Remove reader status and handle errors more precisely

7 files changed, 36 insertions, 43 deletions

diff --git a/serd/serd.h b/serd/serd.h
index 6bdb9164..7082901e 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/byte_source.h b/src/byte_source.h
index beb9d695..1d2c672f 100644
--- a/src/byte_source.h
+++ b/src/byte_source.h
@@ -88,9 +88,9 @@ serd_byte_source_advance(SerdByteSource* source)
 
 	const bool was_eof = source->eof;
 	if (source->from_stream) {
-		source->eof = false;
 		if (source->page_size > 1) {
 			if (++source->read_head == source->page_size) {
+				source->eof = false;
 				st = serd_byte_source_page(source);
 			} else if (source->read_head == source->buf_size) {
 				source->eof = true;
diff --git a/src/n3.c b/src/n3.c
index a3075cac..695ad9aa 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -203,9 +203,10 @@ read_utf8_character(SerdReader* reader, SerdNode* dest, uint8_t c)
 	uint32_t   size;
 	uint8_t    bytes[4];
 	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);
@@ -218,8 +219,8 @@ 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))) {
@@ -307,7 +308,7 @@ static SerdStatus
 read_STRING_LITERAL_LONG(SerdReader* reader, SerdNode* dest, uint8_t q)
 {
 	SerdStatus st = SERD_SUCCESS;
-	while (!reader->status && !(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);
@@ -320,6 +321,8 @@ read_STRING_LITERAL_LONG(SerdReader* reader, SerdNode* dest, uint8_t q)
 				      peek_byte(reader));
 				return SERD_ERR_BAD_SYNTAX;
 			}
+		} 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));
@@ -346,7 +349,7 @@ static SerdStatus
 read_STRING_LITERAL(SerdReader* reader, SerdNode* dest, uint8_t q)
 {
 	SerdStatus st = SERD_SUCCESS;
-	while (!reader->status && !(st && reader->strict)) {
+	while (!st || (st == SERD_ERR_BAD_SYNTAX && !reader->strict)) {
 		const int c    = peek_byte(reader);
 		uint32_t  code = 0;
 		switch (c) {
@@ -354,19 +357,20 @@ read_STRING_LITERAL(SerdReader* reader, SerdNode* dest, uint8_t q)
 			r_err(reader, SERD_ERR_BAD_SYNTAX, "end of file in short string\n");
 			return SERD_ERR_BAD_SYNTAX;
 		case '\n': case '\r':
-			r_err(reader, SERD_ERR_BAD_SYNTAX, "line end in short string\n");
-			return SERD_ERR_BAD_SYNTAX;
+			st = r_err(reader, SERD_ERR_NO_DATA,
+			           "line end in short string\n");
+			break;
 		case '\\':
 			eat_byte_safe(reader, c);
 			if ((st = read_ECHAR(reader, dest, &dest->flags)) &&
 			    (st = read_UCHAR(reader, dest, &code))) {
-				return r_err(reader, st,
-				             "invalid escape `\\%c'\n", peek_byte(reader));
+				st = r_err(reader, st,
+				           "invalid escape `\\%c'\n", peek_byte(reader));
 			}
 			break;
 		default:
 			if (c == q) {
-				eat_byte_check(reader, q);
+				eat_byte_safe(reader, q);
 				return SERD_SUCCESS;
 			} else {
 				st = read_character(
@@ -375,8 +379,7 @@ read_STRING_LITERAL(SerdReader* reader, SerdNode* dest, uint8_t q)
 		}
 	}
 
