mirror of
https://github.com/FirebirdSQL/firebird.git
synced 2025-01-26 06:43:04 +01:00
893 lines
27 KiB
C++
893 lines
27 KiB
C++
/*
|
|
* PROGRAM: JRD Intl
|
|
* MODULE: Collation.cpp
|
|
* DESCRIPTION: International text support routines
|
|
*
|
|
* copyright (c) 1992, 1993 by Borland International
|
|
*/
|
|
|
|
/************* history ************
|
|
*
|
|
* COMPONENT: JRD MODULE: INTL.CPP
|
|
* generated by Marion V2.5 2/6/90
|
|
* from dev db on 4-JAN-1995
|
|
*****************************************************************
|
|
*
|
|
* PR 2002-06-02 Added ugly c hack in
|
|
* intl_back_compat_alloc_func_lookup.
|
|
* When someone has time we need to change the references to
|
|
* return (void*) function to something more C++ like
|
|
*
|
|
* 42 4711 3 11 17 tamlin 2001
|
|
* Added silly numbers before my name, and converted it to C++.
|
|
*
|
|
* 18850 daves 4-JAN-1995
|
|
* Fix gds__alloc usage
|
|
*
|
|
* 18837 deej 31-DEC-1994
|
|
* fixing up HARBOR_MERGE
|
|
*
|
|
* 18821 deej 27-DEC-1994
|
|
* HARBOR MERGE
|
|
*
|
|
* 18789 jdavid 19-DEC-1994
|
|
* Cast some functions
|
|
*
|
|
* 17508 jdavid 15-JUL-1994
|
|
* Bring it up to date
|
|
*
|
|
* 17500 daves 13-JUL-1994
|
|
* Bug 6645: Different calculation of partial keys
|
|
*
|
|
* 17202 katz 24-MAY-1994
|
|
* PC_PLATFORM requires the .dll extension
|
|
*
|
|
* 17191 katz 23-MAY-1994
|
|
* OS/2 requires the .dll extension
|
|
*
|
|
* 17180 katz 23-MAY-1994
|
|
* Define location of DLL on OS/2
|
|
*
|
|
* 17149 katz 20-MAY-1994
|
|
* In JRD, isc_arg_number arguments are SLONG's not int's
|
|
*
|
|
* 16633 daves 19-APR-1994
|
|
* Bug 6202: International licensing uses INTERNATIONAL product code
|
|
*
|
|
* 16555 katz 17-APR-1994
|
|
* The last argument of calls to ERR_post should be 0
|
|
*
|
|
* 16521 katz 14-APR-1994
|
|
* Borland C needs a decorated symbol to lookup
|
|
*
|
|
* 16403 daves 8-APR-1994
|
|
* Bug 6441: Emit an error whenever transliteration from ttype_binary attempted
|
|
*
|
|
* 16141 katz 28-MAR-1994
|
|
* Don't declare return value from ISC_lookup_entrypoint as API_ROUTINE
|
|
*
|
|
* The contents of this file are subject to the Interbase Public
|
|
* License Version 1.0 (the "License"); you may not use this file
|
|
* except in compliance with the License. You may obtain a copy
|
|
* of the License at http://www.Inprise.com/IPL.html
|
|
*
|
|
* Software distributed under the License is distributed on an
|
|
* "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, either express
|
|
* or implied. See the License for the specific language governing
|
|
* rights and limitations under the License.
|
|
*
|
|
* The Original Code was created by Inprise Corporation
|
|
* and its predecessors. Portions created by Inprise Corporation are
|
|
* Copyright (C) Inprise Corporation.
|
|
*
|
|
* All Rights Reserved.
|
|
* Contributor(s): ______________________________________.
|
|
*
|
|
* 2002.10.29 Sean Leyne - Removed obsolete "Netware" port
|
|
*
|
|
* 2002.10.30 Sean Leyne - Removed support for obsolete "PC_PLATFORM" define
|
|
*
|
|
* 2006.10.10 Adriano dos Santos Fernandes - refactored from intl.cpp
|
|
*
|
|
*/
|
|
|
|
#include "firebird.h"
|
|
#include "gen/iberror.h"
|
|
#include "../jrd/jrd.h"
|
|
#include "../jrd/err_proto.h"
|
|
#include "../jrd/evl_string.h"
|
|
#include "../jrd/intl_classes.h"
|
|
#include "../jrd/lck_proto.h"
|
|
|
|
#include "../jrd/intl_classes.h"
|
|
#include "../jrd/TextType.h"
|
|
|
|
using namespace Jrd;
|
|
|
|
|
|
namespace {
|
|
|
|
// Below is former evl_like_h file
|
|
// It was not used except Collation.cpp
|
|
|
|
#define SLEUTH_insensitive 1
|
|
|
|
static const UCHAR special[256] = {
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 0, 0, /* $%*+- (dollar, percent, star, plus, minus) */
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, /* ? (question) */
|
|
1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* @ (at-sign) */
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, /* [ (open square) */
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, /* ~ (tilde) */
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
|
|
};
|
|
|
|
template <class MATCHESTYPE>
|
|
bool MATCHESNAME(Jrd::thread_db* tdbb, Jrd::TextType* obj,
|
|
const MATCHESTYPE* p1, SLONG l1_bytes,
|
|
const MATCHESTYPE* p2, SLONG l2_bytes)
|
|
{
|
|
/**************************************
|
|
*
|
|
* E V L _ ? ? _ m a t c h e s
|
|
* E V L _ w c _ m a t c h e s
|
|
* E V L _ n c _ m a t c h e s
|
|
*
|
|
**************************************
|
|
*
|
|
* Functional description
|
|
* Return true if a string (p1, l1) matches a given pattern (p2, l2).
