Doc/tools/sgmlconv/esis2sgml.py

   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
   9 # This should have an explicit option to indicate whether the *INPUT* was
  10 # generated from an SGML or an XML application.
  11
  12 __version__ = '$Revision$'
  13
  14 import errno
  15 import esistools
  16 import os
  17 import re
  18 import string
  19
  20 from xml.utils import escape
  21
  22
  23 AUTOCLOSE = ()
  24
  25 EMPTIES_FILENAME = "../sgml/empties.dat"
  26 LIST_EMPTIES = 0
  27
  28
  29 _elem_map = {}
  30 _attr_map = {}
  31 _token_map = {}
  32
  33 _normalize_case = str
  34
  35 def map_gi(sgmlgi, map):
  36     uncased = _normalize_case(sgmlgi)
  37     try:
  38         return map[uncased]
  39     except IndexError:
  40         map[uncased] = sgmlgi
  41         return sgmlgi
  42
  43 def null_map_gi(sgmlgi, map):
  44     return sgmlgi
  45
  46
  47 def format_attrs(attrs, xml=0):
  48     attrs = attrs.items()
  49     attrs.sort()
  50     parts = []
  51     append = parts.append
  52     for name, value in attrs:
  53         if xml:
  54             append('%s="%s"' % (name, escape(value)))
  55         else:
  56             # this is a little bogus, but should do for now
  57             if name == value and isnmtoken(value):
  58                 append(value)
  59             elif istoken(value):
  60                 if value == "no" + name:
  61                     append(value)
  62                 else:
  63                     append("%s=%s" % (name, value))
  64             else:
  65                 append('%s="%s"' % (name, escape(value)))
  66     if parts:
  67         parts.insert(0, '')
  68     return string.join(parts)
  69
  70
  71 _nmtoken_rx = re.compile("[a-z][-._a-z0-9]*$", re.IGNORECASE)
  72 def isnmtoken(s):
  73     return _nmtoken_rx.match(s) is not None
  74
  75 _token_rx = re.compile("[a-z0-9][-._a-z0-9]*$", re.IGNORECASE)
  76 def istoken(s):
  77     return _token_rx.match(s) is not None
  78
  79
  80 def convert(ifp, ofp, xml=0, autoclose=(), verbatims=()):
  81     if xml:
  82         autoclose = ()
  83     attrs = {}
  84     lastopened = None
  85     knownempties = []
  86     knownempty = 0
  87     lastempty = 0
  88     inverbatim = 0
  89     while 1:
  90         line = ifp.readline()
  91         if not line:
  92             break
  93
  94         type = line[0]
  95         data = line[1:]
  96         if data and data[-1] == "\n":
  97             data = data[:-1]
  98         if type == "-":
  99             data = esistools.decode(data)
 100             data = escape(data)
 101             if not inverbatim:
 102                 data = string.replace(data, "---", "&mdash;")
 103             ofp.write(data)
 104             if "\n" in data:
 105                 lastopened = None
 106             knownempty = 0
 107             lastempty = 0
 108         elif type == "(":
 109             if data == "COMMENT":
 110                 ofp.write("<!--")
 111                 continue
 112             data = map_gi(data, _elem_map)
 113             if knownempty and xml:
 114                 ofp.write("<%s%s/>" % (data, format_attrs(attrs, xml)))
 115             else:
 116                 ofp.write("<%s%s>" % (data, format_attrs(attrs, xml)))
 117             if knownempty and data not in knownempties:
 118                 # accumulate knowledge!
