Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Commit 607aed7

Browse files
freddrakefreddrake
committed
Some adjustments, mostly to make it more general.
1 parent 4cc902f commit 607aed7

1 file changed

Lines changed: 97 additions & 21 deletions

File tree

‎Doc/tools/sgmlconv/esis2sgml.py‎

Lines changed: 97 additions & 21 deletions
Original file line numberDiff line numberDiff line change
@@ -5,6 +5,10 @@
55
This is limited, but seems sufficient for the ESIS generated by the
66
latex2esis.py script when run over the Python documentation.
77
"""
8+
9+
# This should have an explicit option to indicate whether the *INPUT* was
10+
# generated from an SGML or an XML application.
11+
812
__version__ = '$Revision$'
913

1014
import errno
@@ -16,29 +20,52 @@
1620
from xml.utils import escape
1721

1822

23+
AUTOCLOSE = ()
24+
1925
EMPTIES_FILENAME = "../sgml/empties.dat"
2026
LIST_EMPTIES = 0
2127

2228

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+
2347
def format_attrs(attrs, xml=0):
2448
attrs = attrs.items()
2549
attrs.sort()
26-
s = ''
50+
parts = []
51+
append = parts.append
2752
for name, value in attrs:
2853
if xml:
29-
s = '%s %s="%s"' % (s, name, escape(value))
54+
append('%s="%s"' % (name, escape(value)))
3055
else:
3156
# this is a little bogus, but should do for now
3257
if name == value and isnmtoken(value):
33-
s = "%s %s" % (s, value)
58+
append(value)
3459
elif istoken(value):
3560
if value == "no" + name:
36-
s = "%s %s" % (s, value)
61+
append(value)
3762
else:
38-
s = "%s %s=%s" % (s, name, value)
63+
append("%s=%s" % (name, value))
3964
else:
40-
s = '%s %s="%s"' % (s, name, escape(value))
41-
return s
65+
append('%s="%s"' % (name, escape(value)))
66+
if parts:
67+
parts.insert(0, '')
68+
return string.join(parts)
4269

4370

4471
_nmtoken_rx = re.compile("[a-z][-._a-z0-9]*$", re.IGNORECASE)
@@ -78,6 +105,7 @@ def do_convert(ifp, ofp, xml=0, autoclose=()):
78105
if data == "COMMENT":
79106
ofp.write("<!--")
80107
continue
108+
data = map_gi(data, _elem_map)
81109
if knownempty and xml:
82110
ofp.write("<%s%s/>" % (data, format_attrs(attrs, xml)))
83111
else:
@@ -93,6 +121,7 @@ def do_convert(ifp, ofp, xml=0, autoclose=()):
93121
if data == "COMMENT":
94122
ofp.write("-->")
95123
continue
124+
data = map_gi(data, _elem_map)
96125
if xml:
97126
if not lastempty:
98127
ofp.write("</%s>" % data)
@@ -107,19 +136,24 @@ def do_convert(ifp, ofp, xml=0, autoclose=()):
107136
lastempty = 0
108137
elif type == "A":
109138
name, type, value = string.split(data, " ", 2)
139+
name = map_gi(name, _attr_map)
110140
attrs[name] = esistools.decode(value)
111141
elif type == "e":
112142
knownempty = 1
113143

114144
if LIST_EMPTIES:
115-
knownempties.append("")
116-
if os.path.isfile(EMPTIES_FILENAME):
117-
mode = "a"
118-
else:
119-
mode = "w"
120-
fp = open(EMPTIES_FILENAME, mode)
121-
fp.write(string.join(knownempties, "\n"))
122-
fp.close()
145+
dump_empty_element_names(knownempties)
146+
147+
148+
def dump_empty_element_names(knownempties):
149+
knownempties.append("")
150+
if os.path.isfile(EMPTIES_FILENAME):
151+
mode = "a"
152+
else:
153+
mode = "w"
154+
fp = open(EMPTIES_FILENAME, mode)
155+
fp.write(string.join(knownempties, "\n"))
156+
fp.close()
123157

124158

125159
def sgml_convert(ifp, ofp, autoclose):
@@ -130,27 +164,53 @@ def xml_convert(ifp, ofp, autoclose):
130164
return do_convert(ifp, ofp, xml=1, autoclose=autoclose)
131165

132166

133-
AUTOCLOSE = ("para", "term",)
167+
def update_gi_map(map, names, fromsgml=1):
168+
for name in string.split(names, ","):
169+
if fromsgml:
170+
uncased = string.lower(name)
171+
else:
172+
uncased = name
173+
map[uncased] = name
134174

135175

136176
def main():
137177
import getopt
138178
import sys
139179
#
140180
autoclose = AUTOCLOSE
141-
convert = sgml_convert
142-
xml = 0
181+
convert = xml_convert
182+
xml = 1
143183
xmldecl = 0
144-
opts, args = getopt.getopt(sys.argv[1:], "adx",
145-
["autoclose", "declare", "xml"])
184+
elem_names = ''
185+
attr_names = ''
186+
value_names = ''
187+
opts, args = getopt.getopt(sys.argv[1:], "adesx",
188+
["autoclose=", "declare", "sgml", "xml",
189+
"elements-map=", "attributes-map",
190+
"values-map="])
146191
for opt, arg in opts:
147192
if opt in ("-d", "--declare"):
148193
xmldecl = 1
194+
elif opt == "-e":
195+
global LIST_EMPTIES
196+
LIST_EMPTIES = 1
197+
elif opt in ("-s", "--sgml"):
198+
xml = 0
199+
convert = sgml_convert
149200
elif opt in ("-x", "--xml"):
150201
xml = 1
151202
convert = xml_convert
152203
elif opt in ("-a", "--autoclose"):
153204
autoclose = string.split(arg, ",")
205+
elif opt == "--elements-map":
206+
elem_names = ("%s,%s" % (elem_names, arg))[1:]
207+
elif opt == "--attributes-map":
208+
attr_names = ("%s,%s" % (attr_names, arg))[1:]
209+
elif opt == "--values-map":
210+
value_names = ("%s,%s" % (value_names, arg))[1:]
211+
#
212+
# open input streams:
213+
#
154214
if len(args) == 0:
155215
ifp = sys.stdin
156216
ofp = sys.stdout
@@ -163,7 +223,23 @@ def main():
163223
else:
164224
usage()
165225
sys.exit(2)
166-
# knownempties is ignored in the XML version
226+
#
227+
# setup the name maps:
228+
#
229+
if elem_names or attr_names or value_names:
230+
# assume the origin was SGML; ignore case of the names from the ESIS
231+
# stream but set up conversion tables to get the case right on output
232+
global _normalize_case
233+
_normalize_case = string.lower
234+
update_gi_map(_elem_map, string.split(elem_names, ","))
235+
update_gi_map(_attr_map, string.split(attr_names, ","))
236+
update_gi_map(_values_map, string.split(value_names, ","))
237+
else:
238+
global map_gi
239+
map_gi = null_map_gi
240+
#
241+
# run the conversion:
242+
#
167243
try:
168244
if xml and xmldecl:
169245
opf.write('<?xml version="1.0" encoding="iso8859-1"?>\n')

0 commit comments

Comments
 (0)