import sys
import string
import time
def makeRange(lst):
ret = []
pos = 0
while pos < len(lst):
try: s = lst[pos:].index(1) except:
break pos += s try:
e = lst[pos:].index(0) e += pos
except: e = len(lst)
ret.append((pos, e-1)) pos = e + 1 return ret
sources = "chvalid.def"
minTableSize = 6
Functs = {}
state = 0
try:
defines = open("chvalid.def", "r")
except:
print "Missing chvalid.def, aborting ..."
sys.exit(1)
for line in defines.readlines():
if line[0] == '#':
continue
line = string.strip(line)
if line == '':
continue
try:
fields = string.split(line, ' ')
if fields[0] == 'name':
name = fields[1]
if state != 0:
print "'name' %s found before previous name" \
"completed" % (fields[1])
continue
state = 1
if Functs.has_key(name):
print "name '%s' already present - may give" \
" wrong results" % (name)
else:
Functs[name] = [ [], [] ]
for v in range(256):
Functs[name][0].append(0)
elif fields[0] == 'end':
if state == 0:
print "'end' found outside of function block"
continue
state = 0
elif fields[0] == 'ur':
if state != 1:
raise ValidationError, "'ur' found outside of 'name' block"
for el in fields[1:]:
pos = string.find(el, '..')
if pos <= 0:
if el[0:2] == '0x':
value = int(el[2:],16)
elif el[0] == "'":
value = ord(el[1])
else:
value = int(el)
if ((value < 0) | (value > 0x1fffff)):
raise ValidationError, 'Illegal value (%s) in ch for'\
' name %s' % (el,name)
currange = (value, value)
else:
(first, last) = string.split(el, "..")
if first[0:2] == '0x':
start = int(first[2:],16)
elif first[0] == "'":
start = ord(first[1])
else:
start = int(first)
if last[0:2] == '0x':
end = int(last[2:],16)
elif last[0] == "'":
end = ord(last[1])
else:
end = int(last)
if (start < 0) | (end > 0x1fffff) | (start > end):
raise ValidationError, "Invalid range '%s'" % el
currange = (start, end)
if currange[1] < 0x100: for ch in range(currange[0],currange[1]+1):
if Functs[name][0][ch]:
msg = "Duplicate ch value '%s' for name '%s'" % (el, name)
raise ValidationError, msg
Functs[name][0][ch] = 1
else: if currange in Functs[name][1]:
raise ValidationError, "range already defined in" \
" function"
else:
Functs[name][1].append(currange)
except:
print "Failed to process line: %s" % (line)
raise
try:
header = open("include/libxml/chvalid.h", "w")
except:
print "Failed to open include/libxml/chvalid.h"
sys.exit(1)
try:
output = open("chvalid.c", "w")
except:
print "Failed to open chvalid.c"
sys.exit(1)
date = time.asctime(time.localtime(time.time()))
header.write(
"""/*
* Summary: Unicode character range checking
* Description: this module exports interfaces for the character
* range validation APIs
*
* This file is automatically generated from the cvs source
* definition files using the genChRanges.py Python script
*
* Generation date: %s
* Sources: %s
* Author: William Brack <wbrack@mmm.com.hk>
*/
#ifndef __XML_CHVALID_H__
#define __XML_CHVALID_H__
#include <libxml/xmlversion.h>
#include <libxml/xmlstring.h>
#ifdef __cplusplus
extern "C" {
#endif
/*
* Define our typedefs and structures
*
*/
typedef struct _xmlChSRange xmlChSRange;
typedef xmlChSRange *xmlChSRangePtr;
struct _xmlChSRange {
unsigned short low;
unsigned short high;
};
typedef struct _xmlChLRange xmlChLRange;
typedef xmlChLRange *xmlChLRangePtr;
struct _xmlChLRange {
unsigned int low;
unsigned int high;
};
typedef struct _xmlChRangeGroup xmlChRangeGroup;
typedef xmlChRangeGroup *xmlChRangeGroupPtr;
struct _xmlChRangeGroup {
int nbShortRange;
int nbLongRange;
const xmlChSRange *shortRange; /* points to an array of ranges */
const xmlChLRange *longRange;
};
/**
* Range checking routine
*/
XMLPUBFUN int XMLCALL
xmlCharInRange(unsigned int val, const xmlChRangeGroup *group);
""" % (date, sources));
output.write(
"""/*
* chvalid.c: this module implements the character range
* validation APIs
*
* This file is automatically generated from the cvs source
* definition files using the genChRanges.py Python script
*
* Generation date: %s
* Sources: %s
* William Brack <wbrack@mmm.com.hk>
*/
#define IN_LIBXML
#include "libxml.h"
#include <libxml/chvalid.h>
/*
* The initial tables ({func_name}_tab) are used to validate whether a
* single-byte character is within the specified group. Each table
* contains 256 bytes, with each byte representing one of the 256
* possible characters. If the table byte is set, the character is
* allowed.
