Module: RuVim::TextMetricsExt

Defined in:

ext/ruvim/ruvim_ext.c

Class Method Summary

• .char_index_for_screen_col(*args) ⇒ Object

  char_index_for_screen_col(line, target_screen_col, tabstop: 2, align: :floor) Returns a character index whose screen column is <= target.

• .clip_cells_for_width(*args) ⇒ Object

  clip_cells_for_width(text, width, source_col_start: 0, tabstop: 2) Returns [cells_array, display_col].

Class Method Details

.char_index_for_screen_col(*args) ⇒ Object

char_index_for_screen_col(line, target_screen_col, tabstop: 2, align: :floor) Returns a character index whose screen column is <= target.

# File 'ext/ruvim/ruvim_ext.c', line 413

static VALUE
rb_char_index_for_screen_col(int argc, VALUE *argv, VALUE self)
{
    VALUE line, v_target, opts;
    rb_scan_args(argc, argv, "2:", &line, &v_target, &opts);

    int tabstop = 2;
    int align_ceil = 0;

    if (!NIL_P(opts)) {
        VALUE v;
        static ID id_tabstop, id_align, id_ceil;
        if (!id_tabstop) {
            id_tabstop = rb_intern("tabstop");
            id_align = rb_intern("align");
            id_ceil = rb_intern("ceil");
        }
        v = rb_hash_lookup2(opts, ID2SYM(id_tabstop), Qnil);
        if (!NIL_P(v)) tabstop = NUM2INT(v);
        v = rb_hash_lookup2(opts, ID2SYM(id_align), Qnil);
        if (!NIL_P(v) && SYM2ID(v) == id_ceil) align_ceil = 1;
    }

    if (NIL_P(line)) line = rb_str_new("", 0);
    if (TYPE(line) != T_STRING) line = rb_String(line);

    int target = NUM2INT(v_target);
    if (target < 0) target = 0;

    const char *ptr = RSTRING_PTR(line);
    const char *end = ptr + RSTRING_LEN(line);
    rb_encoding *enc = rb_utf8_encoding();
    int screen_col = 0;
    int char_index = 0;

    /* Walk grapheme clusters using oniguruma regex \X */
    /* Simplified: walk codepoints, treating combining marks as part of
       the previous character (width 0 doesn't advance screen_col). */
    while (ptr < end) {
        /* Measure one grapheme cluster: base char + combining marks */
        int cluster_width = 0;
        int cluster_chars = 0;
        int first = 1;

        while (ptr < end) {
            int clen = rb_enc_precise_mbclen(ptr, end, enc);
            if (!MBCLEN_CHARFOUND_P(clen)) {
                if (first) { ptr++; cluster_chars++; cluster_width = 1; }
                break;
            }
            clen = MBCLEN_CHARFOUND_LEN(clen);
            unsigned int code = rb_enc_codepoint(ptr, end, enc);

            if (!first) {
                /* Check if combining/zero-width 

.clip_cells_for_width(*args) ⇒ Object

clip_cells_for_width(text, width, source_col_start: 0, tabstop: 2) Returns [cells_array, display_col]

# File 'ext/ruvim/ruvim_ext.c', line 301

static VALUE
rb_clip_cells_for_width(int argc, VALUE *argv, VALUE self)
{
    VALUE text, v_width, opts;
    rb_scan_args(argc, argv, "2:", &text, &v_width, &opts);

    int max_width = NUM2INT(v_width);
    if (max_width < 0) max_width = 0;
    int source_col_start = 0, tabstop = 2;

    if (!NIL_P(opts)) {
        VALUE v;
        static ID id_source_col_start, id_tabstop;
        if (!id_source_col_start) {
            id_source_col_start = rb_intern("source_col_start");
            id_tabstop = rb_intern("tabstop");
        }
        v = rb_hash_lookup2(opts, ID2SYM(id_source_col_start), Qnil);
        if (!NIL_P(v)) source_col_start = NUM2INT(v);
        v = rb_hash_lookup2(opts, ID2SYM(id_tabstop), Qnil);
        if (!NIL_P(v)) tabstop = NUM2INT(v);
    }

    if (NIL_P(text)) text = rb_str_new("", 0);
    if (TYPE(text) != T_STRING) text = rb_String(text);

    const char *ptr = RSTRING_PTR(text);
    const char *end = ptr + RSTRING_LEN(text);
    rb_encoding *enc = rb_utf8_encoding();

    VALUE cell_class = get_cell_class();
    static ID id_new = 0;
    if (!id_new) id_new = rb_intern("new");

    VALUE cells = rb_ary_new();
    int display_col = 0;
    int source_col = source_col_start;
    VALUE space_str = rb_str_new(" ", 1);
    VALUE question_str = rb_str_new("?", 1);

    while (ptr < end) {
        int clen = rb_enc_precise_mbclen(ptr, end, enc);
        if (!MBCLEN_CHARFOUND_P(clen)) {
            /* invalid byte */
            if (display_col >= max_width) break;
            rb_ary_push(cells, rb_funcall(cell_class, id_new, 3,
                        question_str, INT2FIX(source_col), INT2FIX(1)));
            display_col++;
            source_col++;
            ptr++;
            continue;
        }
        clen = MBCLEN_CHARFOUND_LEN(clen);
        unsigned int code = rb_enc_codepoint(ptr, end, enc);

        /* Printable ASCII fast path */
        if (code >= 0x20 && code <= 0x7E) {
            if (display_col >= max_width) break;
            VALUE ch = rb_str_new(ptr, 1);
            rb_ary_push(cells, rb_funcall(cell_class, id_new, 3,
                        ch, INT2FIX(source_col), INT2FIX(1)));
            display_col++;
            source_col++;
            ptr += 1;
            continue;
        }

        /* Tab */
        if (code == '\t') {
            int w = tabstop - (display_col % tabstop);
            if (w == 0) w = tabstop;
            if (display_col + w > max_width) break;
            for (int i = 0; i < w; i++) {
                rb_ary_push(cells, rb_funcall(cell_class, id_new, 3,
                            space_str, INT2FIX(source_col), INT2FIX(1)));
            }
            display_col += w;
            source_col++;
            ptr += 1;
            continue;
        }

        /* Control chars */
        if (code < 0x20 || code == 0x7F || (code >= 0x80 && code <= 0x9F)) {
            if (display_col >= max_width) break;
            rb_ary_push(cells, rb_funcall(cell_class, id_new, 3,
                        question_str, INT2FIX(source_col), INT2FIX(1)));
            display_col++;
            source_col++;
            ptr += clen;
            continue;
        }

        /* Multi-byte character */
        int w = codepoint_width(code);
        if (display_col + w > max_width) break;
        VALUE ch = rb_enc_str_new(ptr, clen, enc);
        rb_ary_push(cells, rb_funcall(cell_class, id_new, 3,
                    ch, INT2FIX(source_col), INT2FIX(w)));
        display_col += w;
        source_col++;
        ptr += clen;
    }

    VALUE result = rb_ary_new_capa(2);
    rb_ary_push(result, cells);
    rb_ary_push(result, INT2FIX(display_col));
    return result;
}