[Lldb-commits] [lldb] Colorize output when searching for symbols in lldb (PR #69422)

[via lldb-commits]

taalhaataahir0102 wrote:

Hi David!
I've updated the feedback. I got your point of somehow embedding the use-color and name in a single argument. Now we're doing the following:
In the very first function (`LookupSymbolInModule`) which calls other dump:

```
bool use_color = interpreter.GetDebugger().GetUseColor();
if (name && use_color){
   DumpAddress(
          interpreter.GetExecutionContext().GetBestExecutionContextScope(),
          symbol->GetAddressRef(), verbose, all_ranges, strm, name);
 }
 else{
      DumpAddress(
         interpreter.GetExecutionContext().GetBestExecutionContextScope(),
          symbol->GetAddressRef(), verbose, all_ranges, strm);
  }
```
As the program should only go inside the re-pattern (new name of printRed ;-) function if use-color is true and name is not null. Hence if this condition 
`if (name && use_color)` 
is not true, I'm using the Dump function which has `nullptr` as the default value of variable name.
Plus I've also updated the test cases and re-pattern function according to the feedback. Just 2 things are left:
1. Writing a test case for no symbol match.
> Also just for sanity checking, add one where you don't match anything at all. It shouldn't do any matching or even attempt to, so the test would just ensure that it doesn't try and end up crashing.
> 
> The classic "nothing in, nothing out" test case. So it will match 0 symbols but that's what we expect.

Was facing some issues while checking for empty string using:
```
# RUN: %lldb %t -b -o 'settings set use-color true' -o 'image lookup -r -s .*o132we$' | FileCheck %s --check-prefix CHECK5
# CHECK5-EMPTY:
```
But getting this error:

```
error: found non-empty check string for empty check with prefix 'CHECK5:'
# CHECK5-EMPTY:

```
2. Avoiding string copy in re-pattern function
> You can avoid some string copies by:
> 
> 1. Passing the StringRef to this function, instead of converting it to std::string and then getting a char* from that.
> 2. Then using StringRef's `.begin` and `.end` in the `next` iterator. If it doesn't like the type you can build the begin/end from `.bytes_begin` and `.bytes_end` instead. Those give you char* instead.
> 
> Basically StringRef is a "view" onto a string, so as long as we know where it ends we can iterate over that instead of a new copy of it.

Was having issues while using both `.begin()` and `.bytes_begin()`

But I'll look into both of these remaining things. For the test case maybe I will try to check the length of the output to be zero and for String reference I'll look more in to the `begin` and `bytes_begin` functions. Will update you shortly on these two.
Other than that can you please provide feedback on this commit and any other stuff which needs to be done :)

https://github.com/llvm/llvm-project/pull/69422