Simplify stack management by popping in bulk at higher levels

Since all memory used by the reader is POD in the stack, there is no benefit to
forcing code to explicitly pop everything pushed to the stack, since any
function can record an offset and pop back down to it regardless of what its
callers pushed if it knows that it does not need those items.

4 files changed, 69 insertions, 102 deletions

diff --git a/src/n3.c b/src/n3.c
index 84816f49..e358ac08 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -319,7 +319,7 @@ read_STRING_LITERAL_LONG(SerdReader* reader, SerdNodeFlags* flags, uint8_t q)
 			    read_UCHAR(reader, ref, &code)) {
 				r_err(reader, SERD_ERR_BAD_SYNTAX,
 				      "invalid escape `\\%c'\n", peek_byte(reader));
-				return pop_node(reader, ref);
+				return NULL;
 			}
 		} else if (c == q) {
 			eat_byte_safe(reader, q);
@@ -334,7 +334,7 @@ read_STRING_LITERAL_LONG(SerdReader* reader, SerdNodeFlags* flags, uint8_t q)
 			st = read_character(reader, ref, flags, (uint8_t)q2);
 		} else if (c == EOF) {
 			r_err(reader, SERD_ERR_BAD_SYNTAX, "end of file in long string\n");
-			return pop_node(reader, ref);
+			return NULL;
 		} else {
 			st = read_character(
 				reader, ref, flags, (uint8_t)eat_byte_safe(reader, c));
@@ -357,17 +357,17 @@ read_STRING_LITERAL(SerdReader* reader, SerdNodeFlags* flags, uint8_t q)
 		switch (c) {
 		case EOF:
 			r_err(reader, SERD_ERR_BAD_SYNTAX, "end of file in short string\n");
-			return pop_node(reader, ref);
+			return NULL;
 		case '\n': case '\r':
 			r_err(reader, SERD_ERR_BAD_SYNTAX, "line end in short string\n");
-			return pop_node(reader, ref);
+			return NULL;
 		case '\\':
 			eat_byte_safe(reader, c);
 			if (read_ECHAR(reader, ref, flags) &&
 			    read_UCHAR(reader, ref, &code)) {
 				r_err(reader, SERD_ERR_BAD_SYNTAX,
 				      "invalid escape `\\%c'\n", peek_byte(reader));
-				return pop_node(reader, ref);
+				return NULL;
 			}
 			break;
 		default:
@@ -383,14 +383,10 @@ read_STRING_LITERAL(SerdReader* reader, SerdNodeFlags* flags, uint8_t q)
 
 	if (st) {
 		reader->status = st;
-		return pop_node(reader, ref);
-	}
-
-	if (!eat_byte_check(reader, q)) {
-		return pop_node(reader, ref);
+		return NULL;
 	}
 
-	return ref;
+	return eat_byte_check(reader, q) ? ref : NULL;
 }
 
 static SerdNode*
@@ -655,7 +651,7 @@ read_IRIREF(SerdReader* reader)
 	TRY_RET(eat_byte_check(reader, '<'));
 	SerdNode* ref = push_node(reader, SERD_URI, "", 0);
 	if (!ref || (!fancy_syntax(reader) && !read_IRIREF_scheme(reader, ref))) {
-		return pop_node(reader, ref);
+		return NULL;
 	}
 
 	SerdStatus st   = SERD_SUCCESS;
@@ -666,19 +662,19 @@ read_IRIREF(SerdReader* reader)
 		case '"': case '<': case '^': case '`': case '{': case '|': case '}':
 			r_err(reader, SERD_ERR_BAD_SYNTAX,
 			      "invalid IRI character `%c'\n", c);
-			return pop_node(reader, ref);
+			return NULL;
 		case '>':
 			return ref;
 		case '\\':
 			if (read_UCHAR(reader, ref, &code)) {
 				r_err(reader, SERD_ERR_BAD_SYNTAX, "invalid IRI escape\n");
-				return pop_node(reader, ref);
+				return NULL;
 			}
 			switch (code) {
 			case 0: case ' ': case '<': case '>':
 				r_err(reader, SERD_ERR_BAD_SYNTAX,
 				      "invalid escaped IRI character %X %c\n", code, code);
-				return pop_node(reader, ref);
+				return NULL;
 			}
 			break;
 		default:
@@ -693,7 +689,7 @@ read_IRIREF(SerdReader* reader)
 					      (unsigned)c);
 				}
 				if (reader->strict) {
-					return pop_node(reader, ref);
+					return NULL;
 				}
 				reader->status = SERD_FAILURE;
 				push_byte(reader, ref, c);
@@ -701,13 +697,12 @@ read_IRIREF(SerdReader* reader)
 				push_byte(reader, ref, c);
 			} else if ((st = read_utf8_character(reader, ref, (uint8_t)c))) {
 				if (reader->strict) {
-					reader->status = SERD_FAILURE;
-					return pop_node(reader, ref);
+					return NULL;
 				}
 			}
 		}
 	}
-	return pop_node(reader, ref);
+	return NULL;
 }
 