*
*/
""" % (date, sources));
fkeys = Functs.keys() fkeys.sort()
for f in fkeys:
if max(Functs[f][0]) > 0: rangeTable = makeRange(Functs[f][0])
numRanges = len(rangeTable)
if numRanges >= minTableSize: header.write("XMLPUBVAR const unsigned char %s_tab[256];\n" % f)
header.write("""
/**
* %s_ch:
* @c: char to validate
*
* Automatically generated by genChRanges.py
*/
""" % f)
header.write("#define %s_ch(c)\t(%s_tab[(c)])\n" % (f, f))
output.write("const unsigned char %s_tab[256] = {\n" % f)
pline = " "
for n in range(255):
pline += " 0x%02x," % Functs[f][0][n]
if len(pline) > 72:
output.write(pline + "\n")
pline = " "
output.write(pline + " 0x%02x };\n\n" % Functs[f][0][255])
else: try:
ix = rangeTable.remove((0x20, 0x20))
rangeTable.insert(0, (0x20, 0x20))
except:
pass
firstFlag = 1
header.write("""
/**
* %s_ch:
* @c: char to validate
*
* Automatically generated by genChRanges.py
*/
""" % f)
pline = "#define %s_ch(c)" % f
ntab = 4 - (len(pline)) / 8
if ntab < 0:
ntab = 0
just = ""
for i in range(ntab):
just += "\t"
pline = pline + just + "("
for rg in rangeTable:
if not firstFlag:
pline += " || \\\n\t\t\t\t "
else:
firstFlag = 0
if rg[0] == rg[1]: pline += "((c) == 0x%x)" % rg[0]
else: if rg[1] != 0xff:
pline += "((0x%x <= (c)) &&" % rg[0]
pline += " ((c) <= 0x%x))" % rg[1]
else:
pline += " (0x%x <= (c))" % rg[0]
pline += ")\n"
header.write(pline)
header.write("""
/**
* %sQ:
* @c: char to validate
*
* Automatically generated by genChRanges.py
*/
""" % f)
pline = "#define %sQ(c)" % f
ntab = 4 - (len(pline)) / 8
if ntab < 0:
ntab = 0
just = ""
for i in range(ntab):
just += "\t"
header.write(pline + just + "(((c) < 0x100) ? \\\n\t\t\t\t ")
if max(Functs[f][0]) > 0:
header.write("%s_ch((c)) :" % f)
else:
header.write("0 :")
numRanges = len(Functs[f][1])
if numRanges == 0:
header.write(" 0)\n\n")
else:
if numRanges >= minTableSize:
header.write(" \\\n\t\t\t\t xmlCharInRange((c), &%sGroup))\n\n" % f)
else: firstFlag = 1
for rg in Functs[f][1]:
if not firstFlag:
pline += " || \\\n\t\t\t\t "
else:
firstFlag = 0
pline = "\\\n\t\t\t\t("
if rg[0] == rg[1]: pline += "((c) == 0x%x)" % rg[0]
else: pline += "((0x%x <= (c)) &&" % rg[0]
pline += " ((c) <= 0x%x))" % rg[1]
pline += "))\n\n"
header.write(pline)
if len(Functs[f][1]) > 0:
header.write("XMLPUBVAR const xmlChRangeGroup %sGroup;\n" % f)
for f in fkeys:
if len(Functs[f][1]) > 0: rangeTable = Functs[f][1]
rangeTable.sort() numShort = 0
numLong = 0
for rg in rangeTable:
