vim-patch:9.1.0120: hard to get visual region using Vim script

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

closes: vim/vim#13998
closes: vim/vim#11579

3f905ab3c4

Cherry-pick changes from patch 9.1.0122, with :echom instead of :echow.

Co-authored-by: Shougo Matsushita <Shougo.Matsu@gmail.com>
Co-authored-by: Jakub Łuczyński <doubleloop@o2.pl>

src/nvim/eval.lua

@@ -4355,6 +4355,49 @@ M.funcs = {
     returns = 'table',
     signature = 'getreginfo([{regname}])',
   },
+  getregion = {
+    args = 3,
+    base = 1,
+    desc = [=[
+      Returns the list of strings from {pos1} to {pos2} as if it's
+      selected in visual mode of {type}.
+      For possible values of {pos1} and {pos2} see |line()|.
+      {type} is the selection type:
+      	"v" for |charwise| mode
+      	"V" for |linewise| mode
+      	"<CTRL-V>" for |blockwise-visual| mode
+      You can get the last selection type by |visualmode()|.
+      If Visual mode is active, use |mode()| to get the Visual mode
+      (e.g., in a |:vmap|).
+      This function uses the line and column number from the
+      specified position.
+      It is useful to get text starting and ending in different
+      columns, such as |charwise-visual| selection.
+
+      Note that:
+      - Order of {pos1} and {pos2} doesn't matter, it will always
+        return content from the upper left position to the lower
+        right position.
+      - If 'virtualedit' is enabled and selection is past the end of
+        line, resulting lines are filled with blanks.
+      - If the selection starts or ends in the middle of a multibyte
+        character, it is not included but its selected part is
+        substituted with spaces.
+      - If {pos1} or {pos2} equals "v" (see |line()|) and it is not in
+        |visual-mode|, an empty list is returned.
+      - If {pos1}, {pos2} or {type} is an invalid string, an empty
+        list is returned.
+
+      Examples: >
+      	:xnoremap <CR>
+      	\ <Cmd>echom getregion('v', '.', mode())<CR>
+      <
+    ]=],
+    name = 'getregion',
+    params = { { 'pos1', 'string' }, { 'pos2', 'string' }, { 'type', 'string' } },
+    returns = 'string[]',
+    signature = 'getregion({pos1}, {pos2}, {type})',
+  },
   getregtype = {
     args = { 0, 1 },
     base = 1,

src/nvim/eval/funcs.c

@@ -130,6 +130,7 @@
 #include "nvim/strings.h"
 #include "nvim/syntax.h"
 #include "nvim/tag.h"
+#include "nvim/types_defs.h"
 #include "nvim/ui.h"
 #include "nvim/version.h"
 #include "nvim/vim_defs.h"
@@ -2801,6 +2802,149 @@ static void f_getpos(typval_T *argvars, typval_T *rettv, EvalFuncData fptr)
   getpos_both(argvars, rettv, false, false);
 }
 