|
|
* The character '?' in the pattern may match any single character
|
|
* in the the string, and the character '*' may match any sequence
|
|
* of characters.
|
|
*
|
|
* Wide SCHAR version operates on short-based buffer,
|
|
* instead of SCHAR-based.
|
|
*
|
|
* Matches is not a case-sensitive operation, thus it has no
|
|
* 8-bit international impact.
|
|
*
|
|
**************************************/
|
|
fb_assert(p1 != NULL);
|
|
fb_assert(p2 != NULL);
|
|
fb_assert((l1_bytes % sizeof(MATCHESTYPE)) == 0);
|
|
fb_assert((l2_bytes % sizeof(MATCHESTYPE)) == 0);
|
|
fb_assert((obj->getCanonicalWidth() == sizeof(MATCHESTYPE)));
|
|
|
|
SLONG l1 = l1_bytes / sizeof(MATCHESTYPE);
|
|
SLONG l2 = l2_bytes / sizeof(MATCHESTYPE);
|
|
|
|
while (l2-- > 0) {
|
|
const MATCHESTYPE c = *p2++;
|
|
if (c == *(MATCHESTYPE*)obj->getGdmlMatchAnyCanonic()) {
|
|
while ((l2 > 0) && (*p2 == *(MATCHESTYPE*)obj->getGdmlMatchAnyCanonic())) {
|
|
l2--;
|
|
p2++;
|
|
}
|
|
if (l2 == 0)
|
|
return true;
|
|
while (l1)
|
|
{
|
|
if (MATCHESNAME(tdbb, obj, p1++,
|
|
l1-- * sizeof(MATCHESTYPE), p2,
|
|
l2 * sizeof(MATCHESTYPE)))
|
|
{
|
|
return true;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
if (l1-- == 0)
|
|
return false;
|
|
if (c != *(MATCHESTYPE*)obj->getGdmlMatchOneCanonic() && c != *p1)
|
|
return false;
|
|
p1++;
|
|
}
|
|
|
|
return !l1;
|
|
}
|
|
|
|
|
|
template <class SLEUTHTYPE>
|
|
bool SLEUTHNAME(Jrd::thread_db* tdbb_dummy, Jrd::TextType* obj, USHORT flags,
|
|
const SLEUTHTYPE* search, SLONG search_len,
|
|
const SLEUTHTYPE* match, SLONG match_len)
|
|
{
|
|
/**************************************
|
|
*
|
|
* E V L _ ? ? _ s l e u t h _ c h e c k
|
|
*
|
|
**************************************
|
|
*
|
|
* Functional description
|
|
* Evaluate the "sleuth" search operator.
|
|
*
|
|
* Turn the (pointer, byte length) input parameters into
|
|
* (pointer, end_pointer) for use in SLEUTH_AUX
|
|
*
|
|
**************************************/
|
|
fb_assert((match_len % sizeof(SLEUTHTYPE)) == 0);
|
|
fb_assert((search_len % sizeof(SLEUTHTYPE)) == 0);
|
|
fb_assert(obj->getCanonicalWidth() == sizeof(SLEUTHTYPE));
|
|
|
|
const SLEUTHTYPE* const end_match = match + (match_len / sizeof(SLEUTHTYPE));
|
|
const SLEUTHTYPE* const end_search = search + (search_len / sizeof(SLEUTHTYPE));
|
|
|
|
return SLEUTH_AUX(obj, flags, search, end_search, match, end_match);
|
|
}
|
|
|
|
|
|
template <class SLEUTHTYPE>
|
|
ULONG SLEUTH_MERGE_NAME(Jrd::thread_db* tdbb_dummy, Jrd::TextType* obj,
|
|
const SLEUTHTYPE* match, SLONG match_bytes,
|
|
const SLEUTHTYPE* control, SLONG control_bytes,
|
|
SLEUTHTYPE* combined, SLONG combined_bytes)
|
|
{
|
|
/**************************************
|
|
*
|
|
* E V L _ ? ? _ s l e u t h _ m e r g e
|
|
*
|
|
**************************************
|
|
*
|
|
* Functional description
|
|
* Merge the matching pattern and control strings to give a cannonical
|
|
* matching pattern. Return the length of the combined string.
|
|
*
|
|
* What this routine does is to take the language template, strip off
|
|
* the prefix and put it in the output string, then parse the definitions
|
|
* into an array of character pointers. The index array is the defined
|
|
* character. The routine then takes the actual match pattern and uses
|
|
* the characters in it to index into the definitions to produce an equivalent
|
|
* pattern in the cannonical language.
