Handle read errors more precisely

5 files changed, 22 insertions, 25 deletions

diff --git a/serd/serd.h b/serd/serd.h
index 9d5f09b4..433f38e8 100644
--- a/serd/serd.h
+++ b/serd/serd.h
@@ -110,7 +110,8 @@ typedef enum {
 	SERD_ERR_BAD_CURIE,   ///< Invalid CURIE (e.g. prefix does not exist)
 	SERD_ERR_INTERNAL,    ///< Unexpected internal error (should not happen)
 	SERD_ERR_OVERFLOW,    ///< Stack overflow
-	SERD_ERR_INVALID      ///< Invalid data
+	SERD_ERR_INVALID,     ///< Invalid data
+	SERD_ERR_NO_DATA      ///< Unexpected end of input
 } SerdStatus;
 
 /// RDF syntax type
diff --git a/src/n3.c b/src/n3.c
index 637e728d..5df6cc30 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -206,9 +206,10 @@ read_utf8_character(SerdReader* reader, SerdNode* dest, uint8_t c)
 	uint32_t   size     = 0;
 	uint8_t    bytes[4] = {0, 0, 0, 0};
 	SerdStatus st       = read_utf8_bytes(reader, bytes, &size, c);
-	if (st) {
-		push_bytes(reader, dest, replacement_char, 3);
+	if (st && reader->strict) {
 		return st;
+	} else if (st) {
+		return push_bytes(reader, dest, replacement_char, 3);
 	}
 
 	return push_bytes(reader, dest, bytes, size);
@@ -221,8 +222,8 @@ read_utf8_code(SerdReader* reader, SerdNode* dest, uint32_t* code, uint8_t c)
 	uint8_t    bytes[4] = {0, 0, 0, 0};
 	SerdStatus st       = read_utf8_bytes(reader, bytes, &size, c);
 	if (st) {
-		push_bytes(reader, dest, replacement_char, 3);
-		return st;
+		return reader->strict ? st
+		                      : push_bytes(reader, dest, replacement_char, 3);
 	}
 
 	if (!(st = push_bytes(reader, dest, bytes, size))) {
@@ -312,8 +313,7 @@ static SerdStatus
 read_STRING_LITERAL_LONG(SerdReader* reader, SerdNode* ref, uint8_t q)
 {
 	SerdStatus st = SERD_SUCCESS;
-
-	while (!(st && reader->strict)) {
+	while (!st || (st == SERD_ERR_BAD_SYNTAX && !reader->strict)) {
 		const int c = peek_byte(reader);
 		if (c == '\\') {
 			eat_byte_safe(reader, c);
@@ -323,6 +323,8 @@ read_STRING_LITERAL_LONG(SerdReader* reader, SerdNode* ref, uint8_t q)
 				return r_err(reader, st,
 				             "invalid escape `\\%c'\n", peek_byte(reader));
 			}
+		} else if (c == EOF) {
+			st = r_err(reader, SERD_ERR_NO_DATA, "unexpected end of file\n");
 		} else if (c == q) {
 			eat_byte_safe(reader, q);
 			const int q2 = eat_byte_safe(reader, peek_byte(reader));
@@ -334,9 +336,6 @@ read_STRING_LITERAL_LONG(SerdReader* reader, SerdNode* ref, uint8_t q)
 			ref->flags |= SERD_HAS_QUOTE;
 			push_byte(reader, ref, c);
 			st = read_character(reader, ref, (uint8_t)q2);
-		} else if (c == EOF) {
-			return r_err(reader, SERD_ERR_BAD_SYNTAX,
-			             "end of file in long string\n");
 		} else {
 			st = read_character(reader, ref, (uint8_t)eat_byte_safe(reader, c));
 		}
@@ -351,8 +350,7 @@ static SerdStatus
 read_STRING_LITERAL(SerdReader* reader, SerdNode* ref, uint8_t q)
 {
 	SerdStatus st = SERD_SUCCESS;
-
-	while (!(st && reader->strict)) {
+	while (!st || (st == SERD_ERR_BAD_SYNTAX && !reader->strict)) {
 		const int c    = peek_byte(reader);
 		uint32_t  code = 0;
 		switch (c) {
@@ -372,7 +370,7 @@ read_STRING_LITERAL(SerdReader* reader, SerdNode* ref, uint8_t q)
 			break;
 		default:
 			if (c == q) {
-				eat_byte_check(reader, q);
+				eat_byte_safe(reader, q);
 				return SERD_SUCCESS;
 			} else {
 				st = read_character(
@@ -381,8 +379,7 @@ read_STRING_LITERAL(SerdReader* reader, SerdNode* ref, uint8_t q)
 		}
 	}
 