+/// Convert from block_def to string
+static char *block_def2str(struct block_def *bd)
+{
+  size_t size = (size_t)bd->startspaces + (size_t)bd->endspaces + (size_t)bd->textlen;
+  char *ret = xmalloc(size + 1);
+  char *p = ret;
+  memset(p, ' ', (size_t)bd->startspaces);
+  p += bd->startspaces;
+  memmove(p, bd->textstart, (size_t)bd->textlen);
+  p += bd->textlen;
+  memset(p, ' ', (size_t)bd->endspaces);
+  *(p + bd->endspaces) = NUL;
+  return ret;
+}
+
+/// "getregion()" function
+static void f_getregion(typval_T *argvars, typval_T *rettv, EvalFuncData fptr)
+{
+  tv_list_alloc_ret(rettv, kListLenMayKnow);
+
+  if (tv_check_for_string_arg(argvars, 0) == FAIL
+      || tv_check_for_string_arg(argvars, 1) == FAIL
+      || tv_check_for_string_arg(argvars, 2) == FAIL) {
+    return;
+  }
+
+  int fnum = -1;
+  // NOTE: var2fpos() returns static pointer.
+  pos_T *fp = var2fpos(&argvars[0], true, &fnum, false);
+  if (fp == NULL) {
+    return;
+  }
+  pos_T p1 = *fp;
+
+  fp = var2fpos(&argvars[1], true, &fnum, false);
+  if (fp == NULL) {
+    return;
+  }
+  pos_T p2 = *fp;
+
+  const char *pos1 = tv_get_string(&argvars[0]);
+  const char *pos2 = tv_get_string(&argvars[1]);
+  const char *type = tv_get_string(&argvars[2]);
+
+  const bool is_visual
+    = (pos1[0] == 'v' && pos1[1] == NUL) || (pos2[0] == 'v' && pos2[1] == NUL);
+
+  if (is_visual && !VIsual_active) {
+    return;
+  }
+
+  MotionType region_type = kMTUnknown;
+  if (type[0] == 'v' && type[1] == NUL) {
+    region_type = kMTCharWise;
+  } else if (type[0] == 'V' && type[1] == NUL) {
+    region_type = kMTLineWise;
+  } else if (type[0] == Ctrl_V && type[1] == NUL) {
+    region_type = kMTBlockWise;
+  } else {
+    return;
+  }
+
+  const TriState save_virtual = virtual_op;
+  virtual_op = virtual_active();
+
+  if (!lt(p1, p2)) {
+    // swap position
+    pos_T p = p1;
+    p1 = p2;
+    p2 = p;
+  }
+
+  oparg_T oap;
+  bool inclusive = true;
+
+  if (region_type == kMTCharWise) {
+    // handle 'selection' == "exclusive"
+    if (*p_sel == 'e' && !equalpos(p1, p2)) {
+      if (p2.coladd > 0) {
+        p2.coladd--;
+      } else if (p2.col > 0) {
+        p2.col--;
+        mark_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--;
+          mark_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 == kMTBlockWise) {
+    colnr_T sc1, ec1, sc2, ec2;
+    getvvcol(curwin, &p1, &sc1, NULL, &ec1);
+    getvvcol(curwin, &p2, &sc2, NULL, &ec2);
+    oap.motion_type = kMTBlockWise;
+    oap.inclusive = true;
+    oap.op_type = OP_NOP;
+    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.
+  int l = utfc_ptr2len(ml_get_pos(&p2));
+  if (l > 1) {
+    p2.col += l - 1;
+  }
+
+  for (linenr_T lnum = p1.lnum; lnum <= p2.lnum; lnum++) {
+    char *akt = NULL;
+
+    if (region_type == kMTLineWise) {
+      akt = xstrdup(ml_get(lnum));
+    } else if (region_type == kMTBlockWise) {
+      struct block_def bd;
+      block_prep(&oap, &bd, lnum, false);
+      akt = block_def2str(&bd);
+    } else if (p1.lnum < lnum && lnum < p2.lnum) {
+      akt = xstrdup(ml_get(lnum));
+    } else {
+      struct block_def bd;
+      charwise_block_prep(p1, p2, &bd, lnum, inclusive);
+      akt = block_def2str(&bd);
+    }
+
+    assert(akt != NULL);
+    tv_list_append_allocated_string(rettv->vval.v_list, akt);
+  }
+
+  virtual_op = save_virtual;
+}
+
 /// Common between getreg(), getreginfo() and getregtype(): get the register
 /// name from the first argument.
 /// Returns zero on error.

src/nvim/ops.c

@@ -84,25 +84,6 @@ static bool clipboard_delay_update = false;  // delay clipboard update
 static bool clipboard_needs_update = false;  // clipboard was updated
 static bool clipboard_didwarn = false;
 
