tui: Fix TUI display issues in non-UTF-8 locales

[GabrielKimm]

This commit addresses the issue #1870 where the TUI displays incorrect characters in non-UTF-8 locales. The fix includes:

• Checking the current locale
• Using ASCII characters for graph display in non-UTF-8 locales
• Adjusting character widths accordingly

Fixes: #1870

Signed-off-by: Gabriel Kim [email protected]

[GabrielKimm]

I've addressed the issue #1870 regarding the TUI display in non-UTF-8 locales. Here's a summary of the changes and their effects:

Before the fix:

In non-UTF-8 locales, the TUI was displaying incorrect characters, making the graph structure unclear and difficult to read.

After the fix:

Now, the TUI correctly detects the locale setting and adjusts its display accordingly. In non-UTF-8 locales, it uses ASCII characters to represent the graph structure, ensuring clarity and readability across different environments.

Key changes:

1. Added locale detection to determine if UTF-8 is supported.
2. Implemented alternative ASCII characters for graph representation in non-UTF-8 locales.
3. Adjusted character width calculations to ensure proper alignment in both UTF-8 and non-UTF-8 environments.

These changes should resolve the display issues while maintaining the functionality and user experience of the TUI across various locale settings.

Please review and let me know if any further adjustments are needed.

[namhyung]

Thanks for the update, only some nitpicks below.

[honggyukim]

Hi @GabrielKimm, I see your name isn't correctly shown in the commit as follows.

commit 8340a73374fb17736ae81761d53acd6d351c5f5e (HEAD -> refs/heads/pull/1974)
Author:     0xGabriel <[email protected]>
AuthorDate: Mon Sep 30 01:51:49 2024 +0900

    Fix TUI display issues in non-UTF-8 locales
    
    This commit addresses the issue #1870 where the TUI displays
    incorrect characters in non-UTF-8 locales. The fix includes:
    
    - Checking the current locale
    - Using ASCII characters for graph display in non-UTF-8 locales
    - Adjusting character widths accordingly
    
    Fixes: #1870
    
    Signed-off-by: 0xGabriel <[email protected]>

Please put your real name rather than 0xGabriel.

You can check it at https://github.com/namhyung/uftrace/pull/1974.patch.

[honggyukim]

Your author name isn't changed yet. Please update it together.

From 39030abbd0de2289fa0600f0395b006176ff4c52 Mon Sep 17 00:00:00 2001
From: 0xGabriel <[email protected]>

As far as I remember, your name is "Gabriel Kim" so please use your full name.

You can update it as follows.

$ git commit --amend --author "Gabriel Kim <[email protected]>" --no-edit -s

[honggyukim]

In addition, you need to put category prefix for the commit title as follows.

tui: Fix TUI display issues in non-UTF-8 locales

[honggyukim]

Please use your name "Gabriel Kim" rather than "GabrielKim".

[honggyukim]

Thanks for the update. I will wait for @namhyung's review.

[GabrielKimm]

Thanks for the update. I will wait for @namhyung's review.

Thank you so much for your invaluable feedback and guidance on this PR.
Your detailed comments and suggestions have been incredibly helpful in improving the quality of my contribution.

I truly appreciate the time and effort you've put into reviewing my work and providing such constructive advice.

[honggyukim]

Thanks. I've just tested this and it looks fine but I have some concerns by checking the value of LANG.

The following can be another way to check if we can print graphical characters, but I'm not sure if it doesn't break other environments such as Android build, musl build, alpine Linux, etc.

#include <locale.h>
#include <wchar.h>

int can_print_utf8_graph_chars() {
    const char *graph_char = "★";
    wchar_t wc;

    setlocale(LC_CTYPE, "");

    int result = mbtowc(&wc, graph_char, MB_CUR_MAX);

    if (result > 0 && wc != L'\0')
        return true;

    return false;
}

[namhyung]

It seems there are display issues still.

1. The cursor line overflows to the next line
2. alignment is broken with child functions - see set_program_name and strrchr.

[GabrielKimm]

It seems there are display issues still.

1. The cursor line overflows to the next line
2. alignment is broken with child functions - see set_program_name and strrchr.

Thank you for your thorough review and for pointing out these remaining display issues. I appreciate your attention to detail.

1. Cursor line overflow:
   I understand the problem with the cursor line overflowing to the next line. This is likely due to an miscalculation in handling the terminal width or character width, especially in non-UTF-8 locales. I will review the code for string length calculations and line wrapping logic to address this issue.

2. Alignment of child functions:
   Thank you for bringing this to my attention. The misalignment of child functions, particularly for set_program_name and strrchr, is indeed a critical issue for readability. I will carefully examine the indentation logic and how it handles various function names, especially longer ones. I'll also double-check the character width calculations for different locales to ensure consistent alignment.

I plan to make the following improvements:

• Implement more robust terminal width detection and output adjustment.
• Refine the indentation and alignment logic, particularly for child function display.
• Conduct thorough testing across various locale settings to ensure consistency.
• Review and optimize the string length and display width calculation logic.

I'll update the PR with these changes soon 😊🙏🏻🙏🏻