly that of GNU utilites.
+
+function fatal(message) {
+	print "// " message
+	print "fatal error: " message > "/dev/stderr"
+	exit 1
+}
+
+# The input model of this script is that function take the next line on $0,
+# read further lines as necessary, and leave the next line in $0 again.
+function readline(    ok) {
+	if ((ok = (Command | getline)) < 0)
+		fatal("read error")
+	if (!ok)
+		exit
+}
+
+function emboldenoptions(line) {
+	# -N, --newer=DATE-OR-FILE, --after-date=DATE-OR-FILE
+	sub(/^-[^-=,[:space:]{[<]/, "*&*", line)
+	while (match(line, /[^-_[:alnum:]*'+]-[^-=,[:space:]{[<]/)) {
+		line = substr(line, 1, RSTART) \
+			"**" substr(line, RSTART + 1, RLENGTH - 1) "**" \
+			substr(line, RSTART + RLENGTH)
+	}
+	sub(/^--[-_[:alnum:]]+/, "*&*", line)
+	while (match(line, /[^-_[:alnum:]*'+]--[-_[:alnum:]]+/)) {
+		line = substr(line, 1, RSTART) \
+			"**" substr(line, RSTART + 1, RLENGTH - 1) "**" \
+			substr(line, RSTART + RLENGTH)
+	}
+	return line
+}
+
+function formatinline(line,    programname, last, i) {
+	# Go the extra step of emboldening the program name at word boundaries.
+	programname = ProgramName
+	gsub(/[][\\.^$(){}|*+?]/, "\\\\&", programname)
+	if (match(line, "^" programname "[^-_[:alnum:]*'+/]")) {
+		line = "**" substr(line, RSTART, RLENGTH - 1) "**" \
+			substr(line, RSTART + RLENGTH - 1)
+	}
+	while (match(line, "[^-_[:alnum:]*'+/]" programname "[^-_[:alnum:]*'+/]")) {
+		line = substr(line, 1, RSTART) \
+			"**" substr(line, RSTART + 1, RLENGTH - 2) "**" \
+			substr(line, RSTART + RLENGTH - 1)
+	}
+	if (match(line, "[^-_[:alnum:]*'+/]" programname "$")) {
+		line = substr(line, 1, RSTART) \
+			"**" substr(line, RSTART + 1, RLENGTH - 1) "**"
+	}
+	return emboldenoptions(line)
+}
+
+function printusage(usage,    description) {
+	gsub(/…/, "...", usage)
+	gsub(/—|–/, "-", usage)
+
+	# --help output will more likely than not simply include argv[0],
+	# or perhaps program_invocation_short_name (not addressed here).
+	if (substr(usage, 1, length(Target) + 1) == Target " ")
+		usage = ProgramName substr(usage, length(Target) + 1)
+
+	# A lot of GNOME software includes the description here.
+	if (match(usage, / +- +/) && usage !~ / - [^[:alnum:]]/) {
+		description = substr(usage, RSTART + RLENGTH)
+		usage = substr(usage, 1, RSTART - 1)
+	}
+
+	while (match(usage, /[^-_[:alnum:]*'+.][[:alnum:]][-_[:alnum:]]+/)) {
+		usage = substr(usage, 1, RSTART) \
+			"__" substr(usage, RSTART + 1, RLENGTH - 1) "__" \
+			substr(usage, RSTART + RLENGTH)
+	}
+	sub(/^[^[:space:]]+/, "*&*", usage)
+	print emboldenoptions(usage)
+	print ""
+
+	if (description) {
+		flushsections()
+		print formatinline(description)
+		print ""
+	}
+}
+
+# We're going with Setext headers, because that's what asciiman.awk supports.
+function printheader(text, underline) {
+	print text
+	gsub(/./, underline, text)
+	print text
+}
+
+BEGIN {
+	if (!Target)
+		fatal("missing Target")
+
+	TargetQuoted = Target
+	gsub(/'/, "'\\''", TargetQuoted)
+	TargetQuoted = "'" TargetQuoted "'"
+
+	# Remaining --version lines could be about copyright (GNU),
+	# or something else entirely.
