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 | 43278f0 | 1999-01-20 20:35:05 +0000 | [diff] [blame] | 50 | def do_convert(ifp, ofp, xml=0, autoclose=()): |
| 51 | if xml: |
| 52 | autoclose = () |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 53 | attrs = {} |
| 54 | lastopened = None |
Fred Drake | 4abcffb | 1998-12-10 18:31:37 +0000 | [diff] [blame] | 55 | knownempties = [] |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 56 | 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 Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 68 | data = esistools.decode(data) |
Fred Drake | 79ad1f1 | 1999-01-14 17:06:09 +0000 | [diff] [blame] | 69 | ofp.write(escape(data)) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 70 | if "\n" in data: |
| 71 | lastopened = None |
| 72 | knownempty = 0 |
| 73 | lastempty = 0 |
| 74 | elif type == "(": |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 75 | if data == "COMMENT": |
| 76 | ofp.write("<!--") |
| 77 | continue |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 78 | if knownempty and xml: |
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 | else: |
Fred Drake | f82e4ab | 1999-01-19 17:10:31 +0000 | [diff] [blame] | 81 | ofp.write("<%s%s>" % (data, format_attrs(attrs, xml))) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 82 | 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 Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 90 | if data == "COMMENT": |
| 91 | ofp.write("-->") |
| 92 | continue |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 93 | if xml: |
| 94 | if not lastempty: |
| 95 | ofp.write("</%s>" % data) |
| 96 | elif data not in knownempties: |
Fred Drake | 43278f0 | 1999-01-20 20:35:05 +0000 | [diff] [blame] | 97 | if data in autoclose: |
| 98 | pass |
| 99 | elif lastopened == data: |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 100 | 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 Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 107 | attrs[name] = esistools.decode(value) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 108 | elif type == "e": |
| 109 | knownempty = 1 |
| 110 | |
Fred Drake | 36dfe58 | 1999-01-19 23:03:04 +0000 | [diff] [blame] | 111 | 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 Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 121 | |
Fred Drake | 43278f0 | 1999-01-20 20:35:05 +0000 | [diff] [blame] | 122 | def sgml_convert(ifp, ofp, autoclose): |
| 123 | return do_convert(ifp, ofp, xml=0, autoclose=autoclose) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 124 | |
| 125 | |
Fred Drake | 43278f0 | 1999-01-20 20:35:05 +0000 | [diff] [blame] | 126 | def xml_convert(ifp, ofp, autoclose): |
| 127 | return do_convert(ifp, ofp, xml=1, autoclose=autoclose) |
| 128 | |
| 129 | |
Fred Drake | a49a4ea | 1999-01-22 22:48:24 +0000 | [diff] [blame] | 130 | AUTOCLOSE = ("para", "term",) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 131 | |
| 132 | |
| 133 | def main(): |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 134 | import getopt |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 135 | import sys |
| 136 | # |
Fred Drake | 43278f0 | 1999-01-20 20:35:05 +0000 | [diff] [blame] | 137 | autoclose = AUTOCLOSE |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 138 | convert = sgml_convert |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 139 | xml = 0 |
| 140 | xmldecl = 0 |
Fred Drake | 43278f0 | 1999-01-20 20:35:05 +0000 | [diff] [blame] | 141 | opts, args = getopt.getopt(sys.argv[1:], "adx", |
| 142 | ["autoclose", "declare", "xml"]) |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 143 | 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 Drake | 43278f0 | 1999-01-20 20:35:05 +0000 | [diff] [blame] | 149 | elif opt in ("-a", "--autoclose"): |
| 150 | autoclose = string.split(arg, ",") |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 151 | if len(args) == 0: |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 152 | ifp = sys.stdin |
| 153 | ofp = sys.stdout |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 154 | elif len(args) == 1: |
| 155 | ifp = open(args[0]) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 156 | ofp = sys.stdout |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 157 | elif len(args) == 2: |
| 158 | ifp = open(args[0]) |
| 159 | ofp = open(args[1], "w") |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 160 | else: |
| 161 | usage() |
| 162 | sys.exit(2) |
| 163 | # knownempties is ignored in the XML version |
| 164 | try: |
Fred Drake | f077b9d | 1998-12-01 19:01:53 +0000 | [diff] [blame] | 165 | if xml and xmldecl: |
| 166 | opf.write('<?xml version="1.0" encoding="iso8859-1"?>\n') |
Fred Drake | 43278f0 | 1999-01-20 20:35:05 +0000 | [diff] [blame] | 167 | convert(ifp, ofp, autoclose) |
Fred Drake | 30a68c7 | 1998-11-23 16:59:39 +0000 | [diff] [blame] | 168 | except IOError, (err, msg): |
| 169 | if err != errno.EPIPE: |
| 170 | raise |
| 171 | |
| 172 | |
| 173 | if __name__ == "__main__": |
| 174 | main() |