Chris@392
|
1 /* -*- c-basic-offset: 4 indent-tabs-mode: nil -*- vi:set ts=8 sts=4 sw=4: */
|
Chris@392
|
2
|
Chris@392
|
3 /*
|
Chris@392
|
4 Sonic Visualiser
|
Chris@392
|
5 An audio file viewer and annotation editor.
|
Chris@392
|
6 Centre for Digital Music, Queen Mary, University of London.
|
Chris@392
|
7 This file copyright 2006 Chris Cannam.
|
Chris@392
|
8
|
Chris@392
|
9 This program is free software; you can redistribute it and/or
|
Chris@392
|
10 modify it under the terms of the GNU General Public License as
|
Chris@392
|
11 published by the Free Software Foundation; either version 2 of the
|
Chris@392
|
12 License, or (at your option) any later version. See the file
|
Chris@392
|
13 COPYING included with this distribution for more information.
|
Chris@392
|
14 */
|
Chris@392
|
15
|
Chris@1362
|
16 #ifndef SV_CSV_FORMAT_H
|
Chris@1362
|
17 #define SV_CSV_FORMAT_H
|
Chris@392
|
18
|
Chris@392
|
19 #include <QString>
|
Chris@392
|
20 #include <QStringList>
|
Chris@392
|
21
|
Chris@1585
|
22 #include <set>
|
Chris@1585
|
23
|
Chris@1047
|
24 #include "base/BaseTypes.h"
|
Chris@1047
|
25
|
Chris@392
|
26 class CSVFormat
|
Chris@392
|
27 {
|
Chris@392
|
28 public:
|
Chris@392
|
29 enum ModelType {
|
Chris@1429
|
30 OneDimensionalModel,
|
Chris@1429
|
31 TwoDimensionalModel,
|
Chris@628
|
32 TwoDimensionalModelWithDuration,
|
Chris@897
|
33 TwoDimensionalModelWithDurationAndPitch,
|
Chris@1793
|
34 TwoDimensionalModelWithDurationAndExtent,
|
Chris@1488
|
35 ThreeDimensionalModel,
|
Chris@1488
|
36 WaveFileModel
|
Chris@392
|
37 };
|
Chris@392
|
38
|
Chris@392
|
39 enum TimingType {
|
Chris@1429
|
40 ExplicitTiming,
|
Chris@1429
|
41 ImplicitTiming
|
Chris@392
|
42 };
|
Chris@628
|
43
|
Chris@392
|
44 enum TimeUnits {
|
Chris@1429
|
45 TimeSeconds,
|
Chris@990
|
46 TimeMilliseconds,
|
Chris@1429
|
47 TimeAudioFrames,
|
Chris@1429
|
48 TimeWindows,
|
Chris@392
|
49 };
|
Chris@392
|
50
|
Chris@629
|
51 enum ColumnPurpose {
|
Chris@629
|
52 ColumnUnknown,
|
Chris@629
|
53 ColumnStartTime,
|
Chris@629
|
54 ColumnEndTime,
|
Chris@629
|
55 ColumnDuration,
|
Chris@629
|
56 ColumnValue,
|
Chris@897
|
57 ColumnPitch,
|
Chris@629
|
58 ColumnLabel
|
Chris@629
|
59 };
|
Chris@629
|
60
|
Chris@629
|
61 enum ColumnQuality {
|
Chris@1512
|
62 ColumnNumeric = 1, // No non-numeric values were seen in sample
|
Chris@1512
|
63 ColumnIntegral = 2, // All sampled values were integers
|
Chris@1512
|
64 ColumnIncreasing = 4, // Sampled values were monotonically increasing
|
Chris@1512
|
65 ColumnSmall = 8, // All sampled values had magnitude < 1
