aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--NEWS1
-rw-r--r--doc/serdi.14
-rw-r--r--serd/serd.h8
-rw-r--r--src/n3.c237
-rw-r--r--src/reader.c58
-rw-r--r--src/reader.h67
-rw-r--r--src/serdi.c19
-rw-r--r--src/stack.h11
-rw-r--r--src/string.c1
-rw-r--r--src/writer.c6
-rw-r--r--tests/read_chunk_test.c2
-rw-r--r--tests/serd_test.c8
-rw-r--r--wscript6
13 files changed, 240 insertions, 188 deletions
diff --git a/NEWS b/NEWS
index 59ec44b7..2191dd03 100644
--- a/NEWS
+++ b/NEWS
@@ -8,6 +8,7 @@ serd (1.0.1) unstable;
* Remove half-baked serd_uri_to_path()
* Remove useless character counting from API
* Rename SerdChunk to SerdStringView
+ * Use a fixed-size reader stack
* Use char* for strings in public API
-- David Robillard <d@drobilla.net> Sat, 19 Jan 2019 12:31:12 +0000
diff --git a/doc/serdi.1 b/doc/serdi.1
index fefafd67..a246ff16 100644
--- a/doc/serdi.1
+++ b/doc/serdi.1
@@ -40,6 +40,10 @@ Read input as \fISYNTAX\fR.
Valid values (case-insensitive): \*(lqturtle\*(rq, \*(lqntriples\*(rq, \*(lqtrig\*(rq, \*(lqnquads\*(rq.
.TP
+.BR \-k " " \fIBYTES\fR
+Parser stack size.
+
+.TP
.BR \-l
Lax (non-strict) parsing.
diff --git a/serd/serd.h b/serd/serd.h
index bf1ac8d5..8aa97ed4 100644
--- a/serd/serd.h
+++ b/serd/serd.h
@@ -102,7 +102,8 @@ typedef enum {
SERD_ERR_NOT_FOUND, /**< Not found */
SERD_ERR_ID_CLASH, /**< Encountered clashing blank node IDs */
SERD_ERR_BAD_CURIE, /**< Invalid CURIE (e.g. prefix does not exist) */
- SERD_ERR_INTERNAL /**< Unexpected internal error (should not happen) */
+ SERD_ERR_INTERNAL, /**< Unexpected internal error (should not happen) */
+ SERD_ERR_OVERFLOW /**< Stack overflow */
} SerdStatus;
/**
@@ -903,7 +904,10 @@ serd_env_foreach(const SerdEnv* env,
*/
SERD_API
SerdReader*
-serd_reader_new(SerdWorld* world, SerdSyntax syntax, const SerdSink* sink);
+serd_reader_new(SerdWorld* world,
+ SerdSyntax syntax,
+ const SerdSink* sink,
+ size_t stack_size);
/**
Enable or disable strict parsing.
diff --git a/src/n3.c b/src/n3.c
index 37976f58..190d7239 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -41,7 +41,7 @@ fancy_syntax(const SerdReader* reader)
}
static SerdStatus
-read_collection(SerdReader* reader, ReadContext ctx, Ref* dest);
+read_collection(SerdReader* reader, ReadContext ctx, SerdNode** dest);
static SerdStatus
read_predicateObjectList(SerdReader* reader, ReadContext ctx, bool* ate_dot);
@@ -60,7 +60,7 @@ read_HEX(SerdReader* reader)
// Read UCHAR escape, initial \ is already eaten by caller
static inline SerdStatus
-read_UCHAR(SerdReader* reader, Ref dest, uint32_t* char_code)
+read_UCHAR(SerdReader* reader, SerdNode* dest, uint32_t* char_code)
{
const int b = peek_byte(reader);
unsigned length = 0;
@@ -99,9 +99,9 @@ read_UCHAR(SerdReader* reader, Ref dest, uint32_t* char_code)
} else {
r_err(reader, SERD_ERR_BAD_SYNTAX,
"unicode character 0x%X out of range\n", code);
- push_bytes(reader, dest, replacement_char, 3);
*char_code = 0xFFFD;
- return SERD_SUCCESS;
+ const SerdStatus st = push_bytes(reader, dest, replacement_char, 3);
+ return st ? st : SERD_SUCCESS;
}
// Build output in buf
@@ -130,42 +130,35 @@ read_UCHAR(SerdReader* reader, Ref dest, uint32_t* char_code)
break;
}
- push_bytes(reader, dest, buf, size);
*char_code = code;
- return SERD_SUCCESS;
+ return push_bytes(reader, dest, buf, size);
}
// Read ECHAR escape, initial \ is already eaten by caller
static inline SerdStatus
-read_ECHAR(SerdReader* reader, Ref dest, SerdNodeFlags* flags)
+read_ECHAR(SerdReader* reader, SerdNode* dest, SerdNodeFlags* flags)
{
const int c = peek_byte(reader);
switch (c) {
case 't':
eat_byte_safe(reader, 't');
- push_byte(reader, dest, '\t');
- return SERD_SUCCESS;
+ return push_byte(reader, dest, '\t');
case 'b':
eat_byte_safe(reader, 'b');
- push_byte(reader, dest, '\b');
- return SERD_SUCCESS;
+ return push_byte(reader, dest, '\b');
case 'n':
*flags |= SERD_HAS_NEWLINE;
eat_byte_safe(reader, 'n');
- push_byte(reader, dest, '\n');
- return SERD_SUCCESS;
+ return push_byte(reader, dest, '\n');
case 'r':
*flags |= SERD_HAS_NEWLINE;
eat_byte_safe(reader, 'r');
- push_byte(reader, dest, '\r');
- return SERD_SUCCESS;
+ return push_byte(reader, dest, '\r');
case 'f':
eat_byte_safe(reader, 'f');
- push_byte(reader, dest, '\f');
- return SERD_SUCCESS;
+ return push_byte(reader, dest, '\f');