-	return st ? st
-	          : eat_byte_check(reader, q) ? SERD_SUCCESS : SERD_ERR_BAD_SYNTAX;
+	return st;
 }
 
 static SerdStatus
@@ -689,9 +692,7 @@ 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;
 			}
 		}
 	}
@@ -1453,8 +1454,6 @@ read_n3_statement(SerdReader* reader)
 			ctx.subject = subj = 0;
 			TRY(st, read_wrappedGraph(reader, &ctx));
 			read_ws_star(reader);
-		} else if (!ctx.subject) {
-			st = r_err(reader, SERD_ERR_BAD_SYNTAX, "bad subject\n");
 		} else if ((st = read_triples(reader, ctx, &ate_dot))) {
 			if (st == SERD_FAILURE && s_type == '[') {
 				return SERD_SUCCESS;
@@ -1489,23 +1488,23 @@ 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;
 			}
 			skip_until(reader, '\n');
-			reader->status = SERD_SUCCESS;
 		}
 		serd_stack_pop_to(&reader->stack, orig_stack_size);
 	}
-	return reader->status;
+	return SERD_SUCCESS;
 }
 
 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;
@@ -1549,11 +1548,8 @@ read_nquadsDoc(SerdReader* reader)
 			}
 		}
 
-		if (emit_statement(reader, ctx, ctx.object)) {
-			break;
-		}
-
+		st = emit_statement(reader, ctx, ctx.object);
 		serd_stack_pop_to(&reader->stack, orig_stack_size);
 	}
-	return reader->status;
+	return st;
 }
diff --git a/src/reader.c b/src/reader.c
index 248eaff7..c22579e3 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -91,7 +91,6 @@ push_node_padded(SerdReader* reader, size_t maxlen,
 		&reader->stack, sizeof(SerdNode) + maxlen + 1, sizeof(SerdNode));
 
 	if (!mem) {
-		reader->status = SERD_ERR_OVERFLOW;
 		return NULL;
 	}
 
@@ -285,15 +284,15 @@ serd_reader_start_string(SerdReader*     reader,
 static SerdStatus
 serd_reader_prepare(SerdReader* reader)
 {
-	reader->status = serd_byte_source_prepare(&reader->source);
-	if (reader->status == SERD_SUCCESS) {
-		reader->status = skip_bom(reader);
-	} else if (reader->status == SERD_FAILURE) {
+	SerdStatus st = serd_byte_source_prepare(&reader->source);
+	if (st == SERD_SUCCESS) {
+		st = skip_bom(reader);
+	} else if (st == SERD_FAILURE) {
 		reader->source.eof = true;
 	} else {
-		r_err(reader, reader->status, "read error: %s\n", strerror(errno));
+		r_err(reader, st, "error preparing read: %s\n", strerror(errno));
 	}
-	return reader->status;
+	return st;
 }
 
 SerdStatus
diff --git a/src/reader.h b/src/reader.h
index 302c8a7b..01462af8 100644
--- a/src/reader.h
+++ b/src/reader.h
@@ -61,7 +61,6 @@ struct SerdReaderImpl {
 	SerdStack                stack;
 	SerdSyntax               syntax;
 	unsigned                 next_id;
-	SerdStatus               status;
 	uint8_t*                 buf;
 	char*                    bprefix;
 	size_t                   bprefix_len;
@@ -108,11 +107,8 @@ peek_byte(SerdReader* reader)
 static inline int
 eat_byte(SerdReader* reader)
 {
-	const int        c  = peek_byte(reader);
-	const SerdStatus st = serd_byte_source_advance(&reader->source);
-	if (st) {
-		reader->status = st;
-	}
+	const int c = peek_byte(reader);
+	serd_byte_source_advance(&reader->source);
 	return c;
 }
 
diff --git a/src/string.c b/src/string.c
index 9f82efb1..77ebd40a 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";
 	}
 	return "Unknown error";  // never reached
 }
diff --git a/tests/serd_test.c b/tests/serd_test.c
index 041cd9ea..4bbdb960 100644
--- a/tests/serd_test.c
+++ b/tests/serd_test.c
@@ -318,7 +318,7 @@ test_strerror(void)
 {
 	const char* msg = NULL;
 	assert(!strcmp((msg = serd_strerror(SERD_SUCCESS)), "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"));
 	}