vtgl

Commit: 2f79cb6313b4316457fe9f2f3333575ec8c69386
Parent: d49dcdf55714cc86063d211aa80d0ee7c541b153
Author: Randy Palamar
Date:   Sun, 11 Aug 2024 15:53:28 -0600

properly ensure utf-8 is added to line_end and add this test case

Diffstat:
M
terminal.c
 | 
19
++++++++++++++-----
M
test.c
 | 
67
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++--

2 files changed, 79 insertions(+), 7 deletions(-)
diff --git a/terminal.c b/terminal.c
@@ -711,7 +711,6 @@ split_raw_input_to_lines(Term *t, s8 raw)
 		lb->buf[lb->widx].has_unicode |= _mm_movemask_epi8(hasutf8);
 		raw = consume(raw, data - raw.data);
 
-		u32 last_c = 0;
 		if (raw.len) {
 			if (peek(raw, 0) == 0x1B) {
 				s8 old = raw;
@@ -727,12 +726,22 @@ split_raw_input_to_lines(Term *t, s8 raw)
 				default: break;
 				}
 			} else {
-				last_c = get_ascii(&raw);
-				if (last_c == '\n') {
+				u32 cp = peek(raw, 0);
+				if (cp == '\n') {
+					raw = consume(raw, 1);
 					parsed_lines++;
 					feed_line(lb, raw.data, t->cursor.state);
-				} else if (last_c & 0x80) {
+				} else if (cp & 0x80) {
 					lb->buf[lb->widx].has_unicode = 1;
+					/* TODO: this is probably slow */
+					size old_len = raw.len;
+					if (get_utf8(&raw) == (u32)-1) {
+						/* NOTE: Need More Bytes! */
+						t->unprocessed_bytes = old_len;
+						return parsed_lines;
+					}
+				} else {
+					raw = consume(raw, 1);
 				}
 			}
 		}
@@ -740,7 +749,7 @@ split_raw_input_to_lines(Term *t, s8 raw)
 		lb->buf[lb->widx].end = raw.data;
 		//if (tv->lines.buf[tv->lines.widx].end < tv->lines.buf[tv->lines.widx].start)
 		//	tv->lines.buf[tv->lines.widx].end += tv->log.cap;
-		if (line_length(lb->buf + lb->widx) > SPLIT_LONG && !(last_c & 0x80)) {
+		if (line_length(lb->buf + lb->widx) > SPLIT_LONG) {
 			parsed_lines++;
 			feed_line(lb, raw.data, t->cursor.state);
 		}
diff --git a/test.c b/test.c
@@ -5,6 +5,8 @@ static const char *glfwGetWindowTitle(GLFWwindow *w) { return "test"; };
 
 #include "util.h"
 
+#include <string.h> /* memcmp */
+
 #define ESC(a) s8("\x1B"#a)
 #define CSI(a) ESC([a)
 
@@ -12,8 +14,9 @@ struct test_result { b32 status; const char *info; };
 #define TEST_FN(name) struct test_result name(Term *term, Arena arena)
 typedef TEST_FN(Test_Fn);
 
-#define TESTS              \
-	X(colour_setting)  \
+#define TESTS                      \
+	X(colour_setting)          \
+	X(cursor_at_line_boundary) \
 	X(cursor_movement)
 
 #define X(fn) static TEST_FN(fn);
@@ -25,6 +28,16 @@ static Test_Fn *tests[] = {
 	TESTS
 };
 
+static void
+buffer_reset(Term *t)
+{
+	t->log.widx          = 0;
+	t->log.filled        = 0;
+	t->unprocessed_bytes = 0;
+	t->log_lines.widx    = 0;
+	t->log_lines.filled  = 0;
+}
+
 static size
 copy_into_ringbuf(RingBuf *rb, s8 raw)
 {
@@ -107,6 +120,55 @@ static TEST_FN(colour_setting)
 	return result;
 }
 
+static TEST_FN(cursor_at_line_boundary)
+{
+	/* NOTE: two tests here
+	 * 1. split_raw_input_to_lines should modify cursor style properties
+	 * 2. split_raw_input_to_lines should not split utf-8 characters into separate lines
+	 */
+	struct test_result result = {.info = __FUNCTION__};
+
+	s8 long_line = s8alloc(&arena, 8192);
+	mem_clear(long_line.data, ' ', long_line.len);
+
+	/* NOTE: change the cursor state in the middle of the line */
+	long_line.data[128 + 0] = 0x1B;
+	long_line.data[128 + 1] = '[';
+	long_line.data[128 + 2] = '5';
+	long_line.data[128 + 3] = 'm';
+
+	/* NOTE: place a non-ASCII character on the long line edge */
+	s8 red_dragon = utf8_encode(0x1F004);
+	long_line.data[SPLIT_LONG - 1] = red_dragon.data[0];
+	long_line.data[SPLIT_LONG + 0] = red_dragon.data[1];
+	long_line.data[SPLIT_LONG + 1] = red_dragon.data[2];
+	long_line.data[SPLIT_LONG + 2] = red_dragon.data[3];
+
+	/* NOTE: shove a newline at the end so that finish parsing the second line */
+	long_line.data[long_line.len - 1] = '\n';
+
+	#if 0
+	long_line.data[SPLIT_LONG + 3] = 0;
+	printf("encoded char: %s\n", (char *)(long_line.data + SPLIT_LONG - 1));
+	long_line.data[SPLIT_LONG + 3] = ' ';
+	#endif
+
+	s8 raw = launder_static_string(term, long_line);
+
+	/* NOTE: check if line actually split and that it was split after red dragon */
+	size parsed_lines = split_raw_input_to_lines(term, raw);
+	result.status  = parsed_lines == 2;
+	result.status &= line_length(term->log_lines.buf) > SPLIT_LONG;
+
+	/* NOTE: check that cursor state was transferred */
+	Cursor line_end = simulate_line(term, term->log_lines.buf);
+	result.status &= !memcmp(&line_end.state,
+	                         &term->log_lines.buf[1].cursor_state,
+	                         sizeof(term->log_lines.buf[0].cursor_state));
+
+	return result;
+}
+
 static u32 failure_count;
 
 int
@@ -125,6 +187,7 @@ main(void)
 	os_alloc_framebuffer(&term.fb, term.size.h, term.size.w);
 
 	for (u32 i = 0; i < ARRAY_COUNT(tests); i++) {
+		buffer_reset(&term);
 		cursor_reset(&term);
 		struct test_result result = tests[i](&term, memory);
 		if (result.status == 0) {