|
|
*
|
|
* The silly loop setting *v++ to zero initializes the array up to the
|
|
* highest character defined (also max_op). Believe it or not, that part
|
|
* is not a bug.
|
|
*
|
|
**************************************/
|
|
fb_assert(match != NULL);
|
|
fb_assert(control != NULL);
|
|
fb_assert(combined != NULL);
|
|
|
|
fb_assert((match_bytes % sizeof(SLEUTHTYPE)) == 0);
|
|
fb_assert((control_bytes % sizeof(SLEUTHTYPE)) == 0);
|
|
fb_assert(obj->getCanonicalWidth() == sizeof(SLEUTHTYPE));
|
|
|
|
const SLEUTHTYPE* const end_match = match + (match_bytes / sizeof(SLEUTHTYPE));
|
|
const SLEUTHTYPE* const end_control = control + (control_bytes / sizeof(SLEUTHTYPE));
|
|
|
|
SLEUTHTYPE max_op = 0;
|
|
SLEUTHTYPE* comb = combined;
|
|
SLEUTHTYPE* vector[256];
|
|
SLEUTHTYPE** v = vector;
|
|
SLEUTHTYPE temp[256];
|
|
SLEUTHTYPE* t = temp;
|
|
|
|
/* Parse control string into substitution strings and initializing string */
|
|
|
|
while (control < end_control) {
|
|
SLEUTHTYPE c = *control++;
|
|
if (*control == *(SLEUTHTYPE*)obj->getGdmlSubstituteCanonic()) {
|
|
/* Note: don't allow substitution characters larger than vector */
|
|
SLEUTHTYPE** const end_vector =
|
|
vector + (((int)c < FB_NELEM(vector)) ? c : 0);
|
|
while (v <= end_vector)
|
|
*v++ = 0;
|
|
*end_vector = t;
|
|
++control;
|
|
while (control < end_control) {
|
|
c = *control++;
|
|
if ((t > temp && t[-1] == *(SLEUTHTYPE*)obj->getGdmlQuoteCanonic())
|
|
|| ((c != *(SLEUTHTYPE*)obj->getGdmlCommaCanonic()) && (c != *(SLEUTHTYPE*)obj->getGdmlRParenCanonic())))
|
|
{
|
|
*t++ = c;
|
|
}
|
|
else
|
|
break;
|
|
}
|
|
*t++ = 0;
|
|
}
|
|
else if (c == *(SLEUTHTYPE*)obj->getGdmlQuoteCanonic() && control < end_control)
|
|
*comb++ = *control++;
|
|
else if (c == *(SLEUTHTYPE*)obj->getGdmlRParenCanonic())
|
|
break;
|
|
else if (c != *(SLEUTHTYPE*)obj->getGdmlLParenCanonic())
|
|
*comb++ = c;
|
|
}
|
|
|
|
max_op = v - vector;
|
|
|
|
/* Interpret matching string, substituting where appropriate */
|
|
|
|
while (match < end_match) {
|
|
const SLEUTHTYPE c = *match++;
|
|
|
|
/* if we've got a defined character, slurp the definition */
|
|
|
|
SLEUTHTYPE* p;
|
|
if (c <= max_op && (p = vector[c])) {
|
|
while (*p)
|
|
*comb++ = *p++;
|
|
|
|
/* if we've got the definition of a quote character,
|
|
slurp the next character too */
|
|
|
|
if (comb > combined && comb[-1] == *(SLEUTHTYPE*)obj->getGdmlQuoteCanonic() && *match)
|
|
*comb++ = *match++;
|
|
}
|
|
|
|
/* at this point we've got a non-match, but as it might be one of ours,
|
|
quote it. */
|
|
|
|
else {
|
|
if ((((size_t) c) < sizeof(special)) && special[c] &&
|
|
comb > combined && comb[-1] != *(SLEUTHTYPE*)obj->getGdmlQuoteCanonic())
|
|
{
|
|
*comb++ = *(SLEUTHTYPE*)obj->getGdmlQuoteCanonic();
|
|
}
|
|
*comb++ = c;
|
|
}
|
|
}
|
|
|
|
/* Put in trailing stuff */
|
|
|
|
while (control < end_control)
|
|
*comb++ = *control++;
|
|
|
|
/* YYY - need to add code watching for overflow of combined */
|
|
|
|
return (comb - combined) * sizeof(SLEUTHTYPE);
|
|
}
|
|
|
|
|
|
template <class SLEUTHTYPE>
|
|
static bool SLEUTH_AUX(
|
|
Jrd::TextType* obj,
|
|
USHORT flags,
|
|
const SLEUTHTYPE* search,
|
|
const SLEUTHTYPE* end_search,
|
|
const SLEUTHTYPE* match, const SLEUTHTYPE* end_match)
|
|
{
|
|
/**************************************
|
|
*
|
|
* s l e u t h _ a u x
|
|
*
|
|
**************************************
|
|
*
|
|
* Functional description
|
|
* Evaluate the "sleuth" search operator.
