Full support edit CRLF(win) files in editor

src/editor/editbuffer.c

@@ -154,6 +154,9 @@ edit_buffer_init (edit_buffer_t *buf, off_t size)
 
     buf->size = size;
     buf->lines = 0;
+
+    buf->lb_detected = LB_ASIS;
+    buf->lb_dirty = TRUE;
 }
 
 /* --------------------------------------------------------------------------------------------- */
@@ -506,6 +509,9 @@ edit_buffer_insert (edit_buffer_t *buf, int c)
 
     // update file length
     buf->size++;
+
+    if (c == '\r' || c == '\n')
+        buf->lb_dirty = TRUE;
 }
 
 /* --------------------------------------------------------------------------------------------- */
@@ -538,6 +544,9 @@ edit_buffer_insert_ahead (edit_buffer_t *buf, int c)
 
     // update file length
     buf->size++;
+
+    if (c == '\r' || c == '\n')
+        buf->lb_dirty = TRUE;
 }
 
 /* --------------------------------------------------------------------------------------------- */
@@ -577,6 +586,9 @@ edit_buffer_delete (edit_buffer_t *buf)
     // update file length
     buf->size--;
 
+    if (c == '\r' || c == '\n')
+        buf->lb_dirty = TRUE;
+
     return c;
 }
 
@@ -617,6 +629,9 @@ edit_buffer_backspace (edit_buffer_t *buf)
     // update file length
     buf->size--;
 
+    if (c == '\r' || c == '\n')
+        buf->lb_dirty = TRUE;
+
     return c;
 }
 
@@ -790,6 +805,9 @@ edit_buffer_read_file (edit_buffer_t *buf, int fd, off_t size,
         }
     }
 
+    // the loaded content may contain line breaks
+    buf->lb_dirty = TRUE;
+
     return ret;
 }
 
@@ -895,3 +913,125 @@ edit_buffer_calc_percent (const edit_buffer_t *buf, off_t offset)
 }
 
 /* --------------------------------------------------------------------------------------------- */
+/**
+ * Get the start offset of the trailing whitespace (spaces and tabs) run of the
+ * line that begins at the given offset. If the line has no trailing whitespace
+ * the end of the line content is returned, so no character of the line is at or
+ * after the result.
+ *
+ * @param buf editor buffer
+ * @param bol offset of the first character of the line
+ *
+ * @return start offset of the trailing whitespace run (or end of line content)
+ */
+
+off_t
+edit_buffer_trailing_ws_start (const edit_buffer_t *buf, off_t bol)
+{
+    off_t tws, eol;
+
+    eol = edit_buffer_get_eol (buf, bol);
+
+    // a CRLF line break's content ends at the "\r", not at the "\n"
+    if (eol > bol && edit_buffer_is_crlf (buf, eol - 1))
+        eol--;
+
+    for (tws = eol; tws > bol; tws--)
+    {
+        int c;
+
+        c = edit_buffer_get_byte (buf, tws - 1);
+        if (c != ' ' && c != '\t')
+            break;
+    }
+
+    return tws;
+}
+
+/* --------------------------------------------------------------------------------------------- */
+/**
+ * Detect the line break type used in the buffer content.
+ *
+ * @param buf editor buffer
+ *
+ * @return LB_WIN if all line breaks are "\r\n"; LB_UNIX if all line breaks
+ *         are "\n" (including buffers without any line breaks); LB_MAC if
+ *         all line breaks are "\r"; LB_ASIS if the buffer uses a mixture of
+ *         different line breaks (such a file can only be saved as-is)
+ */
+
+LineBreaks
+edit_buffer_detect_line_breaks (const edit_buffer_t *buf)
+{
+    off_t crlf = 0;  // "\r\n" line breaks
+    off_t cr = 0;    // "\r" line breaks not followed by "\n"
+    off_t lf = 0;    // "\n" line breaks not preceded by "\r"
+    off_t i;
+
+    for (i = 0; i < buf->size; i++)
+    {
+        if (edit_buffer_get_byte (buf, i) == '\r')
+        {
+            if (edit_buffer_is_crlf (buf, i))
+                crlf++;
+            else
+                cr++;
+        }
+        else if (edit_buffer_get_byte (buf, i) == '\n'
+                 && (i == 0 || edit_buffer_get_byte (buf, i - 1) != '\r'))
+            lf++;
+    }
+
+    if (crlf > 0)
+    {
+        if (cr > 0 || lf > 0)
+            return LB_ASIS;  // mixture of line breaks
+        return LB_WIN;
+    }
+
+    if (cr > 0)
+    {
+        if (lf > 0)
+            return LB_ASIS;  // mixture of line breaks
+        return LB_MAC;
+    }
+
+    return LB_UNIX;
+}
+
+/* --------------------------------------------------------------------------------------------- */
+/**
+ * Recompute the cached line break type if the line breaks have changed.
+ * Cheap (O(1)) unless a line break was inserted or deleted since the last
+ * recomputation.
+ *
+ * @param buf pointer to editor buffer
+ */
+
+void
+edit_buffer_refresh_line_breaks (edit_buffer_t *buf)
+{
+    if (buf->lb_dirty)
+    {
+        buf->lb_detected = edit_buffer_detect_line_breaks (buf);
+        buf->lb_dirty = FALSE;
+    }
+}
+
+/* --------------------------------------------------------------------------------------------- */
+/**
+ * Get the cached line break type of the buffer content.
+ * Call edit_buffer_refresh_line_breaks() first if the result must be current.
+ *
+ * @param buf pointer to editor buffer
+ *
+ * @return cached detect_line_breaks() result
+ */
+
+LineBreaks
+edit_buffer_get_line_breaks (const edit_buffer_t *buf)
+{
+    return buf->lb_detected;
+}
+
+/* --------------------------------------------------------------------------------------------- */

