Ticket #1652: autodetect line-endings

* On opening file, detect line-endings used by sampling some initial content.
    * If it happen to be CR or CRLF, skip fast load path, and in edit_insert_file()
      convert such line endings to '\n'.
    * Save detected line ending type for editor.

Signed-off-by: Slava Zanko <slavazanko@gmail.com>

src/editor/edit-impl.h

@@ -258,7 +258,7 @@ int edit_insert_column_of_text_from_file (WEdit * edit, int file,
                                           long *start_pos, long *end_pos, int *col1, int *col2);
 
 char *edit_get_word_from_pos (WEdit * edit, long start_pos, long *start, gsize * len, gsize * cut);
-long edit_insert_file (WEdit * edit, const vfs_path_t * filename_vpath);
+long edit_insert_file (WEdit * edit, const vfs_path_t * filename_vpath, LineBreaks lb_type);
 gboolean edit_load_back_cmd (WEdit * edit);
 gboolean edit_load_forward_cmd (WEdit * edit);
 void edit_block_process_cmd (WEdit * edit, int macro_number);

src/editor/edit.c

@@ -114,6 +114,8 @@ const char VERTICAL_MAGIC[] = { '\1', '\1', '\1', '\1', '\n' };
 
 #define space_width 1
 
+#define DETECT_LB_TYPE_BUFLEN BUF_MEDIUM
+
 /*** file scope type declarations ****************************************************************/
 
 /*** file scope variables ************************************************************************/
@@ -415,6 +417,37 @@ check_file_access (WEdit * edit, const vfs_path_t * filename_vpath, struct stat
 
 /* --------------------------------------------------------------------------------------------- */
 
+/**
+ * detect type of line breaks
+ *
+ */
+/* --------------------------------------------------------------------------------------------- */
+
+static LineBreaks
+detect_lb_type (const vfs_path_t *filename_vpath)
+{
+    char buf[BUF_MEDIUM];
+    ssize_t file, sz;
+
+    file = mc_open (filename_vpath, O_RDONLY | O_BINARY);
+    if (file == -1)
+        return LB_ASIS;
+
+    sz = mc_read (file, buf, sizeof (buf) - 1);
+    mc_close (file);
+
+    if (sz <= 0)
+        return LB_ASIS;
+
+    buf[(size_t) sz] = '\0';
+    if (strstr (buf, "\r\n") != NULL)
+        return LB_WIN;
+    if (strchr (buf, '\r') != NULL)
+        return LB_MAC;
+    return LB_ASIS;
+}
+
+/* --------------------------------------------------------------------------------------------- */
 /**
  * Open the file and load it into the buffers, either directly or using
  * a filter.  Return TRUE on success, FALSE on error.
@@ -431,6 +464,7 @@ static gboolean
 edit_load_file (WEdit * edit)
 {
     gboolean fast_load = TRUE;
+    LineBreaks lb_type = LB_ASIS;
 
     /* Cannot do fast load if a filter is used */
     if (edit_find_filter (edit->filename_vpath) >= 0)
@@ -455,6 +489,10 @@ edit_load_file (WEdit * edit)
             edit_clean (edit);
             return FALSE;
         }
+        lb_type = detect_lb_type (edit->filename_vpath);
+
+        if (lb_type != LB_ASIS && lb_type != LB_UNIX)
+            fast_load = FALSE;
     }
     else
     {
@@ -478,15 +516,16 @@ edit_load_file (WEdit * edit)
             && *(vfs_path_get_by_index (edit->filename_vpath, 0)->path) != '\0')
         {
             edit->undo_stack_disable = 1;
-            if (edit_insert_file (edit, edit->filename_vpath) < 0)
+            if (edit_insert_file (edit, edit->filename_vpath, lb_type) < 0)
             {
                 edit_clean (edit);
                 return FALSE;
             }
+            edit_set_markers (edit, 0, 0, 0, 0);
             edit->undo_stack_disable = 0;
         }
     }
-    edit->lb = LB_ASIS;
+    edit->lb = lb_type;
     return TRUE;
 }
 
@@ -1774,7 +1813,7 @@ user_menu (WEdit * edit, const char *menu_file, int selected_entry)
         {
             long ins_len;
 
-            ins_len = edit_insert_file (edit, block_file_vpath);
+            ins_len = edit_insert_file (edit, block_file_vpath, LB_ASIS);
             if (nomark == 0 && ins_len > 0)
                 edit_set_markers (edit, start_mark, start_mark + ins_len, 0, 0);
         }
@@ -2040,7 +2079,7 @@ edit_get_word_from_pos (WEdit * edit, long start_pos, long *start, gsize * len,
 /** inserts a file at the cursor, returns count of inserted bytes on success */
 
 long
-edit_insert_file (WEdit * edit, const vfs_path_t * filename_vpath)
+edit_insert_file (WEdit * edit, const vfs_path_t * filename_vpath, LineBreaks lb_type)
 {
     char *p;
     long ins_len = 0;
@@ -2120,7 +2159,19 @@ edit_insert_file (WEdit * edit, const vfs_path_t * filename_vpath)
             while ((blocklen = mc_read (file, (char *) buf, TEMP_BUF_LEN)) > 0)
             {
                 for (i = 0; i < blocklen; i++)
-                    edit_insert (edit, buf[i]);
+                {
+                    if (buf[i] == '\r')
+                    {
+                        if (lb_type == LB_MAC)
+                            edit_insert (edit, '\n');
+                        else if (lb_type == LB_WIN)
+                            /* just skip */ ;
+                        else
+                            edit_insert (edit, '\r');
+                    }
+                    else
+                        edit_insert (edit, buf[i]);
+                }
             }
             /* highlight inserted text then not persistent blocks */
             if (!option_persistent_selections && edit->modified)
@@ -2225,6 +2276,7 @@ edit_init (WEdit * edit, int y, int x, int lines, int cols, const vfs_path_t * f
     edit->redo_stack_size_mask = START_STACK_SIZE - 1;
     edit->redo_stack = g_malloc0 ((edit->redo_stack_size + 10) * sizeof (long));
 