case '\\': case '"': case '\'':
- push_byte(reader, dest, eat_byte_safe(reader, c));
- return SERD_SUCCESS;
+ return push_byte(reader, dest, eat_byte_safe(reader, c));
default:
return SERD_ERR_BAD_SYNTAX;
}
@@ -208,21 +201,21 @@ read_utf8_bytes(SerdReader* reader, uint8_t bytes[4], uint32_t* size, uint8_t c)
}
static SerdStatus
-read_utf8_character(SerdReader* reader, Ref dest, uint8_t c)
+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);
- } else {
- push_bytes(reader, dest, bytes, size);
+ return st;
}
- return st;
+
+ return push_bytes(reader, dest, bytes, size);
}
static SerdStatus
-read_utf8_code(SerdReader* reader, Ref dest, uint32_t* code, uint8_t c)
+read_utf8_code(SerdReader* reader, SerdNode* dest, uint32_t* code, uint8_t c)
{
uint32_t size = 0;
uint8_t bytes[4] = {0, 0, 0, 0};
@@ -232,15 +225,17 @@ read_utf8_code(SerdReader* reader, Ref dest, uint32_t* code, uint8_t c)
return st;
}
- push_bytes(reader, dest, bytes, size);
- *code = parse_counted_utf8_char(bytes, size);
+ if (!(st = push_bytes(reader, dest, bytes, size))) {
+ *code = parse_counted_utf8_char(bytes, size);
+ }
+
return st;
}
// Read one character (possibly multi-byte)
// The first byte, c, has already been eaten by caller
static inline SerdStatus
-read_character(SerdReader* reader, Ref dest, SerdNodeFlags* flags, uint8_t c)
+read_character(SerdReader* reader, SerdNode* dest, SerdNodeFlags* flags, uint8_t c)
{
if (!(c & 0x80)) {
switch (c) {
@@ -253,6 +248,7 @@ read_character(SerdReader* reader, Ref dest, SerdNodeFlags* flags, uint8_t c)
default:
break;
}
+
return push_byte(reader, dest, c);
}
return read_utf8_character(reader, dest, c);
@@ -314,7 +310,7 @@ eat_delim(SerdReader* reader, const char delim)
// Initial triple quotes are already eaten by caller
static SerdStatus
read_STRING_LITERAL_LONG(SerdReader* reader,
- Ref ref,
+ SerdNode* ref,
SerdNodeFlags* flags,
uint8_t q)
{
@@ -357,7 +353,7 @@ read_STRING_LITERAL_LONG(SerdReader* reader,
// Initial quote is already eaten by caller
static SerdStatus
read_STRING_LITERAL(SerdReader* reader,
- Ref ref,
+ SerdNode* ref,
SerdNodeFlags* flags,
uint8_t q)
{
@@ -397,7 +393,7 @@ read_STRING_LITERAL(SerdReader* reader,
}
static SerdStatus
-read_String(SerdReader* reader, Ref node, SerdNodeFlags* flags)
+read_String(SerdReader* reader, SerdNode* node, SerdNodeFlags* flags)
{
const int q1 = peek_byte(reader);
eat_byte_safe(reader, q1);
@@ -438,7 +434,7 @@ is_PN_CHARS_BASE(const uint32_t c)
}
static SerdStatus
-read_PN_CHARS_BASE(SerdReader* reader, Ref dest)
+read_PN_CHARS_BASE(SerdReader* reader, SerdNode* dest)
{
uint32_t code = 0;
const int c = peek_byte(reader);
@@ -468,7 +464,7 @@ is_PN_CHARS(const uint32_t c)
}
static SerdStatus
-read_PN_CHARS(SerdReader* reader, Ref dest)
+read_PN_CHARS(SerdReader* reader, SerdNode* dest)
{
uint32_t code = 0;
const int c = peek_byte(reader);
@@ -488,7 +484,7 @@ read_PN_CHARS(SerdReader* reader, Ref dest)
}
static SerdStatus
-read_PERCENT(SerdReader* reader, Ref dest)
+read_PERCENT(SerdReader* reader, SerdNode* dest)
{
push_byte(reader, dest, eat_byte_safe(reader, '%'));
const uint8_t h1 = read_HEX(reader);
@@ -501,7 +497,7 @@ read_PERCENT(SerdReader* reader, Ref dest)
}
static SerdStatus
-read_PN_LOCAL_ESC(SerdReader* reader, Ref dest)
+read_PN_LOCAL_ESC(SerdReader* reader, SerdNode* dest)
{
eat_byte_safe(reader, '\\');
@@ -537,7 +533,7 @@ read_PN_LOCAL_ESC(SerdReader* reader, Ref dest)
}
static SerdStatus
-read_PLX(SerdReader* reader, Ref dest)
+read_PLX(SerdReader* reader, SerdNode* dest)
{
const int c = peek_byte(reader);
switch (c) {
@@ -551,7 +547,7 @@ read_PLX(SerdReader* reader, Ref dest)
}
static SerdStatus
-read_PN_LOCAL(SerdReader* reader, Ref dest, bool* ate_dot)
+read_PN_LOCAL(SerdReader* reader, SerdNode* dest, bool* ate_dot)
{
int c = peek_byte(reader);
SerdStatus st = SERD_SUCCESS;
@@ -580,10 +576,9 @@ read_PN_LOCAL(SerdReader* reader, Ref dest, bool* ate_dot)
trailing_unescaped_dot = (c == '.');
}
- SerdNode* const n = deref(reader, dest);
if (trailing_unescaped_dot) {
// Ate trailing dot, pop it from stack/node and inform caller
- --n->n_bytes;
+ --dest->n_bytes;
serd_stack_pop(&reader->stack, 1);
*ate_dot = true;
}
@@ -593,7 +588,7 @@ read_PN_LOCAL(SerdReader* reader, Ref dest, bool* ate_dot)
// Read the remainder of a PN_PREFIX after some initial characters
