Mercurial > dedupe

annotate HuffmanSet.cpp @ 60:60c1e0a2cacf

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Fix compile problem.
author Tom Fredrik Blenning Klaussen <bfg@blenning.no>
date Fri, 14 Sep 2012 20:23:45 +0200
parents 19b2a2d98788
children e5fa379d4030
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
1 #include "HuffmanString.hpp"
40
f711ddb56ae7 Sort up includes.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 37
diff changeset
2
49
f8d0ea827db3 Use BitArray.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 40
diff changeset
3 #include "FastBitDecoder.hpp"
f8d0ea827db3 Use BitArray.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 40
diff changeset
4 #include "BitDecoder.hpp"
f8d0ea827db3 Use BitArray.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 40
diff changeset
5
40
f711ddb56ae7 Sort up includes.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 37
diff changeset
6 #include "Exception/InvalidDataException.hpp"
28
b2c2c2bf2bbd Refactor Exceptions into a separate directory.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 26
diff changeset
7 #include "Exception/NoSuchValueException.hpp"
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
8
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
9 #include <QtCore/QHash>
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
10
55
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
11 HuffmanSet::HuffmanSet() : cutoff(1024), numInserts(0), lut(0)
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
12 {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
13 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
14
55
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
15 HuffmanSet::~HuffmanSet()
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
16 {
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
17 delete lut;
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
18 }
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
19
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
20 void HuffmanSet::setCutoff(uint cutoff)
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
21 {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
22 this->cutoff = cutoff;
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
23 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
24
55
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
25 void HuffmanSet::setAutoRebuild(bool autoRebuild)
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
26 {
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
27 this->autoRebuild = autoRebuild;
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
28 }
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
29
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
30 QStringList HuffmanSet::chunks(const QString& str)
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
31 {
37
c52a0627337c BUGFIX: Chunking got extra values at beginning and end.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 34
diff changeset
32 return str.split("", QString::SkipEmptyParts);
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
33 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
34
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
35 BitDecoder* HuffmanSet::createLut(const QMap<QString, uint>& freqTable)
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
36 {
34
fda70a362ed5 Remove whitespace.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 28
diff changeset
37 QMultiMap<uint, BitDecoder* > freqs;
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
38 for(QMap<QString, uint>::const_iterator it = freqTable.begin();
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
39 it != freqTable.end(); ++it) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
40 freqs.insert(it.value(), new BitDecoder(it.key()));
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
41 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
42
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
43 if (freqs.size() == 1) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
44 QList<uint> keys = freqs.keys();
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
45 return freqs.take(keys[0]);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
46 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
47
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
48 QList<uint> keys = freqs.keys();
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
49
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
50 while (freqs.size() >= 2) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
51 QList<uint> keys = freqs.keys();
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
52
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
53 BitDecoder* v0 = freqs.take(keys[0]);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
54 BitDecoder* v1 = freqs.take(keys[1]);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
55
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
56 BitDecoder* n = BitDecoder::merge(v0, v1);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
57
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
58 freqs.insert(keys[0] + keys[1], n);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
59 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
60 BitDecoder* retVal = freqs.values()[0];
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
61 return retVal;
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
62 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
63
49
f8d0ea827db3 Use BitArray.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 40
diff changeset
64 QString HuffmanSet::decode(const BitArray& bits) const
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
65 {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
66 return lut->decode(bits);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
67 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
68
49
f8d0ea827db3 Use BitArray.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 40
diff changeset
69 BitArray HuffmanSet::encode(const QString& string, const QMap<QString, BitArray>& encoder)
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
70 {
49
f8d0ea827db3 Use BitArray.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 40
diff changeset
71 BitArray retVal;
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
72 QStringList c = chunks(string);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
73 foreach(const QString& fragment, c) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
74 if (encoder.contains(fragment))
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
75 retVal = BitDecoder::unite(retVal, encoder[fragment]);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
76 else
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
77 throw InvalidDataException();
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
78 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
79 return retVal;
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
80 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
81
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
82 uint HuffmanSet::totalElements() const
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
83 {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
84 return newStrings.size() + map.size();
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
85 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
86
55
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
87 void HuffmanSet::generateFreqTable(QMap<QString, uint>& freqTable) const
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
88 {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
89 foreach(key_t key, map.keys()) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
90 foreach(const QString& chunk, chunks(decode(map.value(key)))) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
91 ++freqTable[chunk];
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
92 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
93 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
94 foreach(key_t key, newStrings.keys()) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
95 foreach(const QString& chunk, chunks(newStrings.value(key))) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
96 ++freqTable[chunk];
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
97 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
98 }
55
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
99 }
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
100
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
101 QMap<QString, uint> HuffmanSet::generateFreqTable() const
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
102 {
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
103 QMap<QString, uint> freqTable;
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
104 generateFreqTable(freqTable);
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
105 return freqTable;
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
106 }
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
107
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
108
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
109 void HuffmanSet::rebuild()
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
110 {
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
111 QMap<QString, uint> freqTable;
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
112
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
113 generateFreqTable(freqTable);
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
114
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
115 BitDecoder* newLut = createLut(freqTable);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
116
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
117 encoder = newLut->createEncoder();
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
118
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
119 foreach(key_t key, map.keys()) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
120 map.insert(key, encode(decode(map.value(key)), encoder));
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
121 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
122 foreach(key_t key, newStrings.keys()) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
123 map.insert(key, encode(newStrings.value(key), encoder));
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
124 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
125 numInserts = 0;
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
126 delete lut;
55
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
127 #if FASTENCODE
49
f8d0ea827db3 Use BitArray.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 40
diff changeset
128 lut = new FastBitDecoder(encoder);
f8d0ea827db3 Use BitArray.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 40
diff changeset
129 delete newLut;
55
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
130 #else
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
131 lut = newLut;
55
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
132 #endif //FASTENCODE
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
133 newStrings.clear();
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
134 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
135
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
136 bool HuffmanSet::contains(key_t key) const
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
137 {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
138 return newStrings.contains(key) || map.contains(key);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
139 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
140
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
141 HuffmanSet::key_t HuffmanSet::hash(const QString& str)
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
142 {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
143 key_t key = qHash(str);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
144 while (contains(key) && value(key) != str)
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
145 ++key;
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
146 return key;
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
147 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
148
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
149 HuffmanSet::key_t HuffmanSet::insert(const QString& str)
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
150 {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
151 key_t key = hash(str);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
152 if (!contains(key)) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
153 try {
49
f8d0ea827db3 Use BitArray.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 40
diff changeset
154 BitArray bits = encode(str, encoder);
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
155 map.insert(key, bits);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
156 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
157 catch (InvalidDataException& e) {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
158 newStrings.insert(key, str);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
159 }
55
19b2a2d98788 Use define to multiplex between different types of decoders.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents: 49
diff changeset
160 if ((++numInserts >= cutoff) && autoRebuild) {
21
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
161 rebuild();
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
162 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
163 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
164 return key;
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
165 }
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
166
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
167 QString HuffmanSet::value(key_t key) const
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
168 {
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
169 if (map.contains(key))
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
170 return decode(map.value(key));
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
171 if (newStrings.contains(key))
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
172 return newStrings.value(key);
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
173 throw NoSuchValueException();
3bcdb8bb6914 Huffman representations.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff changeset
174 }