 static bool
@@ -804,8 +799,6 @@ read_number(SerdReader*    reader,
 	*dest = ref;
 	return true;
 except:
-	pop_node(reader, *datatype);
-	pop_node(reader, ref);
 	return r_err(reader, SERD_ERR_BAD_SYNTAX, "bad number syntax\n");
 }
 
@@ -853,15 +846,15 @@ read_literal(SerdReader*    reader,
 	*dest = str;
 	return true;
 except:
-	*datatype = pop_node(reader, *datatype);
-	*lang     = pop_node(reader, *lang);
-	pop_node(reader, str);
+	*datatype = NULL;
+	*lang     = NULL;
 	return r_err(reader, SERD_ERR_BAD_SYNTAX, "bad literal syntax\n");
 }
 
 static bool
 read_verb(SerdReader* reader, SerdNode** dest)
 {
+	const size_t orig_stack_size = reader->stack.size;
 	if (peek_byte(reader) == '<') {
 		return (*dest = read_IRIREF(reader));
 	}
@@ -880,12 +873,12 @@ read_verb(SerdReader* reader, SerdNode** dest)
 	if (!st && node->n_bytes == 1 &&
 	    serd_node_get_string(node)[0] == 'a' &&
 	    next != ':' && !is_PN_CHARS_BASE((uint32_t)next)) {
-		pop_node(reader, *dest);
+		serd_stack_pop_to(&reader->stack, orig_stack_size);
 		return (*dest = push_node(reader, SERD_URI, NS_RDF "type", 47));
 	} else if (st > SERD_FAILURE ||
 	           !read_PrefixedName(reader, *dest, false, &ate_dot) ||
 	           ate_dot) {
-		*dest = pop_node(reader, *dest);
+		*dest = NULL;
 		return r_err(reader, SERD_ERR_BAD_SYNTAX, "bad verb\n");
 	}
 
@@ -909,7 +902,7 @@ read_BLANK_NODE_LABEL(SerdReader* reader, bool* ate_dot)
 		push_byte(reader, n, eat_byte_safe(reader, c));
 	} else if (read_PN_CHARS(reader, n)) {
 		r_err(reader, SERD_ERR_BAD_SYNTAX, "invalid name start character\n");
-		return pop_node(reader, n);
+		return NULL;
 	}
 
 	while ((c = peek_byte(reader))) {  // Middle: (PN_CHARS | '.')*
@@ -936,7 +929,7 @@ read_BLANK_NODE_LABEL(SerdReader* reader, bool* ate_dot)
 			} else if (reader->seen_genid && buf[reader->bprefix_len] == 'B') {
 				r_err(reader, SERD_ERR_ID_CLASH,
 				      "found both `b' and `B' blank IDs, prefix required\n");
-				return pop_node(reader, n);
+				return NULL;
 			}
 		}
 	}
@@ -1013,9 +1006,7 @@ read_object(SerdReader* reader, ReadContext* ctx, bool emit, bool* ate_dot)
 	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
 
 	bool      ret      = false;
 	bool      simple   = (ctx->subject != 0);
@@ -1094,9 +1085,7 @@ read_object(SerdReader* reader, ReadContext* ctx, bool emit, bool* ate_dot)
 	}
 
 except:
-	pop_node(reader, lang);
-	pop_node(reader, datatype);
-	pop_node(reader, o);
+	serd_stack_pop_to(&reader->stack, orig_stack_size);
 #ifndef NDEBUG
 	assert(reader->stack.size == orig_stack_size);
 #endif
@@ -1121,11 +1110,13 @@ read_objectList(SerdReader* reader, ReadContext ctx, bool* ate_dot)
 static bool
 read_predicateObjectList(SerdReader* reader, ReadContext ctx, bool* ate_dot)
 {
+	const size_t orig_stack_size = reader->stack.size;
+
 	while (read_verb(reader, &ctx.predicate) &&
 	       read_ws_star(reader) &&
 	       read_objectList(reader, ctx, ate_dot)) {
-		ctx.predicate = pop_node(reader, ctx.predicate);
 		if (*ate_dot) {
+			serd_stack_pop_to(&reader->stack, orig_stack_size);
 			return true;
 		}
 