static SerdStatus
-read_PN_PREFIX_tail(SerdReader* reader, Ref dest)
+read_PN_PREFIX_tail(SerdReader* reader, SerdNode* dest)
{
int c = 0;
while ((c = peek_byte(reader))) { // Middle: (PN_CHARS | '.')*
@@ -604,8 +599,7 @@ read_PN_PREFIX_tail(SerdReader* reader, Ref dest)
}
}
- const SerdNode* const n = deref(reader, dest);
- if (serd_node_string(n)[n->n_bytes - 1] == '.' &&
+ if (serd_node_string(dest)[dest->n_bytes - 1] == '.' &&
read_PN_CHARS(reader, dest)) {
return r_err(reader, SERD_ERR_BAD_SYNTAX, "prefix ends with `.'\n");
}
@@ -614,7 +608,7 @@ read_PN_PREFIX_tail(SerdReader* reader, Ref dest)
}
static SerdStatus
-read_PN_PREFIX(SerdReader* reader, Ref dest)
+read_PN_PREFIX(SerdReader* reader, SerdNode* dest)
{
if (!read_PN_CHARS_BASE(reader, dest)) {
return read_PN_PREFIX_tail(reader, dest);
@@ -623,14 +617,16 @@ read_PN_PREFIX(SerdReader* reader, Ref dest)
}
static SerdStatus
-read_LANGTAG(SerdReader* reader, Ref* dest)
+read_LANGTAG(SerdReader* reader, SerdNode** dest)
{
int c = peek_byte(reader);
if (!is_alpha(c)) {
return r_err(reader, SERD_ERR_BAD_SYNTAX, "unexpected `%c'\n", c);
}
- *dest = push_node(reader, SERD_LITERAL, "", 0);
+ if (!(*dest = push_node(reader, SERD_LITERAL, "", 0))) {
+ return SERD_ERR_OVERFLOW;
+ }
SerdStatus st = SERD_SUCCESS;
TRY(st, push_byte(reader, *dest, eat_byte_safe(reader, c)));
@@ -647,7 +643,7 @@ read_LANGTAG(SerdReader* reader, Ref* dest)
}
static SerdStatus
-read_IRIREF_scheme(SerdReader* reader, Ref dest)
+read_IRIREF_scheme(SerdReader* reader, SerdNode* dest)
{
int c = peek_byte(reader);
if (!is_alpha(c)) {
@@ -675,7 +671,7 @@ read_IRIREF_scheme(SerdReader* reader, Ref dest)
}
static SerdStatus
-read_IRIREF(SerdReader* reader, Ref* dest)
+read_IRIREF(SerdReader* reader, SerdNode** dest)
{
if (!eat_byte_check(reader, '<')) {
return SERD_ERR_BAD_SYNTAX;
@@ -747,7 +743,7 @@ read_IRIREF(SerdReader* reader, Ref* dest)
}
static SerdStatus
-read_PrefixedName(SerdReader* reader, Ref dest, bool read_prefix, bool* ate_dot)
+read_PrefixedName(SerdReader* reader, SerdNode* dest, bool read_prefix, bool* ate_dot)
{
SerdStatus st = SERD_SUCCESS;
if (read_prefix && ((st = read_PN_PREFIX(reader, dest)) > SERD_FAILURE)) {
@@ -764,7 +760,7 @@ read_PrefixedName(SerdReader* reader, Ref dest, bool read_prefix, bool* ate_dot)
}
static SerdStatus
-read_0_9(SerdReader* reader, Ref str, bool at_least_one)
+read_0_9(SerdReader* reader, SerdNode* str, bool at_least_one)
{
unsigned count = 0;
SerdStatus st = SERD_SUCCESS;
@@ -779,8 +775,8 @@ read_0_9(SerdReader* reader, Ref str, bool at_least_one)
static SerdStatus
read_number(SerdReader* reader,
- Ref* dest,
- Ref* datatype,
+ SerdNode** dest,
+ SerdNode** datatype,
SerdNodeFlags* flags,
bool* ate_dot)
{
@@ -793,9 +789,12 @@ read_number(SerdReader* reader,
SerdStatus st = SERD_SUCCESS;
int c = peek_byte(reader);
bool has_decimal = false;
- if (c == '-' || c == '+') {
+ if (!*dest) {
+ return SERD_ERR_OVERFLOW;
+ } else if (c == '-' || c == '+') {
push_byte(reader, *dest, eat_byte_safe(reader, c));
}
+
if ((c = peek_byte(reader)) == '.') {
has_decimal = true;
// decimal case 2 (e.g. '.0' or `-.0' or `+.0')
@@ -847,20 +846,26 @@ read_number(SerdReader* reader,
}
static SerdStatus
-read_iri(SerdReader* reader, Ref* dest, bool* ate_dot)
+read_iri(SerdReader* reader, SerdNode** dest, bool* ate_dot)
{
switch (peek_byte(reader)) {
case '<':
return read_IRIREF(reader, dest);
default:
- *dest = push_node(reader, SERD_CURIE, "", 0);
+ if (!(*dest = push_node(reader, SERD_CURIE, "", 0))) {
+ return SERD_ERR_OVERFLOW;
+ }
return read_PrefixedName(reader, *dest, true, ate_dot);
}
}
static SerdStatus
-read_literal(SerdReader* reader, Ref* dest,
- Ref* datatype, Ref* lang, SerdNodeFlags* flags, bool* ate_dot)
+read_literal(SerdReader* reader,
+ SerdNode** dest,
+ SerdNode** datatype,
+ SerdNode** lang,
+ SerdNodeFlags* flags,
+ bool* ate_dot)
{
*dest = push_node(reader, SERD_LITERAL, "", 0);
@@ -897,7 +902,7 @@ read_literal(SerdReader* reader, Ref* dest,
}
static SerdStatus
-read_verb(SerdReader* reader, Ref* dest)
+read_verb(SerdReader* reader, SerdNode** dest)
{
if (peek_byte(reader) == '<') {
return read_IRIREF(reader, dest);
@@ -906,11 +911,13 @@ read_verb(SerdReader* reader, Ref* dest)
/* Either a qname, or "a". Read the prefix first, and if it is in fact
"a", produce that instead.
