Find and replace optimization for large files

[rebornix]

When users run find in a single file, we will run regex line by line originally. Now we have a text buffer which stores content in chunks instead of lines, running regex on chunks directly can avoid the overhead of getLineContent.

Things to note while doing this optimization

• When a line sits across Piece Tree nodes/buffers, we fall back to getLineContent then search.
• When users search by multiline regex, we need to get the whole content and then search, like before.
• When users search by single line regex, but the regex contains ^, $ or similar, executing this kind of regex on chunks will miss matches when there are line breaks.

[rebornix]

Benchmark on 1GB file

Line Array

Search is slow, but deltaDecorations is just operating on the interval tree

[AlCalzone]

This is probably related, but feel free to open another issue if its not:
When opening a large log file (30k lines in my case), I can reliably freeze the window just using search. Steps to reproduce:

• Open large file
• Enter the following in the search box with regex enabled: $\n

As soon as the n is entered, VSCode freezes and asks me a couple of seconds later if I want to restart the current window.

Version 1.21.1
Commit 79b44aa704ce542d8ca4a3cc44cfca566e7720f1
Datum 2018-03-14T14:47:13.351Z
Shell 1.7.9
Renderer 58.0.3029.110
Node 7.9.0
Architektur ia32

[rebornix]

@AlCalzone we didn't optimize regex searching yet but it's on my palette.