patch 9.1.0120: hard to get visual region using Vim scriptv9.1.0120

Problem:  hard to get visual region using Vim script
Solution: Add getregion() Vim script function
          (Shougo Matsushita, Jakub Łuczyński)

closes: #13998
closes: #11579

Co-authored-by: =?UTF-8?q?Jakub=20=C5=81uczy=C5=84ski?= <doubleloop@o2.pl>
Co-authored-by: Shougo Matsushita <Shougo.Matsu@gmail.com>
Signed-off-by: Shougo Matsushita <Shougo.Matsu@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>

7 files changed, 431 insertions, 67 deletions

diff --git a/src/evalfunc.c b/src/evalfunc.c
index d13eb1c9f5..81f747f843 100644
--- a/src/evalfunc.c
+++ b/src/evalfunc.c
@@ -63,15 +63,16 @@ static void f_get(typval_T *argvars, typval_T *rettv);
 static void f_getchangelist(typval_T *argvars, typval_T *rettv);
 static void f_getcharpos(typval_T *argvars, typval_T *rettv);
 static void f_getcharsearch(typval_T *argvars, typval_T *rettv);
+static void f_getcurpos(typval_T *argvars, typval_T *rettv);
+static void f_getcursorcharpos(typval_T *argvars, typval_T *rettv);
 static void f_getenv(typval_T *argvars, typval_T *rettv);
 static void f_getfontname(typval_T *argvars, typval_T *rettv);
 static void f_getjumplist(typval_T *argvars, typval_T *rettv);
 static void f_getpid(typval_T *argvars, typval_T *rettv);
-static void f_getcurpos(typval_T *argvars, typval_T *rettv);
-static void f_getcursorcharpos(typval_T *argvars, typval_T *rettv);
 static void f_getpos(typval_T *argvars, typval_T *rettv);
 static void f_getreg(typval_T *argvars, typval_T *rettv);
 static void f_getreginfo(typval_T *argvars, typval_T *rettv);
+static void f_getregion(typval_T *argvars, typval_T *rettv);
 static void f_getregtype(typval_T *argvars, typval_T *rettv);
 static void f_gettagstack(typval_T *argvars, typval_T *rettv);
 static void f_gettext(typval_T *argvars, typval_T *rettv);
@@ -2131,6 +2132,8 @@ static funcentry_T global_functions[] =
 			ret_getreg,	    f_getreg},
     {"getreginfo",	0, 1, FEARG_1,	    arg1_string,
 			ret_dict_any,	    f_getreginfo},
+    {"getregion",	3, 3, FEARG_1,	    arg3_string,
+			ret_list_string,    f_getregion},
     {"getregtype",	0, 1, FEARG_1,	    arg1_string,
 			ret_string,	    f_getregtype},
     {"getscriptinfo",	0, 1, 0,	    arg1_dict_any,
@@ -5453,6 +5456,186 @@ f_getpos(typval_T *argvars, typval_T *rettv)
 }
 
 /*
+ * Convert from block_def to string
+ */
+   static char_u *
+block_def2str(struct block_def *bd)
+{
+    char_u *p, *ret;
+    size_t size = bd->startspaces + bd->endspaces + bd->textlen;
+
+    ret = alloc(size + 1);
+    if (ret != NULL)
+    {
+	p = ret;
+	vim_memset(p, ' ', bd->startspaces);
+	p += bd->startspaces;
+	mch_memmove(p, bd->textstart, bd->textlen);
+	p += bd->textlen;
+	vim_memset(p, ' ', bd->endspaces);
+	*(p + bd->endspaces) = NUL;
+    }
+    return ret;
+}
+
+/*
+ * "getregion()" function
+ */
+    static void
+f_getregion(typval_T *argvars, typval_T *rettv)
+{
+    linenr_T		lnum;
+    oparg_T		oap;
+    struct block_def	bd;
+    char_u		*akt = NULL;
+    int			inclusive = TRUE;
+    int			fnum = -1;
+    pos_T		p1, p2;
+    pos_T		*fp = NULL;
+    char_u		*pos1, *pos2, *type;
+    int			save_virtual = -1;
+    int			l;
+    int			region_type = -1;
+    int			is_visual;
+
+    if (rettv_list_alloc(rettv) == FAIL)
+	return;
+
+    if (check_for_string_arg(argvars, 0) == FAIL
+	    || check_for_string_arg(argvars, 1) == FAIL
+	    || check_for_string_arg(argvars, 2) == FAIL)
+	return;
+
+    // NOTE: var2fpos() returns static pointer.
