A municipal utility ingests hourly meter readings into a raw staging table where every value is written out as a 64-bit floating-point decimal string, even though every reading in the table is guaranteed to represent an exact whole number (its fractional part is always written as a single trailing 0 after the decimal point). Before the readings can move into the compact archival table, each one must have its data type changed from a floating-point representation to a native 32-bit signed integer.

Given n such readings, output each one recast to its plain integer value, in the same order they were given.

Input format

The first line contains a single integer n. Each of the next n lines contains one reading, written as a decimal string of the form <optional -><digits>.0 (for example 182056.0 or -91.0), with no leading zeros in the integer part other than a single 0 for the value zero.

Output format

Print n lines. The i-th line must contain the i-th reading recast as a plain integer, with no decimal point and no leading +. If the recast value equals zero, print 0 with no sign, even if the input reading carried a leading - (e.g. -0.0 must print as 0).

Constraints

• 1 <= n <= 1000
• Once its trailing .0 is dropped, every reading's integer value lies in the inclusive range [-2^31, 2^31 - 1].

2
182056.0
-91.0

182056
-91

1
-0.0

0