src/editor/editbuffer.h

@@ -20,6 +20,9 @@ typedef struct edit_buffer_struct
     off_t size;      // file size
     long lines;      // total lines in the file
     long curs_line;  // line number of the cursor.
+
+    LineBreaks lb_detected;  // cached detect_line_breaks() result
+    gboolean lb_dirty;       // line breaks have changed, need detect_line_breaks()
 } edit_buffer_t;
 
 typedef struct edit_buffer_read_file_status_msg_struct
@@ -63,6 +66,11 @@ off_t edit_buffer_write_file (edit_buffer_t *buf, int fd);
 
 int edit_buffer_calc_percent (const edit_buffer_t *buf, off_t offset);
 
+off_t edit_buffer_trailing_ws_start (const edit_buffer_t *buf, off_t bol);
+LineBreaks edit_buffer_detect_line_breaks (const edit_buffer_t *buf);
+void edit_buffer_refresh_line_breaks (edit_buffer_t *buf);
+LineBreaks edit_buffer_get_line_breaks (const edit_buffer_t *buf);
+
 /*** inline functions ****************************************************************************/
 
 static inline int
@@ -109,6 +117,25 @@ edit_buffer_get_current_eol (const edit_buffer_t *buf)
     return edit_buffer_get_eol (buf, buf->curs1);
 }
 
+/* --------------------------------------------------------------------------------------------- */
+/**
+ * Check whether the buffer contains a Windows ("\r\n") line break
+ * starting at the specified position.
+ *
+ * @param buf editor buffer
+ * @param byte_index position of the potential "\r" character
+ *
+ * @return TRUE if the two bytes at byte_index and byte_index + 1 are "\r" and "\n"
+ */
+
+static inline gboolean
+edit_buffer_is_crlf (const edit_buffer_t *buf, off_t byte_index)
+{
+    return (byte_index >= 0 && byte_index + 1 < buf->size
+            && edit_buffer_get_byte (buf, byte_index) == '\r'
+            && edit_buffer_get_byte (buf, byte_index + 1) == '\n');
+}
+
 /* --------------------------------------------------------------------------------------------- */
 
 #endif

src/editor/editdraw.c

@@ -142,18 +142,59 @@ format_character_code (WEdit *edit)
 
 #undef CHAR_CODE_BUF_SIZE
 
+/* --------------------------------------------------------------------------------------------- */
+/**
+ * Get the cached line break type of the buffer content.
+ * The result is recomputed only when line breaks have changed.
+ */
+
+static inline LineBreaks
+edit_get_detected_line_breaks (edit_buffer_t *buf)
+{
+    edit_buffer_refresh_line_breaks (buf);
+    return edit_buffer_get_line_breaks (buf);
+}
+
+/* --------------------------------------------------------------------------------------------- */
+/**
+ * Get the status line sign describing the line break type
+ * of the buffer content:
+ * <CRLF> - Windows ("\r\n")
+ * <CR>   - Macintosh ("\r")
+ * <LF>   - Unix ("\n")
+ * <?>    - mixture of line breaks or no line breaks
+ */
+
+static inline const char *
+edit_line_breaks_status_sign (edit_buffer_t *buf)
+{
+    switch (edit_get_detected_line_breaks (buf))
+    {
+    case LB_WIN:
+        return "<CRLF>";
+    case LB_MAC:
+        return "<CR>";
+    case LB_UNIX:
+        return "<LF>";
+    case LB_ASIS:
+    default:
+        return "<?>";
+    }
+}
+
 /* --------------------------------------------------------------------------------------------- */
 
 static inline void
 status_string (WEdit *edit, char *s, int w)
 {
     char *character_code;
+    const char *lb = edit_line_breaks_status_sign (&edit->buffer);
 
     character_code = format_character_code (edit);
 
