Precision of a Spam Classifier

The support inbox is why Larkspur Mail cares about this one. Every time the filter quarantines a real email, somebody eventually notices and writes in, so the team needs to know how much the junk folder can be trusted as a verdict. The bake-off runs a fresh batch of hand-labelled test emails against every candidate build, so the figure is recomputed from that build's own counts.

Throughout, spam is the positive class - a "positive" is the filter raising the alarm, nothing more.

Task: Print the filter's precision for the spam class, as a single number rounded to 4 decimals. Round once, at the end.

Input

Four lines, each holding one whole number from a single evaluation run, in this order:

1. spam, and the filter flagged it.
2. real mail, and the filter flagged it anyway.
3. real mail, and the filter let it through.
4. spam, and the filter let it through.

Every run flags at least one email.

Output

One line holding the precision rounded to 4 decimals, printed the way Python prints a float - 0.8, never 0.8000.

Example:

Input:
40
10
45
5

Output:
0.8