blob: c43ddbc56da5fc5312111665d628da1f9e23dfee [file] [log] [blame]
Fred Drake30a68c71998-11-23 16:59:39 +00001#! /usr/bin/env python
2
3"""Convert ESIS events to SGML or XML markup.
4
5This is limited, but seems sufficient for the ESIS generated by the
6latex2esis.py script when run over the Python documentation.
7"""
8__version__ = '$Revision$'
9
10import errno
Fred Drakef077b9d1998-12-01 19:01:53 +000011import esistools
Fred Drake36dfe581999-01-19 23:03:04 +000012import os
Fred Drake30a68c71998-11-23 16:59:39 +000013import re
14import string
15
Fred Drake79ad1f11999-01-14 17:06:09 +000016from xml.utils import escape
17
Fred Drake30a68c71998-11-23 16:59:39 +000018
Fred Drake36dfe581999-01-19 23:03:04 +000019EMPTIES_FILENAME = "../sgml/empties.dat"
20LIST_EMPTIES = 0
21
22
Fred Drakef82e4ab1999-01-19 17:10:31 +000023def format_attrs(attrs, xml=0):
Fred Drake30a68c71998-11-23 16:59:39 +000024 attrs = attrs.items()
25 attrs.sort()
26 s = ''
27 for name, value in attrs:
Fred Drakef82e4ab1999-01-19 17:10:31 +000028 if xml:
29 s = '%s %s="%s"' % (s, name, escape(value))
30 else:
31 # this is a little bogus, but should do for now
32 if name == value and isnmtoken(value):
33 s = "%s %s" % (s, value)
34 elif istoken(value):
35 s = "%s %s=%s" % (s, name, value)
36 else:
37 s = '%s %s="%s"' % (s, name, escape(value))
Fred Drake30a68c71998-11-23 16:59:39 +000038 return s
39
40
Fred Drake36dfe581999-01-19 23:03:04 +000041_nmtoken_rx = re.compile("[a-z][-._a-z0-9]*$", re.IGNORECASE)
Fred Drakef82e4ab1999-01-19 17:10:31 +000042def isnmtoken(s):
43 return _nmtoken_rx.match(s) is not None
44
Fred Drake36dfe581999-01-19 23:03:04 +000045_token_rx = re.compile("[a-z0-9][-._a-z0-9]*$", re.IGNORECASE)
Fred Drakef82e4ab1999-01-19 17:10:31 +000046def istoken(s):
47 return _token_rx.match(s) is not None
48
49
Fred Drake43278f01999-01-20 20:35:05 +000050def do_convert(ifp, ofp, xml=0, autoclose=()):
51 if xml:
52 autoclose = ()
Fred Drake30a68c71998-11-23 16:59:39 +000053 attrs = {}
54 lastopened = None
Fred Drake4abcffb1998-12-10 18:31:37 +000055 knownempties = []
Fred Drake30a68c71998-11-23 16:59:39 +000056 knownempty = 0
57 lastempty = 0
58 while 1:
59 line = ifp.readline()
60 if not line:
61 break
62
63 type = line[0]
64 data = line[1:]
65 if data and data[-1] == "\n":
66 data = data[:-1]
67 if type == "-":
Fred Drakef077b9d1998-12-01 19:01:53 +000068 data = esistools.decode(data)
Fred Drake79ad1f11999-01-14 17:06:09 +000069 ofp.write(escape(data))
Fred Drake30a68c71998-11-23 16:59:39 +000070 if "\n" in data:
71 lastopened = None
72 knownempty = 0
73 lastempty = 0
74 elif type == "(":
Fred Drakef077b9d1998-12-01 19:01:53 +000075 if data == "COMMENT":
76 ofp.write("<!--")
77 continue
Fred Drake30a68c71998-11-23 16:59:39 +000078 if knownempty and xml:
Fred Drakef82e4ab1999-01-19 17:10:31 +000079 ofp.write("<%s%s/>" % (data, format_attrs(attrs, xml)))
Fred Drake30a68c71998-11-23 16:59:39 +000080 else:
Fred Drakef82e4ab1999-01-19 17:10:31 +000081 ofp.write("<%s%s>" % (data, format_attrs(attrs, xml)))
Fred Drake30a68c71998-11-23 16:59:39 +000082 if knownempty and data not in knownempties:
83 # accumulate knowledge!
