annotate data/fileio/CSVFormat.cpp @ 1247:8f076d02569a piper

Make SVDEBUG always write to a log file -- formerly this was disabled in NDEBUG builds. I think there's little use to that, it just means that we keep adding more cerr debug output because we aren't getting the log we need. And SVDEBUG logging is not usually used in tight loops, I don't think the performance overhead is too serious.

Also update the About box.

/* -*- c-basic-offset: 4 indent-tabs-mode: nil -*-  vi:set ts=8 sts=4 sw=4: */

/*
    Sonic Visualiser
    An audio file viewer and annotation editor.
    Centre for Digital Music, Queen Mary, University of London.
    This file copyright 2006 Chris Cannam.
    
    This program is free software; you can redistribute it and/or
    modify it under the terms of the GNU General Public License as
    published by the Free Software Foundation; either version 2 of the
    License, or (at your option) any later version.  See the file
    COPYING included with this distribution for more information.
*/

#include "CSVFormat.h"

#include "base/StringBits.h"

#include <QFile>
#include <QString>
#include <QRegExp>
#include <QStringList>
#include <QTextStream>

#include <iostream>

CSVFormat::CSVFormat(QString path) :
    m_separator(""),
    m_sampleRate(44100),
    m_windowSize(1024),
    m_allowQuoting(true)
{
    guessFormatFor(path);
}

void
CSVFormat::guessFormatFor(QString path)
{
    m_modelType = TwoDimensionalModel;
    m_timingType = ExplicitTiming;
    m_timeUnits = TimeSeconds;

    m_maxExampleCols = 0;
    m_columnCount = 0;
    m_variableColumnCount = false;

    m_example.clear();
    m_columnQualities.clear();
    m_columnPurposes.clear();
    m_prevValues.clear();

    QFile file(path);
    if (!file.exists()) return;
    if (!file.open(QIODevice::ReadOnly | QIODevice::Text)) return;

    QTextStream in(&file);
    in.seek(0);

    int lineno = 0;

    while (!in.atEnd()) {

        // See comment about line endings in CSVFileReader::load() 

        QString chunk = in.readLine();
        QStringList lines = chunk.split('\r', QString::SkipEmptyParts);

        for (int li = 0; li < lines.size(); ++li) {

            QString line = lines[li];
            if (line.startsWith("#") || line == "") continue;

            guessQualities(line, lineno);

            ++lineno;
        }

        if (lineno >= 50) break;
    }

    guessPurposes();
}

void
CSVFormat::guessSeparator(QString line)
{
    char candidates[] = { ',', '\t', ' ', '|', '/', ':' };
    for (int i = 0; i < int(sizeof(candidates)/sizeof(candidates[0])); ++i) {
        if (StringBits::split(line, candidates[i], m_allowQuoting).size() >= 2) {
            m_separator = candidates[i];
            return;
        }
    }
    m_separator = " ";
}

void
CSVFormat::guessQualities(QString line, int lineno)
{
    if (m_separator == "") guessSeparator(line);

    QStringList list = StringBits::split(line, m_separator[0], m_allowQuoting);

    int cols = list.size();
    if (lineno == 0 || (cols > m_columnCount)) m_columnCount = cols;
    if (cols != m_columnCount) m_variableColumnCount = true;

    // All columns are regarded as having these qualities until we see
    // something that indicates otherwise:

    ColumnQualities defaultQualities =
        ColumnNumeric | ColumnIntegral | ColumnIncreasing | ColumnNearEmpty;
    
    for (int i = 0; i < cols; ++i) {
	    
        while (m_columnQualities.size() <= i) {
            m_columnQualities.push_back(defaultQualities);
            m_prevValues.push_back(0.f);
        }

        QString s(list[i]);
        bool ok = false;

        ColumnQualities qualities = m_columnQualities[i];

        bool numeric    = (qualities & ColumnNumeric);
        bool integral   = (qualities & ColumnIntegral);
        bool increasing = (qualities & ColumnIncreasing);
        bool large      = (qualities & ColumnLarge); // this one defaults to off
        bool emptyish   = (qualities & ColumnNearEmpty);

        if (lineno > 1 && s.trimmed() != "") {
            emptyish = false;
        }
        
        float value = 0.f;

        //!!! how to take into account headers?

        if (numeric) {
            value = s.toFloat(&ok);
            if (!ok) {
                value = (float)StringBits::stringToDoubleLocaleFree(s, &ok);
            }
            if (ok) {
                if (lineno < 2 && value > 1000.f) large = true;
            } else {
                numeric = false;
            }
        }

        if (numeric) {

            if (integral) {
                if (s.contains('.') || s.contains(',')) {
                    integral = false;
                }
            }

            if (increasing) {
                if (lineno > 0 && value <= m_prevValues[i]) {
                    increasing = false;
                }
            }

            m_prevValues[i] = value;
        }

        m_columnQualities[i] =
            (numeric    ? ColumnNumeric : 0) |
            (integral   ? ColumnIntegral : 0) |
            (increasing ? ColumnIncreasing : 0) |
            (large      ? ColumnLarge : 0) |
            (emptyish   ? ColumnNearEmpty : 0);
    }

    if (lineno < 10) {
        m_example.push_back(list);
        if (lineno == 0 || cols > m_maxExampleCols) {
            m_maxExampleCols = cols;
        }
    }

