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 | 36dfe58 | 1999-01-19 23:03:04 +0000 | [diff] [blame^] | 12 | import os |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 13 | import re |
| 14 | import string |
| 15 | |
Fred Drake | 79ad1f1 | 1999-01-14 17:06:09 +0000 | [diff] [blame] | 16 | from xml.utils import escape |
| 17 | |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 18 | |
Fred Drake | 36dfe58 | 1999-01-19 23:03:04 +0000 | [diff] [blame^] | 19 | EMPTIES_FILENAME = "../sgml/empties.dat" |
| 20 | LIST_EMPTIES = 0 |
| 21 | |
| 22 | |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame] | 23 | def format_attrs(attrs, xml=0): |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 24 | attrs = attrs.items() |
| 25 | attrs.sort() |
| 26 | s = '' |
| 27 | for name, value in attrs: |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame] | 28 | 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 Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 38 | return s |
| 39 | |
| 40 | |
Fred Drake | 36dfe58 | 1999-01-19 23:03:04 +0000 | [diff] [blame^] | 41 | _nmtoken_rx = re.compile("[a-z][-._a-z0-9]*$", re.IGNORECASE) |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame] | 42 | def isnmtoken(s): |
| 43 | return _nmtoken_rx.match(s) is not None |
| 44 | |
Fred Drake | 36dfe58 | 1999-01-19 23:03:04 +0000 | [diff] [blame^] | 45 | _token_rx = re.compile("[a-z0-9][-._a-z0-9]*$", re.IGNORECASE) |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame] | 46 | def istoken(s): |
| 47 | return _token_rx.match(s) is not None |
| 48 | |
| 49 | |
Fred Drake | 4abcffb | 1998-12-10 18:31:37 +0000 | [diff] [blame] | 50 | def do_convert(ifp, ofp, xml=0): |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 51 | attrs = {} |
| 52 | lastopened = None |
Fred Drake | 4abcffb | 1998-12-10 18:31:37 +0000 | [diff] [blame] | 53 | knownempties = [] |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 54 | knownempty = 0 |
| 55 | lastempty = 0 |
| 56 | while 1: |
| 57 | line = ifp.readline() |
| 58 | if not line: |
| 59 | break |
| 60 | |
| 61 | type = line[0] |
| 62 | data = line[1:] |
| 63 | if data and data[-1] == "\n": |
| 64 | data = data[:-1] |
| 65 | if type == "-": |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 66 | data = esistools.decode(data) |
Fred Drake | 79ad1f1 | 1999-01-14 17:06:09 +0000 | [diff] [blame] | 67 | ofp.write(escape(data)) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 68 | if "\n" in data: |
| 69 | lastopened = None |
| 70 | knownempty = 0 |
| 71 | lastempty = 0 |
| 72 | elif type == "(": |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 73 | if data == "COMMENT": |
| 74 | ofp.write("<!--") |
| 75 | continue |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 76 | if knownempty and xml: |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame] | 77 | ofp.write("<%s%s/>" % (data, format_attrs(attrs, xml))) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 78 | else: |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame] | 79 | ofp.write("<%s%s>" % (data, format_attrs(attrs, xml))) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 80 | if knownempty and data not in knownempties: |
| 81 | # accumulate knowledge! |
| 82 | knownempties.append(data) |
| 83 | attrs = {} |
| 84 | lastopened = data |
| 85 | lastempty = knownempty |
| 86 | knownempty = 0 |
| 87 | elif type == ")": |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 88 | if data == "COMMENT": |
| 89 | ofp.write("-->") |
| 90 | continue |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 91 | if xml: |
| 92 | if not lastempty: |
| 93 | ofp.write("</%s>" % data) |
| 94 | elif data not in knownempties: |
| 95 | if lastopened == data: |
| 96 | ofp.write("</>") |
| 97 | else: |
| 98 | ofp.write("</%s>" % data) |
| 99 | lastopened = None |
| 100 | lastempty = 0 |
| 101 | elif type == "A": |
| 102 | name, type, value = string.split(data, " ", 2) |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 103 | attrs[name] = esistools.decode(value) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 104 | elif type == "e": |
| 105 | knownempty = 1 |
| 106 | |
Fred Drake | 36dfe58 | 1999-01-19 23:03:04 +0000 | [diff] [blame^] | 107 | if LIST_EMPTIES: |
| 108 | knownempties.append("") |
| 109 | if os.path.isfile(EMPTIES_FILENAME): |
| 110 | mode = "a" |
| 111 | else: |
| 112 | mode = "w" |
| 113 | fp = open(EMPTIES_FILENAME, mode) |
| 114 | fp.write(string.join(knownempties, "\n")) |
| 115 | fp.close() |
| 116 | |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 117 | |
Fred Drake | 4abcffb | 1998-12-10 18:31:37 +0000 | [diff] [blame] | 118 | def sgml_convert(ifp, ofp): |
| 119 | return do_convert(ifp, ofp, xml=0) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 120 | |
| 121 | |
Fred Drake | 4abcffb | 1998-12-10 18:31:37 +0000 | [diff] [blame] | 122 | def xml_convert(ifp, ofp): |
| 123 | return do_convert(ifp, ofp, xml=1) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 124 | |
| 125 | |
| 126 | def main(): |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 127 | import getopt |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 128 | import sys |
| 129 | # |
| 130 | convert = sgml_convert |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 131 | xml = 0 |
| 132 | xmldecl = 0 |
| 133 | opts, args = getopt.getopt(sys.argv[1:], "dx", ["declare", "xml"]) |
| 134 | for opt, arg in opts: |
| 135 | if opt in ("-d", "--declare"): |
| 136 | xmldecl = 1 |
| 137 | elif opt in ("-x", "--xml"): |
| 138 | xml = 1 |
| 139 | convert = xml_convert |
| 140 | if len(args) == 0: |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 141 | ifp = sys.stdin |
| 142 | ofp = sys.stdout |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 143 | elif len(args) == 1: |
| 144 | ifp = open(args[0]) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 145 | ofp = sys.stdout |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 146 | elif len(args) == 2: |
| 147 | ifp = open(args[0]) |
| 148 | ofp = open(args[1], "w") |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 149 | else: |
| 150 | usage() |
| 151 | sys.exit(2) |
| 152 | # knownempties is ignored in the XML version |
| 153 | try: |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 154 | if xml and xmldecl: |
| 155 | opf.write('<?xml version="1.0" encoding="iso8859-1"?>\n') |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 156 | convert(ifp, ofp) |
| 157 | except IOError, (err, msg): |
| 158 | if err != errno.EPIPE: |
| 159 | raise |
| 160 | |
| 161 | |
| 162 | if __name__ == "__main__": |
| 163 | main() |