Mercurial > dedupe
annotate FileDBLink.cpp @ 85:1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
| author | Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no> |
|---|---|
| date | Thu, 10 Oct 2013 15:55:30 +0200 |
| parents | 9744ec195be3 |
| children | af7962f3274b |
| rev | line source |
|---|---|
|
40
f711ddb56ae7
Sort up includes.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
34
diff
changeset
|
1 #include "FileDBLink.hpp" |
|
f711ddb56ae7
Sort up includes.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
34
diff
changeset
|
2 |
|
78
9744ec195be3
Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
64
diff
changeset
|
3 #include "CachedEditDistance.hpp" |
|
40
f711ddb56ae7
Sort up includes.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
34
diff
changeset
|
4 |
|
28
b2c2c2bf2bbd
Refactor Exceptions into a separate directory.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
15
diff
changeset
|
5 #include "Exception/PermissionException.hpp" |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
6 |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
7 #include <QtCore/QtConcurrentMap> |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
8 |
|
40
f711ddb56ae7
Sort up includes.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
34
diff
changeset
|
9 #include <boost/bind.hpp> |
|
f711ddb56ae7
Sort up includes.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
34
diff
changeset
|
10 |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
11 void FileDBLink::updateIfModified(const QString& path, bool lazy) |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
12 { |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
13 QFileInfo fileinfo(path); |
|
2
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
14 FileDBLink::DBStatus status = existsWithMtime(path, fileinfo.lastModified()); |
|
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
15 |
|
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
16 switch (status) { |
|
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
17 case FileDBLink::NONE: { |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
18 addFile(fileinfo, lazy); |
|
2
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
19 break; |
|
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
20 } |
|
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
21 case FileDBLink::MTIME_DIFFERENT: { |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
22 updateFile(fileinfo, lazy); |
|
2
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
23 } |
|
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
24 default: { |
|
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
25 } |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
26 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
27 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
28 |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
29 void FileDBLink::addFile(const QFileInfo& fileinfo, bool lazy) |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
30 { |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
31 addFile(fileinfo.absoluteFilePath(), fileinfo.size(), |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
32 fileinfo.lastModified(), lazy); |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
33 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
34 |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
35 QByteArray FileDBLink::computeHash(const QString& path, |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
36 QCryptographicHash::Algorithm algorithm) |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
37 { |
|
15
199fc63c60c1
Increase buffersize for filehash computing.
Tom Fredrik Blenning Klaussen <bfg@sim.no>
parents:
13
diff
changeset
|
38 const static uint buffersize = 32768; |
|
13
9463c0c22969
Refactor computeHash into separate function.
Tom Fredrik Blenning Klaussen <bfg@sim.no>
parents:
9
diff
changeset
|
39 QCryptographicHash hash(algorithm); |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
40 QFile file(path); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
41 if ( file.open( QIODevice::ReadOnly ) ) { |
|
13
9463c0c22969
Refactor computeHash into separate function.
Tom Fredrik Blenning Klaussen <bfg@sim.no>
parents:
9
diff
changeset
|
42 while(!file.atEnd()){ |
|
9463c0c22969
Refactor computeHash into separate function.
Tom Fredrik Blenning Klaussen <bfg@sim.no>
parents:
9
diff
changeset
|
43 hash.addData(file.read(buffersize)); |
|
9463c0c22969
Refactor computeHash into separate function.
Tom Fredrik Blenning Klaussen <bfg@sim.no>
parents:
9
diff
changeset
|
44 } |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
45 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
46 else { |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
47 QString errorMsg = path + ": " + file.errorString(); |
|
9
b5943e4bf676
Fix up header includes.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
2
diff
changeset
|
48 |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
49 switch (file.error()) { |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
50 case QFile::PermissionsError: |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
51 throw PermissionException(errorMsg); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
52 default: |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
53 throw IOException(errorMsg); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
54 } |
|
34
fda70a362ed5
Remove whitespace.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
31
diff
changeset
|
55 } |
|
13
9463c0c22969
Refactor computeHash into separate function.