+    fp = var2fpos(&argvars[0], TRUE, &fnum, FALSE);
+    if (fp == NULL)
+	return;
+    p1 = *fp;
+
+    fp = var2fpos(&argvars[1], TRUE, &fnum, FALSE);
+    if (fp == NULL)
+	return;
+    p2 = *fp;
+
+    pos1 = tv_get_string(&argvars[0]);
+    pos2 = tv_get_string(&argvars[1]);
+    type = tv_get_string(&argvars[2]);
+
+    is_visual = (pos1[0] == 'v' && pos1[1] == NUL)
+	|| (pos2[0] == 'v' && pos2[1] == NUL);
+
+    if (is_visual && !VIsual_active)
+	return;
+
+    if (type[0] == 'v' && type[1] == NUL)
+	region_type = MCHAR;
+    else if (type[0] == 'V' && type[1] == NUL)
+	region_type = MLINE;
+    else if (type[0] == Ctrl_V && type[1] == NUL)
+	region_type = MBLOCK;
+    else
+	return;
+
+    save_virtual = virtual_op;
+    virtual_op = virtual_active();
+
+    if (!LT_POS(p1, p2))
+    {
+	// swap position
+	pos_T p;
+
+	p = p1;
+	p1 = p2;
+	p2 = p;
+    }
+
+    if (region_type == MCHAR)
+    {
+	// handle 'selection' == "exclusive"
+	if (*p_sel == 'e' && !EQUAL_POS(p1, p2))
+	{
+	    if (p2.coladd > 0)
+		p2.coladd--;
+	    else if (p2.col > 0)
+	    {
+		p2.col--;
+
+		mb_adjustpos(curbuf, &p2);
+	    }
+	    else if (p2.lnum > 1)
+	    {
+		p2.lnum--;
+		p2.col = (colnr_T)STRLEN(ml_get(p2.lnum));
+		if (p2.col > 0)
+		{
+		    p2.col--;
+
+		    mb_adjustpos(curbuf, &p2);
+		}
+	    }
+	}
+	// if fp2 is on NUL (empty line) inclusive becomes false
+	if (*ml_get_pos(&p2) == NUL && !virtual_op)
+	    inclusive = FALSE;
+    }
+    else if (region_type == MBLOCK)
+    {
+	colnr_T sc1, ec1, sc2, ec2;
+
+	getvvcol(curwin, &p1, &sc1, NULL, &ec1);
+	getvvcol(curwin, &p2, &sc2, NULL, &ec2);
+	oap.motion_type = OP_NOP;
+	oap.inclusive = TRUE;
+	oap.start = p1;
+	oap.end = p2;
+	oap.start_vcol = MIN(sc1, sc2);
+	if (*p_sel == 'e' && ec1 < sc2 && 0 < sc2 && ec2 > ec1)
+	    oap.end_vcol = sc2 - 1;
+	else
+	    oap.end_vcol = MAX(ec1, ec2);
+    }
+
+    // Include the trailing byte of a multi-byte char.
+    l = utfc_ptr2len((char_u *)ml_get_pos(&p2));
+    if (l > 1)
+	p2.col += l - 1;
+
+    for (lnum = p1.lnum; lnum <= p2.lnum; lnum++)
+    {
+	int ret = 0;
+
+	if (region_type == MLINE)
+	    akt = vim_strsave(ml_get(lnum));
+	else if (region_type == MBLOCK)
+	{
+	    block_prep(&oap, &bd, lnum, FALSE);
+	    akt = block_def2str(&bd);
+	}
+	else if (p1.lnum < lnum && lnum < p2.lnum)
+	    akt = vim_strsave(ml_get(lnum));
+	else
+	{
+	    charwise_block_prep(p1, p2, &bd, lnum, inclusive);
+	    akt = block_def2str(&bd);
+	}
+
+	if (akt)
+	{
+	    ret = list_append_string(rettv->vval.v_list, akt, -1);
+	    vim_free(akt);
+	}
+
+	if (akt == NULL || ret == FAIL)
+	{
+	    list_free(rettv->vval.v_list);
+	    break;
+	}
+    }
+
+    virtual_op = save_virtual;
+}
+
+/*
  * Common between getreg(), getreginfo() and getregtype(): get the register
  * name from the first argument.