@@ -1135,9 +1126,11 @@ read_predicateObjectList(SerdReader* reader, ReadContext ctx, bool* ate_dot)
 			read_ws_star(reader);
 			switch (c = peek_byte(reader)) {
 			case EOF: case '\0':
+				serd_stack_pop_to(&reader->stack, orig_stack_size);
 				return r_err(reader, SERD_ERR_BAD_SYNTAX,
 				             "unexpected end of file\n");
 			case '.': case ']': case '}':
+				serd_stack_pop_to(&reader->stack, orig_stack_size);
 				return true;
 			case ';':
 				eat_byte_safe(reader, c);
@@ -1146,18 +1139,19 @@ read_predicateObjectList(SerdReader* reader, ReadContext ctx, bool* ate_dot)
 		} while (c == ';');
 
 		if (!ate_semi) {
+			serd_stack_pop_to(&reader->stack, orig_stack_size);
 			return r_err(reader, SERD_ERR_BAD_SYNTAX, "missing ';' or '.'\n");
 		}
 	}
 
-	return pop_node(reader, ctx.predicate);
+	serd_stack_pop_to(&reader->stack, orig_stack_size);
+	ctx.predicate = 0;
+	return false;
 }
 
 static bool
-end_collection(SerdReader* reader, ReadContext ctx, SerdNode* n1, SerdNode* n2, bool ret)
+end_collection(SerdReader* reader, ReadContext ctx, bool ret)
 {
-	pop_node(reader, n2);
-	pop_node(reader, n1);
 	*ctx.flags &= ~(unsigned)SERD_LIST_CONT;
 	return ret && (eat_byte_safe(reader, ')') == ')');
 }
@@ -1178,7 +1172,7 @@ read_collection(SerdReader* reader, ReadContext ctx, SerdNode** dest)
 	}
 
 	if (end) {
-		return end_collection(reader, ctx, 0, 0, true);
+		return end_collection(reader, ctx, true);
 	}
 
 	/* The order of node allocation here is necessarily not in stack order,
@@ -1198,7 +1192,7 @@ read_collection(SerdReader* reader, ReadContext ctx, SerdNode** dest)
 		ctx.predicate = reader->rdf_first;
 		bool ate_dot = false;
 		if (!read_object(reader, &ctx, true, &ate_dot) || ate_dot) {
-			return end_collection(reader, ctx, n1, n2, false);
+			return end_collection(reader, ctx, false);
 		}
 
 		if (!(end = peek_delim(reader, ')'))) {
@@ -1221,7 +1215,7 @@ read_collection(SerdReader* reader, ReadContext ctx, SerdNode** dest)
 		node        = ctx.subject;  // invariant
 	}
 
-	return end_collection(reader, ctx, n1, n2, true);
+	return end_collection(reader, ctx, true);
 }
 
 static SerdNode*
@@ -1241,7 +1235,7 @@ read_subject(SerdReader* reader, ReadContext ctx, SerdNode** dest, int* s_type)
 	default:
 		TRY_RET(read_iri(reader, dest, &ate_dot));
 	}
-	return ate_dot ? pop_node(reader, *dest) : *dest;
+	return ate_dot ? NULL : *dest;
 }
 
 static SerdNode*
@@ -1296,7 +1290,6 @@ read_base(SerdReader* reader, bool sparql, bool token)
 	} else if (reader->sink->base) {
 		reader->sink->base(reader->sink->handle, uri);
 	}
-	pop_node(reader, uri);
 
 	read_ws_star(reader);
 	if (!sparql) {
@@ -1321,25 +1314,22 @@ read_prefixID(SerdReader* reader, bool sparql, bool token)
 	if (!name) {
 		return false;
 	} else if (read_PN_PREFIX(reader, name) > SERD_FAILURE) {
-		return pop_node(reader, name);
+		return false;
 	}
 
 	if (eat_byte_check(reader, ':') != ':') {
-		return pop_node(reader, name);
+		return false;
 	}
 
 	read_ws_star(reader);
 	const SerdNode* uri = read_IRIREF(reader);
 	if (!uri) {
-		pop_node(reader, name);
 		return false;
 	}
 
