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