#!/usr/bin/python
# Human XML: convert to and from a more human-friendly syntax for XML.
# Adam Sampson <ats@offog.org>

# This is based on simple textual transformations.
# Hux        XML            Notes
# <A>        <A/>
# <A/B>      <A>B</A>       B is expanded recursively
# <--A-->    <!--A-->       A cannot contain --
# <!A>       <!A>           <> must balance in A; A cannot start --
# <?A?>      <?A?>          Only ?> is recognised in A
# <[[A]]>    <[!CDATA[A]]>  Only ]]> is recognised in A

import sys, getopt

class BufferedInput:
	def __init__(self, f):
		self.f = f
		self.buf = ""
		self.pos = 0
	def getc(self):
		if self.pos == len(self.buf):
			self.buf = self.f.read(4096)
			self.pos = 0
			if self.buf == "":
				return ""
		c = self.buf[self.pos]
		self.pos += 1
		return c

class BufferedOutput:
	def __init__(self, f):
		self.f = f
		self.buf = []
	def flush(self):
		self.f.write("".join(self.buf))
		self.buf = []
	def put(self, s):
		self.buf.append(s)
		if len(self.buf) > 100:
			self.flush()

def die(*s):
	print >>sys.stderr, "Fatal: " + "".join(map(str, s))
	sys.exit(1)

def copy_bang(bi, bo):
	level = 1
	c = None
	while not (c == ">" and level == 0):
		c = bi.getc()
		if c == "":
			die("EOF inside <!...>")
		elif c == "<":
			level += 1
		elif c == ">":
			level -= 1
		bo.put(c)

def copy_until(bi, bo, watch):
	history = " " * len(watch)
	while history != watch:
		c = bi.getc()
		if c == "":
			die("EOF while looking for " + watch)
		bo.put(c)
		history = history[1:] + c

def is_whitespace(c):
	return c in "\x20\x09\x0D\x0A"

def hux_to_xml(bi, bo):
	stack = []
	while 1:
		c = bi.getc()
		if c == "":
			break
		elif c == "<":
			bo.put(c)
			c = bi.getc()
			if c == "":
				die("Trailing <")
			elif c == ">":
				die("<> makes no sense")
			elif c == "!":
				bo.put(c)
				copy_bang(bi, bo)
			elif c == "?":
				bo.put(c)
				copy_until(bi, bo, "?>")
			elif c == "-":
				c = bi.getc()
				if c != "-":
					die("Expected <--")
				bo.put("!--")
				copy_until(bi, bo, "-->")
			elif c == "[":
				c = bi.getc()
				if c != "[":
					die("Expected <[[")
				copy_until(bi, bo, "]]>")
			else:
				bo.put(c)
				element = c
				name_done = False
				while 1:
					c = bi.getc()
					if c == "":
						die("EOF inside start tag")
					elif c == "/":
						bo.put(">")
						stack.append(element)
						break
					elif c == ">":
						bo.put("/>")
						break
					elif not name_done:
						bo.put(c)
						if is_whitespace(c):
							name_done = True
						else:
							element += c
					elif c in "\"'":
						bo.put(c)
						copy_until(bi, bo, c)
					else:
						bo.put(c)
		elif c == ">":
			if stack == []:
				die("Found > at top level")
			bo.put("</" + stack.pop() + ">")
		else:
			bo.put(c)
	if stack != []:
		die("Missing >s at end of file for: " + " ".join(stack))

def xml_to_hux(bi, bo):
	while 1:
		c = bi.getc()
		if c == "":
			break
		elif c == "<":
			c = bi.getc()
			if c == "":
				die("Trailing <")
			elif c == ">":
				die("<> makes no sense")
			elif c == "/":
				while c != ">":
					c = bi.getc()
					if c == "":
						die("EOF in closing tag")
				bo.put(">")
			elif c == "!":
				c = bi.getc()
				if c == "":
					die("EOF inside <!")
				elif c == ">":
					die("<!> makes no sense")
				elif c == "-":
					c = bi.getc()
					if c != "-":
						die("Expected <!--")
					bo.put("<--")
					copy_until(bi, bo, "-->")
				else:
					bo.put("<!")
					bo.put(c)
					copy_bang(bi, bo)
			elif c == "[":
				for w in "!CDATA[":
					c = bi.getc()
					if c != w:
						die("Expected <[!CDATA[")
				bo.put("<[[")
				copy_until(bi, bo, "]]>")
			elif c == "?":
				bo.put("<?")
				copy_until(bi, bo, "?>")
			else:
				bo.put("<")
				bo.put(c)
				while 1:
					c = bi.getc()
					if c == "":
						die("EOF inside start tag")
					elif c == "/":
						c = bi.getc()
						if c != ">":
							die("Expected />")
						bo.put(">")
						break
					elif c == ">":
						bo.put("/")
						break
					elif c in "\"'":
						bo.put(c)
						copy_until(bi, bo, c)
					else:
						bo.put(c)
		elif c == ">":
			bo.put("&gt;")
		else:
			bo.put(c)

def usage():
	print "Usage: hux [OPTION]... [FILE]..."
	print "Convert between the hux and standard XML encodings."
	print
	print "  -d             convert XML to hux (default: hux to XML)"
	print "  -o FILENAME    write output to FILENAME (default: stdout)"
	print "  --help         display this help and exit"
	print
	print "Report bugs to <ats@offog.org>."

if __name__ == "__main__":
	try:
		opts, args = getopt.getopt(sys.argv[1:], "o:dh", ["help"])
	except getopt.GetoptError:
		usage()
		sys.exit(1)

	outfn = None
	decode = False
	for (o, a) in opts:
		if o in ("-h", "--help"):
			usage()
			sys.exit(0)
		elif o == "-o":
			outfn = a
		elif o == "-d":
			decode = True

	def process(bi, bo):
		if decode:
			xml_to_hux(bi, bo)
		else:
			hux_to_xml(bi, bo)

	if outfn is None:
		out = sys.stdout
	else:
		out = open(outfn, "w")
	bo = BufferedOutput(out)
	if args == []:
		process(BufferedInput(sys.stdin), bo)
	else:
		for infn in args:
			f = open(infn)
			bi = BufferedInput(f)
			process(bi, bo)
			f.close()
	bo.flush()
	out.close()