|
Chris@1512
|
66 ColumnLarge = 16, // Values "quickly" grew to over 1000
|
Chris@1512
|
67 ColumnSigned = 32, // Some negative values were seen
|
Chris@1512
|
68 ColumnNearEmpty = 64, // Nothing in this column beyond first row
|
Chris@629
|
69 };
|
Chris@629
|
70 typedef unsigned int ColumnQualities;
|
Chris@392
|
71
|
Chris@1515
|
72 enum AudioSampleRange {
|
Chris@1515
|
73 SampleRangeSigned1 = 0, // -1 .. 1
|
Chris@1515
|
74 SampleRangeUnsigned255, // 0 .. 255
|
Chris@1515
|
75 SampleRangeSigned32767, // -32768 .. 32767
|
Chris@1515
|
76 SampleRangeOther // Other/unknown: Normalise on load
|
Chris@1515
|
77 };
|
Chris@1515
|
78
|
Chris@392
|
79 CSVFormat() : // arbitrary defaults
|
Chris@392
|
80 m_modelType(TwoDimensionalModel),
|
Chris@392
|
81 m_timingType(ExplicitTiming),
|
Chris@392
|
82 m_timeUnits(TimeSeconds),
|
Chris@1585
|
83 m_separator(""),
|
Chris@392
|
84 m_sampleRate(44100),
|
Chris@392
|
85 m_windowSize(1024),
|
Chris@629
|
86 m_columnCount(0),
|
Chris@629
|
87 m_variableColumnCount(false),
|
Chris@1516
|
88 m_audioSampleRange(SampleRangeOther),
|
Chris@629
|
89 m_allowQuoting(true),
|
Chris@629
|
90 m_maxExampleCols(0)
|
Chris@392
|
91 { }
|
Chris@629
|
92
|
Chris@629
|
93 CSVFormat(QString path); // guess format
|
Chris@629
|
94
|
Chris@629
|
95 /**
|
Chris@629
|
96 * Guess the format of the given CSV file, setting the fields in
|
Chris@629
|
97 * this object accordingly. If the current separator is the empty
|
Chris@629
|
98 * string, the separator character will also be guessed; otherwise
|
Chris@629
|
99 * the current separator will be used. The other properties of
|
Chris@629
|
100 * this object will be set according to guesses from the file.
|
Chris@1524
|
101 *
|
Chris@1524
|
102 * The properties that are guessed from the file contents are:
|
Chris@1524
|
103 * separator, column count, variable-column-count flag, audio
|
Chris@1524
|
104 * sample range, timing type, time units, column qualities, column
|
Chris@1524
|
105 * purposes, and model type. The sample rate and window size
|
Chris@1524
|
106 * cannot be guessed and will not be changed by this function.
|
Chris@1524
|
107 * Note also that this function will never guess WaveFileModel for
|
Chris@1524
|
108 * the model type.
|
Chris@1524
|
109 *
|
Chris@1524
|
110 * Return false if there is some fundamental error, e.g. the file
|
Chris@1524
|
111 * could not be opened at all. Return true otherwise. Note that
|
Chris@1524
|
112 * this function returns true even if the file doesn't appear to
|
Chris@1524
|
113 * make much sense as a data format.