|
|
*
|
|
**************************************/
|
|
fb_assert(search != NULL);
|
|
fb_assert(end_search != NULL);
|
|
fb_assert(match != NULL);
|
|
fb_assert(end_match != NULL);
|
|
fb_assert(search <= end_search);
|
|
fb_assert(match <= end_match);
|
|
fb_assert(obj->getCanonicalWidth() == sizeof(SLEUTHTYPE));
|
|
|
|
while (match < end_match) {
|
|
SLEUTHTYPE c = *match++;
|
|
if ((c == *(SLEUTHTYPE*)obj->getGdmlQuoteCanonic() && (c = *match++)) ||
|
|
((((size_t) c) < sizeof(special)) && !special[c]))
|
|
{
|
|
if (match >= end_match || *match != *(SLEUTHTYPE*)obj->getGdmlMatchAnyCanonic()) {
|
|
if (search >= end_search)
|
|
return false;
|
|
const SLEUTHTYPE d = *search++;
|
|
if (c != d)
|
|
return false;
|
|
}
|
|
else {
|
|
++match;
|
|
for (;;)
|
|
if (SLEUTH_AUX
|
|
(obj, flags, search, end_search, match, end_match))
|
|
{
|
|
return true;
|
|
}
|
|
else if (search < end_search) {
|
|
const SLEUTHTYPE d = *search++;
|
|
if (c != d)
|
|
return false;
|
|
}
|
|
else
|
|
return false;
|
|
}
|
|
}
|
|
else if (c == *(SLEUTHTYPE*)obj->getGdmlMatchOneCanonic())
|
|
if (match >= end_match || *match != *(SLEUTHTYPE*)obj->getGdmlMatchAnyCanonic()) {
|
|
if (search >= end_search)
|
|
return false;
|
|
search++;
|
|
}
|
|
else {
|
|
if (++match >= end_match)
|
|
return true;
|
|
for (;;)
|
|
if (SLEUTH_AUX
|
|
(obj, flags, search, end_search, match,
|
|
end_match))
|
|
{
|
|
return true;
|
|
}
|
|
else if (++search >= end_search)
|
|
return false;
|
|
}
|
|
else if (c == *(SLEUTHTYPE*)obj->getGdmlClassStartCanonic()) {
|
|
const SLEUTHTYPE* const char_class = match;
|
|
while (*match++ != *(SLEUTHTYPE*)obj->getGdmlClassEndCanonic()) {
|
|
if (match >= end_match)
|
|
return false;
|
|
}
|
|
const SLEUTHTYPE* const end_class = match - 1;
|
|
if (match >= end_match || *match != *(SLEUTHTYPE*)obj->getGdmlMatchAnyCanonic()) {
|
|
if (!SLEUTH_CLASS_NAME
|
|
(obj, flags, char_class, end_class, *search++))
|
|
{
|
|
return false;
|
|
}
|
|
}
|
|
else {
|
|
++match;
|
|
for (;;)
|
|
if (SLEUTH_AUX
|
|
(obj, flags, search, end_search, match,
|
|
end_match))
|
|
{
|
|
return true;
|
|
}
|
|
else if (search < end_search) {
|
|
if (!SLEUTH_CLASS_NAME
|
|
(obj, flags, char_class, end_class, *search++))
|
|
{
|
|
return false;
|
|
}
|
|
}
|
|
else
|
|
return false;
|
|
}
|
|
}
|
|
else if (c == *(SLEUTHTYPE*)obj->getGdmlFlagSetCanonic()) {
|
|
c = *match++;
|
|
if (c == *(SLEUTHTYPE*)obj->getGdmlLowerSCanonic() || c == *(SLEUTHTYPE*)obj->getGdmlUpperSCanonic())
|
|
flags &= ~SLEUTH_insensitive;
|
|
}
|
|
else if (c == *(SLEUTHTYPE*)obj->getGdmlFlagClearCanonic()) {
|
|
c = *match++;
|
|
if (c == *(SLEUTHTYPE*)obj->getGdmlLowerSCanonic() || c == *(SLEUTHTYPE*)obj->getGdmlUpperSCanonic())
|
|
flags |= SLEUTH_insensitive;
|
|
}
|
|
}
|
|
|
|
if (search < end_search)
|
|
return false;
|
|
|
|
return true;
|
|
}
|
|
|
|
|
|
template <class SLEUTHTYPE>
|
|
static bool SLEUTH_CLASS_NAME(
|
|
Jrd::TextType* obj,
|
|
USHORT flags,
|
|
const SLEUTHTYPE* char_class,
|
|
const SLEUTHTYPE* const end_class,
|
|
SLEUTHTYPE character)
|
|
{
|
|
/**************************************
|
|
*
|
|
* s l e u t h _ c l a s s
|
|
*
|
|
**************************************
|
|
*
|
|
* Functional description
|
|
* See if a character is a member of a class.
|
|
* Japanese version operates on short-based buffer,
|
|
* instead of SCHAR-based.