if rg[1] < 0x10000: if numShort == 0: pline = "static const xmlChSRange %s_srng[] = { " % f
else:
pline += ", "
numShort += 1
if len(pline) > 60:
output.write(pline + "\n")
pline = " "
pline += "{0x%x, 0x%x}" % (rg[0], rg[1])
else: if numLong == 0: if numShort > 0: output.write(pline + "};\n")
pline = "static const xmlChLRange %s_lrng[] = { " % f
else:
pline += ", "
numLong += 1
if len(pline) > 60:
output.write(pline + "\n")
pline = " "
pline += "{0x%x, 0x%x}" % (rg[0], rg[1])
output.write(pline + "};\n")
pline = "const xmlChRangeGroup %sGroup =\n\t{%d, %d, " % (f, numShort, numLong)
if numShort > 0:
pline += "%s_srng" % f
else:
pline += "(xmlChSRangePtr)0"
if numLong > 0:
pline += ", %s_lrng" % f
else:
pline += ", (xmlChLRangePtr)0"
output.write(pline + "};\n\n")
output.write(
"""
/**
* xmlCharInRange:
* @val: character to be validated
* @rptr: pointer to range to be used to validate
*
* Does a binary search of the range table to determine if char
* is valid
*
* Returns: true if character valid, false otherwise
*/
int
xmlCharInRange (unsigned int val, const xmlChRangeGroup *rptr) {
int low, high, mid;
const xmlChSRange *sptr;
const xmlChLRange *lptr;
if (rptr == NULL) return(0);
if (val < 0x10000) { /* is val in 'short' or 'long' array? */
if (rptr->nbShortRange == 0)
return 0;
low = 0;
high = rptr->nbShortRange - 1;
sptr = rptr->shortRange;
while (low <= high) {
mid = (low + high) / 2;
if ((unsigned short) val < sptr[mid].low) {
high = mid - 1;
} else {
if ((unsigned short) val > sptr[mid].high) {
low = mid + 1;
} else {
return 1;
}
}
}
} else {
if (rptr->nbLongRange == 0) {
return 0;
}
low = 0;
high = rptr->nbLongRange - 1;
lptr = rptr->longRange;
while (low <= high) {
mid = (low + high) / 2;
if (val < lptr[mid].low) {
high = mid - 1;
} else {
if (val > lptr[mid].high) {
low = mid + 1;
} else {
return 1;
}
}
}
}
return 0;
}
""");
for f in fkeys:
output.write("""
/**
* %s:
* @ch: character to validate
*
* This function is DEPRECATED.
""" % f);
if max(Functs[f][0]) > 0:
output.write(" * Use %s_ch or %sQ instead" % (f, f))
else:
output.write(" * Use %sQ instead" % f)
output.write("""
*
* Returns true if argument valid, false otherwise
*/
""")
output.write("int\n%s(unsigned int ch) {\n return(%sQ(ch));\n}\n\n" % (f,f))
header.write("XMLPUBFUN int XMLCALL\n\t\t%s(unsigned int ch);\n" % f);
header.write("""
#ifdef __cplusplus
}
#endif
#endif /* __XML_CHVALID_H__ */
""")
header.close()
output.write("""#define bottom_chvalid
#include "elfgcchack.h"
""")
output.close()