Tom Fredrik Blenning Klaussen <bfg@sim.no>
parents:
9
diff
changeset
|
56 return hash.result(); |
|
9463c0c22969
Refactor computeHash into separate function.
Tom Fredrik Blenning Klaussen <bfg@sim.no>
parents:
9
diff
changeset
|
57 } |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
58 |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
59 bool FileDBLink::updateAllWithSize(qint64 size) |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
60 { |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
61 const QList<dbinf_ptr_t> others = filesWithSize(size); |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
62 if (!others.empty()) { |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
63 foreach( const dbinf_ptr_t other, others) { |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
64 if (other->checksum().isEmpty()) { |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
65 QByteArray ohash = computeHash(other->path()); |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
66 updateFile(other->path(), other->size(), other->mtime(), ohash); |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
67 } |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
68 } |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
69 return true; |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
70 } |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
71 return false; |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
72 } |
|
13
9463c0c22969
Refactor computeHash into separate function.
Tom Fredrik Blenning Klaussen <bfg@sim.no>
parents:
9
diff
changeset
|
73 |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
74 void FileDBLink::addFile(const QString& path, qint64 size, |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
75 const QDateTime& lastModified, bool lazy) |
|
13
9463c0c22969
Refactor computeHash into separate function.
Tom Fredrik Blenning Klaussen <bfg@sim.no>
parents:
9
diff
changeset
|
76 { |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
77 QByteArray hash; |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
78 // std::cout << path.toStdString() << "::" << lazy << std::endl; |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
79 if (!lazy || updateAllWithSize(size)) |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
80 hash = computeHash(path); |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
81 addFile(path, size, lastModified, hash); |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
82 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
83 |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
84 void FileDBLink::updateFile(const QFileInfo& fileinfo, bool lazy) |
|
2
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
85 { |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
86 updateFile(fileinfo.absoluteFilePath(), fileinfo.size(), |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
87 fileinfo.lastModified(), lazy); |
|
2
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
88 } |
|
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
89 |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
90 void FileDBLink::updateFile(const QString& path, qint64 size, |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
91 const QDateTime& lastModified, bool lazy) |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
92 { |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
93 QByteArray hash; |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
94 if (!lazy || updateAllWithSize(size)) |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
95 hash = computeHash(path); |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
96 updateFile(path, size, lastModified, hash); |
|
2
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
97 } |
|
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
98 |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
99 const QList<FileDBLink::dbinf_ptr_t > |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
100 FileDBLink::sortOn(const QString& prefix, SORTORDER order, bool extended) |
|
2
2833b7f8884a
Sql backend is working.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
1
diff
changeset
|
101 { |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
102 QList<dbinf_ptr_t > list = |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
103 (extended) ? computedValues(prefix) : values(prefix); |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
104 |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
105 switch (order) { |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
106 case PATH: |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
107 { |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
108 QList<FileDBLink::dbinf_ptr_t > oList; |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
109 foreach(dbinf_ptr_t info, list) { |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
110 oList.push_back(info); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
111 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
112 return oList; |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
113 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
114 case SIZE: |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
115 { |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
116 QMultiMap<quint64, dbinf_ptr_t > oList; |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
117 foreach(dbinf_ptr_t info, list) { |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
118 oList.insert(info->size(), info); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
119 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
120 return oList.values(); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
121 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
122 case MTIME: |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
123 { |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
124 QMultiMap<QDateTime, dbinf_ptr_t > oList; |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
125 foreach(dbinf_ptr_t info, list) { |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
126 oList.insert(info->mtime(), info); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
127 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
128 return oList.values(); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
129 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
130 case CHECKSUM: |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
131 { |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
132 QMultiMap<QByteArray, dbinf_ptr_t > oList; |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
133 foreach(dbinf_ptr_t info, list) { |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
134 oList.insert(info->checksum(), info); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
135 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
136 return oList.values(); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
137 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
138 case EDIT: |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