|
|
*
|
|
**************************************/
|
|
fb_assert(char_class != NULL);
|
|
fb_assert(end_class != NULL);
|
|
fb_assert(char_class <= end_class);
|
|
fb_assert(obj->getCanonicalWidth() == sizeof(SLEUTHTYPE));
|
|
|
|
bool result = true;
|
|
|
|
if (*char_class == *(SLEUTHTYPE*)obj->getGdmlNotCanonic()) {
|
|
++char_class;
|
|
result = false;
|
|
}
|
|
|
|
while (char_class < end_class) {
|
|
const SLEUTHTYPE c = *char_class++;
|
|
if (c == *(SLEUTHTYPE*)obj->getGdmlQuoteCanonic()) {
|
|
if (*char_class++ == character)
|
|
return true;
|
|
}
|
|
else if (*char_class == *(SLEUTHTYPE*)obj->getGdmlRangeCanonic()) {
|
|
char_class += 2;
|
|
if (character >= c && character <= char_class[-1])
|
|
return result;
|
|
}
|
|
else if (character == c)
|
|
return result;
|
|
}
|
|
|
|
return !result;
|
|
}
|
|
|
|
|
|
// Below are templates for functions used in Collation implementation
|
|
|
|
class NullStrConverter {
|
|
public:
|
|
NullStrConverter(thread_db* tdbb, const TextType* obj, const UCHAR *str, SLONG len) { }
|
|
};
|
|
|
|
template <typename PrevConverter>
|
|
class UpcaseConverter : public PrevConverter {
|
|
public:
|
|
UpcaseConverter(thread_db* tdbb, TextType* obj, const UCHAR* &str, SLONG &len) :
|
|
PrevConverter(tdbb, obj, str, len)
|
|
{
|
|
if (len > (int) sizeof(tempBuffer))
|
|
out_str = FB_NEW(*tdbb->getDefaultPool()) UCHAR[len];
|
|
else
|
|
out_str = tempBuffer;
|
|
obj->str_to_upper(len, str, len, out_str);
|
|
str = out_str;
|
|
}
|
|
~UpcaseConverter() {
|
|
if (out_str != tempBuffer)
|
|
delete[] out_str;
|
|
}
|
|
private:
|
|
UCHAR tempBuffer[100], *out_str;
|
|
};
|
|
|
|
template <typename PrevConverter>
|
|
class CanonicalConverter : public PrevConverter {
|
|
public:
|
|
CanonicalConverter(thread_db* tdbb, TextType* obj, const UCHAR* &str, SLONG &len) :
|
|
PrevConverter(tdbb, obj, str, len)
|
|
{
|
|
SLONG out_len = len / obj->getCharSet()->minBytesPerChar() * obj->getCanonicalWidth();
|
|
|
|
if (out_len > (int) sizeof(tempBuffer))
|
|
out_str = FB_NEW(*tdbb->getDefaultPool()) UCHAR[out_len];
|
|
else
|
|
out_str = tempBuffer;
|
|
|
|
if (str)
|
|
{
|
|
len = obj->canonical(len, str, out_len, out_str) * obj->getCanonicalWidth();
|
|
str = out_str;
|
|
}
|
|
else
|
|
len = 0;
|
|
}
|
|
~CanonicalConverter() {
|
|
if (out_str != tempBuffer)
|
|
delete[] out_str;
|
|
}
|
|
private:
|
|
UCHAR tempBuffer[100], *out_str;
|
|
};
|
|
|
|
template <typename StrConverter, typename CharType>
|
|
class LikeObjectImpl : public LikeObject {
|
|
public:
|
|
LikeObjectImpl(MemoryPool& pool, const CharType* str, SLONG str_len,
|
|
CharType escape, bool use_escape,
|
|
CharType sql_match_any, CharType sql_match_one)
|
|
: evaluator(pool, str, str_len, escape, use_escape, sql_match_any, sql_match_one)
|
|
{ }
|
|
|
|
void reset() { evaluator.reset(); }
|
|
|
|
bool result() { return evaluator.getResult(); }
|
|
|
|
bool process(thread_db* tdbb, Jrd::TextType* ttype, const UCHAR* str, SLONG length) {
|
|
StrConverter cvt(tdbb, ttype, str, length);
|
|
fb_assert(length % sizeof(CharType) == 0);
|
|
return evaluator.processNextChunk(
|
|
reinterpret_cast<const CharType*>(str), length / sizeof(CharType));
|
|
}
|
|
|
|
~LikeObjectImpl() {}
|
|
|
|
static LikeObject* create(thread_db* tdbb, TextType* ttype, const UCHAR* str, SLONG length,
|
|
const UCHAR* escape, SLONG escape_length,
|
|
const UCHAR* sql_match_any, SLONG match_any_length,
|
|
const UCHAR* sql_match_one, SLONG match_one_length)
|
|
{
|
|
StrConverter cvt(tdbb, ttype, str, length),
|
|
cvt_escape(tdbb, ttype, escape, escape_length),
|
|
