parsing interview question

[Lucas Smith]

Hi,
You have the String array:
String[] array = new String[]{....};
The size of the array is dynamic as well as the size of the strings in that array.
All of the strings are concatenated to create one long string.
array = null;
How would you create the array which structure will be the same?
You can not save the sizes of the strings. In that strings every sign can be used.

[Campbell Ritchie]

Not sure I understand the question. Tell us what you said, then maybe we can understand it.

[David Newton]

My understanding: Re-create the original array precisely given only the string.

[Campbell Ritchie]

Does that mean you use a particular sign or set of signs to catenate the Strings? If those signs are distinct from the remainder of the String, that actually sounds an easy exercise.

[D. Ogranos]

If "you can not save the lengths of the strings" only means that you must not save the lengths in external variables, you could also code each string length (in a certain, fixed length format ;) before each string. That way you wouldn't have to worry about separator chars or sequences.

[David Newton]

My take was that it wasn't delimited (which is why I didn't post an answer ;)

[Lucas Smith]

David Newton wrote:My understanding: Re-create the original array precisely given only the string.

Exactly.

Ogranos, what do you have on your mind? You can not save the length of particular strings.
Let's say that we have an array of Strings, then we create one long string, send it using RMI and the receiver has to create original array.

[Rob Spoor]

In C the '\0' character is used to indicate the end of each string. It is also used as the end of the array itself. For instance, string array {"abc", "def", "ghi"} would be represented as "abc\0def\0ghi\0\0".

In Java you won't need that final array terminator, but you will need something similar between each elements. You could use '\0', or perhaps '\036' or '\037' - the record and unit separator characters. In the end it doesn't matter, as long as you choose a character that is not part of the strings themselves.

Now, if you don't have control over how this string is built then you're screwed. From just "abcdefghi" you can never know what the original was. It could have been {"abc", "def", "ghi"}, or {"ab", "cd", "ef", "gh", "i"}, or any other combination.

For RMI you won't need this though; RMI can send arrays without any problem, provided that each element is serializable.

[D. Ogranos]

Lucas Smith wrote:Ogranos, what do you have on your mind? You can not save the length of particular strings.
Let's say that we have an array of Strings, then we create one long string, send it using RMI and the receiver has to create original array.

Well you didn't say what you had to use this for, so I assumed that it was a (homework) assignment where you were said "don't save the string lenghts in any extra variables". But you could still save the length of each string by coding it into the long string, for example

String[] ar = { "Test", "Huh" }

could become the String "0004Test0003Huh" if you code the length of each string as a 4 char hexadecimal number (it is also helpful to code the length of the array in the same way). Then the receiver just has to read out the lenghts and can determine how many chars of actual text follow.

But as Rob said, this probably won't be needed if you can send arrays over RMI.