Ticket #6531: patch.checkExistingChannelForXmlTvGrabber.diff

channeldata.

@@ -124 +124 @@
     return(chanid);
 }
 
+
+//
+// This is an implementation of JaroWinkler : http://en.wikipedia.org/wiki/Jaro-Winkler
+//
+// Jaro is working with a formula where we look to the:
+//      Number of matching characters : characters that are both in t1 & t2 with an index position diff < (max(t1.length, t2.length)/2) -1
+//      Number of transpose characters : number of characters that are not in the same place in t1 & t2
+// JaroWinkler is an extension of Jaro where we apply a coefficient to the start of the string
+//
+// the code is transpose from : http://www.cppfrance.com/codes/DISTANCE-JARO-WINKLER_49753.aspx
+//
+
+QString TrouverMatches(const QString &txt,int bl[])
+{
+    QString res;
+    QChar ctmp='a';
+
+    int i=0;
+    while (i<txt.length())
+    {
+        ctmp=txt[i];
+        if (bl[i]==true)
+        {
+            res.append(ctmp);
+        }
+        i++;
+    }
+    return res;
+}
+
+double JaroWinkler(const QString &t1,const QString &t2)
+{
+    int ecartMax,l1,l2,compteMatching,compteTransposition,longueurPrefix,i,j;
+    QString t1Matche,t2Matche;
+    double distanceJaro;
+
+    l1=t1.length();
+    l2=t2.length();
+    ecartMax=(int)(((double)max(l1,l2))/2.00);
+    compteMatching=0;
+
+    VERBOSE(VB_XMLTV, QString("input (%1) (%2)").arg(t1).arg(t2));
+    VERBOSE(VB_XMLTV, QString("ecartMax (%1)").arg(ecartMax));
+
+    int b1[l1+2],b2[l2+2];
+
+    for (i=0;i<l1;i++)
+        b1[i]=false;
+    for (i=0;i<l2;i++)
+        b2[i]=false;
+
+    for (i=0;i<l1;i++)
+    {
+        for (j=max(i-ecartMax,0);j<min(i+ecartMax,l2);j++)
+        {
+            if (t1[i]==t2[j])
+            {
+                b1[i]=true; 
+                b2[i]=true;
+                compteMatching++; 
+                break;
+            }
+        }
+    }
+
+    VERBOSE(VB_XMLTV, QString("compteMatching (%1)").arg(compteMatching));
+    if (compteMatching==0)
+        return 0.0;
+
+    t1Matche=TrouverMatches(t1,b1); 
+    t2Matche=TrouverMatches(t2,b2);
+
+    VERBOSE(VB_XMLTV, QString("t1Matche (%1)").arg(t1Matche));
+    VERBOSE(VB_XMLTV, QString("t2Matche (%1)").arg(t2Matche));
+
+    compteTransposition=0;
+    if (strcmp(t1Matche,t2Matche)!=0)
+    {
+        for (i=0;i<t1Matche.length();i++)
+            if (t1Matche[i]!=t2Matche[i])
+                compteTransposition++; 
+    }
+    else
+        compteTransposition=0;
+
+    
+    VERBOSE(VB_XMLTV, QString("compteTransposition (%1)").arg(compteTransposition));
+
+    distanceJaro=(((double)compteMatching/l1)+((double)compteMatching/l2)+((compteMatching-compteTransposition/2.0)/compteMatching))/3.0;
+
+    VERBOSE(VB_XMLTV, QString("dist JARO: (%1)").arg(distanceJaro));
+    longueurPrefix=0;
+    for (i=0;i<min(3,min(l1,l2))+1;i++) //longueur max : 4  
+    {
+        if (i<t1.length() && i<t2.length() && t1[i]==t2[i])
+            longueurPrefix++;
+        else
+            break;
+    }
+
+    VERBOSE(VB_XMLTV, QString("longueurPrefix (%1)").arg(longueurPrefix));
+
+    return distanceJaro+(longueurPrefix*0.1*(1-distanceJaro));
+} 
+
+
+ChanInfo* _findChanMatch(QList<ChanInfo> *oldchanlist, ChanInfo chan)
+{
+    QList<ChanInfo>::iterator i = oldchanlist->begin();
+    QString chanNameUp = chan.name.toUpper();
+    double maxJaro = 0.0;
+    ChanInfo *maxJaroChanInfo;
+
+    VERBOSE(VB_XMLTV,QString("============================================"));
+    VERBOSE(VB_XMLTV,QString("searching for (%1)").arg(chanNameUp));
+
+    for (; i != oldchanlist->end(); i++)
+    {
+        double d1 = 0;
+        double d2 = 0;
+        d1 = JaroWinkler(i->name, chanNameUp);
+        d2 = JaroWinkler(i->callsign, chanNameUp);
+
+        if (d1>0.9 && d2>0.9)
+        {
+            VERBOSE(VB_XMLTV,
+            QString("Looking for a match (%1), (%2), (%3) <==> (%4) = d1 (%5), d2 (%6)")
+            .arg((*i).name).arg((*i).callsign).arg((*i).chanstr).arg(chanNameUp).arg(d1).arg(d2));
+        }
+
+        // Look to the best match
+        if (d1>0.5 && d1>maxJaro)
+        {
+            maxJaro = d1;
+            maxJaroChanInfo = &(*i);
+        }
+
+        if (d2>0.5 && d2>maxJaro)
+        {
+            maxJaro = d2;
+            maxJaroChanInfo = &(*i);
+        }
+    }
+    
+    // only keep match with score > 0.93, under may be a false positive
+    if (maxJaro>0.93)
+    {
+        VERBOSE(VB_XMLTV,
+            QString("match is (%1), (%2), (%3) <==> (%4) = d1 (%5)")
+            .arg(maxJaroChanInfo->name).arg(maxJaroChanInfo->callsign).arg(maxJaroChanInfo->chanstr).arg(chanNameUp).arg(maxJaro));
+        return maxJaroChanInfo;
+    }
+    else
+    {
+        VERBOSE(VB_XMLTV,
+            QString("no match"));
+
+        return NULL;
+    }
+}
+
 void ChannelData::handleChannels(int id, QList<ChanInfo> *chanlist)
 {
     QString fileprefix = SetupIconCacheDirectory();
@@ -132 +293 @@
 