     // The field lengths just prevent the status line from shortening too much
     if (edit_options.simple_statusbar)
-        g_snprintf (s, w, "%c%c%c%c %3ld %5ld/%ld %6ld/%ld [%s] %s",
+        g_snprintf (s, w, "%c%c%c%c %3ld %5ld/%ld %6ld/%ld [%s] %s %s",
                     edit->mark1 != edit->mark2 ? (edit->column_highlight ? 'C' : 'B') : '-',  //
                     edit->modified != 0 ? 'M' : '-',                                          //
                     macro_index < 0 ? '-' : 'R',                                              //
@@ -165,9 +206,11 @@ status_string (WEdit *edit, char *s, int w)
                     (long) edit->buffer.size,                                                 //
                     character_code,
                     mc_global.source_codepage >= 0 ? get_codepage_id (mc_global.source_codepage)
-                                                   : "");
+                                                   : "",
+                    lb  //
+        );
     else
-        g_snprintf (s, w, "[%c%c%c%c] %2ld L:[%3ld+%2ld %3ld/%3ld] *(%-4ld/%4ldb) [%s]  %s",
+        g_snprintf (s, w, "[%c%c%c%c] %2ld L:[%3ld+%2ld %3ld/%3ld] *(%-4ld/%4ldb) [%s]  %s  %s",
                     edit->mark1 != edit->mark2 ? (edit->column_highlight ? 'C' : 'B') : '-',  //
                     edit->modified != 0 ? 'M' : '-',                                          //
                     macro_index < 0 ? '-' : 'R',                                              //
@@ -181,7 +224,9 @@ status_string (WEdit *edit, char *s, int w)
                     (long) edit->buffer.size,                                                 //
                     character_code,
                     mc_global.source_codepage >= 0 ? get_codepage_id (mc_global.source_codepage)
-                                                   : "");
+                                                   : "",
+                    lb  //
+        );
 
     g_free (character_code);
 }
@@ -297,6 +342,15 @@ edit_status_window (WEdit *edit)
                     edit->overwrite == 0 ? '-' : 'O');
     }
 
+    tty_getyx (&y, &x);
+    x -= w->rect.x;
+    x += 4;
+    if (x + 6 <= cols - 2 - 6)
+    {
+        edit_move (x, 0);
+        tty_printf ("%s", edit_line_breaks_status_sign (&edit->buffer));
+    }
+
     if (cols > 30)
     {
         edit_move (2, w->rect.lines - 1);
@@ -497,11 +551,16 @@ edit_draw_this_line (WEdit *edit, off_t b, long row, long start_col, long end_co
     int col, start_col_real;
     int abn_style;
     int book_mark = 0;
+    gboolean crlf_hidden;
     char line_stat[LINE_STATE_WIDTH + 1] = "\0";
 
     if (row > w->rect.lines - 1 - EDIT_TEXT_VERTICAL_OFFSET - 2 * (edit->fullscreen != 0 ? 0 : 1))
         return;
 
+    // in a pure Windows file the "\r" of a "\r\n" line break is a line break
+    // (hidden); in any other file it is shown as "^M"
+    crlf_hidden = (edit_buffer_get_line_breaks (&edit->buffer) == LB_WIN);
+
     if (book_mark_query_color (edit, edit->start_line + row, EDITOR_BOOKMARK_COLOR))
         book_mark = EDITOR_BOOKMARK_COLOR;
     else if (book_mark_query_color (edit, edit->start_line + row, EDITOR_BOOKMARK_FOUND_COLOR))
@@ -554,14 +613,7 @@ edit_draw_this_line (WEdit *edit, off_t b, long row, long start_col, long end_co
             off_t tws = 0;
 
             if (edit_options.visible_tws && tty_use_colors ())
-                for (tws = edit_buffer_get_eol (&edit->buffer, b); tws > b; tws--)
-                {
-                    unsigned int c;
-
-                    c = edit_buffer_get_byte (&edit->buffer, tws - 1);
-                    if (!whitespace (c))
-                        break;
-                }
+                tws = edit_buffer_trailing_ws_start (&edit->buffer, b);
 
             while (col <= end_col - edit->start_col)
             {
@@ -603,6 +655,13 @@ edit_draw_this_line (WEdit *edit, off_t b, long row, long start_col, long end_co
                 else
                     c = edit_buffer_get_byte (&edit->buffer, q);
 
+                // the "\r" of a hidden "\r\n" line break is not shown
+                if (c == '\r' && crlf_hidden && edit_buffer_is_crlf (&edit->buffer, q))
+                {
+                    q++;
+                    continue;
+                }
+
                 // we don't use bg for mc - fg contains both
                 if (book_mark != 0)
                     p->style |= book_mark << 16;
@@ -817,6 +876,10 @@ render_edit_text (WEdit *edit, long start_row, long start_column, long end_row,
     int y1, x1, y2, x2;
     int last_line, last_column;
 
+    // make sure the cached line break type (used to decide whether a "\r" is
+    // shown as "^M" or hidden) is up to date before drawing
+    edit_buffer_refresh_line_breaks (&edit->buffer);
+
     // draw only visible region
 
     last_line = wh->rect.y + wh->rect.lines - 1;