+	Command = TargetQuoted " --version"
+	if ((Command | getline) > 0) {
+		# GNU --version output can place the package name in parentheses.
+		Package = $0
+		if (match($0, /[[:space:]][(][^)]*[)]/)) {
+			Package = substr($0, RSTART + 2, RLENGTH - 3) \
+				substr($0, RSTART + RLENGTH)
+			sub(/[[:space:]]+[(][^)]*[)]/, "")
+		}
+
+		Version = $0
+		sub(/[[:space:]]+[^[:space:]]+$/, "")
+		Name = $0
+	} else {
+		fatal("failed to get --version output")
+	}
+
+	if (Name !~ /[[:space:]]/)
+		ProgramName = Name
+	else if (match(Target, /[^/]+$/))
+		ProgramName = substr(Target, RSTART, RLENGTH)
+
+	printheader(ProgramName "(1)", "=")
+	print ":doctype: manpage"
+	print ":manmanual: " Name " Manual"
+	print ":mansource: " Package
+	print ""
+	printheader("Name", "-")
+	print ProgramName " - manual page for " Version
+	print ""
+
+	close(Command)
+	Command = TargetQuoted " --help"
+	if ((Command | getline) <= 0)
+		fatal("failed to get --help output")
+
+	NextSection = "Description"
+	NextSubsection = ""
+
+	# The SYNOPSIS section is mandatory, so just put it there.
+	printheader("Synopsis", "-")
+	while (1) {
+		if (match($0, /^[Uu]sage:[[:space:]]*/)) {
+			if (($0 = substr($0, RSTART + RLENGTH)))
+				printusage($0)
+		} else if (match($0, /^[[:space:]]+/) && !/^[[:space:]]*-/) {
+			if (($0 = substr($0, RSTART + RLENGTH)))
+				printusage($0)
+		} else if ($0) {
+			break
+		}
+		readline()
+	}
+	while (1) {
+		if (match($0, /^[[:alpha:]][-[:alnum:][:space:]]+:$/)) {
+			# We don't flush sections here,
+			# so that we don't unnecessarily enforce DESCRIPTION first.
+			NextSection = substr($0, RSTART, RLENGTH - 1)
+		} else if (match($0, /^ [[:alpha:]][-[:alnum:][:space:]]+:$/)) {
+			flushsections()
+			NextSubsection = substr($0, RSTART + 1, RLENGTH - 2)
+		} else if (match($0, /^ +-/)) {
+			flushsections()
+			parseoption(substr($0, RSTART + RLENGTH - 1))
+			continue
+		} else if ($0) {
+			flushsections()
+
+			# That will be probably interpreted as a literal block.
+			if (!/^[[:space:]]/)
+				$0 = formatinline($0)
+			print
+		} else {
+			print
+		}
+		readline()
+	}
+}
+
+function flushsections() {
+	if (NextSection) {
+		print ""
+		printheader(NextSection, "-")
+		NextSection = ""
+	}
+	if (NextSubsection) {
+		print ""
+		printheader(NextSubsection, "~")
+		NextSubsection = ""
+	}
+}
+
+function parseoption(line,    usage) {
+	# Often enough you will see it separated with only one space,
+	# which will simply not work for us.
+	if (match(line, /[[:space:]]{2,}/)) {
+		usage = substr(line, 1, RSTART - 1)
+		line = substr(line, RSTART + RLENGTH)
+	} else {
+		usage = line
+		line = ""
+	}
+
+	usage = emboldenoptions(usage)
+	while (match(usage, /[=<, ][[:alnum:]][-_[:alnum:]]*/)) {
+		usage = substr(usage, 1, RSTART) \
+			"__" substr(usage, RSTART + 1, RLENGTH - 1) "__" \
+			substr(usage, RSTART + RLENGTH)
+	}
+
+	print ""
+	print usage "::"
+	if (line)
+		print "\t" formatinline(line)
+
+	readline()
+	while (match($0, /^ +[^-[:space:]]|^ {7,}./)) {
+		print "\t" formatinline(substr($0, RSTART + RLENGTH - 1))
+		readline()
+	}
+}