 	if (reader->sink->prefix) {
 		ret = !reader->sink->prefix(reader->sink->handle, name, uri);
 	}
-	pop_node(reader, uri);
-	pop_node(reader, name);
 	if (!sparql) {
 		read_ws_star(reader);
 		return eat_byte_check(reader, '.');
@@ -1376,8 +1366,10 @@ read_wrappedGraph(SerdReader* reader, ReadContext* ctx)
 	TRY_RET(eat_byte_check(reader, '{'));
 	read_ws_star(reader);
 	while (peek_byte(reader) != '}') {
-		bool ate_dot = false;
-		int  s_type  = 0;
+		const size_t orig_stack_size = reader->stack.size;
+		bool         ate_dot         = false;
+		int          s_type          = 0;
+
 		ctx->subject = 0;
 		SerdNode* subj = read_subject(reader, *ctx, &ctx->subject, &s_type);
 		if (!subj && ctx->subject) {
@@ -1388,7 +1380,7 @@ read_wrappedGraph(SerdReader* reader, ReadContext* ctx)
 			return r_err(reader, SERD_ERR_BAD_SYNTAX,
 			             "missing predicate object list\n");
 		}
-		pop_node(reader, subj);
+		serd_stack_pop_to(&reader->stack, orig_stack_size);
 		read_ws_star(reader);
 		if (peek_byte(reader) == '.') {
 			eat_byte_safe(reader, '.');
@@ -1447,7 +1439,6 @@ read_n3_statement(SerdReader* reader)
 			TRY_RET((ctx.graph = read_labelOrSubject(reader)));
 			read_ws_star(reader);
 			TRY_RET(read_wrappedGraph(reader, &ctx));
-			pop_node(reader, ctx.graph);
 			ctx.graph = 0;
 			read_ws_star(reader);
 		} else if (read_ws_star(reader) && peek_byte(reader) == '{') {
@@ -1458,7 +1449,6 @@ read_n3_statement(SerdReader* reader)
 			ctx.graph   = subj;
 			ctx.subject = subj = 0;
 			TRY_RET(read_wrappedGraph(reader, &ctx));
-			pop_node(reader, ctx.graph);
 			read_ws_star(reader);
 		} else if (!subj) {
 			ret = r_err(reader, SERD_ERR_BAD_SYNTAX, "bad subject\n");
@@ -1471,7 +1461,6 @@ read_n3_statement(SerdReader* reader)
 			read_ws_star(reader);
 			ret = (eat_byte_check(reader, '.') == '.');
 		}
-		pop_node(reader, subj);
 		break;
 	}
 	return ret;
@@ -1489,13 +1478,16 @@ SerdStatus
 read_turtleTrigDoc(SerdReader* reader)
 {
 	while (!reader->source.eof) {
+		const size_t orig_stack_size = reader->stack.size;
 		if (!read_n3_statement(reader)) {
 			if (reader->strict) {
+				serd_stack_pop_to(&reader->stack, orig_stack_size);
 				return SERD_ERR_UNKNOWN;
 			}
 			skip_until(reader, '\n');
 			reader->status = SERD_SUCCESS;
 		}
+		serd_stack_pop_to(&reader->stack, orig_stack_size);
 	}
 	return reader->status;
 }
@@ -1504,6 +1496,8 @@ SerdStatus
 read_nquadsDoc(SerdReader* reader)
 {
 	while (!reader->source.eof) {
+		const size_t orig_stack_size = reader->stack.size;
+
 		SerdStatementFlags flags   = 0;
 		ReadContext        ctx     = { 0, 0, 0, 0, 0, 0, &flags };
 		bool               ate_dot = false;
@@ -1549,10 +1543,7 @@ read_nquadsDoc(SerdReader* reader)
 			break;
 		}
 
-		pop_node(reader, ctx.graph);
-		pop_node(reader, ctx.lang);
-		pop_node(reader, ctx.datatype);
-		pop_node(reader, ctx.object);
+		serd_stack_pop_to(&reader->stack, orig_stack_size);
 	}
 	return reader->status;
 }
diff --git a/src/reader.c b/src/reader.c
index 585a1cd2..3eccacc7 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -76,6 +76,13 @@ SerdNode*
 push_node_padded(SerdReader* reader, size_t maxlen,
                  SerdType type, const char* str, size_t n_bytes)
 {
+	// Push a null byte to ensure the previous node was null terminated
+	char* terminator = (char*)serd_stack_push(&reader->stack, 1);
+	if (!terminator) {
+		return NULL;
+	}
+	*terminator = 0;
+
 	void* mem = serd_stack_push_aligned(
 		&reader->stack, sizeof(SerdNode) + maxlen + 1, sizeof(SerdNode));
 