  * Returns zero on error.
diff --git a/src/ops.c b/src/ops.c
index 3fefbc8a40..e66b4b5ee7 100644
--- a/src/ops.c
+++ b/src/ops.c
@@ -2415,6 +2415,84 @@ block_prep(
 }
 
 /*
+ * Get block text from "start" to "end"
+ */
+    void
+charwise_block_prep(
+    pos_T		start,
+    pos_T		end,
+    struct block_def	*bdp,
+    linenr_T		lnum,
+    int			inclusive)
+{
+    colnr_T startcol = 0, endcol = MAXCOL;
+    int	    is_oneChar = FALSE;
+    colnr_T cs, ce;
+    char_u *p;
+
+    p = ml_get(lnum);
+    bdp->startspaces = 0;
+    bdp->endspaces = 0;
+
+    if (lnum == start.lnum)
+    {
+	startcol = start.col;
+	if (virtual_op)
+	{
+	    getvcol(curwin, &start, &cs, NULL, &ce);
+	    if (ce != cs && start.coladd > 0)
+	    {
+		// Part of a tab selected -- but don't
+		// double-count it.
+		bdp->startspaces = (ce - cs + 1)
+		    - start.coladd;
+		if (bdp->startspaces < 0)
+		    bdp->startspaces = 0;
+		startcol++;
+	    }
+	}
+    }
+
+    if (lnum == end.lnum)
+    {
+	endcol = end.col;
+	if (virtual_op)
+	{
+	    getvcol(curwin, &end, &cs, NULL, &ce);
+	    if (p[endcol] == NUL || (cs + end.coladd < ce
+			// Don't add space for double-wide
+			// char; endcol will be on last byte
+			// of multi-byte char.
+			&& (*mb_head_off)(p, p + endcol) == 0))
+	    {
+		if (start.lnum == end.lnum
+			&& start.col == end.col)
+		{
+		    // Special case: inside a single char
+		    is_oneChar = TRUE;
+		    bdp->startspaces = end.coladd
+			- start.coladd + inclusive;
+		    endcol = startcol;
+		}
+		else
+		{
+		    bdp->endspaces = end.coladd
+			+ inclusive;
+		    endcol -= inclusive;
+		}
+	    }
+	}
+    }
+    if (endcol == MAXCOL)
+	endcol = (colnr_T)STRLEN(p);
+    if (startcol > endcol || is_oneChar)
+	bdp->textlen = 0;
+    else
+	bdp->textlen = endcol - startcol + inclusive;
+    bdp->textstart = p + startcol;
+}
+
+/*
  * Handle the add/subtract operator.
  */
     void
diff --git a/src/proto/ops.pro b/src/proto/ops.pro
index 193e89430e..bebe63553b 100644
--- a/src/proto/ops.pro
+++ b/src/proto/ops.pro
@@ -14,6 +14,7 @@ void adjust_cursor_eol(void);
 char_u *skip_comment(char_u *line, int process, int include_space, int *is_comment);
 int do_join(long count, int insert_space, int save_undo, int use_formatoptions, int setmark);
 void block_prep(oparg_T *oap, struct block_def *bdp, linenr_T lnum, int is_del);
+void charwise_block_prep(pos_T start, pos_T end, struct block_def *bdp, linenr_T lnum, int inclusive);
 void op_addsub(oparg_T *oap, linenr_T Prenum1, int g_cmd);
 void clear_oparg(oparg_T *oap);
 void cursor_pos_info(dict_T *dict);
diff --git a/src/register.c b/src/register.c
index 3f1506a5d8..98015e05ba 100644
--- a/src/register.c
+++ b/src/register.c
@@ -1148,7 +1148,6 @@ op_yank(oparg_T *oap, int deleting, int mess)
     int			yanktype = oap->motion_type;
     long		yanklines = oap->line_count;
     linenr_T		yankendlnum = oap->end.lnum;
-    char_u		*p;
     char_u		*pnew;
     struct block_def	bd;
 #if defined(FEAT_CLIPBOARD) && defined(FEAT_X11)
@@ -1240,70 +1239,7 @@ op_yank(oparg_T *oap, int deleting, int mess)
 
 	    case MCHAR:
 		{
-		    colnr_T startcol = 0, endcol = MAXCOL;
-		    int	    is_oneChar = FALSE;
-		    colnr_T cs, ce;
-
-		    p = ml_get(lnum);
-		    bd.startspaces = 0;
-		    bd.endspaces = 0;
-
-		    if (lnum == oap->start.lnum)
-		    {
-			startcol = oap->start.col;
-			if (virtual_op)
-			{
-			    getvcol(curwin, &oap->start, &cs, NULL, &ce);
-			    if (ce != cs && oap->start.coladd > 0)
-			    {
-				// Part of a tab selected -- but don't
-				// double-count it.