84 knownempties.append(data)
85 attrs = {}
86 lastopened = data
87 lastempty = knownempty
88 knownempty = 0
89 elif type == ")":
Fred Drakef077b9d1998-12-01 19:01:53 +000090 if data == "COMMENT":
91 ofp.write("-->")
92 continue
Fred Drake30a68c71998-11-23 16:59:39 +000093 if xml:
94 if not lastempty:
95 ofp.write("</%s>" % data)
96 elif data not in knownempties:
Fred Drake43278f01999-01-20 20:35:05 +000097 if data in autoclose:
98 pass
99 elif lastopened == data:
Fred Drake30a68c71998-11-23 16:59:39 +0000100 ofp.write("</>")
101 else:
102 ofp.write("</%s>" % data)
103 lastopened = None
104 lastempty = 0
105 elif type == "A":
106 name, type, value = string.split(data, " ", 2)
Fred Drakef077b9d1998-12-01 19:01:53 +0000107 attrs[name] = esistools.decode(value)
Fred Drake30a68c71998-11-23 16:59:39 +0000108 elif type == "e":
109 knownempty = 1
110
Fred Drake36dfe581999-01-19 23:03:04 +0000111 if LIST_EMPTIES:
112 knownempties.append("")
113 if os.path.isfile(EMPTIES_FILENAME):
114 mode = "a"
115 else:
116 mode = "w"
117 fp = open(EMPTIES_FILENAME, mode)
118 fp.write(string.join(knownempties, "\n"))
119 fp.close()
120
Fred Drake30a68c71998-11-23 16:59:39 +0000121
Fred Drake43278f01999-01-20 20:35:05 +0000122def sgml_convert(ifp, ofp, autoclose):
123 return do_convert(ifp, ofp, xml=0, autoclose=autoclose)
Fred Drake30a68c71998-11-23 16:59:39 +0000124
125
Fred Drake43278f01999-01-20 20:35:05 +0000126def xml_convert(ifp, ofp, autoclose):
127 return do_convert(ifp, ofp, xml=1, autoclose=autoclose)
128
129
Fred Drakea49a4ea1999-01-22 22:48:24 +0000130AUTOCLOSE = ("para", "term",)
Fred Drake30a68c71998-11-23 16:59:39 +0000131
132
133def main():
Fred Drakef077b9d1998-12-01 19:01:53 +0000134 import getopt
Fred Drake30a68c71998-11-23 16:59:39 +0000135 import sys
136 #
Fred Drake43278f01999-01-20 20:35:05 +0000137 autoclose = AUTOCLOSE
Fred Drake30a68c71998-11-23 16:59:39 +0000138 convert = sgml_convert
Fred Drakef077b9d1998-12-01 19:01:53 +0000139 xml = 0
140 xmldecl = 0
Fred Drake43278f01999-01-20 20:35:05 +0000141 opts, args = getopt.getopt(sys.argv[1:], "adx",
142 ["autoclose", "declare", "xml"])
Fred Drakef077b9d1998-12-01 19:01:53 +0000143 for opt, arg in opts:
144 if opt in ("-d", "--declare"):
145 xmldecl = 1
146 elif opt in ("-x", "--xml"):
147 xml = 1
148 convert = xml_convert
Fred Drake43278f01999-01-20 20:35:05 +0000149 elif opt in ("-a", "--autoclose"):
150 autoclose = string.split(arg, ",")
Fred Drakef077b9d1998-12-01 19:01:53 +0000151 if len(args) == 0:
Fred Drake30a68c71998-11-23 16:59:39 +0000152 ifp = sys.stdin
153 ofp = sys.stdout
Fred Drakef077b9d1998-12-01 19:01:53 +0000154 elif len(args) == 1:
155 ifp = open(args[0])
Fred Drake30a68c71998-11-23 16:59:39 +0000156 ofp = sys.stdout
Fred Drakef077b9d1998-12-01 19:01:53 +0000157 elif len(args) == 2:
158 ifp = open(args[0])
159 ofp = open(args[1], "w")
Fred Drake30a68c71998-11-23 16:59:39 +0000160 else:
161 usage()
162 sys.exit(2)
163 # knownempties is ignored in the XML version
164 try:
Fred Drakef077b9d1998-12-01 19:01:53 +0000165 if xml and xmldecl:
166 opf.write('<?xml version="1.0" encoding="iso8859-1"?>\n')
Fred Drake43278f01999-01-20 20:35:05 +0000167 convert(ifp, ofp, autoclose)
Fred Drake30a68c71998-11-23 16:59:39 +0000168 except IOError, (err, msg):
169 if err != errno.EPIPE:
170 raise
171
172
173if __name__ == "__main__":
174 main()