feat: implement forward search functionality

[rayworks]

• Implement search_fwd() function in ted_buffer.c

  • Search from current cursor position forward through buffer
  • Move cursor to first match found

• Add search command to ted_commands.c

  • Prompt user for search keyword
  • Integrate with search_fwd() function

• Add ctrl+r keybinding for search in ted_input.c

• Add command hints for search functionality

[rayworks]

Hi @arthurbacci, any feedback for this PR?

[arthurbacci]

Hi @arthurbacci, any feedback for this PR?

Hi. It seems like a good adition, thank you for contributing. But I believe something is wrong with the way x_width is used but I am not sure. Unfortunately I don't have much time this week so I will probably take some days to review it and merge.

[arthurbacci]

Non optimal behaviour with width vs bytes

[arthurbacci]

I just don't remember how to recalculate x from x_bytes, it may be a function from libgrapheme, idk. I will take a better look at this later.

[rayworks]

Thanks for reviewing it. I've updated the part to calculate the cursor position.

[arthurbacci]

Thanks for reviewing it. I've updated the part to calculate the cursor position.

Thank you. Now you just need to also calculate .x using wi_to_gi with .x_bytes

[rayworks]

Thanks for reviewing it. I've updated the part to calculate the cursor position.

Thank you. Now you just need to also calculate .x using wi_to_gi with .x_bytes

👍 Updated.

[arthurbacci]

Thanks for reviewing it. I've updated the part to calculate the cursor position.

Thank you. Now you just need to also calculate .x using wi_to_gi with .x_bytes

👍 Updated.

Actually, all that code you deleted in your last commit was probably right.

Basically:

• x is the index of the grapheme cluster the cursor is at
• x_bytes is the index of the byte the cursor is at
• x_width is the graphical horizontal position of the x cursor in the screen relative to the start of the line

In 90% of the cases they will all be the same.

I don't remember very well what recalc_cur does and what is does not, but I believe you need to set all those three. Since no functions in ted work with bytes, I don't remember how one would set both from x_bytes, but I think what you did with that while loop was the way. You should keep that, since it calculates x from x_bytes, and then you should only do:

buf->cursor.x_width = wi_to_gi(buf->cursor.x, curr_line);

This finishes setting both three.

[rayworks]

typedef struct {
    size_t lx_width;
    size_t x_width;
    size_t x_bytes;
    
    size_t y;
} Cursor;

void truncate_cur(Buffer *buf) {
    Line *ln = &buf->lines[buf->cursor.y];
    char *s = ln->data;
    
    buf->cursor.x_width -= index_by_width(buf->cursor.x_width, &s);
    buf->cursor.x_bytes = s - ln->data;
}

void recalc_cur(Buffer *buf) {
    truncate_cur(buf);
    buf->cursor.lx_width = buf->cursor.x_width;
}

It seems as long as x_width is in place, lx_width will be set as well after calling recalc_cur.

[arthurbacci]

What you are doing now seems right, except that you should use gi_to_wi (grapheme index to width index) instead of wi_to_gi

[rayworks]

What you are doing now seems right, except that you should use gi_to_wi (grapheme index to width index) instead of wi_to_gi

Nice catch.

[arthurbacci]

Seems to work well, tested with the unicode-test.txt file and it seems to correctly set the x position for complex grapheme sequences. Thank you a lot for your contribution.

I will just check for more details this weekend so that I can be sure it's doing everything right before merging into main, since I don't remember very well what's the correctly way of dealing with changing the x cursor. Maybe I should create a utility function to set it from width, grapheme index, or byte index so that this proccess isn't needed.