author | Eckhart Koeppen <eckhart.koppen@nokia.com> |
Wed, 21 Apr 2010 11:15:19 +0300 | |
branch | RCL_3 |
changeset 11 | 25a739ee40f4 |
parent 4 | 3b1da2848fc7 |
permissions | -rw-r--r-- |
0 | 1 |
/**************************************************************************** |
2 |
** |
|
4
3b1da2848fc7
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
3 |
** Copyright (C) 2010 Nokia Corporation and/or its subsidiary(-ies). |
0 | 4 |
** All rights reserved. |
5 |
** Contact: Nokia Corporation (qt-info@nokia.com) |
|
6 |
** |
|
7 |
** This file is part of the QtCore module of the Qt Toolkit. |
|
8 |
** |
|
9 |
** $QT_BEGIN_LICENSE:LGPL$ |
|
10 |
** No Commercial Usage |
|
11 |
** This file contains pre-release code and may not be distributed. |
|
12 |
** You may use this file in accordance with the terms and conditions |
|
13 |
** contained in the Technology Preview License Agreement accompanying |
|
14 |
** this package. |
|
15 |
** |
|
16 |
** GNU Lesser General Public License Usage |
|
17 |
** Alternatively, this file may be used under the terms of the GNU Lesser |
|
18 |
** General Public License version 2.1 as published by the Free Software |
|
19 |
** Foundation and appearing in the file LICENSE.LGPL included in the |
|
20 |
** packaging of this file. Please review the following information to |
|
21 |
** ensure the GNU Lesser General Public License version 2.1 requirements |
|
22 |
** will be met: http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html. |
|
23 |
** |
|
24 |
** In addition, as a special exception, Nokia gives you certain additional |
|
25 |
** rights. These rights are described in the Nokia Qt LGPL Exception |
|
26 |
** version 1.1, included in the file LGPL_EXCEPTION.txt in this package. |
|
27 |
** |
|
28 |
** If you have questions regarding the use of this file, please contact |
|
29 |
** Nokia at qt-info@nokia.com. |
|
30 |
** |
|
31 |
** |
|
32 |
** |
|
33 |
** |
|
34 |
** |
|
35 |
** |
|
36 |
** |
|
37 |
** |
|
38 |
** $QT_END_LICENSE$ |
|
39 |
** |
|
40 |
****************************************************************************/ |
|
41 |
||
42 |
#include "qbytearraymatcher.h" |
|
43 |
||
44 |
#include <limits.h> |
|
45 |
||
46 |
QT_BEGIN_NAMESPACE |
|
47 |
||
48 |
static inline void bm_init_skiptable(const uchar *cc, int len, uchar *skiptable) |
|
49 |
{ |
|
50 |
int l = qMin(len, 255); |
|
51 |
memset(skiptable, l, 256*sizeof(uchar)); |
|
52 |
cc += len - l; |
|
53 |
while (l--) |
|
54 |
skiptable[*cc++] = l; |
|
55 |
} |
|
56 |
||
57 |
static inline int bm_find(const uchar *cc, int l, int index, const uchar *puc, uint pl, |
|
58 |
const uchar *skiptable) |
|
59 |
{ |
|
60 |
if (pl == 0) |
|
61 |
return index > l ? -1 : index; |
|
62 |
const uint pl_minus_one = pl - 1; |
|
63 |
||
64 |
register const uchar *current = cc + index + pl_minus_one; |
|
65 |
const uchar *end = cc + l; |
|
66 |
while (current < end) { |
|
67 |
uint skip = skiptable[*current]; |
|
68 |
if (!skip) { |
|
69 |
// possible match |
|
70 |
while (skip < pl) { |
|
71 |
if (*(current - skip) != puc[pl_minus_one - skip]) |
|
72 |
break; |
|
73 |
skip++; |
|
74 |
} |
|
75 |
if (skip > pl_minus_one) // we have a match |
|
76 |
return (current - cc) - skip + 1; |
|
77 |
||
78 |
// in case we don't have a match we are a bit inefficient as we only skip by one |
|
79 |
// when we have the non matching char in the string. |
|
80 |
if (skiptable[*(current - skip)] == pl) |
|
81 |
skip = pl - skip; |
|
82 |
else |
|
83 |
skip = 1; |
|
84 |
} |
|
85 |
if (current > end - skip) |
|
86 |
break; |
|
87 |
current += skip; |
|
88 |
} |
|
89 |
return -1; // not found |
|
90 |
} |
|
91 |
||
92 |
/*! \class QByteArrayMatcher |
|
93 |
\brief The QByteArrayMatcher class holds a sequence of bytes that |
|
94 |
can be quickly matched in a byte array. |
|
95 |
||
96 |
\ingroup tools |
|
97 |
\ingroup string-processing |
|
98 |
||
99 |
This class is useful when you have a sequence of bytes that you |
