gen-punycode-examples.py [plain text]
import re
import string
import sys
import generate
if len(sys.argv) != 3:
print "usage: %s rfc3492.txt" % sys.argv[0]
sys.exit(1)
f = open(sys.argv[1], 'r')
examples_h = generate.Header('%s/punycode_examples.h' % sys.argv[2])
examples_c = generate.Header('%s/punycode_examples.c' % sys.argv[2])
start = False
while True:
l = f.readline()
if not l:
break
if l[-2:] == "\\\n":
l2 = f.readline()
if not l2:
raise Exception("EOF in backslash escape")
l2 = re.sub('^ *', '', l2)
l = l[:-2] + l2
if start:
if re.match('7\.2', l):
start = False
else:
m = re.search('^ *\([A-Z]\) *(.*)$', l);
if m:
desc = m.group(1)
codes = []
else:
m = re.search('^ *([uU]+.*) *$', l)
if m:
codes.extend(string.split(m.group(1), ' '))
else:
m = re.search('^ *Punycode: (.*) *$', l)
if m:
cases.append([codes, m.group(1), desc])
else:
if re.match('^7\.1', l):
start = True
cases = []
f.close()
examples_h.file.write(
'''
#include <krb5-types.h>
#define MAX_LENGTH 40
struct punycode_example {
size_t len;
uint32_t val[MAX_LENGTH];
const char *pc;
const char *description;
};
extern const struct punycode_example punycode_examples[];
extern const size_t punycode_examples_size;
''')
examples_c.file.write(
'''
#include <stdlib.h>
#include "punycode_examples.h"
const struct punycode_example punycode_examples[] = {
''')
for x in cases:
[cp, pc, desc] = x
examples_c.file.write(
" {%u, {%s}, \"%s\", \"%s\"},\n" %
(len(cp),
string.join([re.sub('[uU]\+', '0x', x) for x in cp], ', '),
pc,
desc))
examples_c.file.write(
'''};
''')
examples_c.file.write(
"const size_t punycode_examples_size = %u;\n\n" % len(cases))
examples_h.close()
examples_c.close()