From 0b04f70d827a850ca5e779819095766194cd0e94 Mon Sep 17 00:00:00 2001 From: David Robillard Date: Sun, 24 Apr 2011 00:49:35 +0000 Subject: Fix reader memory stack use and verify read_object fully clears its stack. Make abbreviation in writer not assume equivalent nodes have equivalent addresses (abbreviate more than on input if triples are sorted correctly, e.g. abbrevate ntriples) git-svn-id: http://svn.drobilla.net/serd/trunk@149 490d8e77-9747-427b-9fa3-0b8f29cee8a0 --- src/node.c | 20 +++++++++++++++++--- src/reader.c | 22 ++++++++++------------ src/writer.c | 57 +++++++++++++++++++++++++++++++++++++++------------------ 3 files changed, 66 insertions(+), 33 deletions(-) (limited to 'src') diff --git a/src/node.c b/src/node.c index 5901f159..173c5358 100644 --- a/src/node.c +++ b/src/node.c @@ -34,12 +34,26 @@ SerdNode serd_node_copy(const SerdNode* node) { SerdNode copy = *node; - uint8_t* buf = malloc(copy.n_bytes); - memcpy(buf, node->buf, copy.n_bytes); - copy.buf = buf; + if (node->buf) { + uint8_t* buf = malloc(copy.n_bytes); + memcpy(buf, node->buf, copy.n_bytes); + copy.buf = buf; + } return copy; } +SERD_API +bool +serd_node_equals(const SerdNode* a, const SerdNode* b) +{ + return (a == b) + || (a->type == b->type + && a->n_bytes == b->n_bytes + && a->n_chars == b->n_chars + && ((a->buf == b->buf) || !strcmp((const char*)a->buf, + (const char*)b->buf))); +} + static size_t serd_uri_string_length(const SerdURI* uri) { diff --git a/src/reader.c b/src/reader.c index a237bbf2..0f250a0c 100644 --- a/src/reader.c +++ b/src/reader.c @@ -204,21 +204,12 @@ stack_is_top_string(SerdReader reader, Ref ref) } #endif -static inline intptr_t -pad_size(intptr_t size) -{ - return (size + 7) & (~7); -} - // Make a new string from a non-UTF-8 C string (internal use only) static Ref push_string(SerdReader reader, const char* c_str, size_t n_bytes) { - // Align strings to 64-bits (assuming malloc/realloc are aligned to 64-bits) - const size_t stack_size = pad_size((intptr_t)reader->stack.size); - const size_t pad = stack_size - reader->stack.size; - uint8_t* mem = serd_stack_push( - &reader->stack, pad + sizeof(SerdString) + n_bytes) + pad; + uint8_t* mem = serd_stack_push(&reader->stack, + sizeof(SerdString) + n_bytes); SerdString* const str = (SerdString*)mem; str->n_bytes = n_bytes; str->n_chars = n_bytes - 1; @@ -278,7 +269,7 @@ pop_string(SerdReader reader, Ref ref) --reader->n_allocs; #endif SerdString* str = deref(reader, ref); - serd_stack_pop(&reader->stack, pad_size(sizeof(SerdString) + str->n_bytes)); + serd_stack_pop(&reader->stack, sizeof(SerdString) + str->n_bytes); } } @@ -1061,6 +1052,10 @@ read_object(SerdReader reader, ReadContext ctx) static const char* const XSD_BOOLEAN = NS_XSD "boolean"; static const size_t XSD_BOOLEAN_LEN = 40; +#ifndef NDEBUG + const size_t orig_stack_size = reader->stack.size; +#endif + uint8_t pre[6]; bool ret = false; bool emit = (ctx.subject != 0); @@ -1118,6 +1113,9 @@ except: pop_string(reader, o.lang); pop_string(reader, o.datatype); pop_string(reader, o.value); +#ifndef NDEBUG + assert(reader->stack.size == orig_stack_size); +#endif return ret; } diff --git a/src/writer.c b/src/writer.c index dd00d1d8..b87ad1b0 100644 --- a/src/writer.c +++ b/src/writer.c @@ -158,6 +158,18 @@ serd_writer_write_delim(SerdWriter writer, const uint8_t delim) } } +static void +reset_context(SerdWriter writer) +{ + if (writer->context.graph.buf) + serd_node_free(&writer->context.graph); + if (writer->context.subject.buf) + serd_node_free(&writer->context.subject); + if (writer->context.predicate.buf) + serd_node_free(&writer->context.predicate); + writer->context = WRITE_CONTEXT_NULL; +} + static bool write_node(SerdWriter writer, const SerdNode* node, @@ -175,9 +187,11 @@ write_node(SerdWriter writer, serd_writer_write_delim(writer, '['); WriteContext* ctx = (WriteContext*)serd_stack_push( &writer->anon_stack, sizeof(WriteContext)); - *ctx = writer->context; - writer->context.subject = *node; - writer->context.predicate = SERD_NODE_NULL; + ctx->graph = serd_node_copy(&writer->context.graph); + ctx->subject = serd_node_copy(&writer->context.subject); + ctx->predicate = serd_node_copy(&writer->context.predicate); + reset_context(writer); + writer->context.subject = serd_node_copy(node); break; } case SERD_ANON: @@ -285,13 +299,15 @@ serd_writer_write_statement(SerdWriter writer, case SERD_TURTLE: break; } - if (subject->buf == writer->context.subject.buf) { - if (predicate->buf == writer->context.predicate.buf) { // Abbreviate S P + if (serd_node_equals(subject, &writer->context.subject)) { + if (serd_node_equals(predicate, &writer->context.predicate)) { + // Abbreviate S P ++writer->indent; serd_writer_write_delim(writer, ','); write_node(writer, object, object_datatype, object_lang); --writer->indent; - } else { // Abbreviate S + } else { + // Abbreviate S if (writer->context.predicate.buf) { serd_writer_write_delim(writer, ';'); } else { @@ -299,7 +315,9 @@ serd_writer_write_statement(SerdWriter writer, serd_writer_write_delim(writer, '\n'); } write_node(writer, predicate, NULL, NULL); - writer->context.predicate = *predicate; + if (writer->context.predicate.buf) + serd_node_free(&writer->context.predicate); + writer->context.predicate = serd_node_copy(predicate); writer->sink(" ", 1, writer->stream); write_node(writer, object, object_datatype, object_lang); } @@ -328,19 +346,21 @@ serd_writer_write_statement(SerdWriter writer, } } - writer->context.subject = *subject; + reset_context(writer); + writer->context.subject = serd_node_copy(subject); writer->context.predicate = SERD_NODE_NULL; write_node(writer, predicate, NULL, NULL); - writer->context.predicate = *predicate; + writer->context.predicate = serd_node_copy(predicate); writer->sink(" ", 1, writer->stream); write_node(writer, object, object_datatype, object_lang); } - const WriteContext new_context = { graph ? *graph : SERD_NODE_NULL, - *subject, - *predicate }; + const WriteContext new_context = { graph ? serd_node_copy(graph) : SERD_NODE_NULL, + serd_node_copy(subject), + serd_node_copy(predicate) }; + reset_context(writer); writer->context = new_context; return true; } @@ -361,10 +381,11 @@ serd_writer_end_anon(SerdWriter writer, --writer->indent; serd_writer_write_delim(writer, '\n'); writer->sink("]", 1, writer->stream); + reset_context(writer); writer->context = *anon_stack_top(writer); serd_stack_pop(&writer->anon_stack, sizeof(WriteContext)); if (!writer->context.subject.buf) { // End of anonymous subject - writer->context.subject = *node; + writer->context.subject = serd_node_copy(node); } return true; } @@ -375,8 +396,8 @@ serd_writer_finish(SerdWriter writer) { if (writer->context.subject.buf) { writer->sink(" .\n", 3, writer->stream); - writer->context.subject.buf = NULL; } + reset_context(writer); } SERD_API @@ -411,13 +432,13 @@ serd_writer_set_base_uri(SerdWriter writer, if (writer->syntax != SERD_NTRIPLES) { if (writer->context.graph.buf || writer->context.subject.buf) { writer->sink(" .\n\n", 4, writer->stream); - writer->context = WRITE_CONTEXT_NULL; + reset_context(writer); } writer->sink("@base <", 7, writer->stream); serd_uri_serialise(uri, writer->sink, writer->stream); writer->sink("> .\n", 4, writer->stream); } - writer->context = WRITE_CONTEXT_NULL; + reset_context(writer); } SERD_API @@ -429,7 +450,7 @@ serd_writer_set_prefix(SerdWriter writer, if (writer->syntax != SERD_NTRIPLES) { if (writer->context.graph.buf || writer->context.subject.buf) { writer->sink(" .\n\n", 4, writer->stream); - writer->context = WRITE_CONTEXT_NULL; + reset_context(writer); } writer->sink("@prefix ", 8, writer->stream); writer->sink(name->buf, name->n_bytes - 1, writer->stream); @@ -437,7 +458,7 @@ serd_writer_set_prefix(SerdWriter writer, write_text(writer, WRITE_URI, uri->buf, uri->n_bytes - 1, '>'); writer->sink("> .\n", 4, writer->stream); } - writer->context = WRITE_CONTEXT_NULL; + reset_context(writer); return true; } -- cgit v1.2.1