*/
- *dest = push_node(reader, SERD_CURIE, "", 0);
+ if (!(*dest = push_node(reader, SERD_CURIE, "", 0))) {
+ return SERD_ERR_OVERFLOW;
+ }
SerdStatus st = read_PN_PREFIX(reader, *dest);
bool ate_dot = false;
- SerdNode* node = deref(reader, *dest);
+ SerdNode* node = *dest;
const int next = peek_byte(reader);
if (!st && node->n_bytes == 1 &&
serd_node_string(node)[0] == 'a' &&
@@ -929,12 +936,12 @@ read_verb(SerdReader* reader, Ref* dest)
}
static SerdStatus
-read_BLANK_NODE_LABEL(SerdReader* reader, Ref* dest, bool* ate_dot)
+read_BLANK_NODE_LABEL(SerdReader* reader, SerdNode** dest, bool* ate_dot)
{
eat_byte_safe(reader, '_');
eat_byte_check(reader, ':');
- const Ref ref = *dest =
+ SerdNode* n = *dest =
push_node(reader,
SERD_BLANK,
reader->bprefix ? reader->bprefix : "",
@@ -942,23 +949,22 @@ read_BLANK_NODE_LABEL(SerdReader* reader, Ref* dest, bool* ate_dot)
int c = peek_byte(reader); // First: (PN_CHARS | '_' | [0-9])
if (is_digit(c) || c == '_') {
- push_byte(reader, ref, eat_byte_safe(reader, c));
- } else if (read_PN_CHARS(reader, ref)) {
- *dest = pop_node(reader, *dest);
+ push_byte(reader, n, eat_byte_safe(reader, c));
+ } else if (read_PN_CHARS(reader, n)) {
+ *dest = pop_node(reader, n);
return r_err(reader, SERD_ERR_BAD_SYNTAX, "invalid name start\n");
}
while ((c = peek_byte(reader))) { // Middle: (PN_CHARS | '.')*
if (c == '.') {
- push_byte(reader, ref, eat_byte_safe(reader, c));
- } else if (read_PN_CHARS(reader, ref)) {
+ push_byte(reader, n, eat_byte_safe(reader, c));
+ } else if (read_PN_CHARS(reader, n)) {
break;
}
}
- SerdNode* n = deref(reader, ref);
- char* buf = serd_node_buffer(n);
- if (buf[n->n_bytes - 1] == '.' && read_PN_CHARS(reader, ref)) {
+ char* buf = serd_node_buffer(n);
+ if (buf[n->n_bytes - 1] == '.' && read_PN_CHARS(reader, n)) {
// Ate trailing dot, pop it from stack/node and inform caller
--n->n_bytes;
serd_stack_pop(&reader->stack, 1);
@@ -972,7 +978,7 @@ read_BLANK_NODE_LABEL(SerdReader* reader, Ref* dest, bool* ate_dot)
reader->seen_genid = true;
} else if (reader->seen_genid &&
buf[reader->bprefix_len] == 'B') {
- *dest = pop_node(reader, *dest);
+ *dest = pop_node(reader, n);
return r_err(
reader, SERD_ERR_ID_CLASH,
"found both `b' and `B' blank IDs, prefix required\n");
@@ -982,24 +988,24 @@ read_BLANK_NODE_LABEL(SerdReader* reader, Ref* dest, bool* ate_dot)
return SERD_SUCCESS;
}
-static Ref
+static SerdNode*
read_blankName(SerdReader* reader)
{
eat_byte_safe(reader, '=');
if (eat_byte_check(reader, '=') != '=') {
r_err(reader, SERD_ERR_BAD_SYNTAX, "expected `='\n");
- return 0;
+ return NULL;
}
- Ref subject = 0;
- bool ate_dot = false;
+ SerdNode* subject = 0;
+ bool ate_dot = false;
read_ws_star(reader);
read_iri(reader, &subject, &ate_dot);
return subject;
}
static SerdStatus
-read_anon(SerdReader* reader, ReadContext ctx, bool subject, Ref* dest)
+read_anon(SerdReader* reader, ReadContext ctx, bool subject, SerdNode** dest)
{
const SerdStatementFlags old_flags = *ctx.flags;
bool empty = false;
@@ -1038,7 +1044,7 @@ read_anon(SerdReader* reader, ReadContext ctx, bool subject, Ref* dest)
}
read_ws_star(reader);
if (reader->sink->end) {
- reader->sink->end(reader->sink->handle, deref(reader, *dest));
+ reader->sink->end(reader->sink->handle, *dest);
}
*ctx.flags = old_flags;
}
@@ -1062,10 +1068,9 @@ read_object(SerdReader* reader, ReadContext* ctx, bool emit, bool* ate_dot)
SerdStatus ret = SERD_FAILURE;
bool simple = (ctx->subject != 0);
- SerdNode* node = NULL;
- Ref o = 0;
- Ref datatype = 0;
- Ref lang = 0;
+ SerdNode* o = 0;
+ SerdNode* datatype = 0;
+ SerdNode* lang = 0;
uint32_t flags = 0;
const int c = peek_byte(reader);
if (!fancy_syntax(reader)) {
@@ -1105,15 +1110,17 @@ read_object(SerdReader* reader, ReadContext* ctx, bool emit, bool* ate_dot)
/* Either a boolean literal, or a qname. Read the prefix first, and if
it is in fact a "true" or "false" literal, produce that instead.
