Ref T97, use fuzzy search for aircraft ICAO

src/blackmisc/aviation/aircrafticaocode.cpp

@@ -290,7 +290,7 @@ namespace BlackMisc
                 if (getEngineCount() != c.digitValue()) { return false; }
             }
             if (et == '*') { return true; }
-            const QString cet = getEngineType();
+            const QString cet = this->getEngineType();
             return cet.length() == 1 && cet.at(0) == et;
         }
 
@@ -417,22 +417,61 @@ namespace BlackMisc
             return hasValidCombinedType() && hasDesignator() && hasValidWtc() && hasManufacturer();
         }
 
-        bool CAircraftIcaoCode::matchesDesignator(const QString &designator) const
+        bool CAircraftIcaoCode::matchesDesignator(const QString &designator, int fuzzyMatch, int *result) const
         {
+            Q_ASSERT_X(fuzzyMatch >= -1 && fuzzyMatch <= 100, Q_FUNC_INFO, "fuzzyMatch range 0..100 or -1");
             if (designator.isEmpty()) { return false; }
-            return designator.trimmed().toUpper() == this->m_designator;
+            const QString d = designator.trimmed().toUpper();
+            if (fuzzyMatch >= 0)
+            {
+                const int r = fuzzyShortStringComparision(this->getDesignator(), d) >= fuzzyMatch;
+                if (result) { *result = r; }
+                return r >= fuzzyMatch;
+            }
+            else
+            {
+                const bool e = this->getDesignator() == d;
+                if (result) { *result = e ? 100 : 0; }
+                return e;
+            }
         }
 
-        bool CAircraftIcaoCode::matchesIataCode(const QString &iata) const
+        bool CAircraftIcaoCode::matchesIataCode(const QString &iata, int fuzzyMatch, int *result) const
         {
+            Q_ASSERT_X(fuzzyMatch >= -1 && fuzzyMatch <= 100, Q_FUNC_INFO, "fuzzyMatch range 0..100 or -1");
             if (iata.isEmpty()) { return false; }
-            return iata.trimmed().toUpper() == this->m_iataCode;
+            const QString i = iata.trimmed().toUpper();
+            if (fuzzyMatch >= 0)
+            {
+                const int r = fuzzyShortStringComparision(this->getIataCode(), i) >= fuzzyMatch;
+                if (result) { *result = r; }
+                return r >= fuzzyMatch;
+            }
+            else
+            {
+                const bool e = this->getIataCode() == i;
+                if (result) { *result = e ? 100 : 0; }
+                return e;
+            }
         }
 
-        bool CAircraftIcaoCode::matchesFamily(const QString &family) const
+        bool CAircraftIcaoCode::matchesFamily(const QString &family, int fuzzyMatch, int *result) const
         {
+            Q_ASSERT_X(fuzzyMatch >= -1 && fuzzyMatch <= 100, Q_FUNC_INFO, "fuzzyMatch range 0..100 or -1");
             if (family.isEmpty()) { return false; }
-            return family.trimmed().toUpper() == this->m_family;
+            const QString f = family.trimmed().toUpper();
+            if (fuzzyMatch >= 0)
+            {
+                const int r = fuzzyShortStringComparision(this->getFamily(), f) >= fuzzyMatch;
+                if (result) { *result = r; }
+                return r >= fuzzyMatch;
+            }
+            else
+            {
+                const bool e = this->getFamily() == f;
+                if (result) { *result = e ? 100 : 0; }
+                return e;
+            }
         }
 
         bool CAircraftIcaoCode::matchesDesignatorOrIata(const QString &icaoOrIata) const

src/blackmisc/aviation/aircrafticaocode.h

@@ -251,13 +251,14 @@ namespace BlackMisc
             bool hasCompleteData() const;
 
             //! Matches designator string?
-            bool matchesDesignator(const QString &designator) const;
+            bool matchesDesignator(const QString &designator, int fuzzyMatch = -1, int *result = nullptr) const;
 
             //! Matches IATA string?
-            bool matchesIataCode(const QString &iata) const;
+            //! \remark IATA codes are only 3 characters, so using fuzzy search might yield bad results
+            bool matchesIataCode(const QString &iata, int fuzzyMatch = -1, int *result = nullptr) const;
 
             //! Matches family?
-            bool matchesFamily(const QString &family) const;
+            bool matchesFamily(const QString &family, int fuzzyMatch = -1, int *result = nullptr) const;
 
             //! Matches ICAO or IATA code
             bool matchesDesignatorOrIata(const QString &icaoOrIata) const;

src/blackmisc/aviation/aircrafticaocodelist.cpp

@@ -25,15 +25,32 @@ namespace BlackMisc
             CSequence<CAircraftIcaoCode>(other)
         { }
 
-        CAircraftIcaoCodeList CAircraftIcaoCodeList::findByDesignator(const QString &designator) const
+        CAircraftIcaoCodeList CAircraftIcaoCodeList::findByDesignator(const QString &designator, int fuzzySearch) const
         {
-            if (!CAircraftIcaoCode::isValidDesignator(designator)) { return CAircraftIcaoCodeList(); }
+            if (!fuzzySearch && !CAircraftIcaoCode::isValidDesignator(designator)) { return CAircraftIcaoCodeList(); }
+            if (fuzzySearch && designator.length() < 3) { return CAircraftIcaoCodeList(); }
             return this->findBy([&](const CAircraftIcaoCode & code)
             {
-                return code.matchesDesignator(designator);
+                return code.matchesDesignator(designator, fuzzySearch);
             });
         }
 