     fileprefix += "/";
 
+    QList<ChanInfo> oldchanlist;
+
+    MSqlQuery query(MSqlQuery::InitCon());
+    query.prepare(
+                "SELECT name, callsign, chanid "
+                "FROM channel "
+                "WHERE sourceid = :SOURCEID");
+    query.bindValue(":SOURCEID", id);
+
+    if (!query.exec())
+    {
+        MythDB::DBError("xmltvid conversion 1", query);
+    }
+    else 
+    {
+        while (query.next())
+        {
+            ChanInfo chanInfo;
+            chanInfo.name = query.value(0).toString().toUpper();
+            chanInfo.callsign = query.value(1).toString().toUpper();
+            chanInfo.chanstr = query.value(2).toString();
+            oldchanlist.push_back(chanInfo);
+        }
+    }
+
     QList<ChanInfo>::iterator i = chanlist->begin();
     for (; i != chanlist->end(); i++)
     {
@@ -151 +337 @@
             }
         }
 
-        MSqlQuery query(MSqlQuery::InitCon());
-
         if (!(*i).old_xmltvid.isEmpty())
         {
             query.prepare(
@@ -197 +381 @@
         {
             MythDB::DBError("handleChannels", query);
         }
-        else if (query.next())
+        
+        bool hasResult = query.next();
+        if (!hasResult)
+        {
+            ChanInfo *chanInfo = _findChanMatch(&oldchanlist, *i);
+
+            if (chanInfo!=NULL)
+            {
+                query.prepare(
+                        "SELECT chanid,   name, callsign, channum, "
+                        "       finetune, icon, freqid,   tvformat "
+                        "FROM channel "
+                        "WHERE chanid = :CHANID");
+                query.bindValue(":CHANID",  chanInfo->chanstr);
+            
+                if (!query.exec())
+                {
+                    MythDB::DBError("handleChannels", query);
+                }
+
+                hasResult = query.next();
+            }
+            else
+            {
+                hasResult = false;
+            }
+
+            
+        }
+        
+        if (hasResult)
         {
+            VERBOSE(VB_GENERAL,
+                QString("update chan."));
+
             QString chanid = query.value(0).toString();
             if (interactive)
             {
@@ -298 +515 @@
                 if (!non_us_updating && localfile != "")
                 {
                     MSqlQuery subquery(MSqlQuery::InitCon());
-                    subquery.prepare("UPDATE channel SET icon = :ICON WHERE "
+                    subquery.prepare("UPDATE channel SET icon = :ICON, "
+                                     "xmltvid = :XMLTVID, "
+                                     "callsign = :CALLSIGN WHERE "
                                      "chanid = :CHANID;");
                     subquery.bindValue(":ICON", localfile);
+                    subquery.bindValue(":XMLTVID", (*i).xmltvid);
+                    subquery.bindValue(":CALLSIGN", (*i).name);
                     subquery.bindValue(":CHANID", chanid);
 
                     if (!subquery.exec())
@@ -387 +608 @@
                 {
                     QStringList words = (*i).name.simplified().toUpper()
                         .split(" ");
+                    
                     QString callsign = "";
                     QString w1 = words.size() > 0 ? words[0] : QString();
                     QString w2 = words.size() > 1 ? words[1] : QString();