*/
- o = push_node(reader, SERD_CURIE, "", 0);
+ if (!(o = push_node(reader, SERD_CURIE, "", 0))) {
+ return SERD_ERR_OVERFLOW;
+ }
+
while (!read_PN_CHARS_BASE(reader, o)) {}
- node = deref(reader, o);
- if ((node->n_bytes == 4 &&
- !memcmp(serd_node_string(node), "true", 4)) ||
- (node->n_bytes == 5 &&
- !memcmp(serd_node_string(node), "false", 5))) {
+ if ((o->n_bytes == 4 &&
+ !memcmp(serd_node_string(o), "true", 4)) ||
+ (o->n_bytes == 5 &&
+ !memcmp(serd_node_string(o), "false", 5))) {
flags = flags | SERD_HAS_DATATYPE;
- node->type = SERD_LITERAL;
+ o->type = SERD_LITERAL;
datatype = push_node(
reader, SERD_URI, XSD_BOOLEAN, XSD_BOOLEAN_LEN);
ret = SERD_SUCCESS;
@@ -1129,7 +1136,7 @@ read_object(SerdReader* reader, ReadContext* ctx, bool emit, bool* ate_dot)
}
if (!ret && simple && o) {
- deref(reader, o)->flags = flags;
+ o->flags = flags;
}
if (!ret && emit && simple) {
@@ -1204,7 +1211,11 @@ read_predicateObjectList(SerdReader* reader, ReadContext ctx, bool* ate_dot)
}
static SerdStatus
-end_collection(SerdReader* reader, ReadContext ctx, Ref n1, Ref n2, SerdStatus st)
+end_collection(SerdReader* reader,
+ ReadContext ctx,
+ SerdNode* n1,
+ SerdNode* n2,
+ SerdStatus st)
{
pop_node(reader, n2);
pop_node(reader, n1);
@@ -1217,7 +1228,7 @@ end_collection(SerdReader* reader, ReadContext ctx, Ref n1, Ref n2, SerdStatus s
}
static SerdStatus
-read_collection(SerdReader* reader, ReadContext ctx, Ref* dest)
+read_collection(SerdReader* reader, ReadContext ctx, SerdNode** dest)
{
SerdStatus st = SERD_SUCCESS;
eat_byte_safe(reader, '(');
@@ -1238,10 +1249,14 @@ read_collection(SerdReader* reader, ReadContext ctx, Ref* dest)
/* The order of node allocation here is necessarily not in stack order,
so we create two nodes and recycle them throughout. */
- Ref n1 = push_node_padded(reader, genid_size(reader), SERD_BLANK, "", 0);
- Ref n2 = 0;
- Ref node = n1;
- Ref rest = 0;
+ SerdNode* n1 = push_node_padded(reader, genid_size(reader), SERD_BLANK, "", 0);
+ SerdNode* n2 = 0;
+ SerdNode* node = n1;
+ SerdNode* rest = 0;
+
+ if (!n1) {
+ return SERD_ERR_OVERFLOW;
+ }
ctx.subject = *dest;
while (!peek_delim(reader, ')')) {
@@ -1276,7 +1291,7 @@ read_collection(SerdReader* reader, ReadContext ctx, Ref* dest)
}
static SerdStatus
-read_subject(SerdReader* reader, ReadContext ctx, Ref* dest, int* s_type)
+read_subject(SerdReader* reader, ReadContext ctx, SerdNode** dest, int* s_type)
{
SerdStatus st = SERD_SUCCESS;
bool ate_dot = false;
@@ -1303,7 +1318,7 @@ read_subject(SerdReader* reader, ReadContext ctx, Ref* dest, int* s_type)
}
static SerdStatus
-read_labelOrSubject(SerdReader* reader, Ref* dest)
+read_labelOrSubject(SerdReader* reader, SerdNode** dest)
{
bool ate_dot = false;
switch (peek_byte(reader)) {
@@ -1356,10 +1371,10 @@ read_base(SerdReader* reader, bool sparql, bool token)
read_ws_star(reader);
- Ref uri = 0;
+ SerdNode* uri = NULL;
TRY(st, read_IRIREF(reader, &uri));
if (reader->sink->base) {
- TRY(st, reader->sink->base(reader->sink->handle, deref(reader, uri)));
+ TRY(st, reader->sink->base(reader->sink->handle, uri));
}
pop_node(reader, uri);
@@ -1382,8 +1397,10 @@ read_prefixID(SerdReader* reader, bool sparql, bool token)
}
read_ws_star(reader);
- Ref name = push_node(reader, SERD_LITERAL, "", 0);
- if ((st = read_PN_PREFIX(reader, name)) > SERD_FAILURE) {
+ SerdNode* name = push_node(reader, SERD_LITERAL, "", 0);
+ if (!name) {
+ return SERD_ERR_OVERFLOW;
+ } else if ((st = read_PN_PREFIX(reader, name)) > SERD_FAILURE) {
return st;
}
@@ -1393,13 +1410,11 @@ read_prefixID(SerdReader* reader, bool sparql, bool token)
}
read_ws_star(reader);
- Ref uri = 0;
+ SerdNode* uri = NULL;
TRY(st, read_IRIREF(reader, &uri));
if (reader->sink->prefix) {
- st = reader->sink->prefix(reader->sink->handle,
- deref(reader, name),
- deref(reader, uri));
+ st = reader->sink->prefix(reader->sink->handle, name, uri);
}
pop_node(reader, uri);
pop_node(reader, name);
@@ -1469,13 +1484,11 @@ read_wrappedGraph(SerdReader* reader, ReadContext* ctx)
}
static int
-tokcmp(SerdReader* reader, Ref ref, const char* tok, size_t n)
+tokcmp(SerdNode* node, const char* tok, size_t n)
{
- SerdNode* node = deref(reader, ref);
- if (!node || node->n_bytes != n) {
- return -1;
- }
- return serd_strncasecmp(serd_node_string(node), tok, n);
+ return ((!node || node->n_bytes != n)
+ ? -1
+ : serd_strncasecmp(serd_node_string(node), tok, n));
}
SerdStatus
@@ -1516,11 +1529,11 @@ read_n3_statement(SerdReader* reader)
return st;
}
- if (!tokcmp(reader, ctx.subject, "base", 4)) {
+ if (!tokcmp(ctx.subject, "base", 4)) {
st = read_base(reader, true, false);
- } else if (!tokcmp(reader, ctx.subject, "prefix", 6)) {
+ } else if (!tokcmp(ctx.subject, "prefix", 6)) {
st = read_prefixID(reader, true, false);
- } else if (!tokcmp(reader, ctx.subject, "graph", 5)) {
+ } else if (!tokcmp(ctx.subject, "graph", 5)) {
read_ws_star(reader);
TRY(st, read_labelOrSubject(reader, &ctx.graph));
read_ws_star(reader);
diff --git a/src/reader.c b/src/reader.c
index 95031e0c..bb71be83 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -41,9 +41,8 @@ r_err(SerdReader* reader, SerdStatus st, const char* fmt, ...)