-// structure used by block_prep, op_delete and op_yank for blockwise operators
-// also op_change, op_shift, op_insert, op_replace - AKelly
-struct block_def {
-  int startspaces;              // 'extra' cols before first char
-  int endspaces;                // 'extra' cols after last char
-  int textlen;                  // chars in block
-  char *textstart;              // pointer to 1st char (partially) in block
-  colnr_T textcol;              // index of chars (partially) in block
-  colnr_T start_vcol;           // start col of 1st char wholly inside block
-  colnr_T end_vcol;             // start col of 1st char wholly after block
-  int is_short;                 // true if line is too short to fit in block
-  int is_MAX;                   // true if curswant==MAXCOL when starting
-  int is_oneChar;               // true if block within one character
-  int pre_whitesp;              // screen cols of ws before block
-  int pre_whitesp_c;            // chars of ws before block
-  colnr_T end_char_vcols;       // number of vcols of post-block char
-  colnr_T start_char_vcols;     // number of vcols of pre-block char
-};
-
 #ifdef INCLUDE_GENERATED_DECLARATIONS
 # include "ops.c.generated.h"
 #endif
@@ -2655,66 +2636,11 @@ static void op_yank_reg(oparg_T *oap, bool message, yankreg_T *reg, bool append)
       reg->y_array[y_idx] = xstrdup(ml_get(lnum));
       break;
 
-    case kMTCharWise: {
-      colnr_T startcol = 0;
-      colnr_T endcol = MAXCOL;
-      bool is_oneChar = false;
-      colnr_T cs, ce;
-      char *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.
-                                   && utf_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;
+    case kMTCharWise:
+      charwise_block_prep(oap->start, oap->end, &bd, lnum, oap->inclusive);
       yank_copy_line(reg, &bd, y_idx, false);
       break;
-    }
+
     // NOTREACHED
     case kMTUnknown:
       abort();
@@ -4203,7 +4129,7 @@ static void restore_lbr(bool lbr_saved)
 /// - textlen includes the first/last char to be wholly yanked
 /// - start/endspaces is the number of columns of the first/last yanked char
 ///   that are to be yanked.
-static void block_prep(oparg_T *oap, struct block_def *bdp, linenr_T lnum, bool is_del)
+void block_prep(oparg_T *oap, struct block_def *bdp, linenr_T lnum, bool is_del)
 {
   int incr = 0;
   // Avoid a problem with unwanted linebreaks in block mode.
@@ -4326,6 +4252,65 @@ static void block_prep(oparg_T *oap, struct block_def *bdp, linenr_T lnum, bool
   restore_lbr(lbr_saved);
 }
 
+/// Get block text from "start" to "end"
+void charwise_block_prep(pos_T start, pos_T end, struct block_def *bdp, linenr_T lnum,
+                         bool inclusive)
+{
+  colnr_T startcol = 0;
+  colnr_T endcol = MAXCOL;
+  bool is_oneChar = false;
+  colnr_T cs, ce;
+  char *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.
+                               && utf_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.
 ///
 /// @param[in]  oap      Arguments of operator.

src/nvim/ops.h

@@ -15,6 +15,25 @@
 #include "nvim/pos_defs.h"
 #include "nvim/types_defs.h"
 
+/// structure used by block_prep, op_delete and op_yank for blockwise operators
+/// also op_change, op_shift, op_insert, op_replace - AKelly
+struct block_def {
+  int startspaces;           ///< 'extra' cols before first char
+  int endspaces;             ///< 'extra' cols after last char
+  int textlen;               ///< chars in block
+  char *textstart;           ///< pointer to 1st char (partially) in block
+  colnr_T textcol;           ///< index of chars (partially) in block
+  colnr_T start_vcol;        ///< start col of 1st char wholly inside block
+  colnr_T end_vcol;          ///< start col of 1st char wholly after block
+  int is_short;              ///< true if line is too short to fit in block
+  int is_MAX;                ///< true if curswant==MAXCOL when starting
+  int is_oneChar;            ///< true if block within one character
+  int pre_whitesp;           ///< screen cols of ws before block
+  int pre_whitesp_c;         ///< chars of ws before block
+  colnr_T end_char_vcols;    ///< number of vcols of post-block char
+  colnr_T start_char_vcols;  ///< number of vcols of pre-block char
+};
+
 typedef int (*Indenter)(void);
 
 /// flags for do_put()