CommandScreen: wrap at wide character boundaries

* create separate functions for UTF-8/ASCII scanning
* expand line buffer to handle UTF-8 combining marks
* skip UTF-8 codepoint decoding for ASCII characters
* add missing include and ifdef guard for CRT_utf8
* check mbrtowc before char width calculation, thanks to @Explorer09
* ensure sufficient room to handle UTF-8 code sequences, especially
  unicode characters with arbitrary combining marks or diacritics.

Author: ryenus

CommandScreen.c

@@ -12,45 +12,124 @@ in the source distribution for its full text.
 
 #include <assert.h>
 #include <stdlib.h>
+#include <stdio.h>
 #include <string.h>
+#include <wchar.h>
+#include <wctype.h>
+#include <locale.h>
 
+#include "CRT.h"
 #include "Macros.h"
 #include "Panel.h"
 #include "ProvideCurses.h"
 
 
-static void CommandScreen_scan(InfoScreen* this) {
-   Panel* panel = this->display;
-   int idx = MAXIMUM(Panel_getSelectedIndex(panel), 0);
-   Panel_prune(panel);
-
-   const char* p = Process_getCommand(this->process);
-   char line[COLS + 1];
-   int line_offset = 0, last_spc = -1, len;
-   for (; *p != '\0'; p++, line_offset++) {
-      assert(line_offset >= 0 && (size_t)line_offset < sizeof(line));
-      line[line_offset] = *p;
-      if (*p == ' ') {
-         last_spc = line_offset;
+static int CommandScreen_scanAscii(InfoScreen* this, const char* p, size_t total, char* line) {
+   int line_offset = 0, line_size = 0, last_spc_offset = -1;
+   for (size_t i = 0; i < total; i++, line_offset++) {
+      assert(line_offset >= 0 && (size_t)line_offset <= total);
+      char c = line[line_offset] = p[i];
+      if (c == ' ') {
+         last_spc_offset = line_offset;
       }
 
       if (line_offset == COLS) {
-         len = (last_spc == -1) ? line_offset : last_spc;
-         line[len] = '\0';
+         line_size = (last_spc_offset == -1) ? line_offset : last_spc_offset;
+         line[line_size] = '\0';
          InfoScreen_addLine(this, line);
 
-         line_offset -= len;
-         last_spc = -1;
-         memcpy(line, p - line_offset, line_offset + 1);
+         line_offset -= line_size;
+         last_spc_offset = -1;
+         memcpy(line, p + i - line_offset, line_offset + 1);
+      }
+   }
+
+   return line_offset;
+}
+
+#ifdef HAVE_LIBNCURSESW
+
+static int CommandScreen_scanWide(InfoScreen* this, const char* p, size_t total, char* line) {
+   mbstate_t state;
+   memset(&state, 0, sizeof(state));
+   int line_cols = 0, line_offset = 0, line_size = 0, width = 1;
+   int last_spc_cols = -1, last_spc_offset = -1;
+   for (size_t i = 0, bytes = 1; i < total; bytes = 1, width = 1) {
+      assert(line_offset >= 0 && (size_t)line_offset <= total);
+      unsigned char c = (unsigned char)p[i];
+      if (c < 0x80) { // skip mbrtowc for ASCII characters
+         line[line_offset] = c;
+         if (c == ' ') {
+            last_spc_offset = line_offset;
+            last_spc_cols = line_cols;
+         }
+      } else {
+         wchar_t wc;
+         bytes = mbrtowc(&wc, p + i, total - i, &state);
+         if (bytes != (size_t)-1 && bytes != (size_t)-2) {
+            width = wcwidth(wc);
+            width = MAXIMUM(width, 1);
+         } else {
+            bytes = 1;
+         }
+         memcpy(line + line_offset, p + i, bytes);
+      }
+
+      i += bytes;
+      line_offset += bytes;
+      line_cols += width;
+      if (line_cols < COLS) {
+         continue;
+      }
+
+      if (last_spc_offset >= 0) {  // wrap by last space
+         line_size = last_spc_offset;
+         line_cols -= last_spc_cols;
+         last_spc_offset = last_spc_cols = -1;
+      } else if (line_cols > COLS) {  // wrap current wide char to next line
+         line_size = line_offset - (int) bytes;
+         line_cols = width;
+      } else { // wrap by current character
+         line_size = line_offset;
+         line_cols = 0;
+      }
+
+      line[line_size] = '\0';
+      InfoScreen_addLine(this, line);
+      line_offset -= line_size;
+      if (line_offset > 0) {
+         memcpy(line, p + i - line_offset, line_offset + 1);
       }
    }
 
+   return line_offset;
+}
+
+#endif // HAVE_LIBNCURSESW
+
+static void CommandScreen_scan(InfoScreen* this) {
+   Panel* panel = this->display;
+   int idx = Panel_getSelectedIndex(panel);
+   Panel_prune(panel);
+
+   const char* p = Process_getCommand(this->process);
+   assert(p != NULL);
+   size_t total = strlen(p);
+   char line[total + 1];
+
+   int line_offset =
+#ifdef HAVE_LIBNCURSESW
+      CRT_utf8 ? CommandScreen_scanWide(this, p, total, line) :
+#endif
+      CommandScreen_scanAscii(this, p, total, line);
+
+   assert(line_offset >= 0 && (size_t)line_offset <= total);
    if (line_offset > 0) {
       line[line_offset] = '\0';
       InfoScreen_addLine(this, line);
    }
 
-   Panel_setSelected(panel, idx);
+   Panel_setSelected(panel, MAXIMUM(idx, 0));
 }
 
 static void CommandScreen_draw(InfoScreen* this) {