better buffer opening and displaying, maybe fixing #40

src/display.c

@@ -96,30 +96,33 @@ void display_buffer(Buffer buf, int len_line_number) {
 
             if (size + gw > screen_sz) {
                 attron(A_REVERSE);
-                while (size < screen_sz) {
+                for (; size < screen_sz; size++)
                     addch('<');
-                    size++;
-                }
                 attroff(A_REVERSE);
 
                 break;
             }
 
+            if (1 == grapheme.sz && !isprint(*grapheme.dt))
+                grapheme = replacement_character();
+
             if (1 == grapheme.sz && '\t' == *grapheme.dt) {
-                /*attron(A_REVERSE);
-                addch('T');
-                for (int k = 0; k < config.tablen - 1; k++)
-                    addch(' ');
-                attroff(A_REVERSE);*/
+                attron(A_REVERSE);
+
                 for (int k = 0; k < config.tablen; k++)
                     addch(' ');
+
                 size += config.tablen;
+            } else if (is_replacement_character(grapheme)) {
+                attron(A_REVERSE);
+
+                addch('X');
             } else {
                 printw("%.*s", (int)grapheme.sz, grapheme.dt);
-
                 size += grapheme_width(grapheme);
             }
 
+
             attroff(A_REVERSE);
         }
 

src/grapheme.c

@@ -39,8 +39,6 @@ Grapheme get_next_grapheme(char **str, size_t len) {
     return grapheme;
 }
 
-void print_grapheme_width_debug() {
-}
 
 size_t grapheme_width(Grapheme g) {
     // TODO: the current mechanism of getting the width of a grapheme cluster
@@ -53,10 +51,6 @@ size_t grapheme_width(Grapheme g) {
     // Resources that may be useful:
     // - https://www.unicode.org/reports/tr51/
 
-    if (1 == g.sz && *g.dt == '\t')
-        return config.tablen;
-
-
     for (size_t off = 0; off < g.sz; ) {
         uint_least32_t cp;
 
@@ -125,4 +119,18 @@ size_t index_by_width(size_t wi, char **s) {
     return wi;
 }
 
+bool is_replacement_character(Grapheme g) {
+    return g.sz == 3
+        && g.dt[0] == (char)0xef
+        && g.dt[1] == (char)0xbf
+        && g.dt[2] == (char)0xbd;
+}
+
+Grapheme replacement_character(void) {
+    static unsigned char dt[3] = {0xef, 0xbf, 0xbd};
+    Grapheme g = {3, (char *)dt};
+
+    return g;
+}
+
 

src/open_and_save.c

@@ -56,31 +56,27 @@ Buffer read_lines(FILE *fp, char *filename, bool can_write) {
     }
 
     b.num_lines = 0;
-    for (size_t i = 0; !feof(fp); i++) {
+    for (size_t ln_num = 0; !feof(fp); ln_num++) {
         b.lines = realloc(b.lines, ++b.num_lines * sizeof(Line));
-        b.lines[i] = blank_line();
-        Line *curln = &b.lines[i];
+        b.lines[ln_num] = blank_line();
+        Line *curln = &b.lines[ln_num];
 
-
-        int c;
-        size_t j;
-        for (j = 0; EOF != (c = fgetc(fp)) && '\n' != c; j++) {
-            if (c == '\r')
+        for (int c; EOF != (c = fgetc(fp)) && '\n' != c; curln->length++) {
+            if ('\r' == c) {
                 b.crlf = true;
+                continue;
+            }
 
-            // TODO: check unicode if file isn't tooooo big
-            // For now the buffer is not being asserted to be encoded correctly
-
-            if (++curln->length >= curln->cap) {
+            if (curln->length + 1 >= curln->cap) {
                 curln->cap *= 2;
-                curln->data = realloc(curln->data, curln->cap * sizeof(Line));
+                curln->data = realloc(curln->data, curln->cap * sizeof(char));
             }
-            curln->data[j] = c;
+            curln->data[curln->length] = c;
         }
-
-        curln->data[j] = '\0';
+        curln->data[curln->length] = '\0';
     }
 
+
     if (fp) fclose(fp);
 
     if (b.num_lines > 0)

src/ted.h

@@ -175,6 +175,8 @@ size_t wi_to_gi(size_t si, char *s);
 size_t gi_to_wi(size_t gi, char *s);
 ssize_t index_by_width_after(size_t _wi, char **s);
 size_t index_by_width(size_t wi, char **s);
+bool is_replacement_character(Grapheme g);
+Grapheme replacement_character(void);
 
 
 extern GlobalCfg config;