How to avoid string copy when spliting the string?

Question

Hi all,
I have some old APIs (which I do not have control over) that expects the caller to pass the "content" of a text file as either String or String[] (each element is a line in the text file). Since the text file can be huge, I want to save memory and try to only hold 1 copy of the file content in memory. So, the question boils down to how can I make a String variable and a String[] variable share one single piece of memory.
After digging into the source code, I notice the String.substring(...) always makes allocate a new memory space for the returning string if the result substring is smaller than 1/4 of the actual String. The String.split() method internally also invokes the String.substring(...) method.
This creates a problem for me. If I first load the content of a file into a String "a". Then I call "String[] array = a.split("\n")", as the result, each element in the resulting array will be a copy of the substring of the original String "a" (since each line in my text file will be smaller than 1/4 of the entire file). StringTokenizer seems to have the same problem. Is there anything else I could do?
I really would like to extends the String class and override the substring method to make it always "share" the same memory content. But since String class is declared as final, I cannot do this. I think maybe the String.split() method implementation should make the return String[] from the split function share the same memory space as the original string. Any workaround anyone can suggest?
Thanks
-- Gary

Answer 1

While you could try to create some sort of custom class, if you want something that is already in java, you need to use one of the instantiations of CharSequence
http://java.sun.com/j2se/1.5.0/docs/api/java/lang/CharSequence.html
Personally, I would recommend CharBuffer. From the look of it, it seems most like what you want.
Hi all,
I have some old APIs (which I do not have control
rol over) that expects the caller to pass the
"content" of a text file as either String or String[]
(each element is a line in the text file). Since the
text file can be huge, I want to save memory and try
to only hold 1 copy of the file content in memory.
So, the question boils down to how can I make a
a String variable and a String[] variable share one
single piece of memory.
After digging into the source code, I notice the
the String.substring(...) always makes allocate a new
memory space for the returning string if the result
substring is smaller than 1/4 of the actual String.
The String.split() method internally also invokes
s the String.substring(...) method.
This creates a problem for me. If I first load
oad the content of a file into a String "a". Then I
call "String[] array = a.split("\n")", as the result,
each element in the resulting array will be a copy of
the substring of the original String "a" (since each
line in my text file will be smaller than 1/4 of the
entire file). StringTokenizer seems to have the same
problem. Is there anything else I could do?
I really would like to extends the String class and
nd override the substring method to make it always
"share" the same memory content. But since String
class is declared as final, I cannot do this. I
think maybe the String.split() method implementation
should make the return String[] from the split
function share the same memory space as the original
string. Any workaround anyone can suggest?
Thanks
-- Gary