-	return st ? st
-	          : eat_byte_check(reader, q) ? SERD_SUCCESS : SERD_ERR_BAD_SYNTAX;
+	return st;
 }
 
 static SerdStatus
@@ -723,9 +720,7 @@ read_IRIREF(SerdReader* reader, SerdNode** dest)
 			} else if (!(c & 0x80)) {
 				push_byte(reader, *dest, c);
 			} else if (read_utf8_character(reader, *dest, (uint8_t)c)) {
-				if (reader->strict) {
-					return SERD_ERR_BAD_SYNTAX;
-				}
+				st = reader->strict ? SERD_ERR_BAD_SYNTAX : SERD_FAILURE;
 			}
 		}
 	}
@@ -1532,7 +1527,8 @@ read_turtleTrigDoc(SerdReader* reader)
 		const size_t     orig_stack_size = reader->stack.size;
 		const SerdStatus st              = read_n3_statement(reader);
 		if (st > SERD_FAILURE) {
-			if (reader->strict) {
+			if (reader->strict || reader->source.eof ||
+			    st == SERD_ERR_OVERFLOW) {
 				serd_stack_pop_to(&reader->stack, orig_stack_size);
 				return st;
 			}
@@ -1547,7 +1543,7 @@ SerdStatus
 read_nquadsDoc(SerdReader* reader)
 {
 	SerdStatus st = SERD_SUCCESS;
-	while (!reader->source.eof) {
+	while (!st && !reader->source.eof) {
 		const size_t orig_stack_size = reader->stack.size;
 
 		SerdStatementFlags flags   = 0;
@@ -1591,9 +1587,8 @@ read_nquadsDoc(SerdReader* reader)
 			}
 		}
 
-		TRY(st, emit_statement(reader, ctx, ctx.object));
-
+		st = emit_statement(reader, ctx, ctx.object);
 		serd_stack_pop_to(&reader->stack, orig_stack_size);
 	}
-	return SERD_SUCCESS;
+	return st;
 }
diff --git a/src/reader.c b/src/reader.c
index a88e128d..d2ce61d3 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -286,7 +286,7 @@ serd_reader_prepare(SerdReader* reader)
 	} else if (st == SERD_FAILURE) {
 		reader->source.eof = true;
 	} else {
-		r_err(reader, st, "read error: %s\n", strerror(errno));
+		r_err(reader, st, "error preparing read: %s\n", strerror(errno));
 	}
 	return st;
 }
diff --git a/src/string.c b/src/string.c
index 8ab735dc..b465bf8d 100644
--- a/src/string.c
+++ b/src/string.c
@@ -45,6 +45,7 @@ serd_strerror(SerdStatus status)
 	case SERD_ERR_INTERNAL:   return "Internal error";
 	case SERD_ERR_OVERFLOW:   return "Stack overflow";
 	case SERD_ERR_INVALID:    return "Invalid data";
+	case SERD_ERR_NO_DATA:    return "Unexpectd end of input";
 	default:                  break;
 	}
 	return "Unknown error";  // never reached
diff --git a/tests/serd_test.c b/tests/serd_test.c
index e5e1ae18..7a9e67bd 100644
--- a/tests/serd_test.c
+++ b/tests/serd_test.c
@@ -355,7 +355,7 @@ test_strerror(void)
 {
 	const char* msg = serd_strerror(SERD_SUCCESS);
 	assert(!strcmp(msg, "Success"));
-	for (int i = SERD_FAILURE; i <= SERD_ERR_INVALID; ++i) {
+	for (int i = SERD_FAILURE; i <= SERD_ERR_NO_DATA; ++i) {
 		msg = serd_strerror((SerdStatus)i);
 		assert(strcmp(msg, "Success"));
 	}