ledit

commit 79d3040398e884808312473b32e65a02a04feeef
parent f08805b3d2b41daefe6a552a569440a3e09b4b65
Author: lumidify <nobody@lumidify.org>
Date:   Thu,  5 Sep 2024 09:27:09 +0200

Handle \r in input files

Diffstat:
M
buffer.c
 | 
54
++++++++++++++++++++----------------------------------
M
ledit.1
 | 
11
++++++++++-

2 files changed, 30 insertions(+), 35 deletions(-)
diff --git a/buffer.c b/buffer.c
@@ -58,11 +58,6 @@ static void resize_and_move_text_gap(ledit_line *line, size_t min_size, size_t i
 static void resize_and_move_line_gap(ledit_buffer *buffer, size_t min_size, size_t index);
 
 /*
- * Similar to strchr, but the length of the text is given separately
- */
-static char *strchr_len(char *text, char c, size_t len);
-
-/*
  * Initialize a line with default values for its struct members.
  */
 static void init_line(ledit_buffer *buffer, ledit_line *line);
@@ -352,9 +347,11 @@ buffer_load_file(ledit_buffer *buffer, char *filename, size_t line, char **errst
 	if (ferror(file)) goto errorclose;
 	file_contents[len + off] = '\0';
 	/* don't generate extra newline at end */
-	if (len + off > 0 && file_contents[len + off - 1] == '\n') {
-		file_contents[len + off - 1] = '\0';
-		len--;
+	for (int i = 0; i < 2; i++) {
+		if (len + off > 0 && (file_contents[len + off - 1] == '\n' || file_contents[len + off - 1] == '\r')) {
+			file_contents[len + off - 1] = '\0';
+			len--;
+		}
 	}
 	if (fclose(file)) goto error;
 
@@ -583,16 +580,6 @@ buffer_insert_text_base(ledit_buffer *buffer, size_t line_index, size_t index, c
 	}
 }
 
-/* FIXME: this isn't optimized like the standard version, but whatever */
-static char *
-strchr_len(char *text, char c, size_t len) {
-	for (size_t i = 0; i < len; i++) {
-		if (text[i] == c)
-			return text + i;
-	}
-	return NULL;
-}
-
 /* FIXME: make these functions that call recalc* also be final as described above */
 static void
 buffer_insert_text_with_newlines(
@@ -613,34 +600,33 @@ buffer_insert_text_with_newlines(
 		buffer_recalc_from_line(buffer, line_index);
 }
 
-/* FIXME: also look for \r */
 static void
 buffer_insert_text_with_newlines_base(
     ledit_buffer *buffer,
     size_t line_index, size_t index,
     char *text, size_t len,
     size_t *end_line_ret, size_t *end_byte_ret) {
-	size_t rem_len = len;
-	char *cur, *last = text;
 	size_t cur_line = line_index;
 	size_t cur_index = index;
-	/* FIXME: strchr_len isn't really needed when the lines are normalized anyways */
-	while ((cur = strchr_len(last, '\n', rem_len)) != NULL) {
-		/* FIXME: this is probably inefficient, but I don't have time to
-		   think about it right now */
-		buffer_append_line_base(buffer, cur_line, cur_index, 1);
-		buffer_insert_text_base(buffer, cur_line, cur_index, last, cur - last);
-		cur_index = 0;
-		cur_line++;
-		rem_len -= cur - last + 1;
-		last = cur + 1;
+	size_t last_pos = 0;
+	for (size_t cur_pos = 0; cur_pos < len; cur_pos++) {
+		if (text[cur_pos] == '\n' || text[cur_pos] == '\r') {
+			buffer_append_line_base(buffer, cur_line, cur_index, 1);
+			buffer_insert_text_base(buffer, cur_line, cur_index, text + last_pos, cur_pos - last_pos);
+			cur_index = 0;
+			cur_line++;
+			/* handle \n\r or \r\n */
+			if (cur_pos + 1 < len && (text[cur_pos + 1] == '\r' || text[cur_pos + 1] == '\n'))
+				cur_pos++;
+			last_pos = cur_pos + 1;
+		}
 	}
-	/* FIXME: check how legal this casting between pointers and size_t's is */
-	buffer_insert_text_base(buffer, cur_line, cur_index, last, text + len - last);
+	if (last_pos < len)
+		buffer_insert_text_base(buffer, cur_line, cur_index, text + last_pos, len - last_pos);
 	if (end_line_ret)
 		*end_line_ret = cur_line;
 	if (end_byte_ret)
-		*end_byte_ret = cur_index + text + len - last;
+		*end_byte_ret = cur_index + len - last_pos;
 }
 
 static void
diff --git a/ledit.1 b/ledit.1
@@ -1,4 +1,4 @@
-.Dd November 4, 2023
+.Dd September 5, 2024
 .Dt LEDIT 1
 .Os
 .Sh NAME
@@ -33,6 +33,15 @@ and the default bindings in
 .Pa leditrc.example .
 .Pp
 WARNING: All input data is assumed to be utf8!
+
+Input lines are split at
+.Sq \en ,
+.Sq \er ,
+.Sq \en\er ,
+or
+.Sq \er\en ,
+while lines written to output files only ever use
+.Sq \en .
 .Sh ANTI-DESCRIPTION
 .Nm
 is not a code editor.