|
100 |
want to repeatedly match against some byte arrays (perhaps in a |
|
101 |
loop), or when you want to search for the same sequence of bytes |
|
102 |
multiple times in the same byte array. Using a matcher object and |
|
103 |
indexIn() is faster than matching a plain QByteArray with |
|
104 |
QByteArray::indexOf() if repeated matching takes place. This |
|
105 |
class offers no benefit if you are doing one-off byte array |
|
106 |
matches. |
|
107 |
||
108 |
Create the QByteArrayMatcher with the QByteArray you want to |
|
109 |
search for. Then call indexIn() on the QByteArray that you want to |
|
110 |
search. |
|
111 |
||
112 |
\sa QByteArray, QStringMatcher |
|
113 |
*/ |
|
114 |
||
115 |
/*! |
|
116 |
Constructs an empty byte array matcher that won't match anything. |
|
117 |
Call setPattern() to give it a pattern to match. |
|
118 |
*/ |
|
119 |
QByteArrayMatcher::QByteArrayMatcher() |
|
120 |
: d(0) |
|
121 |
{ |
|
122 |
p.p = 0; |
|
123 |
p.l = 0; |
|
124 |
qMemSet(p.q_skiptable, 0, sizeof(p.q_skiptable)); |
|
125 |
} |
|
126 |
||
127 |
/*! |
|
128 |
Constructs a byte array matcher from \a pattern. \a pattern |
|
129 |
has the given \a length. \a pattern must remain in scope, but |
|
130 |
the destructor does not delete \a pattern. |
|
131 |
*/ |
|
132 |
QByteArrayMatcher::QByteArrayMatcher(const char *pattern, int length) |
|
133 |
: d(0) |
|
134 |
{ |
|
135 |
p.p = reinterpret_cast<const uchar *>(pattern); |
|
136 |
p.l = length; |
|
137 |
bm_init_skiptable(p.p, p.l, p.q_skiptable); |
|
138 |
} |
|
139 |
||
140 |
/*! |
|
141 |
Constructs a byte array matcher that will search for \a pattern. |
|
142 |
Call indexIn() to perform a search. |
|
143 |
*/ |
|
144 |
QByteArrayMatcher::QByteArrayMatcher(const QByteArray &pattern) |
|
145 |
: d(0), q_pattern(pattern) |
|
146 |
{ |
|
147 |
p.p = reinterpret_cast<const uchar *>(pattern.constData()); |
|
148 |
p.l = pattern.size(); |
|
149 |
bm_init_skiptable(p.p, p.l, p.q_skiptable); |
|
150 |
} |
|
151 |
||
152 |
/*! |
|
153 |
Copies the \a other byte array matcher to this byte array matcher. |
|
154 |
*/ |
|
155 |
QByteArrayMatcher::QByteArrayMatcher(const QByteArrayMatcher &other) |
|
156 |
: d(0) |
|
157 |
{ |
|
158 |
operator=(other); |
|
159 |
} |
|
160 |
||
161 |
/*! |
|
162 |
Destroys the byte array matcher. |
|
163 |
*/ |
|
164 |
QByteArrayMatcher::~QByteArrayMatcher() |
|
165 |
{ |
|
166 |
} |
|
167 |
||
168 |
/*! |
|
169 |
Assigns the \a other byte array matcher to this byte array matcher. |
|
170 |
*/ |
|
171 |
QByteArrayMatcher &QByteArrayMatcher::operator=(const QByteArrayMatcher &other) |
|
172 |
{ |
|
173 |
q_pattern = other.q_pattern; |
|
174 |
qMemCopy(&p, &other.p, sizeof(p)); |
|
175 |
return *this; |
|
176 |
} |
|
177 |
||
178 |
/*! |
|
179 |
Sets the byte array that this byte array matcher will search for |
|
180 |
to \a pattern. |
|
181 |
||
182 |
\sa pattern(), indexIn() |
|
183 |
*/ |
|
184 |
void QByteArrayMatcher::setPattern(const QByteArray &pattern) |
|
185 |
{ |
|
186 |
q_pattern = pattern; |
|
187 |
p.p = reinterpret_cast<const uchar *>(pattern.constData()); |
|
188 |
p.l = pattern.size(); |
|
189 |
bm_init_skiptable(p.p, p.l, p.q_skiptable); |
|
190 |
} |
|
191 |
||
192 |
/*! |
|
193 |
Searches the byte array \a ba, from byte position \a from (default |
|
194 |
0, i.e. from the first byte), for the byte array pattern() that |
|
195 |
was set in the constructor or in the most recent call to |
|
196 |
setPattern(). Returns the position where the pattern() matched in |
|
197 |
\a ba, or -1 if no match was found. |
|
198 |
*/ |
|
199 |
int QByteArrayMatcher::indexIn(const QByteArray &ba, int from) const |
|
200 |
{ |
|
201 |
if (from < 0) |
|
202 |
from = 0; |
|
203 |
return bm_find(reinterpret_cast<const uchar *>(ba.constData()), ba.size(), from, |
|
204 |
p.p, p.l, p.q_skiptable); |
|
205 |
} |
|
206 |
||
207 |
/*! |
|
208 |
Searches the char string \a str, which has length \a len, from |
