cpython/Doc/tools/sgmlconv/esis2sgml.py

254 lines
6.7 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.
"""
# This should have an explicit option to indicate whether the *INPUT* was
# generated from an SGML or an XML application.
1998-11-23 12:59:39 -04:00
__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
AUTOCLOSE = ()
EMPTIES_FILENAME = "../sgml/empties.dat"
LIST_EMPTIES = 0
_elem_map = {}
_attr_map = {}
_token_map = {}
_normalize_case = str
def map_gi(sgmlgi, map):
uncased = _normalize_case(sgmlgi)
try:
return map[uncased]
except IndexError:
map[uncased] = sgmlgi
return sgmlgi
def null_map_gi(sgmlgi, map):
return sgmlgi
def format_attrs(attrs, xml=0):
1998-11-23 12:59:39 -04:00
attrs = attrs.items()
attrs.sort()
parts = []
append = parts.append
1998-11-23 12:59:39 -04:00
for name, value in attrs:
if xml:
append('%s="%s"' % (name, escape(value)))
else:
# this is a little bogus, but should do for now
if name == value and isnmtoken(value):
append(value)
elif istoken(value):
if value == "no" + name:
append(value)
else:
append("%s=%s" % (name, value))
else:
append('%s="%s"' % (name, escape(value)))
if parts:
parts.insert(0, '')
return string.join(parts)
1998-11-23 12:59:39 -04:00
_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
data = map_gi(data, _elem_map)
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
data = map_gi(data, _elem_map)
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)
name = map_gi(name, _attr_map)
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:
dump_empty_element_names(knownempties)
def dump_empty_element_names(knownempties):
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)
def update_gi_map(map, names, fromsgml=1):
for name in string.split(names, ","):
if fromsgml:
uncased = string.lower(name)
else:
uncased = name
map[uncased] = name
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
convert = xml_convert
xml = 1
1998-12-01 15:01:53 -04:00
xmldecl = 0
elem_names = ''
attr_names = ''
value_names = ''
opts, args = getopt.getopt(sys.argv[1:], "adesx",
["autoclose=", "declare", "sgml", "xml",
"elements-map=", "attributes-map",
"values-map="])
1998-12-01 15:01:53 -04:00
for opt, arg in opts:
if opt in ("-d", "--declare"):
xmldecl = 1
elif opt == "-e":
global LIST_EMPTIES
LIST_EMPTIES = 1
elif opt in ("-s", "--sgml"):
xml = 0
convert = sgml_convert
1998-12-01 15:01:53 -04:00
elif opt in ("-x", "--xml"):
xml = 1
convert = xml_convert
elif opt in ("-a", "--autoclose"):
autoclose = string.split(arg, ",")
elif opt == "--elements-map":
elem_names = ("%s,%s" % (elem_names, arg))[1:]
elif opt == "--attributes-map":
attr_names = ("%s,%s" % (attr_names, arg))[1:]
elif opt == "--values-map":
value_names = ("%s,%s" % (value_names, arg))[1:]
#
# open input streams:
#
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)
#
# setup the name maps:
#
if elem_names or attr_names or value_names:
# assume the origin was SGML; ignore case of the names from the ESIS
# stream but set up conversion tables to get the case right on output
global _normalize_case
_normalize_case = string.lower
update_gi_map(_elem_map, string.split(elem_names, ","))
update_gi_map(_attr_map, string.split(attr_names, ","))
update_gi_map(_values_map, string.split(value_names, ","))
else:
global map_gi
map_gi = null_map_gi
#
# run the conversion:
#
1998-11-23 12:59:39 -04:00
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()