nvi: import version 2.2.1

CMakeLists.txt

@@ -37,6 +37,7 @@ if (NOT APPLE)
 endif()
 add_compile_options($<$<CONFIG:Release>:-Wuninitialized>)
 add_compile_options($<$<CONFIG:Release>:-Wno-dangling-else>)
+add_compile_options(-Wno-string-compare)
 add_compile_options(-Wstack-protector -fstack-protector)
 add_compile_options(-Wstrict-aliasing -fstrict-aliasing)
 
@@ -148,6 +149,9 @@ if(USE_WIDECHAR)
     target_sources(regex PRIVATE ${REGEX_SRCS})
     target_include_directories(regex PUBLIC regex)
     target_compile_definitions(regex PUBLIC __REGEX_PRIVATE)
+    # The macro _XOPEN_SOURCE_EXTENDED is needed to get the waddnwstr()
+    # definition on at least FreeBSD and recent macOS.
+    target_compile_definitions(nvi PRIVATE _XOPEN_SOURCE_EXTENDED)
     target_link_libraries(nvi PRIVATE regex)
 else()
     find_library(CURSES_LIBRARY NAMES ncurses curses HINTS /usr/lib)

INSTALL.md

@@ -0,0 +1,47 @@
+# Install from source
+
+For instructions to bring nvi2 as a part of your operating system's base system, see [Porting](https://github.com/lichray/nvi2/wiki/Porting) in the Wiki. This document is an overview of the build process that allows you to give nvi2 a try.
+
+## Prerequisites
+
+- CMake >= 3.17;
+- Ninja build system;
+- libiconv (for `USE_ICONV`);
+- libncursesw (for `USE_WIDECHAR`);
+
+Anything required by a minimal nvi, notably:
+
+- Berkeley DB1 in libc;
+- /var/tmp/vi.recover/ with mode 41777.
+
+## Building
+
+Nvi2 uses CMake build system generator. By specifying "Ninja Multi-Config" as the build system to generate, you can compile the project in both Debug and Release modes without re-running CMake. Under the project root directory, run
+
+```sh
+cmake -G "Ninja Multi-Config" -B build
+```
+
+Now `build` becomes your build directory to hold the artifacts. To build nvi2 in Debug mode, run
+
+```sh
+ninja -C build
+```
+
+Upon finishing, the nvi2 executable will be available as `build/Debug/nvi`. To launch it in `ex` mode, you can create a symlink
+
+```sh
+ln -s nvi build/Debug/ex
+```
+
+and run `./build/Debug/ex` rather than `./build/Debug/nvi`.
+
+To build nvi2 in Release mode, use the following command instead:
+
+```sh
+ninja -C build -f build-Release.ninja
+```
+
+Upon finishing, you will be able to edit files with `./build/Release/nvi`.
+
+To change configure-time options, such as disabling wide character support, use `ccmake build`.

README

@@ -1,4 +1,4 @@
-This is version 2.2.0 (2020-08-01) of nex/nvi, a reimplementation of the ex/vi
+This is version 2.2.1 (2023-09-25) of nex/nvi, a reimplementation of the ex/vi
 text editors originally distributed as part of the Fourth Berkeley
 Software Distribution (4BSD), by the University of California, Berkeley.
 
@@ -24,6 +24,30 @@ o Nvi was written by Keith Bostic, and the last version is 1.79.  After that,
 	Jun-ichiro itojun Hagino developed the file encoding detection
 	techniques in his nvi-m17n.
 