}
void
-set_blank_id(SerdReader* reader, Ref ref, size_t buf_size)
+set_blank_id(SerdReader* reader, SerdNode* node, size_t buf_size)
{
- SerdNode* node = deref(reader, ref);
char* buf = (char*)(node + 1);
const char* prefix = reader->bprefix ? (const char*)reader->bprefix : "";
@@ -57,21 +56,28 @@ genid_size(SerdReader* reader)
return reader->bprefix_len + 1 + 10 + 1; // + "b" + UINT32_MAX + \0
}
-Ref
+SerdNode*
blank_id(SerdReader* reader)
{
- Ref ref = push_node_padded(reader, genid_size(reader), SERD_BLANK, "", 0);
- set_blank_id(reader, ref, genid_size(reader));
+ SerdNode* ref = push_node_padded(
+ reader, genid_size(reader), SERD_BLANK, "", 0);
+ if (ref) {
+ set_blank_id(reader, ref, genid_size(reader));
+ }
return ref;
}
-Ref
+SerdNode*
push_node_padded(SerdReader* reader, size_t maxlen,
SerdType type, const char* str, size_t n_bytes)
{
void* mem = serd_stack_push_aligned(
&reader->stack, sizeof(SerdNode) + maxlen + 1, sizeof(SerdNode));
+ if (!mem) {
+ return NULL;
+ }
+
SerdNode* const node = (SerdNode*)mem;
node->n_bytes = n_bytes;
node->flags = 0;
@@ -85,41 +91,34 @@ push_node_padded(SerdReader* reader, size_t maxlen,
reader->allocs, sizeof(reader->allocs) * (++reader->n_allocs));
reader->allocs[reader->n_allocs - 1] = (mem - reader->stack.buf);
#endif
- return (Ref)((char*)node - reader->stack.buf);
+ return node;
}
-Ref
+SerdNode*
push_node(SerdReader* reader, SerdType type, const char* str, size_t n_bytes)
{
return push_node_padded(reader, n_bytes, type, str, n_bytes);
}
SerdNode*
-deref(SerdReader* reader, const Ref ref)
-{
- return ref ? (SerdNode*)(reader->stack.buf + ref) : NULL;
-}
-
-Ref
-pop_node(SerdReader* reader, Ref ref)
+pop_node(SerdReader* reader, const SerdNode* node)
{
- if (ref && ref != reader->rdf_first && ref != reader->rdf_rest
- && ref != reader->rdf_nil) {
+ if (node && node != reader->rdf_first && node != reader->rdf_rest
+ && node != reader->rdf_nil) {
#ifdef SERD_STACK_CHECK
- SERD_STACK_ASSERT_TOP(reader, ref);
+ SERD_STACK_ASSERT_TOP(reader, node);
--reader->n_allocs;
#endif
- SerdNode* const node = deref(reader, ref);
- char* const top = reader->stack.buf + reader->stack.size;
+ char* const top = reader->stack.buf + reader->stack.size;
serd_stack_pop_aligned(&reader->stack, (size_t)(top - (char*)node));
}
- return 0;
+ return NULL;
}
SerdStatus
-emit_statement(SerdReader* reader, ReadContext ctx, Ref o)
+emit_statement(SerdReader* reader, ReadContext ctx, SerdNode* o)
{
- SerdNode* graph = deref(reader, ctx.graph);
+ SerdNode* graph = ctx.graph;
if (!graph && reader->default_graph) {
graph = reader->default_graph;
}
@@ -130,9 +129,9 @@ emit_statement(SerdReader* reader, ReadContext ctx, Ref o)
: reader->sink->statement(reader->sink->handle,
*ctx.flags,
graph,
- deref(reader, ctx.subject),
- deref(reader, ctx.predicate),
- deref(reader, o));
+ ctx.subject,
+ ctx.predicate,
+ o);
*ctx.flags &= SERD_ANON_CONT|SERD_LIST_CONT; // Preserve only cont flags
return st;
@@ -159,14 +158,17 @@ serd_reader_read_document(SerdReader* reader)
}
SerdReader*
-serd_reader_new(SerdWorld* world, SerdSyntax syntax, const SerdSink* sink)
+serd_reader_new(SerdWorld* world,
+ SerdSyntax syntax,
+ const SerdSink* sink,
+ size_t stack_size)
{
SerdReader* me = (SerdReader*)calloc(1, sizeof(SerdReader));
me->world = world;
me->sink = sink;
me->default_graph = NULL;
- me->stack = serd_stack_new(SERD_PAGE_SIZE);
+ me->stack = serd_stack_new(stack_size);
me->syntax = syntax;
me->next_id = 1;
me->strict = true;
diff --git a/src/reader.h b/src/reader.h
index d3d27f39..4925dbda 100644
--- a/src/reader.h
+++ b/src/reader.h
@@ -41,18 +41,13 @@
# define SERD_STACK_ASSERT_TOP(reader, ref)
#endif
-/* Reference to a node in the stack (we can not use pointers since the
- stack may be reallocated, invalidating any pointers to elements).
