Bot or not

Fake news spreads mostly through bot accounts. A bot follows other accounts and reposts their stories so that it looks like a person. Bots and people repost in different patterns. Use that difference to decide which accounts are human.

There are two kinds of accounts: $p$ primary accounts that post stories, and $s$ secondary accounts that repost stories from primary accounts. For each secondary account you get the primary accounts it follows, whether it is known to be human, and the list of posts it reposted. Assume a secondary account sees every post made by every primary account it follows. Leaving a post it saw without a repost is therefore a deliberate choice.

The similarity score of two secondary accounts is set by three integers $w_0$, $w_1$, $w_2$ with $w_1 < 0 < w_0, w_2$. Only the posts of primary accounts that both secondary accounts follow enter the score. For each such post, if $j$ of the two accounts reposted it, add $w_j$. Both reposting or neither reposting gives a positive score, and a repost by exactly one of them gives the negative score $w_1$. If the two accounts follow no primary account in common, their score is 0.

Two accounts are similar if their total score is at least the threshold $t$. A secondary account is judged human if it is similar to an account known to be human. Being similar to an account that was judged human does not make an account human, so the judgement applies only one step.

For each data set, count the secondary accounts that are known human or judged human.

The first line contains the number of data sets $K$. The $K$ data sets follow, each in this format.

The first line of a data set contains six integers $p$, $s$, $w_0$, $w_1$, $w_2$, $t$. Here $1 \le p \le 100$ is the number of primary accounts, $1 \le s \le 100$ is the number of secondary accounts, $-100 \le w_1 < 0 < w_0, w_2 \le 100$ are the similarity scores, and $0 \le t < 100000$ is the similarity threshold.

The next line contains $p$ integers $m_i$ ($0 \le m_i \le 1000$), where $m_i$ is the number of posts made by primary account $i$. Posts are numbered from 1: the first $m_1$ posts belong to primary account 1, the next $m_2$ posts to primary account 2, and so on.

The next $s$ lines describe one secondary account $j$ each. A line starts with the integer $h_j$, which is 1 if the account is known to be human and 0 if it is not known whether it is human or a bot. Next comes the integer $f_j$ ($0 \le f_j \le p$), the number of primary accounts that $j$ follows, followed by $f_j$ distinct integers between $1$ and $p$, the indices of those primary accounts. Next comes the integer $r_j$ ($0 \le r_j \le \sum_i m_i$), the number of reposts made by $j$, followed by $r_j$ distinct integers between $1$ and $\sum_i m_i$, the posts that $j$ reposted. No input has an account reposting a post of a primary account it does not follow.

For each data set, print Data Set x: on a line of its own, where $x$ is the number of the data set counting from 1. On the next line print the number of secondary accounts that are known human or judged human. Print one blank line after each data set.