|Just another Perl shrine|
Displaying/buffering huge text filesby spurperl (Priest)
|on Feb 23, 2005 at 06:19 UTC||Need Help??|
spurperl has asked for the wisdom of the Perl Monks concerning the following question:
My GUI application contains multiple windows. One of them displays a text file - read only. The user can scroll the file, selects portions of it, run "Find" on it, simple stuff.
The problems begin when this file is very big - 100s of MBs. The application natually tries to load it wholly into memory, and BOOM.
Web searches brought up surprisingly little results. Even most of the popular text editors ignore this problem and just collapse on files too big.
But I know it's possible, because some editors do it, and it sounds possible in theory.
The problem with text files is that we can't seek in them freely. In binary files it's possible, in text files not.
I can display a single page to the user (perhaps padded by a buffer from above and below) - as far as he is concerned all the rest is virtual. But there's a problem - say a user drags a scroll bar to some far away location in the file - line 999999, for example. How do I get there quickly ?
One solution is just read the file line by line until 999999. This is slow.
Another solution: when the file is initially opened, I read it and create an index table: line -> byte. Say, line 225 starts at byte 1069 in the file. Then I can immediately go to the desired line by seeking in the file.
There's a problem: 1 million lines => about 8 million bytes to store the index. Still, quite a lot of memory. (There can also be 10 mln lines, as far as I'm concerned).
So, I can keep this index in a separate, binary file. When the user asks line 999999, I go to my index binary file, quickly seek to record 999999, read the byte start in the text file and jump there.
Does this sound logical ? Can you think of simpler solutions ?
Thanks in advance