-				bd.startspaces = (ce - cs + 1)
-							  - oap->start.coladd;
-				if (bd.startspaces < 0)
-				    bd.startspaces = 0;
-				startcol++;
-			    }
-			}
-		    }
-
-		    if (lnum == oap->end.lnum)
-		    {
-			endcol = oap->end.col;
-			if (virtual_op)
-			{
-			    getvcol(curwin, &oap->end, &cs, NULL, &ce);
-			    if (p[endcol] == NUL || (cs + oap->end.coladd < ce
-					// Don't add space for double-wide
-					// char; endcol will be on last byte
-					// of multi-byte char.
-					&& (*mb_head_off)(p, p + endcol) == 0))
-			    {
-				if (oap->start.lnum == oap->end.lnum
-					    && oap->start.col == oap->end.col)
-				{
-				    // Special case: inside a single char
-				    is_oneChar = TRUE;
-				    bd.startspaces = oap->end.coladd
-					 - oap->start.coladd + oap->inclusive;
-				    endcol = startcol;
-				}
-				else
-				{
-				    bd.endspaces = oap->end.coladd
-							     + oap->inclusive;
-				    endcol -= oap->inclusive;
-				}
-			    }
-			}
-		    }
-		    if (endcol == MAXCOL)
-			endcol = (colnr_T)STRLEN(p);
-		    if (startcol > endcol || is_oneChar)
-			bd.textlen = 0;
-		    else
-			bd.textlen = endcol - startcol + oap->inclusive;
-		    bd.textstart = p + startcol;
+		    charwise_block_prep(oap->start, oap->end, &bd, lnum, oap->inclusive);
 		    if (yank_copy_line(&bd, y_idx, FALSE) == FAIL)
 			goto fail;
 		    break;
diff --git a/src/testdir/test_vim9_builtin.vim b/src/testdir/test_vim9_builtin.vim
index aa819811cc..e2585d4f69 100644
--- a/src/testdir/test_vim9_builtin.vim
+++ b/src/testdir/test_vim9_builtin.vim
@@ -5197,4 +5197,13 @@ def Test_passing_type_to_builtin()
   v9.CheckScriptFailure(lines, 'E1405: Class "C" cannot be used as a value')
 enddef
 
+def Test_getregion()
+  assert_equal(['x'], getregion('.', '.', 'v')->map((_, _) => 'x'))
+
+  v9.CheckDefAndScriptFailure(['getregion(10, ".", "v")'], ['E1013: Argument 1: type mismatch, expected string but got number', 'E1174: String required for argument 1'])
+  assert_equal([''], getregion('.', '.', 'v'))
+  v9.CheckDefExecFailure(['getregion("a", ".", "v")'], 'E1209:')
+  v9.CheckDefExecAndScriptFailure(['getregion("", ".", "v")'], 'E1209: Invalid value for a line number')
+enddef
+
 " vim: ts=8 sw=2 sts=2 expandtab tw=80 fdm=marker
diff --git a/src/testdir/test_visual.vim b/src/testdir/test_visual.vim
index 17f0fd0687..34c572e617 100644
--- a/src/testdir/test_visual.vim
+++ b/src/testdir/test_visual.vim
@@ -1630,4 +1630,159 @@ func Test_visual_substitute_visual()
   bwipe!