 119                 knownempties.append(data)
 120             attrs = {}
 121             lastopened = data
 122             lastempty = knownempty
 123             knownempty = 0
 124             inverbatim = data in verbatims
 125         elif type == ")":
 126             if data == "COMMENT":
 127                 ofp.write("-->")
 128                 continue
 129             data = map_gi(data, _elem_map)
 130             if xml:
 131                 if not lastempty:
 132                     ofp.write("</%s>" % data)
 133             elif data not in knownempties:
 134                 if data in autoclose:
 135                     pass
 136                 elif lastopened == data:
 137                     ofp.write("</>")
 138                 else:
 139                     ofp.write("</%s>" % data)
 140             lastopened = None
 141             lastempty = 0
 142             inverbatim = 0
 143         elif type == "A":
 144             name, type, value = string.split(data, " ", 2)
 145             name = map_gi(name, _attr_map)
 146             attrs[name] = esistools.decode(value)
 147         elif type == "e":
 148             knownempty = 1
 149         elif type == "&":
 150             ofp.write("&%s;" % data)
 151             knownempty = 0
 152         else:
 153             raise RuntimeError, "unrecognized ESIS event type: '%s'" % type
 154
 155     if LIST_EMPTIES:
 156         dump_empty_element_names(knownempties)
 157
 158
 159 def dump_empty_element_names(knownempties):
 160     d = {}
 161     for gi in knownempties:
 162         d[gi] = gi
 163     knownempties.append("")
 164     if os.path.isfile(EMPTIES_FILENAME):
 165         fp = open(EMPTIES_FILENAME)
 166         while 1:
 167             line = fp.readline()
 168             if not line:
 169                 break
 170             gi = string.strip(line)
 171             if gi:
 172                 d[gi] = gi
 173     fp = open(EMPTIES_FILENAME, "w")
 174     gilist = d.keys()
 175     gilist.sort()
 176     fp.write(string.join(gilist, "\n"))
 177     fp.write("\n")
 178     fp.close()
 179
 180
 181 def update_gi_map(map, names, fromsgml=1):
 182     for name in string.split(names, ","):
 183         if fromsgml:
 184             uncased = string.lower(name)
 185         else:
 186             uncased = name
 187         map[uncased] = name
 188
 189
 190 def main():
 191     import getopt
 192     import sys
 193     #
 194     autoclose = AUTOCLOSE
 195     xml = 1
 196     xmldecl = 0
 197     elem_names = ''
 198     attr_names = ''
 199     value_names = ''
 200     verbatims = ('verbatim', 'interactive-session')
 201     opts, args = getopt.getopt(sys.argv[1:], "adesx",
 202                                ["autoclose=", "declare", "sgml", "xml",
 203                                 "elements-map=", "attributes-map",
 204                                 "values-map="])
 205     for opt, arg in opts:
 206         if opt in ("-d", "--declare"):
 207             xmldecl = 1
 208         elif opt == "-e":
 209             global LIST_EMPTIES
 210             LIST_EMPTIES = 1
 211         elif opt in ("-s", "--sgml"):
 212             xml = 0
 213         elif opt in ("-x", "--xml"):
 214             xml = 1
 215         elif opt in ("-a", "--autoclose"):
 216             autoclose = string.split(arg, ",")
 217         elif opt == "--elements-map":
 218             elem_names = ("%s,%s" % (elem_names, arg))[1:]
 219         elif opt == "--attributes-map":
 220             attr_names = ("%s,%s" % (attr_names, arg))[1:]
 221         elif opt == "--values-map":
 222             value_names = ("%s,%s" % (value_names, arg))[1:]
 223     #
 224     # open input streams:
 225     #
 226     if len(args) == 0:
 227         ifp = sys.stdin
 228         ofp = sys.stdout
 229     elif len(args) == 1:
 230         ifp = open(args[0])
 231         ofp = sys.stdout
 232     elif len(args) == 2:
 233         ifp = open(args[0])
 234         ofp = open(args[1], "w")
 235     else:
 236         usage()
 237         sys.exit(2)
 238     #
 239     # setup the name maps:
 240     #
 241     if elem_names or attr_names or value_names:
 242         # assume the origin was SGML; ignore case of the names from the ESIS
 243         # stream but set up conversion tables to get the case right on output
 244         global _normalize_case
 245         _normalize_case = string.lower
 246         update_gi_map(_elem_map, string.split(elem_names, ","))
 247         update_gi_map(_attr_map, string.split(attr_names, ","))
 248         update_gi_map(_values_map, string.split(value_names, ","))
 249     else:
 250         global map_gi
 251         map_gi = null_map_gi
 252     #
 253     # run the conversion:
 254     #
 255     try:
 256         if xml and xmldecl:
 257             opf.write('<?xml version="1.0" encoding="iso8859-1"?>\n')
 258         convert(ifp, ofp, xml=xml, autoclose=autoclose, verbatims=verbatims)
 259     except IOError, (err, msg):
 260         if err != errno.EPIPE:
 261             raise
 262
 263
 264 if __name__ == "__main__":
 265     main()