Guess the Algorithms That Will Appear at KCPC

Time limit1sMemory limit1024 MB

Summary

Given N problems, each with a list of tags, print the tag that appears on the most problems, or -1 if several tags tie for the top count.

Level: Easy3 of 10
Topics: Hash map, String
Solved: No attempts yet

Problem

You are taking part in KCPC and want to use big data to predict which algorithms will appear in this contest.

Given each problem number, the number of tags attached to it, and the list of tags, print the tag that appears most often.

Input

The first line contains the number of problems $N$ ( $1 \leq N \leq 1000$ ).

From the second line onward, $N$ lines give the information for each problem in the following format.

$s$ $t$ $a_1$ $a_2$ $\cdots$ $a_t$
- $s$ is an integer for the problem number ( $1000 \leq s \leq 32196$ ).
- $t$ is an integer for the number of tags on the problem ( $1 \leq t \leq 100$ ).
- $a_i$ is a tag of problem $s$ , a string of lowercase English letters and _ ( $1 \leq |a_i| \leq 31$ ).
- The $t$ tags on each problem are all distinct.

The problems are given in increasing order of $s$ , and all $N$ problem numbers are different.

Output

Print the most frequently appearing tag on the first line. If several tags tie for the most appearances, print -1.

Hint

The test data may not match real online judge problems or solved.ac tags.

Examples2

Example 1

Input

6
30701 2 greedy sorting
30702 4 bfs dfs graphs graph_traversal
30705 2 data_structures disjoint_set
30706 2 geometry math
30707 4 constructive graphs greedy mst
30708 3 ad_hoc constructive greedy

Expected output

greedy

Example 2

Input

6
30701 2 greedy sorting
30702 4 bfs dfs graphs graph_traversal
30705 2 data_structures disjoint_set
30706 2 geometry math
30707 4 constructive graphs greedy mst
30708 2 ad_hoc constructive

Expected output

-1