 endfunc
 
+func Test_visual_getregion()
+  new
+
+  call setline(1, ['one', 'two', 'three'])
+
+  " Visual mode
+  call cursor(1, 1)
+  call feedkeys("\<ESC>vjl", 'tx')
+  call assert_equal(['one', 'tw'], 'v'->getregion('.', 'v'))
+  call assert_equal(['one', 'tw'], '.'->getregion('v', 'v'))
+  call assert_equal(['o'], 'v'->getregion('v', 'v'))
+  call assert_equal(['w'], '.'->getregion('.', 'v'))
+  call assert_equal(['one', 'two'], '.'->getregion('v', 'V'))
+  call assert_equal(['on', 'tw'], '.'->getregion('v', "\<C-v>"))
+
+  " Line visual mode
+  call cursor(1, 1)
+  call feedkeys("\<ESC>Vl", 'tx')
+  call assert_equal(['one'], getregion('v', '.', 'V'))
+  call assert_equal(['one'], getregion('.', 'v', 'V'))
+  call assert_equal(['one'], getregion('v', 'v', 'V'))
+  call assert_equal(['one'], getregion('.', '.', 'V'))
+  call assert_equal(['on'], '.'->getregion('v', 'v'))
+  call assert_equal(['on'], '.'->getregion('v', "\<C-v>"))
+
+  " Block visual mode
+  call cursor(1, 1)
+  call feedkeys("\<ESC>\<C-v>ll", 'tx')
+  call assert_equal(['one'], getregion('v', '.', "\<C-v>"))
+  call assert_equal(['one'], getregion('.', 'v', "\<C-v>"))
+  call assert_equal(['o'], getregion('v', 'v', "\<C-v>"))
+  call assert_equal(['e'], getregion('.', '.', "\<C-v>"))
+  call assert_equal(['one'], '.'->getregion('v', 'V'))
+  call assert_equal(['one'], '.'->getregion('v', 'v'))
+
+  " Using Marks
+  call setpos("'a", [0, 2, 3, 0])
+  call cursor(1, 1)
+  call assert_equal(['one', 'two'], "'a"->getregion('.', 'v'))
+  call assert_equal(['one', 'two'], "."->getregion("'a", 'v'))
+  call assert_equal(['one', 'two'], "."->getregion("'a", 'V'))
+  call assert_equal(['two'], "'a"->getregion("'a", 'V'))
+  call assert_equal(['one', 'two'], "."->getregion("'a", "\<c-v>"))
+
+  " Multiline with line visual mode
+  call cursor(1, 1)
+  call feedkeys("\<ESC>Vjj", 'tx')
+  call assert_equal(['one', 'two', 'three'], getregion('v', '.', 'V'))
+
+  " Multiline with block visual mode
+  call cursor(1, 1)
+  call feedkeys("\<ESC>\<C-v>jj", 'tx')
+  call assert_equal(['o', 't', 't'], getregion('v', '.', "\<C-v>"))
+
+  call cursor(1, 1)
+  call feedkeys("\<ESC>\<C-v>jj$", 'tx')
+  call assert_equal(['one', 'two', 'three'], getregion('v', '.', "\<C-v>"))
+
+  " 'virtualedit'
+  set virtualedit=all
+  call cursor(1, 1)
+  call feedkeys("\<ESC>\<C-v>10ljj$", 'tx')
+  call assert_equal(['one   ', 'two   ', 'three '],
+        \ getregion('v', '.', "\<C-v>"))
+  set virtualedit&
+
+  " Invalid position
+  call cursor(1, 1)
+  call feedkeys("\<ESC>vjj$", 'tx')
+  call assert_fails("call getregion(1, 2, 'v')", 'E1174:')
+  call assert_fails("call getregion('.', {}, 'v')", 'E1174:')
+  call assert_equal([], getregion('', '.', 'v'))
+  call assert_equal([], getregion('.', '.', ''))
+  call feedkeys("\<ESC>", 'tx')
+  call assert_equal([], getregion('v', '.', 'v'))
+
+  " using an unset mark
+  call assert_equal([], "'z"->getregion(".", 'V'))
+  " using the wrong type
+  call assert_fails(':echo "."->getregion([],"V")', 'E1174:')
+  call assert_fails(':echo "."->getregion("$", {})', 'E1174:')
+  call assert_fails(':echo [0, 1, 1, 0]->getregion("$", "v")', 'E1174:')
+
+
+  bwipe!