//    cerr << "Estimated column qualities: ";
//    for (int i = 0; i < m_columnCount; ++i) {
//        cerr << int(m_columnQualities[i]) << " ";
//    }
//    cerr << endl;
}

void
CSVFormat::guessPurposes()
{
    m_timingType = CSVFormat::ImplicitTiming;
    m_timeUnits = CSVFormat::TimeWindows;
	
    int timingColumnCount = 0;

    // if our first column has zero or one entries in it and the rest
    // have more, then we'll default to ignoring the first column and
    // counting the next one as primary. (e.g. Sonic Annotator output
    // with filename at start of first column.)

    int primaryColumnNo = 0;

    if (m_columnCount >= 2) {
        if ( (m_columnQualities[0] & ColumnNearEmpty) &&
            !(m_columnQualities[1] & ColumnNearEmpty)) {
            primaryColumnNo = 1;
        }
    }
    
    for (int i = 0; i < m_columnCount; ++i) {
        
        ColumnPurpose purpose = ColumnUnknown;

        if (i < primaryColumnNo) {
            setColumnPurpose(i, purpose);
            continue;
        }
        
        bool primary = (i == primaryColumnNo);

        ColumnQualities qualities = m_columnQualities[i];

        bool numeric    = (qualities & ColumnNumeric);
        bool integral   = (qualities & ColumnIntegral);
        bool increasing = (qualities & ColumnIncreasing);
        bool large      = (qualities & ColumnLarge);

        bool timingColumn = (numeric && increasing);

        if (timingColumn) {

            ++timingColumnCount;
                              
            if (primary) {

                purpose = ColumnStartTime;

                m_timingType = ExplicitTiming;

                if (integral && large) {
                    m_timeUnits = TimeAudioFrames;
                } else {
                    m_timeUnits = TimeSeconds;
                }

            } else {

                if (timingColumnCount == 2 && m_timingType == ExplicitTiming) {
                    purpose = ColumnEndTime;
                }
            }
        }

        if (purpose == ColumnUnknown) {
            if (numeric) {
                purpose = ColumnValue;
            } else {
                purpose = ColumnLabel;
            }
        }

        setColumnPurpose(i, purpose);
    }            

    int valueCount = 0;
    for (int i = 0; i < m_columnCount; ++i) {
        if (m_columnPurposes[i] == ColumnValue) ++valueCount;
    }

    if (valueCount == 2 && timingColumnCount == 1) {
        // If we have exactly two apparent value columns and only one
        // timing column, but one value column is integral and the
        // other is not, guess that whichever one matches the integral
        // status of the time column is either duration or end time
        if (m_timingType == ExplicitTiming) {
            int a = -1, b = -1;
            for (int i = 0; i < m_columnCount; ++i) {
                if (m_columnPurposes[i] == ColumnValue) {
                    if (a == -1) a = i;
                    else b = i;
                }
            }
            if ((m_columnQualities[a] & ColumnIntegral) !=
                (m_columnQualities[b] & ColumnIntegral)) {
                int timecol = a;
                if ((m_columnQualities[a] & ColumnIntegral) !=
                    (m_columnQualities[0] & ColumnIntegral)) {
                    timecol = b;
                }
                if (m_columnQualities[timecol] & ColumnIncreasing) {
                    // This shouldn't happen; should have been settled above
                    m_columnPurposes[timecol] = ColumnEndTime;
                } else {
                    m_columnPurposes[timecol] = ColumnDuration;
                }
                --valueCount;
            }
        }
    }

    if (timingColumnCount > 1) {
        m_modelType = TwoDimensionalModelWithDuration;
    } else {
        if (valueCount == 0) {
            m_modelType = OneDimensionalModel;
        } else if (valueCount == 1) {
            m_modelType = TwoDimensionalModel;
        } else {
            m_modelType = ThreeDimensionalModel;
        }
    }

//    cerr << "Estimated column purposes: ";
//    for (int i = 0; i < m_columnCount; ++i) {
//        cerr << int(m_columnPurposes[i]) << " ";
//    }
//    cerr << endl;

//    cerr << "Estimated model type: " << m_modelType << endl;
//    cerr << "Estimated timing type: " << m_timingType << endl;
//    cerr << "Estimated units: " << m_timeUnits << endl;
}

CSVFormat::ColumnPurpose
CSVFormat::getColumnPurpose(int i)
{
    while (m_columnPurposes.size() <= i) {
        m_columnPurposes.push_back(ColumnUnknown);
    }
    return m_columnPurposes[i];
}

CSVFormat::ColumnPurpose
CSVFormat::getColumnPurpose(int i) const
{
    if (m_columnPurposes.size() <= i) {
        return ColumnUnknown;
    }
    return m_columnPurposes[i];
}

void
CSVFormat::setColumnPurpose(int i, ColumnPurpose p)
{
    while (m_columnPurposes.size() <= i) {
        m_columnPurposes.push_back(ColumnUnknown);
    }
    m_columnPurposes[i] = p;
}