/* -*- 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.
*/

#ifndef SV_MODEL_H
#define SV_MODEL_H

#include <vector>
#include <atomic>

#include <QObject>
#include <QMutex>

#include "base/ById.h"
#include "base/XmlExportable.h"
#include "base/Playable.h"
#include "base/BaseTypes.h"
#include "base/DataExportOptions.h"

namespace sv {

class ZoomConstraint;
class AlignmentModel;

/** 
 * Model is the base class for all data models that represent any sort
 * of data on a time scale based on an audio frame rate.
 *
 * Model classes are expected to be thread-safe, particularly with
 * regard to content rather than metadata (e.g. populating a model
 * from a transform running in a background thread while displaying it
 * in a UI layer).
 *
 * Never store a pointer to a model unless it is completely private to
 * the code owning it. Models should be referred to using their
 * ModelId id and looked up from the ById pool when needed. This
 * returns a shared pointer, which ensures a sufficient lifespan for a
 * transient operation locally. See notes in ById.h. The document
 * unregisters models when it is closed, and then they are deleted
 * when the last shared pointer instance expires.
 */
class Model : public QObject,
              public WithTypedId<Model>,
              public XmlExportable,
              public Playable
{
    Q_OBJECT

public:
    typedef Id ModelId;
    
    virtual ~Model();

    /**
     * Return true if the model was constructed successfully.  Classes
     * that refer to the model should always test this before use.
     */
    virtual bool isOK() const = 0;

    /**
     * Return the first audio frame spanned by the model.
     */
    virtual sv_frame_t getStartFrame() const = 0;

    /**
     * Return the audio frame at the end of the model, i.e. the final
     * frame contained within the model plus 1 (rounded up to the
     * model's "resolution" granularity, if more than 1). The end
     * frame minus the start frame should yield the total duration in
     * frames (as a multiple of the resolution) spanned by the
     * model. This is broadly consistent with the definition of the
     * end frame of a Selection object.
     *
     * If the end has been extended by extendEndFrame() beyond the
     * true end frame, return the extended end instead. This is
     * usually the behaviour you want.
     */
    sv_frame_t getEndFrame() const {
        sv_frame_t trueEnd = getTrueEndFrame();
        if (m_extendTo > trueEnd) {
            return m_extendTo;
        } else {
            return trueEnd;
        }
    }

    /**
     * Return the audio frame at the end of the model. This is
     * identical to getEndFrame(), except that it ignores any extended
     * duration set with extendEndFrame().
     */
    virtual sv_frame_t getTrueEndFrame() const = 0;

    /**
     * Extend the end of the model. If this is set to something beyond
     * the true end of the data within the model, then getEndFrame()
     * will return this value instead of the true end. (This is used
     * by the Tony application.)
     */
    void extendEndFrame(sv_frame_t to) {
        m_extendTo = to;
    }

    /**
     * Return the frame rate in frames per second.
     */
    virtual sv_samplerate_t getSampleRate() const = 0;

    /**
     * Return the frame rate of the underlying material, if the model
     * itself has already been resampled.
     */
    virtual sv_samplerate_t getNativeRate() const { return getSampleRate(); }

    /**
     * Return the "work title" of the model, if known.
     */
    virtual QString getTitle() const;

    /**
     * Return the "artist" or "maker" of the model, if known.
     */
    virtual QString getMaker() const;

    /**
     * Return the location of the data in this model (e.g. source
     * URL).  This should not normally be returned for editable models
     * that have been edited.
     */
    virtual QString getLocation() const;

    /**
     * Return the type of the model.  For display purposes only.
     */
    virtual QString getTypeName() const = 0;

    /**
     * Return true if this is a sparse model.
     */
    virtual bool isSparse() const { return false; }

    /**
     * Return true if the model has finished loading or calculating
     * all its data, for a model that is capable of calculating in a
     * background thread.
     *
     * If "completion" is non-NULL, return through it an estimated
     * percentage value showing how far through the background
     * operation it thinks it is (for progress reporting). This should
     * be identical to the value returned by getCompletion().
     *
     * A model that carries out all its calculation from the
     * constructor or accessor functions would typically return true
     * (and completion == 100) as long as isOK() is true. Other models
     * may make the return value here depend on the internal
     * completion status.
     *
     * See also getCompletion().
     */
    virtual bool isReady(int *cp = nullptr) const {
        int c = getCompletion();
        if (cp) *cp = c;
        if (!isOK()) return false;
        else return (c == 100);
    }
    
    /**
     * Return an estimated percentage value showing how far through
     * any background operation used to calculate or load the model
     * data the model thinks it is. Must return 100 when the model is
     * complete.
     *
     * A model that carries out all its calculation from the
     * constructor or accessor functions might return 0 if isOK() is
     * false and 100 if isOK() is true. Other models may make the
     * return value here depend on the internal completion status.
     *
     * See also isReady().
     */
    virtual int getCompletion() const = 0;