139 { |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
140 assert(extended); |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
141 QMultiMap<int, dbinf_ptr_t > oList; |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
142 foreach(dbinf_ptr_t info, list) { |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
143 QSharedPointer<ExtendedDBInfo> ptr; |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
144 ptr = info.dynamicCast<ExtendedDBInfo>(); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
145 oList.insert(ptr->editDistance(), info); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
146 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
147 return oList.values(); |
|
34
fda70a362ed5
Remove whitespace.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
31
diff
changeset
|
148 |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
149 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
150 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
151 abort(); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
152 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
153 |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
154 FileDBLink::dbinf_ptr_t |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
155 FileDBLink::computedValue(const dbinf_ptr_t& info, |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
156 const QList<dbinf_ptr_t >& entries) |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
157 { |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
158 QString p1 = info->name(); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
159 int minDist = 100000; |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
160 QString other; |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
161 for (QList<dbinf_ptr_t >::const_iterator it2 = entries.begin(); |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
162 it2 != entries.end(); ++it2) { |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
163 if (info == *it2) |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
164 continue; |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
165 QString p2 = (*it2)->name(); |
|
78
9744ec195be3
Encapsulate EditDistance with caching.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
64
diff
changeset
|
166 int dist = CachedEditDistance::Compute(p1, p2, false); |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
167 if (dist < minDist) { |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
168 minDist = dist; |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
169 other = (*it2)->path(); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
170 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
171 } |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
172 return dbinf_ptr_t(new ExtendedDBInfo(*info, other, minDist)); |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
173 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
174 |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
175 const QList<FileDBLink::dbinf_ptr_t> |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
176 FileDBLink::computedValues(const QString& prefix) const |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
177 { |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
178 QList<dbinf_ptr_t > list; |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
179 QList<dbinf_ptr_t > entries = values(prefix); |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
180 |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
181 #if 1 |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
182 list = |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
183 QtConcurrent::blockingMapped(entries, |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
184 boost::bind( &FileDBLink::computedValue, |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
185 _1, |
|
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
186 entries)); |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
187 #else |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
188 for (QList<dbinf_ptr_t >::const_iterator it1 = entries.begin(); |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
189 it1 != entries.end(); ++it1) { |
|
64
b9515dc35fe4
Make sure no file has greater linewidth than 80.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
62
diff
changeset
|
190 dbinf_ptr_t ext = computedValue(*it1, entries); |
|
0
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
191 list.push_back(ext); |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
192 } |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
193 #endif |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
194 return list; |
|
a3834af36579
Working with memory backend.
Tom Fredrik Blenning Klaussen <bfg@blenning.no>
parents:
diff
changeset
|
195 } |
|
85
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
196 |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
197 |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
198 const QList<FileDBLink::dbinf_ptr_t> |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
199 FileDBLink::filesWithSize(qint64 size, const QString& prefix) const |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
200 { |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
201 QList<dbinf_ptr_t> retVal; |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
202 const QList<dbinf_ptr_t> vals = values(prefix); |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
203 foreach (const dbinf_ptr_t val, vals) { |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
204 QString path = val->path(); |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
205 if (val->size() == size) { |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
206 retVal << val; |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
207 } |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
208 } |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
209 return retVal; |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
210 } |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
211 |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
212 FileDBLink::dbinf_ptr_t FileDBLink::value(const QString& path) const |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
213 { |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
214 const QList<dbinf_ptr_t> vals = values(); |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
215 foreach (const dbinf_ptr_t val, vals) { |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
216 QString vpath = val->path(); |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
217 if (vpath == path) { |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
218 return val; |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
219 } |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
220 } |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
221 return dbinf_ptr_t(); |
|
1f9e27a0bd7f
Allow for lazy calculation of checksums, ignore them, if only one file of given size.
Tom Fredrik Blenning Klaussen <bfg@bfgconsult.no>
parents:
78
diff
changeset
|
222 } |