-*/
-typedef size_t Ref;
-
typedef struct {
- Ref graph;
- Ref subject;
- Ref predicate;
- Ref object;
- Ref datatype;
- Ref lang;
+ SerdNode* graph;
+ SerdNode* subject;
+ SerdNode* predicate;
+ SerdNode* object;
+ SerdNode* datatype;
+ SerdNode* lang;
SerdStatementFlags* flags;
} ReadContext;
@@ -61,9 +56,9 @@ struct SerdReaderImpl {
const SerdSink* sink;
SerdErrorSink error_sink;
void* error_handle;
- Ref rdf_first;
- Ref rdf_rest;
- Ref rdf_nil;
+ SerdNode* rdf_first;
+ SerdNode* rdf_rest;
+ SerdNode* rdf_nil;
SerdNode* default_graph;
SerdByteSource source;
SerdStack stack;
@@ -75,8 +70,8 @@ struct SerdReaderImpl {
bool strict; ///< True iff strict parsing
bool seen_genid;
#ifdef SERD_STACK_CHECK
- Ref* allocs; ///< Stack of push offsets
- size_t n_allocs; ///< Number of stack pushes
+ SerdNode** allocs; ///< Stack of push offsets
+ size_t n_allocs; ///< Number of stack pushes
#endif
};
@@ -84,27 +79,25 @@ SERD_LOG_FUNC(3, 4)
SerdStatus
r_err(SerdReader* reader, SerdStatus st, const char* fmt, ...);
-Ref push_node_padded(SerdReader* reader,
+SerdNode* push_node_padded(SerdReader* reader,
size_t maxlen,
SerdType type,
const char* str,
size_t n_bytes);
-Ref push_node(SerdReader* reader,
- SerdType type,
- const char* str,
- size_t n_bytes);
-
-size_t genid_size(SerdReader* reader);
-Ref blank_id(SerdReader* reader);
-void set_blank_id(SerdReader* reader, Ref ref, size_t buf_size);
+SerdNode* push_node(SerdReader* reader,
+ SerdType type,
+ const char* str,
+ size_t n_bytes);
-SerdNode* deref(SerdReader* reader, Ref ref);
+size_t genid_size(SerdReader* reader);
+SerdNode* blank_id(SerdReader* reader);
+void set_blank_id(SerdReader* reader, SerdNode* node, size_t buf_size);
-Ref pop_node(SerdReader* reader, Ref ref);
+SerdNode* pop_node(SerdReader* reader, const SerdNode* node);
SerdStatus
-emit_statement(SerdReader* reader, ReadContext ctx, Ref o);
+emit_statement(SerdReader* reader, ReadContext ctx, SerdNode* o);
SerdStatus read_n3_statement(SerdReader* reader);
SerdStatus read_nquadsDoc(SerdReader* reader);
@@ -154,25 +147,33 @@ eat_string(SerdReader* reader, const char* str, unsigned n)
}
static inline SerdStatus
-push_byte(SerdReader* reader, Ref ref, const int c)
+push_byte(SerdReader* reader, SerdNode* node, const int c)
{
assert(c != EOF);
SERD_STACK_ASSERT_TOP(reader, ref);
- char* const s = (char*)serd_stack_push(&reader->stack, 1);
- SerdNode* const node = (SerdNode*)(reader->stack.buf + ref);
+ char* const s = (char*)serd_stack_push(&reader->stack, 1);
+ if (!s) {
+ return SERD_ERR_OVERFLOW;
+ }
+
++node->n_bytes;
*(s - 1) = (uint8_t)c;
*s = '\0';
return SERD_SUCCESS;
}
-static inline void
-push_bytes(SerdReader* reader, Ref ref, const uint8_t* bytes, unsigned len)
+static inline SerdStatus
+push_bytes(SerdReader* reader, SerdNode* ref, const uint8_t* bytes, unsigned len)
{
+ if (reader->stack.buf_size < reader->stack.size + len) {
+ return SERD_ERR_OVERFLOW;
+ }
+
for (unsigned i = 0; i < len; ++i) {
push_byte(reader, ref, bytes[i]);
}
+ return SERD_SUCCESS;
}
#endif // SERD_READER_H
diff --git a/src/serdi.c b/src/serdi.c
index ef620c3c..fcb4274e 100644
--- a/src/serdi.c
+++ b/src/serdi.c
@@ -25,8 +25,10 @@
#include <io.h>
#endif
+#include <limits.h>
#include <stdbool.h>
#include <stdio.h>
+#include <stdlib.h>
#include <string.h>
#define SERDI_ERROR(msg) fprintf(stderr, "serdi: " msg)
@@ -98,6 +100,7 @@ print_usage(const char* name, bool error)
fprintf(os, " -f Keep full URIs in input (don't qualify).\n");
fprintf(os, " -h Display this help and exit.\n");
fprintf(os, " -i SYNTAX Input syntax: turtle/ntriples/trig/nquads.\n");
+ fprintf(os, " -k BYTES Parser stack size.\n");
fprintf(os, " -l Lax (non-strict) parsing.\n");
fprintf(os, " -o SYNTAX Output syntax: turtle/ntriples/nquads.\n");
fprintf(os, " -p PREFIX Add PREFIX to blank node IDs.\n");
@@ -140,6 +143,7 @@ main(int argc, char** argv)
bool full_uris = false;
bool lax = false;
bool quiet = false;
+ size_t stack_size = 4194304;
const char* add_prefix = NULL;
const char* chop_prefix = NULL;
const char* root_uri = NULL;
@@ -174,6 +178,17 @@ main(int argc, char** argv)
} else if (!(input_syntax = get_syntax(argv[a]))) {
return print_usage(argv[0], true);
}
+ } else if (argv[a][1] == 'k') {
+ if (++a == argc) {
+ return missing_arg(argv[0], 'k');
+ }
+ char* endptr = NULL;
+ const long size = strtol(argv[a], &endptr, 10);
+ if (size <= 0 || size == LONG_MAX || *endptr != '\0') {
+ SERDI_ERRORF("invalid stack size `%s'\n", argv[a]);
+ return 1;
+ }
+ stack_size = (size_t)size;
} else if (argv[a][1] == 'o') {
if (++a == argc) {
return missing_arg(argv[0], 'o');
@@ -262,8 +277,8 @@ main(int argc, char** argv)
(SerdWriteFunc)fwrite,
out_fd);
- SerdReader* reader =
- serd_reader_new(world, input_syntax, serd_writer_get_sink(writer));
+ SerdReader* reader = serd_reader_new(
+ world, input_syntax, serd_writer_get_sink(writer), stack_size);
serd_reader_set_strict(reader, !lax);
if (quiet) {
diff --git a/src/stack.h b/src/stack.h