    /**
     * If this model imposes a zoom constraint, i.e. some limit to the
     * set of resolutions at which its data can meaningfully be
     * displayed, then return it.
     */
    virtual const ZoomConstraint *getZoomConstraint() const {
        return 0;
    }

    /**
     * If this model was derived from another, return the id of the
     * model it was derived from.  The assumption is that the source
     * model's alignment will also apply to this model, unless some
     * other property (such as a specific alignment model set on this
     * model) indicates otherwise.
     */
    virtual ModelId getSourceModel() const {
        return m_sourceModel;
    }

    /**
     * Set the source model for this model.
     */
    virtual void setSourceModel(ModelId model);

    /**
     * Specify an alignment between this model's timeline and that of
     * a reference model. The alignment model, of type AlignmentModel,
     * records both the reference and the alignment.
     */
    virtual void setAlignment(ModelId alignmentModel);

    /**
     * Retrieve the alignment model for this model.  This is not a
     * generally useful function, as the alignment you really want may
     * be performed by the source model instead.  You should normally
     * use getAlignmentReference, alignToReference and
     * alignFromReference instead of this.  The main intended
     * application for this function is in streaming out alignments to
     * the session file.
     */
    virtual const ModelId getAlignment() const;

    /**
     * Return the reference model for the current alignment timeline,
     * if any.
     */
    virtual const ModelId getAlignmentReference() const;

    /**
     * Return the frame number of the reference model that corresponds
     * to the given frame number in this model.
     */
    virtual sv_frame_t alignToReference(sv_frame_t frame) const;

    /**
     * Return the frame number in this model that corresponds to the
     * given frame number of the reference model.
     */
    virtual sv_frame_t alignFromReference(sv_frame_t referenceFrame) const;

    /**
     * Return the completion percentage for the alignment model: 100
     * if there is no alignment model or it has been entirely
     * calculated, or less than 100 if it is still being calculated.
     */
    virtual int getAlignmentCompletion() const;

    /**
     * Set the event, feature, or signal type URI for the features
     * contained in this model, according to the Audio Features RDF
     * ontology.
     */
    void setRDFTypeURI(QString uri) { m_typeUri = uri; }

    /**
     * Retrieve the event, feature, or signal type URI for the
     * features contained in this model, if previously set with
     * setRDFTypeURI.
     */
    QString getRDFTypeURI() const { return m_typeUri; }

    void toXml(QTextStream &stream,
               QString indent = "",
               QString extraAttributes = "") const override;

    /**
     * Return a label for each column that would be written by
     * toStringExportRows.
     */
    virtual QVector<QString>
    getStringExportHeaders(DataExportOptions options) const = 0;
    
    /**
     * Emit events starting within the given range as string rows
     * ready for conversion to an e.g. comma-separated data format.
     */
    virtual QVector<QVector<QString>>
    toStringExportRows(DataExportOptions options,
                       sv_frame_t startFrame,
                       sv_frame_t duration) const = 0;

signals:
    /**
     * Emitted when a model has been edited (or more data retrieved
     * from cache, in the case of a cached model that generates slowly)
     */
    void modelChanged(ModelId myId);

    /**
     * Emitted when a model has been edited (or more data retrieved
     * from cache, in the case of a cached model that generates slowly)
     */
    void modelChangedWithin(ModelId myId, sv_frame_t startFrame, sv_frame_t endFrame);

    /**
     * Emitted when some internal processing has advanced a stage, but
     * the model has not changed externally.  Views should respond by
     * updating any progress meters or other monitoring, but not
     * refreshing the actual view.
     */
    void completionChanged(ModelId myId);

    /**
     * Emitted when internal processing is complete (i.e. when
     * isReady() would return true, with completion at 100).
     */
    void ready(ModelId myId);

    /**
     * Emitted when the completion percentage changes for the
     * calculation of this model's alignment model. (The ModelId
     * provided is that of this model, not the alignment model.)
     */
    void alignmentCompletionChanged(ModelId myId);

private slots:
    void alignmentModelCompletionChanged(ModelId);
    
protected:
    Model() : m_extendTo(0) { }

    // Not provided.
    Model(const Model &) =delete;
    Model &operator=(const Model &) =delete;

    mutable QMutex m_mutex;
    ModelId m_sourceModel;
    ModelId m_alignmentModel;
    QString m_typeUri;
    std::atomic<sv_frame_t> m_extendTo;
};

typedef Model::Id ModelId;
typedef TypedById<Model, Model::Id> ModelById;

} // end namespace sv

#endif