comparison common/Objects.h @ 10:d35e5d769c87 classical-rdf

* some experiments with composer name matching

 
     struct Ordering {
         bool operator()(Work *, Work *);
     };
 
+    /**
+     * Compare the ordering of two strings that are known to contain
+     * catalogue number texts, such as "Op. 1 no 4" and "Op. 3 no 2"
+     * (which should compare in that order).  Return value is as for
+     * strcmp.
+     */
+    //!!! todo: unit tests
+    static int compareCatalogueNumberTexts(QString a, QString b);
+
 private:
     QString m_key;
     QString m_opus;
     QString m_catalogue;
     QString m_number;

     Q_PROPERTY(QSet<QUrl> geonameURIs READ geonameURIs WRITE setGeonameURIs STORED true)
     Q_PROPERTY(QString period READ period WRITE setPeriod STORED true)
     Q_PROPERTY(ClassicalData::Birth *birth READ birth WRITE setBirth STORED true)
     Q_PROPERTY(ClassicalData::Death *death READ death WRITE setDeath STORED true)
 
+    Q_PROPERTY(QString surname READ getSurname STORED false)
+    Q_PROPERTY(QString forenames READ getForenames STORED false)
+
 public:
     Composer(QObject *parent = 0) : NamedEntity(parent), m_birth(0), m_death(0) { }
 
     QString gender() const { return m_gender; }
     void setGender(QString n) { m_gender = n; }

 
     Death *death() { return m_death; }
     const Death *death() const { return m_death; }
     void setDeath(Death *d) { m_death = d; }
 
-    bool datesMatch(const Composer *other) const; // "well enough"
+    QString getSurname() const;
+    QString getForenames() const;
     QString getSortName(bool caps) const;
     QString getDisplayDates() const;
+
+    /**
+     * Given another composer, return true if the other composer's
+     * dates match outs.  This is mostly intended (like
+     * matchCatalogueName) for use in merging distinct catalogues.
+     * Matching is somewhat fuzzy; more slack is cut when the dates
+     * are very long ago or are marked as approximate.
+     */
+    bool matchDates(const Composer *other) const; // "well enough"
+
+    /**
+     * Given another name which is intended to be a well-formatted
+     * catalogue name for a composer (but which may differ in
+     * ordering, number of forenames, and perhaps in spelling), test
+     * whether the name is a plausible match for our own.  This is
+     * mostly intended (like matchDates) for use in merging distinct
+     * catalogues.  Return true if the given name is highly likely to
+     * match our own.
+     */
+    bool matchCatalogueName(QString otherName) const;
+
+    /**
+     * Given another name which is believed to be a user-entered
+     * composer name with unpredictable formatting and spelling (and
+     * probably incomplete), return an estimate for the likelihood
+     * that the intended composer was this one.  Higher return values
+     * indicate greater confidence.
+     */
+    int matchFuzzyName(QString name) const;
+
+    /**
+     * Return the supplied name reduced into a "simplified" form,
+     * eliminating many of the differences often found particularly in
+     * European language names that have been anglicised.  Used in
+     * catalogue and fuzzy name matching.
+     */
+    static QString reduceName(QString name);
 
 private:
     QString m_gender;
     QSet<QString> m_nationality;
     QSet<QUrl> m_geonameURIs;