+    edit->highlight = 0;
     edit->utf8 = 0;
     edit->converter = str_cnv_from_term;
     edit_set_codeset (edit);

src/editor/editcmd.c

@@ -449,7 +449,9 @@ edit_get_save_file_as (WEdit * edit)
         char *fname;
         vfs_path_t *ret_vpath;
 
-        edit->lb = cur_lb;
+        /* Don't change current LB type (possibly autodetected), unless user asked to. */
+        if (cur_lb != LB_ASIS)
+            edit->lb = cur_lb;
         fname = tilde_expand (filename_res);
         g_free (filename_res);
         ret_vpath = vfs_path_from_str (fname);
@@ -3021,7 +3023,7 @@ edit_paste_from_X_buf_cmd (WEdit * edit)
     /* try use external clipboard utility */
     mc_event_raise (MCEVENT_GROUP_CORE, "clipboard_file_from_ext_clip", NULL);
     tmp = mc_config_get_full_vpath (EDIT_CLIP_FILE);
-    edit_insert_file (edit, tmp);
+    edit_insert_file (edit, tmp, LB_ASIS);
     vfs_path_free (tmp);
 }
 
@@ -3136,7 +3138,7 @@ edit_insert_file_cmd (WEdit * edit)
         vfs_path_t *exp_vpath;
 
         exp_vpath = vfs_path_from_str (exp);
-        ret = (edit_insert_file (edit, exp_vpath) >= 0);
+        ret = (edit_insert_file (edit, exp_vpath, LB_ASIS) >= 0);
         vfs_path_free (exp_vpath);
 
         if (!ret)
@@ -3209,12 +3211,12 @@ edit_sort_cmd (WEdit * edit)
 
     if (edit_block_delete_cmd (edit))
         return 1;
-
+    else
     {
         vfs_path_t *tmp_vpath;
 
         tmp_vpath = mc_config_get_full_vpath (EDIT_TEMP_FILE);
-        edit_insert_file (edit, tmp_vpath);
+        edit_insert_file (edit, tmp_vpath, LB_ASIS);
         vfs_path_free (tmp_vpath);
     }
     return 0;
@@ -3251,16 +3253,15 @@ edit_ext_cmd (WEdit * edit)
         edit_error_dialog (_("External command"), get_sys_error (_("Cannot execute command")));
         return -1;
     }
-
+    else
-    edit->force |= REDRAW_COMPLETELY;
-
     {
         vfs_path_t *tmp_vpath;
 
         tmp_vpath = mc_config_get_full_vpath (EDIT_TEMP_FILE);
-        edit_insert_file (edit, tmp_vpath);
+        edit_insert_file (edit, tmp_vpath, LB_ASIS);
         vfs_path_free (tmp_vpath);
     }
+    edit->force |= REDRAW_COMPLETELY;
     return 0;
 }
 

src/editor/editdraw.c

@@ -112,6 +112,13 @@ status_string (WEdit * edit, char *s, int w)
     unsigned int cur_utf = 0;
     int cw = 1;
 
+    static const char *lb_names[LB_NAMES] = {
+        "",
+        "LF",
+        "CRLF",
+        "CR"
+    };
+
     /*
      * If we are at the end of file, print <EOF>,
      * otherwise print the current character as is (if printable),
@@ -151,7 +158,7 @@ status_string (WEdit * edit, char *s, int w)
     /* The field lengths just prevent the status line from shortening too much */
     if (simple_statusbar)
         g_snprintf (s, w,
-                    "%c%c%c%c %3ld %5ld/%ld %6ld/%ld %s %s",
+                    "%c%c%c%c %3ld %5ld/%ld %6ld/%ld %s %s %s",
                     edit->mark1 != edit->mark2 ? (edit->column_highlight ? 'C' : 'B') : '-',
                     edit->modified ? 'M' : '-',
                     macro_index < 0 ? '-' : 'R',
@@ -161,14 +168,14 @@ status_string (WEdit * edit, char *s, int w)
                     edit->total_lines + 1, edit->curs1, edit->last_byte, byte_str,
 #ifdef HAVE_CHARSET
                     mc_global.source_codepage >=
-                    0 ? get_codepage_id (mc_global.source_codepage) : ""
+                    0 ? get_codepage_id (mc_global.source_codepage) : "",
 #else
-                    ""
+                    "",
 #endif
-            );
+                    lb_names[edit->lb]);
     else
         g_snprintf (s, w,
-                    "[%c%c%c%c] %2ld L:[%3ld+%2ld %3ld/%3ld] *(%-4ld/%4ldb) %s  %s",
+                    "[%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 ? 'M' : '-',
                     macro_index < 0 ? '-' : 'R',
@@ -180,11 +187,11 @@ status_string (WEdit * edit, char *s, int w)
                     edit->total_lines + 1, edit->curs1, edit->last_byte, byte_str,
 #ifdef HAVE_CHARSET
                     mc_global.source_codepage >=
-                    0 ? get_codepage_id (mc_global.source_codepage) : ""
+                    0 ? get_codepage_id (mc_global.source_codepage) : "",
 #else
-                    ""
+                    "",
 #endif
-            );
+                    lb_names[edit->lb]);
 }
 
 /* --------------------------------------------------------------------------------------------- */