cvt_match_any(tdbb, ttype, sql_match_any, match_any_length),
|
|
cvt_match_one(tdbb, ttype, sql_match_one, match_one_length);
|
|
|
|
fb_assert(length % sizeof(CharType) == 0);
|
|
return FB_NEW(*tdbb->getDefaultPool()) LikeObjectImpl(*tdbb->getDefaultPool(),
|
|
reinterpret_cast<const CharType*>(str), length / sizeof(CharType),
|
|
(escape ? *reinterpret_cast<const CharType*>(escape) : 0), escape_length != 0,
|
|
*reinterpret_cast<const CharType*>(sql_match_any),
|
|
*reinterpret_cast<const CharType*>(sql_match_one));
|
|
}
|
|
|
|
static bool evaluate(thread_db* tdbb, TextType* ttype, const UCHAR* s, SLONG sl,
|
|
const UCHAR* p, SLONG pl, const UCHAR* escape, SLONG escape_length, const UCHAR* sql_match_any, SLONG match_any_length, const UCHAR* sql_match_one, SLONG match_one_length)
|
|
{
|
|
StrConverter cvt1(tdbb, ttype, p, pl),
|
|
cvt2(tdbb, ttype, s, sl),
|
|
cvt_escape(tdbb, ttype, escape, escape_length),
|
|
cvt_match_any(tdbb, ttype, sql_match_any, match_any_length),
|
|
cvt_match_one(tdbb, ttype, sql_match_one, match_one_length);
|
|
|
|
fb_assert(pl % sizeof(CharType) == 0);
|
|
fb_assert(sl % sizeof(CharType) == 0);
|
|
Firebird::LikeEvaluator<CharType> evaluator(*tdbb->getDefaultPool(),
|
|
reinterpret_cast<const CharType*>(p), pl / sizeof(CharType),
|
|
(escape ? *reinterpret_cast<const CharType*>(escape) : 0), escape_length != 0,
|
|
*reinterpret_cast<const CharType*>(sql_match_any),
|
|
*reinterpret_cast<const CharType*>(sql_match_one));
|
|
evaluator.processNextChunk(reinterpret_cast<const CharType*>(s), sl / sizeof(CharType));
|
|
return evaluator.getResult();
|
|
}
|
|
|
|
private:
|
|
Firebird::LikeEvaluator<CharType> evaluator;
|
|
};
|
|
|
|
template <typename StrConverter, typename CharType>
|
|
class ContainsObjectImpl : public ContainsObject
|
|
{
|
|
public:
|
|
ContainsObjectImpl(MemoryPool& pool, const CharType* str, SLONG str_len)
|
|
: evaluator(pool, str, str_len)
|
|
{ }
|
|
|
|
void reset() { evaluator.reset(); }
|
|
|
|
bool result() { return evaluator.getResult(); }
|
|
|
|
bool process(thread_db* tdbb, Jrd::TextType* ttype, const UCHAR* str, SLONG length) {
|
|
StrConverter cvt(tdbb, ttype, str, length);
|
|
fb_assert(length % sizeof(CharType) == 0);
|
|
return evaluator.processNextChunk(
|
|
reinterpret_cast<const CharType*>(str), length / sizeof(CharType));
|
|
}
|
|
|
|
~ContainsObjectImpl() {}
|
|
|
|
static ContainsObject* create(thread_db* tdbb, TextType* ttype, const UCHAR* str, SLONG length) {
|
|
StrConverter cvt(tdbb, ttype, str, length);
|
|
fb_assert(length % sizeof(CharType) == 0);
|
|
return FB_NEW(*tdbb->getDefaultPool()) ContainsObjectImpl(*tdbb->getDefaultPool(),
|
|
reinterpret_cast<const CharType*>(str), length / sizeof(CharType));
|
|
}
|
|
|
|
static bool evaluate(thread_db* tdbb, TextType* ttype, const UCHAR* s, SLONG sl,
|
|
const UCHAR* p, SLONG pl)
|
|
{
|
|
StrConverter cvt1(tdbb, ttype, p, pl), cvt2(tdbb, ttype, s, sl);
|
|
fb_assert(pl % sizeof(CharType) == 0);
|
|
fb_assert(sl % sizeof(CharType) == 0);
|
|
Firebird::ContainsEvaluator<CharType> evaluator(*tdbb->getDefaultPool(),
|
|
reinterpret_cast<const CharType*>(p), pl / sizeof(CharType));
|
|
evaluator.processNextChunk(reinterpret_cast<const CharType*>(s), sl / sizeof(CharType));
|
|
return evaluator.getResult();
|
|
}
|
|
|
|
private:
|
|
Firebird::ContainsEvaluator<CharType> evaluator;
|
|
};
|
|
|
|
template <typename StrConverter, typename CharType>
|
|
class MatchesObjectImpl
|
|
{
|
|
public:
|
|
static bool evaluate(thread_db* tdbb, TextType* ttype, const UCHAR* s, SLONG sl,
|
|
const UCHAR* p, SLONG pl)
|
|
{