+o In nvi2, Zhihao Yuan incorporated the multibyte encoding support onto DB1.
+
+	It was not possible without great support from Alexander Leidinger,
+	Peter Wemm, and the FreeBSD community.
+
+	Last but not least, money from Google Summer of Code.
+
+o Since then,
+
+	Todd C. Miller and Craig Leres adopted and refined the NetBSD-style
+	expandtab option.
+
+	Yamamoto Takashi, Matija Skala, and Jessica Clarke ported the
+	software to macOS and Linux.
+
+	Anthony J. Bentley made heroic efforts to modernize the code base
+	and documentation, leveraging experience from OpenBSD to improve the
+	quality of the project.
+
+	...and many others, including Michael McConville, Marc Simpson,
+	Jeffrey H. Johnson, Bosco García, Anton Konyahin, Walter Alejandro
+	Iglesias, and those who tried hard to keep anonymous on GitHub :)
+	Their insights render the software usable, secure, and sustainable.
+
 The following acknowledgments were written by Keith Bostic:
 
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

common/key.c

@@ -31,6 +31,7 @@ static int	v_event_grow(SCR *, int);
 static int	v_key_cmp(const void *, const void *);
 static void	v_keyval(SCR *, int, scr_keyval_t);
 static void	v_sync(SCR *, int);
+static const char *alt_key_notation(int ch);
 
 /*
  * !!!
@@ -252,6 +253,10 @@ v_key_name(SCR *sp, ARG_CHAR_T ach)
 	 * followed by the character offset from the '@' character in the ASCII
 	 * character set.  Del (0x7f) is represented as '^' followed by '?'.
 	 *
+	 * If set O_ALTNOTATION, control characters less than 0x20 are
+	 * represented in <C-char> notations.  Carriage feed, escape, and
+	 * delete are marked as <Enter>, <Esc>, and <Del>, respectively.
+	 *
 	 * XXX
 	 * The following code depends on the current locale being identical to
 	 * the ASCII map from 0x40 to 0x5f (since 0x1f + 0x40 == 0x5f).  I'm
@@ -264,9 +269,14 @@ v_key_name(SCR *sp, ARG_CHAR_T ach)
 	if (CAN_PRINT(sp, ach))
 		goto done;
 nopr:	if (iscntrl(ch) && (ch < 0x20 || ch == 0x7f)) {
-		sp->cname[0] = '^';
-		sp->cname[1] = ch == 0x7f ? '?' : '@' + ch;
-		len = 2;
+		if (O_ISSET(sp, O_ALTNOTATION)) {
+			const char *notation = alt_key_notation(ch);
+			len = strlcpy(sp->cname, notation, sizeof(sp->cname));
+		} else {
+			sp->cname[0] = '^';
+			sp->cname[1] = ch == 0x7f ? '?' : '@' + ch;
+			len = 2;
+		}
 		goto done;
 	}
 #ifdef USE_WIDECHAR
@@ -745,6 +755,85 @@ v_sync(SCR *sp, int flags)
 		rcv_sync(sp, flags);
 }
 
+/*
+ * alt_key_notation --
+ *	Lookup for alternative notations of control characters.
+ */
+static const char*
+alt_key_notation(int ch)
+{
+	switch (ch) {
+	case 0x00:
+		return "<C-@>";
+	case 0x01:
+		return "<C-a>";
+	case 0x02:
+		return "<C-b>";
+	case 0x03:
+		return "<C-c>";
+	case 0x04:
+		return "<C-d>";
+	case 0x05:
+		return "<C-e>";
+	case 0x06:
+		return "<C-f>";
+	case 0x07:
+		return "<C-g>";
+	case 0x08:
+		return "<C-h>";
+	case 0x09:
+		return "<Tab>";
+	case 0x0A:
+		return "<NL>";
+	case 0x0B:
+		return "<C-k>";
+	case 0x0C:
+		return "<C-l>";
+	case 0x0D:
+		return "<Enter>";
+	case 0x0E:
+		return "<C-n>";
+	case 0x0F:
+		return "<C-o>";
+	case 0x10:
+		return "<C-p>";
+	case 0x11:
+		return "<C-q>";
+	case 0x12:
+		return "<C-r>";
+	case 0x13:
+		return "<C-s>";
+	case 0x14:
+		return "<C-t>";
+	case 0x15:
+		return "<C-u>";
+	case 0x16:
+		return "<C-v>";
+	case 0x17:
+		return "<C-w>";
+	case 0x18:
+		return "<C-x>";
+	case 0x19:
+		return "<C-y>";
+	case 0x1A:
+		return "<C-z>";
+	case 0x1B:
+		return "<Esc>";
+	case 0x1C:
+		return "<C-\\>";
+	case 0x1D:
+		return "<C-]>";
+	case 0x1E:
+		return "<C-^>";
+	case 0x1F:
+		return "<C-_>";
+	case 0x7f:
+		return "<Del>";
+	default:
+		__builtin_unreachable();
+	}
+}
+
 /*
  * v_event_err --
  *	Unexpected event.

