Ticket #3259: mc-3259-hex-update-v2.patch

src/viewer/hex.c

@@ -73 +73 @@
 /*** file scope functions ************************************************************************/
 /* --------------------------------------------------------------------------------------------- */
 
-#ifdef HAVE_CHARSET
-static int
-utf8_to_int (char *str, int *char_width, gboolean * result)
-{
-    int res = -1;
-    gunichar ch;
-    gchar *next_ch = NULL;
-    int width = 0;
-
-    *result = TRUE;
-
-    if (str == NULL)
-    {
-        *result = FALSE;
-        return 0;
-    }
-
-    res = g_utf8_get_char_validated (str, -1);
-
-    if (res < 0)
-        ch = *str;
-    else
-    {
-        ch = res;
-        /* Calculate UTF-8 char width */
-        next_ch = g_utf8_next_char (str);
-        if (next_ch)
-            width = next_ch - str;
-        else
-            ch = 0;
-    }
-    *char_width = width;
-    return ch;
-}
-#endif /* HAVE_CHARSET */
-
 /* --------------------------------------------------------------------------------------------- */
 /** Determine the state of the current byte.
  *
@@ -146 +110 @@
     const screen_dimen text_start = 8 + 13 * ngroups +
         ((width < 80) ? 0 : (width == 80) ? (ngroups - 1) : (ngroups - 1 + 1));
 
-    screen_dimen row;
+    int row;
     off_t from;
     int c;
     mark_t boldflag = MARK_NORMAL;
     struct hexedit_change_node *curr = view->change_list;
 #ifdef HAVE_CHARSET
     int ch = 0;
+    int cont_bytes = 0;  /* number of continuation bytes remanining from current UTF-8 */
+    gboolean cjk_right = FALSE;  /* whether the second byte of a CJK is to be processed */
 #endif /* HAVE_CHARSET */
 
     char hex_buff[10];          /* A temporary buffer for sprintf and mvwaddstr */
@@ -161 +127 @@
     mcview_display_clean (view);
 
     /* Find the first displayable changed byte */
+    /* In UTF-8 mode, go back by 1 or maybe 2 lines to handle continuation bytes properly. */
     from = view->dpy_start;
+    row = 0;
+#ifdef HAVE_CHARSET
+    if (view->utf8)
+    {
+        if (from >= view->bytes_per_line) {
+            row--;
+            from -= view->bytes_per_line;
+        }
+        if (view->bytes_per_line == 4 && from >= view->bytes_per_line) {
+            row--;
+            from -= view->bytes_per_line;
+        }
+    }
+#endif /* HAVE_CHARSET */
     while (curr && (curr->offset < from))
     {
         curr = curr->next;
     }
 
-    for (row = 0; mcview_get_byte (view, from, NULL) == TRUE && row < height; row++)
+    for (; mcview_get_byte (view, from, NULL) == TRUE && row < (int) height; row++)
     {
         screen_dimen col = 0;
         size_t i;
@@ -175 +156 @@
         col = 0;
 
         /* Print the hex offset */
-        g_snprintf (hex_buff, sizeof (hex_buff), "%08" PRIXMAX " ", (uintmax_t) from);
-        widget_move (view, top + row, left);
-        tty_setcolor (VIEW_BOLD_COLOR);
-        for (i = 0; col < width && hex_buff[i] != '\0'; i++)
-        {
-            tty_print_char (hex_buff[i]);
-            /*              tty_print_char(hex_buff[i]); */
-            col += 1;
+        if (row >= 0) {
+            g_snprintf (hex_buff, sizeof (hex_buff), "%08" PRIXMAX " ", (uintmax_t) from);
+            widget_move (view, top + row, left);
+            tty_setcolor (VIEW_BOLD_COLOR);
+            for (i = 0; col < width && hex_buff[i] != '\0'; i++)
+            {
+                tty_print_char (hex_buff[i]);
+                col += 1;
+            }
+            tty_setcolor (VIEW_NORMAL_COLOR);
         }
-        tty_setcolor (VIEW_NORMAL_COLOR);
 
         for (bytes = 0; bytes < view->bytes_per_line; bytes++, from++)
         {
@@ -192 +174 @@
 #ifdef HAVE_CHARSET
             if (view->utf8)
             {
-                int cw = 1;
-                gboolean read_res = TRUE;
-
-                ch = mcview_get_utf (view, from, &cw, &read_res);
-                if (!read_res)
-                    break;
-                /* char width is greater 0 bytes */
-                if (cw != 0)
-                {
-                    int cnt;
-                    char corr_buf[UTF8_CHAR_LEN + 1];
-                    struct hexedit_change_node *corr = curr;
-                    int res;
-
-                    res = g_unichar_to_utf8 (ch, (char *) corr_buf);
-
-                    for (cnt = 0; cnt < cw; cnt++)
-                    {
-                        if (curr != NULL && from + cnt == curr->offset)
-                        {
-                            /* replace only changed bytes in array of multibyte char */
-                            corr_buf[cnt] = curr->value;
-                            curr = curr->next;
-                        }
+                int res;
+                int j;
+                struct hexedit_change_node *corr = curr;
+                gchar utf8buf[UTF8_CHAR_LEN + 1];
+
+                if (cont_bytes) {
+                    /* UTF-8 continuation bytes, print a space (with proper attributes)... */
+                    cont_bytes--;
+                    ch = ' ';
+                    if (cjk_right) {
+                        /* ... except when it'd wipe out the right half of a CJK, then print nothing */
+                        cjk_right = FALSE;
+                        ch = -1;
                     }
-                    corr_buf[res] = '\0';
-                    /* Determine the state of the current multibyte char */
-                    ch = utf8_to_int ((char *) corr_buf, &cw, &read_res);
-                    curr = corr;
-                }
+                } else {
+                    for (j = 0; j < UTF8_CHAR_LEN; j++) {
+                        if (mcview_get_byte (view, from + j, &res))
+                            utf8buf[j] = res;
+                        else
+                        {
+                            utf8buf[j] = '\0';
+                            break;
+                        }
+                        if (curr != NULL && from + j == curr->offset)
+                            utf8buf[j] = curr->value;
+                        if (curr != NULL && from + j >= curr->offset)
+                            curr = curr->next;
+                    }
+                    utf8buf[UTF8_CHAR_LEN] = '\0';
+                    /* Determine the state of the current multibyte char */
+                    ch = g_utf8_get_char_validated (utf8buf, -1);
+                    if (ch == -1 || ch == -2) {
+                        ch = '.';
+                    } else {
+                        gchar *next_ch = g_utf8_next_char (utf8buf);
+                        cont_bytes = next_ch - utf8buf - 1;
+                        if (g_unichar_iswide(ch))
+                            cjk_right = TRUE;
+                    }
+                    curr = corr;
+                }
             }
 #endif /* HAVE_CHARSET */
+
+            /* For negative rows, the only thing we care about is overflowing
+             * UTF-8 continuation bytes which were handled above. */
+            if (row < 0)
+            {
+                if (curr != NULL && from == curr->offset)
+                    curr = curr->next;
+                continue;
+            }
+
             if (!mcview_get_byte (view, from, &c))
                 break;