|
|
StrConverter cvt1(tdbb, ttype, p, pl), cvt2(tdbb, ttype, s, sl);
|
|
fb_assert(pl % sizeof(CharType) == 0);
|
|
fb_assert(sl % sizeof(CharType) == 0);
|
|
return MATCHESNAME(tdbb, ttype, reinterpret_cast<const CharType*>(s), sl,
|
|
reinterpret_cast<const CharType*>(p), pl);
|
|
}
|
|
};
|
|
|
|
template <typename StrConverter, typename CharType>
|
|
class SleuthObjectImpl
|
|
{
|
|
public:
|
|
static bool check(thread_db* tdbb, TextType* ttype, USHORT flags,
|
|
const UCHAR* search, SLONG search_len,
|
|
const UCHAR* match, SLONG match_len)
|
|
{
|
|
StrConverter cvt1(tdbb, ttype, search, search_len);//, cvt2(tdbb, ttype, match, match_len);
|
|
fb_assert(search_len % sizeof(CharType) == 0);
|
|
fb_assert(match_len % sizeof(CharType) == 0);
|
|
return SLEUTHNAME(tdbb, ttype, flags,
|
|
reinterpret_cast<const CharType*>(search), search_len,
|
|
reinterpret_cast<const CharType*>(match), match_len);
|
|
}
|
|
|
|
static ULONG merge(thread_db* tdbb, TextType* ttype,
|
|
const UCHAR* match, SLONG match_bytes,
|
|
const UCHAR* control, SLONG control_bytes,
|
|
UCHAR* combined, SLONG combined_bytes)
|
|
{
|
|
StrConverter cvt1(tdbb, ttype, match, match_bytes), cvt2(tdbb, ttype, control, control_bytes);
|
|
fb_assert(match_bytes % sizeof(CharType) == 0);
|
|
fb_assert(control_bytes % sizeof(CharType) == 0);
|
|
return SLEUTH_MERGE_NAME(tdbb, ttype,
|
|
reinterpret_cast<const CharType*>(match), match_bytes,
|
|
reinterpret_cast<const CharType*>(control), control_bytes,
|
|
reinterpret_cast<CharType*>(combined), combined_bytes);
|
|
}
|
|
};
|
|
|
|
|
|
template <typename pContainsObjectImpl, typename pLikeObjectImpl,
|
|
typename pMatchesObjectImpl, typename pSleuthObjectImpl>
|
|
class CollationImpl : public Collation
|
|
{
|
|
public:
|
|
CollationImpl(TTYPE_ID a_type, texttype* a_tt, CharSet* a_cs) : Collation(a_type, a_tt, a_cs) {}
|
|
|
|
virtual bool matches(thread_db* tdbb, const UCHAR* a, SLONG b, const UCHAR* c, SLONG d)
|
|
{
|
|
return pMatchesObjectImpl::evaluate(tdbb, this, a, b, c, d);
|
|
}
|
|
|
|
virtual bool sleuth_check(thread_db* tdbb, USHORT a, const UCHAR* b, SLONG c, const UCHAR* d, SLONG e)
|
|
{
|
|
return pSleuthObjectImpl::check(tdbb, this, a, b, c, d, e);
|
|
}
|
|
|
|
virtual ULONG sleuth_merge(thread_db* tdbb, const UCHAR* a, SLONG b, const UCHAR* c, SLONG d, UCHAR* e, SLONG f)
|
|
{
|
|
return pSleuthObjectImpl::merge(tdbb, this, a, b, c, d, e, f);
|
|
}
|
|
|
|
virtual bool like(thread_db* tdbb, const UCHAR* s, SLONG sl, const UCHAR* p, SLONG pl, const UCHAR* escape, SLONG escape_length)
|
|
{
|
|
return pLikeObjectImpl::evaluate(tdbb, this, s, sl, p, pl, escape, escape_length, getCharSet()->getSqlMatchAny(), getCharSet()->getSqlMatchAnyLength(), getCharSet()->getSqlMatchOne(), getCharSet()->getSqlMatchOneLength());
|
|
}
|
|
|
|
virtual LikeObject *like_create(thread_db* tdbb, const UCHAR* p, SLONG pl, const UCHAR* escape, SLONG escape_length)
|
|
{
|
|
return pLikeObjectImpl::create(tdbb, this, p, pl, escape, escape_length, getCharSet()->getSqlMatchAny(), getCharSet()->getSqlMatchAnyLength(), getCharSet()->getSqlMatchOne(), getCharSet()->getSqlMatchOneLength());
|
|
}
|
|
|
|
virtual bool contains(thread_db* tdbb, const UCHAR* s, SLONG sl, const UCHAR* p, SLONG pl)
|
|
{
|
|
return pContainsObjectImpl::evaluate(tdbb, this, s, sl, p, pl);
|
|
}
|
|
|
|
virtual ContainsObject *contains_create(thread_db* tdbb, const UCHAR* p, SLONG pl)
|
|
{
|
|
return pContainsObjectImpl::create(tdbb, this, p, pl);
|
|
}
|
|
};
|
|
|
|
typedef ContainsObjectImpl<UpcaseConverter<NullStrConverter>, UCHAR> uchar_contains_direct;