common/main.c

@@ -419,8 +419,7 @@ err:		rval = 1;
  * PUBLIC: void v_end(GS *);
  */
 void
-v_end(gp)
-	GS *gp;
+v_end(GS *gp)
 {
 	MSGS *mp;
 	SCR *sp;

common/options.c

@@ -46,6 +46,8 @@ static int	 	 opts_print(SCR *, OPTLIST const *);
  * VI and EX Text Editors", 1990.
  */
 OPTLIST const optlist[] = {
+/* O_ALTNOTATION */
+	{L("altnotation"),	f_print,	OPT_0BOOL,	0},
 /* O_ALTWERASE	  4.4BSD */
 	{L("altwerase"),	f_altwerase,	OPT_0BOOL,	0},
 /* O_AUTOINDENT	    4BSD */

common/options_f.c

@@ -147,7 +147,7 @@ f_print(SCR *sp, OPTION *op, char *str, u_long *valp)
 	int offset = op - sp->opts;
 
 	/* Preset the value, needed for reinitialization of lookup table. */
-	if (offset == O_OCTAL) {
+	if (offset == O_OCTAL || offset == O_ALTNOTATION) {
 		if (*valp)
 			O_SET(sp, offset);
 		else

common/search.c

@@ -103,9 +103,14 @@ prev:			if (sp->re == NULL) {
 					++p;
 				break;
 			}
-			if (plen > 1 && p[0] == '\\' && p[1] == delim) {
-				++p;
-				--plen;
+			if (plen > 1 && p[0] == '\\') {
+				if (p[1] == delim) {
+					++p;
+					--plen;
+				} else if ( p[1] == '\\') {
+					*t++ = *p++;
+					--plen;
+				}
 			}
 		}
 		if (epp != NULL)

ex/ex.c

@@ -154,6 +154,10 @@ ex(SCR **spp)
 			if (file_end(sp, NULL, F_ISSET(sp, SC_EXIT_FORCE)))
 				return (1);
 			*spp = screen_next(sp);
+			if (*spp) {
+				F_CLR(*spp, SC_SCR_VI);
+				F_SET(*spp, SC_SCR_EX);
+			}
 			return (screen_end(sp));
 		}
 	}
@@ -1463,8 +1467,13 @@ usage:		msgq(sp, M_ERR, "086|Usage: %s", ecp->cmd->usage);
 		LF_INIT(FL_ISSET(ecp->iflags, E_C_HASH | E_C_LIST | E_C_PRINT));
 		if (!LF_ISSET(E_C_HASH | E_C_LIST | E_C_PRINT | E_NOAUTO) &&
 		    !F_ISSET(sp, SC_EX_GLOBAL) &&
-		    O_ISSET(sp, O_AUTOPRINT) && F_ISSET(ecp, E_AUTOPRINT))
-			LF_INIT(E_C_PRINT);
+		    O_ISSET(sp, O_AUTOPRINT) && F_ISSET(ecp, E_AUTOPRINT)) {
+			/* Honor the number option if autoprint is set. */
+			if (F_ISSET(ecp, E_OPTNUM))
+				LF_INIT(E_C_HASH);
+			else
+				LF_INIT(E_C_PRINT);
+		}
 
 		if (LF_ISSET(E_C_HASH | E_C_LIST | E_C_PRINT)) {
 			cur.lno = sp->lno;

ex/ex_argv.c

@@ -764,11 +764,11 @@ err:		if (ifp != NULL)
 	 * shell that does that is broken.
 	 */
 	for (p = bp, len = 0, ch = EOF;
-	    (ch = GETC(ifp)) != EOF; *p++ = ch, blen-=sizeof(CHAR_T), ++len)
+	    (ch = GETC(ifp)) != EOF; *p++ = ch, blen -= sizeof(CHAR_T), ++len)
 		if (blen < 5) {
-			ADD_SPACE_GOTOW(sp, bp, *blenp, *blenp * 2);
+			ADD_SPACE_GOTO(sp, CHAR_T, bp, *blenp, *blenp * 2);
 			p = bp + len;
-			blen = *blenp - len;
+			blen = *blenp - len * sizeof(CHAR_T);
 		}
 