+        CAircraftIcaoCode CAircraftIcaoCodeList::findBestFuzzyMatchOrDefault(const QString &designator, int cutoff) const
+        {
+            if (designator.length() < 3) { return CAircraftIcaoCode(); }
+            int best = 0;
+            int current = 0;
+            CAircraftIcaoCode found;
+            const QString d(designator.trimmed().toUpper());
+            for (const CAircraftIcaoCode &code : * this)
+            {
+                if (!code.matchesDesignator(d, cutoff, &current)) { continue; }
+                if (current == 100.0) { return code; }
+                if (best < current) { found = code; }
+            }
+            return found;
+        }
+
         CAircraftIcaoCodeList CAircraftIcaoCodeList::findByValidDesignator() const
         {
             return this->findBy([](const CAircraftIcaoCode & code)
@@ -88,21 +105,21 @@ namespace BlackMisc
             return icaosDesignator.isEmpty() ? icaosFamily : icaosDesignator;
         }
 
-        CAircraftIcaoCodeList CAircraftIcaoCodeList::findByIataCode(const QString &iata) const
+        CAircraftIcaoCodeList CAircraftIcaoCodeList::findByIataCode(const QString &iata, int fuzzySearch) const
         {
             if (iata.isEmpty()) { return CAircraftIcaoCodeList(); }
             return this->findBy([&](const CAircraftIcaoCode & code)
             {
-                return code.matchesIataCode(iata);
+                return code.matchesIataCode(iata, fuzzySearch);
             });
         }
 
-        CAircraftIcaoCodeList CAircraftIcaoCodeList::findByFamily(const QString &family) const
+        CAircraftIcaoCodeList CAircraftIcaoCodeList::findByFamily(const QString &family, int fuzzySearch) const
         {
             if (family.isEmpty()) { return CAircraftIcaoCodeList(); }
             return this->findBy([&](const CAircraftIcaoCode & code)
             {
-                return code.matchesFamily(family);
+                return code.matchesFamily(family, fuzzySearch);
             });
         }
 
@@ -338,22 +355,15 @@ namespace BlackMisc
                 codes = this->findByIataCode(designator);
                 if (!codes.isEmpty()) break;
 
-                // search by reduced length
+                // search fuzzy and restrict length
-                if (designator.length() >= 4)
+                const CAircraftIcaoCode bestMatch = this->findBestFuzzyMatchOrDefault(designator.length() < 5 ? designator : designator.left(5), 70);
-                {
+                if (bestMatch.hasValidDesignator()) { return bestMatch; }
-                    codes = this->findByDesignator(designator.left(4));
-                    if (!codes.isEmpty()) break;
-                }
 
                 // still empty, try to find by family
                 codes = this->findByFamily(designator);
                 if (!codes.isEmpty()) break;
 
-                // now try to find as ending
+                // by any description
-                codes = this->findEndingWith(designator);
-                if (!codes.isEmpty()) break;
-
-                // by any descriptionS
                 codes = this->findMatchingByAnyDescription(designator);
             }
             while (false);

src/blackmisc/aviation/aircrafticaocodelist.h

@@ -46,13 +46,16 @@ namespace BlackMisc
             CAircraftIcaoCodeList(const CSequence<CAircraftIcaoCode> &other);
 
             //! Find by designator
-            CAircraftIcaoCodeList findByDesignator(const QString &designator) const;
+            CAircraftIcaoCodeList findByDesignator(const QString &designator, int fuzzySearch = -1) const;
+
+            //! Find by designator
+            CAircraftIcaoCode findBestFuzzyMatchOrDefault(const QString &designator, int cutoff = 50) const;
 
             //! Find by IATA code
-            CAircraftIcaoCodeList findByIataCode(const QString &iata) const;
+            CAircraftIcaoCodeList findByIataCode(const QString &iata, int fuzzySearch = -1) const;
 
             //! Find by family
-            CAircraftIcaoCodeList findByFamily(const QString &family) const;
+            CAircraftIcaoCodeList findByFamily(const QString &family, int fuzzySearch = -1) const;
 
             //! Ones with a valid designator
             CAircraftIcaoCodeList findByValidDesignator() const;
@@ -67,7 +70,8 @@ namespace BlackMisc
             CAircraftIcaoCodeList findByDesignatorIataOrFamily(const QString &icaoIataOrFamily) const;
 
             //! Find code ending with string, e.g. "738" finds "B738"
-            //! \remark many users use wrong ICAO designators, one typical mistake is "738" for  "B737"
+            //! \remark many users use wrong ICAO designators, one typical mistake is "738" for "B737"
+            //! \remark consider to use findBestFuzzyMatchOrDefault
             CAircraftIcaoCodeList findEndingWith(const QString &icaoEnding) const;
 
             //! Find by manufacturer