|
209 |
byte position \a from (default 0, i.e. from the first byte), for |
|
210 |
the byte array pattern() that was set in the constructor or in the |
|
211 |
most recent call to setPattern(). Returns the position where the |
|
212 |
pattern() matched in \a str, or -1 if no match was found. |
|
213 |
*/ |
|
214 |
int QByteArrayMatcher::indexIn(const char *str, int len, int from) const |
|
215 |
{ |
|
216 |
if (from < 0) |
|
217 |
from = 0; |
|
218 |
return bm_find(reinterpret_cast<const uchar *>(str), len, from, |
|
219 |
p.p, p.l, p.q_skiptable); |
|
220 |
} |
|
221 |
||
222 |
/*! |
|
223 |
\fn QByteArray QByteArrayMatcher::pattern() const |
|
224 |
||
225 |
Returns the byte array pattern that this byte array matcher will |
|
226 |
search for. |
|
227 |
||
228 |
\sa setPattern() |
|
229 |
*/ |
|
230 |
||
231 |
||
232 |
static int findChar(const char *str, int len, char ch, int from) |
|
233 |
{ |
|
234 |
const uchar *s = (const uchar *)str; |
|
235 |
uchar c = (uchar)ch; |
|
236 |
if (from < 0) |
|
237 |
from = qMax(from + len, 0); |
|
238 |
if (from < len) { |
|
239 |
const uchar *n = s + from - 1; |
|
240 |
const uchar *e = s + len; |
|
241 |
while (++n != e) |
|
242 |
if (*n == c) |
|
243 |
return n - s; |
|
244 |
} |
|
245 |
return -1; |
|
246 |
} |
|
247 |
||
248 |
/*! \internal |
|
249 |
*/ |
|
250 |
static int qFindByteArrayBoyerMoore( |
|
251 |
const char *haystack, int haystackLen, int haystackOffset, |
|
252 |
const char *needle, int needleLen) |
|
253 |
{ |
|
254 |
uchar skiptable[256]; |
|
255 |
bm_init_skiptable((const uchar *)needle, needleLen, skiptable); |
|
256 |
if (haystackOffset < 0) |
|
257 |
haystackOffset = 0; |
|
258 |
return bm_find((const uchar *)haystack, haystackLen, haystackOffset, |
|
259 |
(const uchar *)needle, needleLen, skiptable); |
|
260 |
} |
|
261 |
||
262 |
#define REHASH(a) \ |
|
263 |
if (sl_minus_1 < sizeof(uint) * CHAR_BIT) \ |
|
264 |
hashHaystack -= (a) << sl_minus_1; \ |
|
265 |
hashHaystack <<= 1 |
|
266 |
||
267 |
/*! \internal |
|
268 |
*/ |
|
269 |
int qFindByteArray( |
|
270 |
const char *haystack0, int haystackLen, int from, |
|
271 |
const char *needle, int needleLen) |
|
272 |
{ |
|
273 |
const int l = haystackLen; |
|
274 |
const int sl = needleLen; |
|
275 |
if (from < 0) |
|
276 |
from += l; |
|
277 |
if (uint(sl + from) > (uint)l) |
|
278 |
return -1; |
|
279 |
if (!sl) |
|
280 |
return from; |
|
281 |
if (!l) |
|
282 |
return -1; |
|
283 |
||
284 |
if (sl == 1) |
|
285 |
return findChar(haystack0, haystackLen, needle[0], from); |
|
286 |
||
287 |
/* |
|
288 |
We use the Boyer-Moore algorithm in cases where the overhead |
|
289 |
for the skip table should pay off, otherwise we use a simple |
|
290 |
hash function. |
|
291 |
*/ |
|
292 |
if (l > 500 && sl > 5) |
|
293 |
return qFindByteArrayBoyerMoore(haystack0, haystackLen, from, |
|
294 |
needle, needleLen); |
|
295 |
||
296 |
/* |
|
297 |
We use some hashing for efficiency's sake. Instead of |
|
298 |
comparing strings, we compare the hash value of str with that |
|
299 |
of a part of this QString. Only if that matches, we call memcmp(). |
|
300 |
*/ |
|
301 |
const char *haystack = haystack0 + from; |
|
302 |
const char *end = haystack0 + (l - sl); |
|
303 |
const uint sl_minus_1 = sl - 1; |
|
304 |
uint hashNeedle = 0, hashHaystack = 0; |
|
305 |
int idx; |
|
306 |
for (idx = 0; idx < sl; ++idx) { |
|
307 |
hashNeedle = ((hashNeedle<<1) + needle[idx]); |
|
308 |
hashHaystack = ((hashHaystack<<1) + haystack[idx]); |
|
309 |
} |
|
310 |
hashHaystack -= *(haystack + sl_minus_1); |
|
311 |
||
312 |
while (haystack <= end) { |
|
313 |
hashHaystack += *(haystack + sl_minus_1); |
|
314 |
if (hashHaystack == hashNeedle && *needle == *haystack |
|
315 |
&& memcmp(needle, haystack, sl) == 0) |
|
316 |
return haystack - haystack0; |
|
317 |
||
318 |
REHASH(*haystack); |
|
319 |
++haystack; |
|
320 |
} |
|
321 |
return -1; |
|
322 |
} |
|
323 |
||
324 |
QT_END_NAMESPACE |