 	/* Delete the final newline, nul terminate the string. */

ex/ex_subst.c

@@ -643,7 +643,9 @@ nextmatch:	match[0].rm_so = 0;
 					goto lquit;
 				}
 			} else {
-				if (ex_print(sp, cmdp, &from, &to, 0) ||
+				const int flags =
+				    O_ISSET(sp, O_NUMBER) ? E_C_HASH : 0;
+				if (ex_print(sp, cmdp, &from, &to, flags) ||
 				    ex_scprint(sp, &from, &to))
 					goto lquit;
 				if (ex_txt(sp, tiq, 0, TXT_CR))
@@ -1195,7 +1197,8 @@ re_tag_conv(SCR *sp, CHAR_T **ptrnp, size_t *plenp, int *replacedp)
 	for (; len > 0; --len) {
 		if (p[0] == '\\' && (p[1] == '/' || p[1] == '?')) {
 			++p;
-			--len;
+			if (len > 1)
+				--len;
 		} else if (STRCHR(L("^.[]$*"), p[0]))
 			*t++ = '\\';
 		*t++ = *p++;

man/vi.1

@@ -2282,6 +2282,10 @@ and
 .Nm vi
 modes, unless otherwise specified.
 .Bl -tag -width Ds
+.It Cm altnotation Bq off
+Display control characters less than 0x20 in <C-char> notations.
+Carriage feed, escape, and delete are marked as <Enter>, <Esc>, and <Del>,
+respectively.
 .It Cm altwerase Bq off
 .Nm vi
 only.

regex/regex.h

@@ -96,6 +96,16 @@ typedef struct {
 #define	REG_LARGE	01000	/* force large representation */
 #define	REG_BACKR	02000	/* force use of backref code */
 
+#ifdef USE_WIDECHAR
+/*
+ * Avoid function name conflicts with the system regex functions.
+ * This is needed e.g. to build with AddressSanitizer.
+ */
+#define regcomp nvi_regcomp
+#define regerror nvi_regerror
+#define regexec nvi_regexec
+#define regfree nvi_regfree
+#endif
 int	regcomp(regex_t *, const RCHAR_T *, int);
 size_t	regerror(int, const regex_t *, char *, size_t);
 int	regexec(const regex_t *,

vi/v_paragraph.c

@@ -39,15 +39,20 @@
 	if (p[0] == '\014') {						\
 		if (!--cnt)						\
 			goto found;					\
+		if (pstate == P_INTEXT && !--cnt)			\
+			goto found;					\
 		continue;						\
 	}								\
 	if (p[0] != '.' || len < 2)					\
 		continue;						\
 	for (lp = VIP(sp)->ps; *lp != '\0'; lp += 2)			\
 		if (lp[0] == p[1] &&					\
-		    (lp[1] == ' ' && len == 2 || lp[1] == p[2]) &&	\
-		    !--cnt)						\
-			goto found;					\
+		    (lp[1] == ' ' && len == 2 || lp[1] == p[2])) {	\
+			if (!--cnt)					\
+				goto found;				\
+			if (pstate == P_INTEXT && !--cnt)		\
+				goto found;				\
+		}							\
 } while (0)
 
 /*

vi/v_redraw.c

@@ -29,5 +29,6 @@
 int
 v_redraw(SCR *sp, VICMD *vp)
 {
+	F_SET(sp, SC_SCR_REFORMAT);
 	return (sp->gp->scr_refresh(sp, 1));
 }