+  " Selection in starts or ends in the middle of a multibyte character
+  new
+  call setline(1, [
+        \   "abcdefghijk\u00ab",
+        \   "\U0001f1e6\u00ab\U0001f1e7\u00ab\U0001f1e8\u00ab\U0001f1e9",
+        \   "1234567890"
+        \ ])
+  call cursor(1, 3)
+  call feedkeys("\<Esc>\<C-v>ljj", 'xt')
+  call assert_equal(['cd', "\u00ab ", '34'],
+        \ getregion('v', '.', "\<C-v>"))
+  call cursor(1, 4)
+  call feedkeys("\<Esc>\<C-v>ljj", 'xt')
+  call assert_equal(['de', "\U0001f1e7", '45'],
+        \ getregion('v', '.', "\<C-v>"))
+  call cursor(1, 5)
+  call feedkeys("\<Esc>\<C-v>jj", 'xt')
+  call assert_equal(['e', ' ', '5'], getregion('v', '.', "\<C-v>"))
+  call cursor(1, 1)
+  call feedkeys("\<Esc>vj", 'xt')
+  call assert_equal(['abcdefghijk«', "\U0001f1e6"], getregion('v', '.', "v"))
+  " marks on multibyte chars
+  set selection=exclusive
+  call setpos("'a", [0, 1, 11, 0])
+  call setpos("'b", [0, 2, 16, 0])
+  call setpos("'c", [0, 2, 0, 0])
+  call cursor(1, 1)
+  call assert_equal(['ghijk', '🇨«🇩'], getregion("'a", "'b", "\<c-v>"))
+  call assert_equal(['k«', '🇦«🇧«🇨'], getregion("'a", "'b", "v"))
+  call assert_equal(['k«'], getregion("'a", "'c", "v"))
+
+  bwipe!
+
+  " Exclusive selection
+  new
+  set selection=exclusive
+  call setline(1, ["a\tc", "x\tz", '', ''])
+  call cursor(1, 1)
+  call feedkeys("\<Esc>v2l", 'xt')
+  call assert_equal(["a\t"], getregion('v', '.', 'v'))
+  call cursor(1, 1)
+  call feedkeys("\<Esc>v$G", 'xt')
+  call assert_equal(["a\tc", "x\tz", ''], getregion('v', '.', 'v'))
+  call cursor(1, 1)
+  call feedkeys("\<Esc>v$j", 'xt')
+  call assert_equal(["a\tc", "x\tz"], getregion('v', '.', 'v'))
+  call cursor(1, 1)
+  call feedkeys("\<Esc>\<C-v>$j", 'xt')
+  call assert_equal(["a\tc", "x\tz"], getregion('v', '.', "\<C-v>"))
+  call cursor(1, 1)
+  call feedkeys("\<Esc>\<C-v>$G", 'xt')
+  call assert_equal(["a", "x", '', ''], getregion('v', '.', "\<C-v>"))
+  call cursor(1, 1)
+  call feedkeys("\<Esc>wv2j", 'xt')
+  call assert_equal(["c", "x\tz"], getregion('v', '.', 'v'))
+
+  " virtualedit
+  set virtualedit=all
+  call cursor(1, 1)
+  call feedkeys("\<Esc>2lv2lj", 'xt')
+  call assert_equal(['      c', 'x   '], getregion('v', '.', 'v'))
+  call cursor(1, 1)
+  call feedkeys("\<Esc>2l\<C-v>2l2j", 'xt')
+  call assert_equal(['  ', '  ', '  '], getregion('v', '.', "\<C-v>"))
+  set virtualedit&
+  set selection&
+
+  bwipe!
+endfunc
+
 " vim: shiftwidth=2 sts=2 expandtab
diff --git a/src/version.c b/src/version.c
index f3d8fd8d1d..d495165516 100644
--- a/src/version.c
+++ b/src/version.c
@@ -705,6 +705,8 @@ static char *(features[]) =
 static int included_patches[] =
 {   /* Add new patch number below this line */
 /**/
+    120,
+/**/
     119,
 /**/
     118,