encode

1 files changed, 175 insertions, 1 deletions

diff --git a/test_creole.c b/test_creole.c
index b8dc008..22a18b1 100644
--- a/test_creole.c
+++ b/test_creole.c
@@ -3,7 +3,7 @@ GREATEST_MAIN_DEFS();
 #include "creole.c"
 
 /**************************************************************************
- * Reader suite
+ * Reader
  *************************************************************************/
 #define reader_lit(r, s) do {     \
 	r.p = (unsigned char *)s; \
@@ -37,8 +37,182 @@ SUITE(reader) {
 	RUN_TEST1(reader_test_basic, &r);
 }
 
+/**************************************************************************
+ * Pseudo UTF-8 sequences
+ *************************************************************************/
+
+TEST no_values(void) {
+	struct creole_reader r;
+	struct word w;
+	r.p = NULL;
+	r.left = 0;
+	ASSERT_EQ(decode_seq(&r, &w), 0);
+
+	PASS();
+}
+
+struct seq {
+	creole_word max;
+	unsigned encode_to;
+	unsigned high_bits;
+
+	unsigned char minbuf[7];
+	unsigned char maxbuf[7];
+};
+
+void bprint(unsigned char c) {
+	int i;
+
+	for (i = 0; i < 8; i++) {
+		printf("%u", (c >> (7 - i)) & 1);
+	}
+}
+
+void bprintb(unsigned char *b, int len) {
+	while (len-- > 0) {
+		bprint(*b++);
+		printf(" ");
+	}
+}
+
+TEST encode_byte_seq(struct seq *s) {
+	creole_word i = 0;
+	int j;
+	unsigned char buf[7];
+
+	for (;;) {
+		/*
+		printf("0x%X ", i);
+		bprintb(s->minbuf, s->encode_to);
+		printf("\n");
+		*/
+
+		ASSERT_EQ(creole_encode(i, s->encode_to, s->high_bits,
+		                        buf), 1);
+		ASSERT_MEM_EQ(s->minbuf, buf, s->encode_to);
+
+		if (i == s->max)
+			break;
+		i++;
+
+		for (j = s->encode_to - 1; j > 0; j--) {
+			if (s->minbuf[j] == 0xBF) {
+				s->minbuf[j] = 0x80;
+			} else {
+				s->minbuf[j]++;
+				break;
+			}
+		}
+
+		if (j == 0)
+			s->minbuf[0]++;
+	}
+	ASSERT_MEM_EQ(s->maxbuf, s->minbuf, s->encode_to);
+	PASS();
+}
+
+TEST encode_decode_byte_seq(struct seq *s) {
+	unsigned char buf[7];
+	struct creole_reader r = {0};
+	struct word w;
+	creole_word i = 0;
+
+	for (;;) {
+		ASSERT_EQ(creole_encode(i, s->encode_to, s->high_bits,
+		                        buf), 1);
+		r.p = buf;
+		r.left = s->encode_to;
+		ASSERT_EQ(decode_seq(&r, &w), 1);
+		ASSERT_EQ(w.len, s->encode_to);
+		ASSERT_EQ(w.high_bits, s->high_bits);
+		ASSERT_EQ(w.word, i);
+
+		if (i == s->max)
+			break;
+		i++;
+	}
+
+	PASS();
+}
+
+SUITE(pseudo_utf8_encode_all) {
+	struct seq s;
+
+	RUN_TEST(no_values);
+
+	s.max = 0x7F;
+	s.encode_to = 1;
+	s.high_bits = 0;
+	s.minbuf[0] = 0x00;
+	s.maxbuf[0] = 0x7F;
+
+	RUN_TEST1(encode_byte_seq, &s);
+
+	for (s.high_bits = 0; s.high_bits < 16; s.high_bits++) {
+		memset(s.maxbuf, 0xBF, sizeof(s.maxbuf));
+
+		s.max = 0x7F;
+		s.encode_to = 2;
+		s.maxbuf[0] = s.minbuf[0] = 0xC0 | (s.high_bits << 1);
+		s.maxbuf[0] = 0xC1 | (s.high_bits << 1);
+		s.minbuf[1] = 0x80;
+		RUN_TEST1(encode_byte_seq, &s);
+
+		s.max = 0xFFF;
+		s.encode_to = 3;
+		s.minbuf[0] = 0xE0 | s.high_bits;
+		s.maxbuf[0] = 0xE0 | s.high_bits;
+		s.minbuf[1] = 0x80;
+		s.minbuf[2] = 0x80;
+		RUN_TEST1(encode_byte_seq, &s);
+
+		s.max = 0x1FFFF;
+		s.encode_to = 4;
+		s.maxbuf[0] = s.minbuf[0] = 0xF0 | (s.high_bits >> 1);
+		s.minbuf[1] = 0x80 | (s.high_bits & 0x1 << 5);
+		s.maxbuf[1] = 0x9F | (s.high_bits & 0x1 << 5);
+		s.minbuf[2] = 0x80;
+		s.minbuf[3] = 0x80;
+		RUN_TEST1(encode_byte_seq, &s);
+
+		s.max = 0x3FFFFF;
+		s.encode_to = 5;
+		s.maxbuf[0] = s.minbuf[0] = 0xF8 | (s.high_bits >> 2);
+		s.minbuf[1] = 0x80 | (s.high_bits & 0x3 << 4);
+		s.maxbuf[1] = 0x8F | (s.high_bits & 0x3 << 4);
+		s.minbuf[2] = 0x80;
+		s.minbuf[3] = 0x80;
+		s.minbuf[4] = 0x80;
+		RUN_TEST1(encode_byte_seq, &s);
+
+		s.max = 0x7FFFFFF;
+		s.encode_to = 6;
+		s.maxbuf[0] = s.minbuf[0] = 0xFC | (s.high_bits >> 3);
+		s.minbuf[1] = 0x80 | (s.high_bits & 0x7 << 3);
+		s.maxbuf[1] = 0x87 | (s.high_bits & 0x7 << 3);
+		s.minbuf[2] = 0x80;
+		s.minbuf[3] = 0x80;
+		s.minbuf[4] = 0x80;
+		s.minbuf[5] = 0x80;
+		RUN_TEST1(encode_byte_seq, &s);
+
+		s.max = 0xFFFFFFFF;
+		s.encode_to = 7;
+		s.maxbuf[0] = s.minbuf[0] = 0xFE;
+		s.minbuf[1] = 0x80 | (s.high_bits << 2);
+		s.maxbuf[1] = 0x83 | (s.high_bits << 2);
+		s.minbuf[2] = 0x80;
+		s.minbuf[3] = 0x80;
+		s.minbuf[4] = 0x80;
+		s.minbuf[5] = 0x80;
+		s.minbuf[6] = 0x80;
+		RUN_TEST1(encode_byte_seq, &s);
+	}
+}
+
 int main(int argc, char *argv[]) {
 	GREATEST_MAIN_BEGIN();
 	RUN_SUITE(reader);
+	RUN_SUITE(pseudo_utf8_encode_all);
 	GREATEST_MAIN_END();
 }