index 8943370c..174c6378 100644
--- a/src/stack.h
+++ b/src/stack.h
@@ -66,8 +66,7 @@ serd_stack_push(SerdStack* stack, size_t n_bytes)
{
const size_t new_size = stack->size + n_bytes;
if (stack->buf_size < new_size) {
- stack->buf_size += (stack->buf_size >> 1); // *= 1.5
- stack->buf = (char*)realloc(stack->buf, stack->buf_size);
+ return NULL;
}
char* const ret = (stack->buf + stack->size);
stack->size = new_size;
@@ -85,12 +84,16 @@ static inline void*
serd_stack_push_aligned(SerdStack* stack, size_t n_bytes, size_t align)
{
// Push one byte to ensure space for a pad count
- serd_stack_push(stack, 1);
+ if (!serd_stack_push(stack, 1)) {
+ return NULL;
+ }
// Push padding if necessary
const size_t pad = align - stack->size % align;
if (pad > 0) {
- serd_stack_push(stack, pad);
+ if (!serd_stack_push(stack, pad)) {
+ return NULL;
+ }
}
// Set top of stack to pad count so we can properly pop later
diff --git a/src/string.c b/src/string.c
index 9a6b4ff6..c755ff97 100644
--- a/src/string.c
+++ b/src/string.c
@@ -42,6 +42,7 @@ serd_strerror(SerdStatus status)
case SERD_ERR_ID_CLASH: return "Blank node ID clash";
case SERD_ERR_BAD_CURIE: return "Invalid CURIE";
case SERD_ERR_INTERNAL: return "Internal error";
+ case SERD_ERR_OVERFLOW: return "Stack overflow";
default: break;
}
return "Unknown error"; // never reached
diff --git a/src/writer.c b/src/writer.c
index 9d7a6790..827f7264 100644
--- a/src/writer.c
+++ b/src/writer.c
@@ -804,6 +804,10 @@ serd_writer_write_statement(SerdWriter* writer,
if (flags & (SERD_ANON_S_BEGIN|SERD_ANON_O_BEGIN)) {
WriteContext* ctx = (WriteContext*)serd_stack_push(
&writer->anon_stack, sizeof(WriteContext));
+ if (!ctx) {
+ return SERD_ERR_OVERFLOW;
+ }
+
*ctx = writer->context;
WriteContext new_context = {
serd_node_copy(graph), serd_node_copy(subject), NULL };
@@ -876,7 +880,7 @@ serd_writer_new(SerdWorld* world,
writer->env = env;
writer->root_node = NULL;
writer->root_uri = SERD_URI_NULL;
- writer->anon_stack = serd_stack_new(4 * sizeof(WriteContext));
+ writer->anon_stack = serd_stack_new(SERD_PAGE_SIZE);
writer->context = context;
writer->list_subj = NULL;
writer->empty = true;
diff --git a/tests/read_chunk_test.c b/tests/read_chunk_test.c
index decfe829..9fa40bab 100644
--- a/tests/read_chunk_test.c
+++ b/tests/read_chunk_test.c
@@ -82,7 +82,7 @@ main(void)
SerdWorld* world = serd_world_new();
const SerdSink sink = {NULL, on_base, on_prefix, on_statement, on_end};
- SerdReader* reader = serd_reader_new(world, SERD_TURTLE, &sink);
+ SerdReader* reader = serd_reader_new(world, SERD_TURTLE, &sink, 4096);
assert(reader);
assert(!serd_reader_start_string(reader,
diff --git a/tests/serd_test.c b/tests/serd_test.c
index 72f5db84..859ba088 100644
--- a/tests/serd_test.c
+++ b/tests/serd_test.c
@@ -133,7 +133,7 @@ test_read_chunks(void)
FILE* const f = tmpfile();
static const char null = 0;
SerdSink sink = {rt, NULL, NULL, test_sink, NULL};
- SerdReader* reader = serd_reader_new(world, SERD_TURTLE, &sink);
+ SerdReader* reader = serd_reader_new(world, SERD_TURTLE, &sink, 4096);
assert(reader);
assert(f);
@@ -213,7 +213,7 @@ test_read_string(void)
SerdWorld* world = serd_world_new();
ReaderTest* rt = (ReaderTest*)calloc(1, sizeof(ReaderTest));
SerdSink sink = {rt, NULL, NULL, test_sink, NULL};
- SerdReader* reader = serd_reader_new(world, SERD_TURTLE, &sink);
+ SerdReader* reader = serd_reader_new(world, SERD_TURTLE, &sink, 4096);
assert(reader);
// Test reading a string that ends exactly at the end of input (no newline)
@@ -360,7 +360,7 @@ test_strerror(void)
{
const char* msg = serd_strerror(SERD_SUCCESS);
assert(!strcmp(msg, "Success"));
- for (int i = SERD_FAILURE; i <= SERD_ERR_INTERNAL; ++i) {
+ for (int i = SERD_FAILURE; i <= SERD_ERR_OVERFLOW; ++i) {
msg = serd_strerror((SerdStatus)i);
assert(strcmp(msg, "Success"));
}
@@ -687,7 +687,7 @@ test_reader(const char* path)
SerdWorld* world = serd_world_new();
ReaderTest rt = { 0, NULL };
SerdSink sink = { &rt, NULL, NULL, test_sink, NULL };
- SerdReader* reader = serd_reader_new(world, SERD_TURTLE, &sink);
+ SerdReader* reader = serd_reader_new(world, SERD_TURTLE, &sink, 4096);
assert(reader);
SerdNode* g = serd_node_new_uri("http://example.org/");
diff --git a/wscript b/wscript
index 09c45d33..f344f1c2 100644
--- a/wscript
+++ b/wscript
@@ -560,7 +560,7 @@ def test(tst):
check([serdi, '%s/tests/good/manifest.ttl' % srcdir])
check([serdi, '-v'])
check([serdi, '-h'])
- check([serdi, '-s', '<foo> a <#Thingie> .'])
+ check([serdi, '-k', '512', '-s', '<foo> a <#Thingie> .'])
check([serdi, os.devnull])
with tempfile.TemporaryFile(mode='r') as stdin:
check([serdi, '-'], stdin=stdin)
@@ -575,6 +575,10 @@ def test(tst):
check([serdi, '-i', 'illegal'])
check([serdi, '-i', 'turtle'])
check([serdi, '-i'])
+ check([serdi, '-k'])
+ check([serdi, '-k', '-1'])
+ check([serdi, '-k', str(2**63 - 1)])
+ check([serdi, '-k', '1024junk'])
check([serdi, '-o', 'illegal'])
check([serdi, '-o'])
check([serdi, '-p'])