@@ -95,7 +102,7 @@ push_node_padded(SerdReader* reader, size_t maxlen,
 #ifdef SERD_STACK_CHECK
 	reader->allocs = realloc(
 		reader->allocs, sizeof(reader->allocs) * (++reader->n_allocs));
-	reader->allocs[reader->n_allocs - 1] = (mem - reader->stack.buf);
+	reader->allocs[reader->n_allocs - 1] = node;
 #endif
 	return node;
 }
@@ -106,21 +113,6 @@ 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*
-pop_node(SerdReader* reader, const SerdNode* node)
-{
-	if (node && node != reader->rdf_first && node != reader->rdf_rest
-	    && node != reader->rdf_nil) {
-#ifdef SERD_STACK_CHECK
-		SERD_STACK_ASSERT_TOP(reader, node);
-		--reader->n_allocs;
-#endif
-		char* const top = reader->stack.buf + reader->stack.size;
-		serd_stack_pop_aligned(&reader->stack, (size_t)(top - (char*)node));
-	}
-	return NULL;
-}
-
 bool
 emit_statement(SerdReader* reader, ReadContext ctx, SerdNode* o)
 {
@@ -190,9 +182,6 @@ serd_reader_set_strict(SerdReader* reader, bool strict)
 void
 serd_reader_free(SerdReader* reader)
 {
-	pop_node(reader, reader->rdf_nil);
-	pop_node(reader, reader->rdf_rest);
-	pop_node(reader, reader->rdf_first);
 	serd_node_free(reader->default_graph);
 
 #ifdef SERD_STACK_CHECK
diff --git a/src/reader.h b/src/reader.h
index 30663210..64e00eb7 100644
--- a/src/reader.h
+++ b/src/reader.h
@@ -94,8 +94,6 @@ size_t    genid_size(SerdReader* reader);
 SerdNode* blank_id(SerdReader* reader);
 void      set_blank_id(SerdReader* reader, SerdNode* node, size_t buf_size);
 
-SerdNode* pop_node(SerdReader* reader, const SerdNode* node);
-
 bool emit_statement(SerdReader* reader, ReadContext ctx, SerdNode* o);
 
 bool read_n3_statement(SerdReader* reader);
@@ -156,7 +154,7 @@ static inline SerdStatus
 push_byte(SerdReader* reader, SerdNode* node, const int c)
 {
 	assert(c != EOF);
-	SERD_STACK_ASSERT_TOP(reader, ref);
+	SERD_STACK_ASSERT_TOP(reader, node);
 
 	char* const s = (char*)serd_stack_push(&reader->stack, 1);
 	if (!s) {
diff --git a/src/stack.h b/src/stack.h
index 2356731b..5bd86f7a 100644
--- a/src/stack.h
+++ b/src/stack.h
@@ -22,6 +22,7 @@
 #include <stddef.h>
 #include <stdint.h>
 #include <stdlib.h>
+#include <string.h>
 
 /** An offset to start the stack at. Note 0 is reserved for NULL. */
 #define SERD_STACK_BOTTOM sizeof(void*)
@@ -77,40 +78,28 @@ serd_stack_pop(SerdStack* stack, size_t n_bytes)
 	stack->size -= n_bytes;
 }
 
+static inline void
+serd_stack_pop_to(SerdStack* stack, size_t n_bytes)
+{
+	assert(stack->size >= n_bytes);
+	stack->size = n_bytes;
+}
+
 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
-	if (!serd_stack_push(stack, 1)) {
-		return NULL;
-	}
-
 	// Push padding if necessary
 	const size_t pad = align - stack->size % align;
 	if (pad > 0) {
-		if (!serd_stack_push(stack, pad)) {
+		void* padding = serd_stack_push(stack, pad);
+		if (!padding) {
 			return NULL;
 		}
+		memset(padding, 0, pad);
 	}
 
-	// Set top of stack to pad count so we can properly pop later
-	stack->buf[stack->size - 1] = (char)pad;
-
 	// Push requested space at aligned location
 	return serd_stack_push(stack, n_bytes);
 }
 
-static inline void
-serd_stack_pop_aligned(SerdStack* stack, size_t n_bytes)
-{
-	// Pop requested space down to aligned location
-	serd_stack_pop(stack, n_bytes);
-
-	// Get amount of padding from top of stack
-	const uint8_t pad = (uint8_t)stack->buf[stack->size - 1];
-
-	// Pop padding and pad count
-	serd_stack_pop(stack, pad + 1u);
-}
-
 #endif  // SERD_STACK_H