|
Chris@629
|
114 */
|
Chris@1524
|
115 bool guessFormatFor(QString path);
|
Chris@628
|
116
|
Chris@628
|
117 ModelType getModelType() const { return m_modelType; }
|
Chris@628
|
118 TimingType getTimingType() const { return m_timingType; }
|
Chris@628
|
119 TimeUnits getTimeUnits() const { return m_timeUnits; }
|
Chris@1047
|
120 sv_samplerate_t getSampleRate() const { return m_sampleRate; }
|
Chris@929
|
121 int getWindowSize() const { return m_windowSize; }
|
Chris@630
|
122 int getColumnCount() const { return m_columnCount; }
|
Chris@1516
|
123 AudioSampleRange getAudioSampleRange() const { return m_audioSampleRange; }
|
Chris@631
|
124 bool getAllowQuoting() const { return m_allowQuoting; }
|
Chris@631
|
125 QChar getSeparator() const {
|
Chris@1585
|
126 if (m_separator == "") return ',';
|
Chris@631
|
127 else return m_separator[0];
|
Chris@631
|
128 }
|
Chris@1585
|
129 // set rather than QSet to ensure a fixed order
|
Chris@1585
|
130 std::set<QChar> getPlausibleSeparators() const {
|
Chris@1585
|
131 return m_plausibleSeparators;
|
Chris@1585
|
132 }
|
Chris@630
|
133
|
Chris@628
|
134 void setModelType(ModelType t) { m_modelType = t; }
|
Chris@628
|
135 void setTimingType(TimingType t) { m_timingType = t; }
|
Chris@628
|
136 void setTimeUnits(TimeUnits t) { m_timeUnits = t; }
|
Chris@631
|
137 void setSeparator(QChar s) { m_separator = s; }
|
Chris@1047
|
138 void setSampleRate(sv_samplerate_t r) { m_sampleRate = r; }
|
Chris@1009
|
139 void setWindowSize(int s) { m_windowSize = s; }
|
Chris@630
|
140 void setColumnCount(int c) { m_columnCount = c; }
|
Chris@1516
|
141 void setAudioSampleRange(AudioSampleRange r) { m_audioSampleRange = r; }
|
Chris@631
|
142 void setAllowQuoting(bool q) { m_allowQuoting = q; }
|
Chris@392
|
143
|
Chris@631
|
144 QList<ColumnPurpose> getColumnPurposes() const { return m_columnPurposes; }
|
Chris@629
|
145 void setColumnPurposes(QList<ColumnPurpose> cl) { m_columnPurposes = cl; }
|
Chris@631
|
146
|
Chris@631
|
147 ColumnPurpose getColumnPurpose(int i);
|
Chris@631
|
148 ColumnPurpose getColumnPurpose(int i) const;
|
Chris@631
|
149 void setColumnPurpose(int i, ColumnPurpose p);
|
Chris@392
|
150
|
Chris@629
|
151 // read-only; only valid if format has been guessed:
|
Chris@1510
|
152 const QList<ColumnQualities> &getColumnQualities() const {
|
Chris@1510
|
153 return m_columnQualities;
|
Chris@1510
|
154 }
|
Chris@629
|
155
|
Chris@629
|
156 // read-only; only valid if format has been guessed:
|
Chris@1510
|
157 const QList<QStringList> &getExample() const {
|
Chris@1510
|
158 return m_example;
|
Chris@1510
|
159 }
|
Chris@1510
|
160
|
Chris@392
|
161 int getMaxExampleCols() const { return m_maxExampleCols; }
|
Chris@1429
|
162
|
Chris@392
|
163 protected:
|
Chris@628
|
164 ModelType m_modelType;
|
Chris@628
|
165 TimingType m_timingType;
|
Chris@628
|
166 TimeUnits m_timeUnits;
|
Chris@1585
|
167 QString m_separator; // "" or a single char - basically QChar option
|
Chris@1585
|
168 std::set<QChar> m_plausibleSeparators;
|
Chris@1047
|
169 sv_samplerate_t m_sampleRate;
|
Chris@929
|
170 int m_windowSize;
|
Chris@392
|
171
|
Chris@629
|
172 int m_columnCount;
|
Chris@629
|
173 bool m_variableColumnCount;
|
Chris@629
|
174
|
Chris@629
|
175 QList<ColumnQualities> m_columnQualities;
|
Chris@629
|
176 QList<ColumnPurpose> m_columnPurposes;
|
Chris@629
|
177
|
Chris@1515
|
178 AudioSampleRange m_audioSampleRange;
|
Chris@1515
|
179
|
Chris@629
|
180 QList<float> m_prevValues;
|
Chris@629
|
181
|
Chris@629
|
182 bool m_allowQuoting;
|
Chris@392
|
183
|
Chris@392
|
184 QList<QStringList> m_example;
|
Chris@392
|
185 int m_maxExampleCols;
|
Chris@629
|
186
|
Chris@629
|
187 void guessSeparator(QString line);
|
Chris@629
|
188 void guessQualities(QString line, int lineno);
|
Chris@629
|
189 void guessPurposes();
|
Chris@1515
|
190 void guessAudioSampleRange();
|
Chris@392
|
191 };
|
Chris@392
|
192
|
Chris@392
|
193 #endif
|