c# - Faster alternative than loop through list to specify non unique values and their position -

i have data set need isolate , feedback non-unique values across multiple columns. (think multi column primary key violation in database table). doing concatenating columns on each row list<string>. doing count of item in list , if greater 1 adding error message another column on same row - bit's important, need able provide feedback on position of duplicate/s, not fact there duplicate.

the problem speed, although technically works, doesn't work practical solution because potentially working data sets of several hundred thousand rows

code:

list<string> mylist = new list<string>(); string thisconcat = ""; (int = 0; < dtlogdata.rows.count-1; i++)     {        foreach (int colnum in colnumlist)        {            thisconcat += dtlogdata.rows[i].field<string>(colnum-1);        }        mylist.add(thisconcat);        thisconcat = "";     } 

then:

for (int = 0; < dtlogdata.rows.count-1; i++) {     int count = mylist.count(j => j == mylist[i]);     if (count > 1)     {       dtlogdata.rows[i][colcnt] = myerrorstring;     }  } 

i'll give different solution. assumes willing add column datatable:

dtlogdata.columns.add("hash"); 

then cast table:

var t = dtlogdata.asenumerable(); 

first compute concatenated string , hold of rows. can in 2 ways. if want limit columns indexes (like original code):

var rows = t.select(row =>     {         stringbuilder builder = new stringbuilder();         colnumlist.foreach(i => builder.append(row[i]));         row["hash"] = builder.tostring();         return row;     } ); 

or if want use columns:

var rows = t.select(row =>     {         row["hash"] = string.join("", row.itemarray.select(i => i.tostring()));         return row;     } ); 

you grab rows duplicates , mark them accordingly:

foreach (var datarow in rows.groupby(r => r["hash"]).where(g => g.count() > 1).selectmany(g => g)) {     datarow[colcnt] = myerrorstring;                 }