|
|
typedef ContainsObjectImpl<UpcaseConverter<NullStrConverter>, USHORT> ushort_contains_direct;
|
|
typedef ContainsObjectImpl<UpcaseConverter<NullStrConverter>, ULONG> ulong_contains_direct;
|
|
|
|
typedef MatchesObjectImpl<CanonicalConverter<NullStrConverter>, UCHAR> uchar_matches_canonical;
|
|
typedef SleuthObjectImpl<CanonicalConverter<NullStrConverter>, UCHAR> uchar_sleuth_canonical;
|
|
typedef LikeObjectImpl<CanonicalConverter<NullStrConverter>, UCHAR> uchar_like_canonical;
|
|
typedef ContainsObjectImpl<CanonicalConverter<UpcaseConverter<NullStrConverter> >, UCHAR> uchar_contains_canonical;
|
|
|
|
typedef MatchesObjectImpl<CanonicalConverter<NullStrConverter>, USHORT> ushort_matches_canonical;
|
|
typedef SleuthObjectImpl<CanonicalConverter<NullStrConverter>, USHORT> ushort_sleuth_canonical;
|
|
typedef LikeObjectImpl<CanonicalConverter<NullStrConverter>, USHORT> ushort_like_canonical;
|
|
typedef ContainsObjectImpl<CanonicalConverter<UpcaseConverter<NullStrConverter> >, USHORT> ushort_contains_canonical;
|
|
|
|
typedef MatchesObjectImpl<CanonicalConverter<NullStrConverter>, ULONG> ulong_matches_canonical;
|
|
typedef SleuthObjectImpl<CanonicalConverter<NullStrConverter>, ULONG> ulong_sleuth_canonical;
|
|
typedef LikeObjectImpl<CanonicalConverter<NullStrConverter>, ULONG> ulong_like_canonical;
|
|
typedef ContainsObjectImpl<CanonicalConverter<UpcaseConverter<NullStrConverter> >, ULONG> ulong_contains_canonical;
|
|
|
|
} // namespace
|
|
|
|
|
|
//-------------
|
|
|
|
|
|
namespace Jrd {
|
|
|
|
|
|
Collation* Collation::createInstance(MemoryPool& pool, TTYPE_ID id, texttype* tt, CharSet* cs)
|
|
{
|
|
fb_assert(tt->texttype_canonical_width == 1 ||
|
|
tt->texttype_canonical_width == 2 ||
|
|
tt->texttype_canonical_width == 4);
|
|
|
|
switch (tt->texttype_canonical_width)
|
|
{
|
|
case 1:
|
|
if (tt->texttype_flags & TEXTTYPE_DIRECT_MATCH)
|
|
{
|
|
return FB_NEW(pool) CollationImpl<uchar_contains_direct, uchar_like_canonical,
|
|
uchar_matches_canonical, uchar_sleuth_canonical>(id, tt, cs);
|
|
}
|
|
else
|
|
{
|
|
return FB_NEW(pool) CollationImpl<uchar_contains_canonical, uchar_like_canonical,
|
|
uchar_matches_canonical, uchar_sleuth_canonical>(id, tt, cs);
|
|
}
|
|
break;
|
|
|
|
case 2:
|
|
if (tt->texttype_flags & TEXTTYPE_DIRECT_MATCH)
|
|
{
|
|
return FB_NEW(pool) CollationImpl<uchar_contains_direct, ushort_like_canonical,
|
|
ushort_matches_canonical, ushort_sleuth_canonical>(id, tt, cs);
|
|
}
|
|
else
|
|
{
|
|
return FB_NEW(pool) CollationImpl<ushort_contains_canonical, ushort_like_canonical,
|
|
ushort_matches_canonical, ushort_sleuth_canonical>(id, tt, cs);
|
|
}
|
|
break;
|
|
|
|
case 4:
|
|
if (tt->texttype_flags & TEXTTYPE_DIRECT_MATCH)
|
|
{
|
|
return FB_NEW(pool) CollationImpl<uchar_contains_direct, ulong_like_canonical,
|
|
ulong_matches_canonical, ulong_sleuth_canonical>(id, tt, cs);
|
|
}
|
|
else
|
|
{
|
|
return FB_NEW(pool) CollationImpl<ulong_contains_canonical, ulong_like_canonical,
|
|
ulong_matches_canonical, ulong_sleuth_canonical>(id, tt, cs);
|
|
}
|
|
break;
|
|
}
|
|
|
|
fb_assert(false);
|
|
return NULL; // compiler silencer
|
|
}
|
|
|
|
|
|
void Collation::destroy()
|
|
{
|
|
if (tt->texttype_fn_destroy)
|
|
tt->texttype_fn_destroy(tt);
|
|
|
|
delete tt;
|
|
delete existenceLock;
|
|
}
|
|
|
|
|
|
void Collation::incUseCount(thread_db* tdbb)
|
|
{
|
|
++useCount;
|
|
}
|
|
|
|
|
|
void Collation::decUseCount(thread_db* tdbb)
|
|
{
|
|
if (--useCount == 0)
|
|
{
|
|
if (existenceLock)
|
|
LCK_re_post(existenceLock);
|
|
}
|
|
}
|
|
|
|
|
|
} // namespace Jrd
|