// Copyright 2013, University of Freiburg,
// Chair of Algorithms and Data Structures.
// Author: Hannah Bast <bast@cs.uni-freiburg.de>.

// Class that realizes error-tolerant prefix search.
class ErrorTolerantSearch {
 // PUBLIC MEMBERS.

 // Read input strings from file, where each line consists of one input string
 // followed by a TAB followed by a score. The input strings must contain
 // neither TABs nor newlines. The scores are used for ranking, see method
 // findMatches below.
 void readStringsFromFile(String fileName);

 // Build k-gram index from the input strings read with readStringsFromFile.
 // Remember to transfor each q-Gram to lowercase.
 void buildQgramIndex(int q);

 // For the given query prefix, find all input strings to which the prefix edit
 // distance is at most the given delta. Rank the matches by the scores read in
 // readStringsFromFile. Return the top-k matches in that order.
 // As explained in the lecture, proceed as follows. First use the q-gram index
 // to compute a set of candidate matches, using method computeUnion below. Then 
 // for each candidate match compute the exact PED to see whether it is really a
 // match.
 Array<String> findMatches(String prefix, int delta, int k);

 // PRIVATE MEMBERS.

 // Compute the prefix edit distance from the given query to the given string.
 // Check if the given string has a PED <= delta.
 // Note that the prefix edit distance is not symmetric.
 bool checkPrefixEditDistance(String prefix, String string, int delta);

 // Compute the union of the given inverted lists from the q-gram index. In the
 // result, along with each input string id, also store the number of q-grams
 // (from the input lists) containing that string.
 // IMPLEMENTATION NOTE 1: the union can be computed in much the same way as the
 // intersection. You can use a sequence of pairwise unions, as done for Exercise
 // Sheet 1 with intersections. Or you can compute the union of all lists at
 // once using a priority queue. The second one will be faster.
 // IMPLEMENTATION NOTE 2: in C++ one has to take care how to pass this
 // While a reference to something complex is good in general,
 // the one callign this method (findMatches) shouldn't copy index lists, either.
 // A proper parameter would be (const vector<const vector<int>*>& list).
 Array<Pair<int,int>> computeUnion(Array<Array<int>> lists);

 // The strings and their scores.
 Array<String> strings;
 Array<int> scores;
 
 // The q used for the index, stored because it is needed at query time.
 int q;

 // The inverted lists of the q-gram index. For each q-gram that occurs in one
 // of the input strings, contains the list of ids of all input strings
 // containing that q-gram.
 Map<String, Array<int>> qgramIndex;
}