cpython/Doc/tools/sgmlconv/esis2sgml.py

178 lines
4.7 KiB
Python
Raw Normal View History

1998-11-23 12:59:39 -04:00
#! /usr/bin/env python
"""Convert ESIS events to SGML or XML markup.
This is limited, but seems sufficient for the ESIS generated by the
latex2esis.py script when run over the Python documentation.
"""
__version__ = '$Revision$'
import errno
1998-12-01 15:01:53 -04:00
import esistools
import os
1998-11-23 12:59:39 -04:00
import re
import string
1999-01-14 13:06:09 -04:00
from xml.utils import escape
1998-11-23 12:59:39 -04:00
EMPTIES_FILENAME = "../sgml/empties.dat"
LIST_EMPTIES = 0
def format_attrs(attrs, xml=0):
1998-11-23 12:59:39 -04:00
attrs = attrs.items()
attrs.sort()
s = ''
for name, value in attrs:
if xml:
s = '%s %s="%s"' % (s, name, escape(value))
else:
# this is a little bogus, but should do for now
if name == value and isnmtoken(value):
s = "%s %s" % (s, value)
elif istoken(value):
if value == "no" + name:
s = "%s %s" % (s, value)
else:
s = "%s %s=%s" % (s, name, value)
else:
s = '%s %s="%s"' % (s, name, escape(value))
1998-11-23 12:59:39 -04:00
return s
_nmtoken_rx = re.compile("[a-z][-._a-z0-9]*$", re.IGNORECASE)
def isnmtoken(s):
return _nmtoken_rx.match(s) is not None
_token_rx = re.compile("[a-z0-9][-._a-z0-9]*$", re.IGNORECASE)
def istoken(s):
return _token_rx.match(s) is not None
def do_convert(ifp, ofp, xml=0, autoclose=()):
if xml:
autoclose = ()
1998-11-23 12:59:39 -04:00
attrs = {}
lastopened = None
knownempties = []
1998-11-23 12:59:39 -04:00
knownempty = 0
lastempty = 0
while 1:
line = ifp.readline()
if not line:
break
type = line[0]
data = line[1:]
if data and data[-1] == "\n":
data = data[:-1]
if type == "-":
1998-12-01 15:01:53 -04:00
data = esistools.decode(data)
1999-01-14 13:06:09 -04:00
ofp.write(escape(data))
1998-11-23 12:59:39 -04:00
if "\n" in data:
lastopened = None
knownempty = 0
lastempty = 0
elif type == "(":
1998-12-01 15:01:53 -04:00
if data == "COMMENT":
ofp.write("<!--")
continue
1998-11-23 12:59:39 -04:00
if knownempty and xml:
ofp.write("<%s%s/>" % (data, format_attrs(attrs, xml)))
1998-11-23 12:59:39 -04:00
else:
ofp.write("<%s%s>" % (data, format_attrs(attrs, xml)))
1998-11-23 12:59:39 -04:00
if knownempty and data not in knownempties:
# accumulate knowledge!
knownempties.append(data)
attrs = {}
lastopened = data
lastempty = knownempty
knownempty = 0
elif type == ")":
1998-12-01 15:01:53 -04:00
if data == "COMMENT":
ofp.write("-->")
continue
1998-11-23 12:59:39 -04:00
if xml:
if not lastempty:
ofp.write("</%s>" % data)
elif data not in knownempties:
if data in autoclose:
pass
elif lastopened == data:
1998-11-23 12:59:39 -04:00
ofp.write("</>")
else:
ofp.write("</%s>" % data)
lastopened = None
lastempty = 0
elif type == "A":
name, type, value = string.split(data, " ", 2)
1998-12-01 15:01:53 -04:00
attrs[name] = esistools.decode(value)
1998-11-23 12:59:39 -04:00
elif type == "e":
knownempty = 1
if LIST_EMPTIES:
knownempties.append("")
if os.path.isfile(EMPTIES_FILENAME):
mode = "a"
else:
mode = "w"
fp = open(EMPTIES_FILENAME, mode)
fp.write(string.join(knownempties, "\n"))
fp.close()
1998-11-23 12:59:39 -04:00
def sgml_convert(ifp, ofp, autoclose):
return do_convert(ifp, ofp, xml=0, autoclose=autoclose)
1998-11-23 12:59:39 -04:00
def xml_convert(ifp, ofp, autoclose):
return do_convert(ifp, ofp, xml=1, autoclose=autoclose)
1999-01-22 18:48:24 -04:00
AUTOCLOSE = ("para", "term",)
1998-11-23 12:59:39 -04:00
def main():
1998-12-01 15:01:53 -04:00
import getopt
1998-11-23 12:59:39 -04:00
import sys
#
autoclose = AUTOCLOSE
1998-11-23 12:59:39 -04:00
convert = sgml_convert
1998-12-01 15:01:53 -04:00
xml = 0
xmldecl = 0
opts, args = getopt.getopt(sys.argv[1:], "adx",
["autoclose", "declare", "xml"])
1998-12-01 15:01:53 -04:00
for opt, arg in opts:
if opt in ("-d", "--declare"):
xmldecl = 1
elif opt in ("-x", "--xml"):
xml = 1
convert = xml_convert
elif opt in ("-a", "--autoclose"):
autoclose = string.split(arg, ",")
1998-12-01 15:01:53 -04:00
if len(args) == 0:
1998-11-23 12:59:39 -04:00
ifp = sys.stdin
ofp = sys.stdout
1998-12-01 15:01:53 -04:00
elif len(args) == 1:
ifp = open(args[0])
1998-11-23 12:59:39 -04:00
ofp = sys.stdout
1998-12-01 15:01:53 -04:00
elif len(args) == 2:
ifp = open(args[0])
ofp = open(args[1], "w")
1998-11-23 12:59:39 -04:00
else:
usage()
sys.exit(2)
# knownempties is ignored in the XML version
try:
1998-12-01 15:01:53 -04:00
if xml and xmldecl:
opf.write('<?xml version="1.0" encoding="iso8859-1"?>\n')
convert(ifp, ofp, autoclose)
1998-11-23 12:59:39 -04:00
except IOError, (err, msg):
if err != errno.EPIPE:
raise
if __name__ == "__main__":
main()