Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 1 | #! /usr/bin/env python |
| 2 | |
| 3 | """Convert ESIS events to SGML or XML markup. |
| 4 | |
| 5 | This is limited, but seems sufficient for the ESIS generated by the |
| 6 | latex2esis.py script when run over the Python documentation. |
| 7 | """ |
| 8 | __version__ = '$Revision$' |
| 9 | |
| 10 | import errno |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 11 | import esistools |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 12 | import re |
| 13 | import string |
| 14 | |
Fred Drake | 79ad1f1 | 1999-01-14 17:06:09 +0000 | [diff] [blame] | 15 | from xml.utils import escape |
| 16 | |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 17 | |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame^] | 18 | def format_attrs(attrs, xml=0): |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 19 | attrs = attrs.items() |
| 20 | attrs.sort() |
| 21 | s = '' |
| 22 | for name, value in attrs: |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame^] | 23 | if xml: |
| 24 | s = '%s %s="%s"' % (s, name, escape(value)) |
| 25 | else: |
| 26 | # this is a little bogus, but should do for now |
| 27 | if name == value and isnmtoken(value): |
| 28 | s = "%s %s" % (s, value) |
| 29 | elif istoken(value): |
| 30 | s = "%s %s=%s" % (s, name, value) |
| 31 | else: |
| 32 | s = '%s %s="%s"' % (s, name, escape(value)) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 33 | return s |
| 34 | |
| 35 | |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame^] | 36 | _nmtoken_rx = re.compile("[a-z][-._a-z0-9]*", re.IGNORECASE) |
| 37 | def isnmtoken(s): |
| 38 | return _nmtoken_rx.match(s) is not None |
| 39 | |
| 40 | _token_rx = re.compile("[a-z0-9][-._a-z0-9]*", re.IGNORECASE) |
| 41 | def istoken(s): |
| 42 | return _token_rx.match(s) is not None |
| 43 | |
| 44 | |
Fred Drake | 4abcffb | 1998-12-10 18:31:37 +0000 | [diff] [blame] | 45 | def do_convert(ifp, ofp, xml=0): |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 46 | attrs = {} |
| 47 | lastopened = None |
Fred Drake | 4abcffb | 1998-12-10 18:31:37 +0000 | [diff] [blame] | 48 | knownempties = [] |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 49 | knownempty = 0 |
| 50 | lastempty = 0 |
| 51 | while 1: |
| 52 | line = ifp.readline() |
| 53 | if not line: |
| 54 | break |
| 55 | |
| 56 | type = line[0] |
| 57 | data = line[1:] |
| 58 | if data and data[-1] == "\n": |
| 59 | data = data[:-1] |
| 60 | if type == "-": |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 61 | data = esistools.decode(data) |
Fred Drake | 79ad1f1 | 1999-01-14 17:06:09 +0000 | [diff] [blame] | 62 | ofp.write(escape(data)) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 63 | if "\n" in data: |
| 64 | lastopened = None |
| 65 | knownempty = 0 |
| 66 | lastempty = 0 |
| 67 | elif type == "(": |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 68 | if data == "COMMENT": |
| 69 | ofp.write("<!--") |
| 70 | continue |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 71 | if knownempty and xml: |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame^] | 72 | ofp.write("<%s%s/>" % (data, format_attrs(attrs, xml))) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 73 | else: |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame^] | 74 | ofp.write("<%s%s>" % (data, format_attrs(attrs, xml))) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 75 | if knownempty and data not in knownempties: |
| 76 | # accumulate knowledge! |
| 77 | knownempties.append(data) |
| 78 | attrs = {} |
| 79 | lastopened = data |
| 80 | lastempty = knownempty |
| 81 | knownempty = 0 |
| 82 | elif type == ")": |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 83 | if data == "COMMENT": |
| 84 | ofp.write("-->") |
| 85 | continue |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 86 | if xml: |
| 87 | if not lastempty: |
| 88 | ofp.write("</%s>" % data) |
| 89 | elif data not in knownempties: |
| 90 | if lastopened == data: |
| 91 | ofp.write("</>") |
| 92 | else: |
| 93 | ofp.write("</%s>" % data) |
| 94 | lastopened = None |
| 95 | lastempty = 0 |
| 96 | elif type == "A": |
| 97 | name, type, value = string.split(data, " ", 2) |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 98 | attrs[name] = esistools.decode(value) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 99 | elif type == "e": |
| 100 | knownempty = 1 |
| 101 | |
| 102 | |
Fred Drake | 4abcffb | 1998-12-10 18:31:37 +0000 | [diff] [blame] | 103 | def sgml_convert(ifp, ofp): |
| 104 | return do_convert(ifp, ofp, xml=0) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 105 | |
| 106 | |
Fred Drake | 4abcffb | 1998-12-10 18:31:37 +0000 | [diff] [blame] | 107 | def xml_convert(ifp, ofp): |
| 108 | return do_convert(ifp, ofp, xml=1) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 109 | |
| 110 | |
| 111 | def main(): |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 112 | import getopt |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 113 | import sys |
| 114 | # |
| 115 | convert = sgml_convert |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 116 | xml = 0 |
| 117 | xmldecl = 0 |
| 118 | opts, args = getopt.getopt(sys.argv[1:], "dx", ["declare", "xml"]) |
| 119 | for opt, arg in opts: |
| 120 | if opt in ("-d", "--declare"): |
| 121 | xmldecl = 1 |
| 122 | elif opt in ("-x", "--xml"): |
| 123 | xml = 1 |
| 124 | convert = xml_convert |
| 125 | if len(args) == 0: |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 126 | ifp = sys.stdin |
| 127 | ofp = sys.stdout |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 128 | elif len(args) == 1: |
| 129 | ifp = open(args[0]) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 130 | ofp = sys.stdout |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 131 | elif len(args) == 2: |
| 132 | ifp = open(args[0]) |
| 133 | ofp = open(args[1], "w") |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 134 | else: |
| 135 | usage() |
| 136 | sys.exit(2) |
| 137 | # knownempties is ignored in the XML version |
| 138 | try: |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 139 | if xml and xmldecl: |
| 140 | opf.write('<?xml version="1.0" encoding="iso8859-1"?>\n') |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 141 | convert(ifp, ofp) |
| 142 | except IOError, (err, msg): |
| 143 | if err != errno.EPIPE: |
| 144 | raise |
| 145 | |
| 146 | |
| 147 | if __name__ == "__main__": |
| 148 | main() |