Add topological sort and cycle detection

scratchpad/dag_man.c

@@ -6,6 +6,8 @@
 #include <stdlib.h>
 #include <string.h>
 
+wp_intern WpLogger _log = { .name = wpStr8LitRo("dag_man") };
+
 /* ---------------------------------------------------------------------------
  * Pool allocator (arena-backed, intrusive free list)
  * -------------------------------------------------------------------------*/
@@ -63,6 +65,7 @@ typedef struct {
 	u64 index;
 	u64 generation;
 } PrNodeId;
+typedef PrNodeId *PrNodeIdArray;
 
 typedef struct {
 	union {
@@ -124,6 +127,11 @@ wp_intern void prGraphInit(PrGraph *graph, WpAllocator *allocator, u64 capacity)
 	graph->max_count_ever = 0;
 	graph->count          = 0;
 
+	if (!graph->nodes || !graph->vertices) {
+		graph->capacity = 0;
+		return;
+	}
+
 	prPoolInit(&graph->vertex_pool, allocator, sizeof(PrEdgeVertex));
 	memset(graph->nodes, 0, capacity * sizeof(PrNode));
 	memset(graph->vertices, 0, capacity * sizeof(PrEdgeVertex));
@@ -134,40 +142,73 @@ wp_intern void prGraphInit(PrGraph *graph, WpAllocator *allocator, u64 capacity)
 	}
 }
 
-wp_intern void prGraphAddEdge(PrGraph *graph, PrNodeId from, PrNodeId to) {
+/* ---------------------------------------------------------------------------
-	if (!prGraphIsActiveNode(graph, from) || !prGraphIsActiveNode(graph, to)) { return; }
+ * Kahn's algorithm — topological sort / cycle detection
-	if (from.index == to.index) { return; }
+ *
+ * Returns a WpArray of PrNodeId (sorted topologically). If the result count
+ * is less than graph->count, the graph contains a cycle.
+ * -------------------------------------------------------------------------*/
 
-	PrEdgeVertex *edge = (PrEdgeVertex *)prPoolAlloc(&graph->vertex_pool);
+wp_intern PrNodeId *prGraphTopologicalSort(PrGraph *graph, const WpAllocator *allocator) {
+	if (graph->count == 0) { return NULL; }
+	if (!graph->nodes || graph->capacity == 0) { return NULL; }
 
-	PrEdgeVertex *src  = &graph->vertices[from.index];
+	PrNodeIdArray result = wpArrayAllocCapacity(PrNodeId, allocator, graph->count, WP_ARRAY_INIT_NONE);
-	PrEdgeVertex *dst  = &graph->vertices[to.index];
+	if (!result) { return NULL; }
 
-	edge->source        = from;
+	WpAllocator local_arena = wpMemArenaAllocatorInitZero(KiB(16));
-	edge->target        = to;
+
-	edge->next_forward  = src->next_forward;
+	WpU64Array in_degree = wpArrayAllocCapacity(u64, &local_arena, graph->capacity, WP_ARRAY_INIT_FILLED);
-	edge->next_backward = dst->next_backward;
+	if (!in_degree) { return result; }
-	src->next_forward   = edge;
+	memset(in_degree, 0, wpArrayCapacity(in_degree) * sizeof(u64));
-	dst->next_backward  = edge;
+
+	for (u64 i = 0; i < graph->max_count_ever; i++) {
+		PrNodeId id = prGraphGetNode(graph, i);
+		if (!prGraphIsActiveNode(graph, id)) { continue; }
+
+		PrEdgeVertex *vertex = &graph->vertices[i];
+		PrEdgeVertex *curr   = vertex->next_forward;
+		while (curr) {
+			in_degree[curr->target.index]++;
+			curr = curr->next_forward;
+		}
+	}
+
+	WpQueue queue = wpQueueAlloc(u64, &local_arena, graph->count);
+
+	for (u64 i = 0; i < graph->max_count_ever; i++) {
+		PrNodeId id = prGraphGetNode(graph, i);
+		if (!prGraphIsActiveNode(graph, id)) { continue; }
+		if (in_degree[i] == 0) {
+			wpQueuePush(u64, &queue, &i);
+		}
+	}
+
+	while (queue.count > 0) {
+		u64 *node_idx = wpQueuePop(u64, &queue);
+		if (!node_idx) { break; }
+
+		PrNodeId id = prGraphGetNode(graph, *node_idx);
+		wpArrayAppendCapped(PrNodeId, result, &id);
+
+		PrEdgeVertex *vertex = &graph->vertices[*node_idx];
+		PrEdgeVertex *curr   = vertex->next_forward;
+		while (curr) {
+			u64 target_idx = curr->target.index;
+			if (in_degree[target_idx] > 0) {
+				in_degree[target_idx]--;
+				if (in_degree[target_idx] == 0) {
+					wpQueuePush(u64, &queue, &target_idx);
+				}
+			}
+			curr = curr->next_forward;
+		}
+	}
+
+	return result;
 }
 
-wp_intern PrNodeId prGraphAddNode(PrGraph *graph, PrNodeType type) {
+/* --- edge management -------------------------------------------------- */
-	u64 idx = graph->free_head;
-	if (idx == INVALID_NODE_INDEX) { return INVALID_NODE_ID; }
-
-	PrNode *node = &graph->nodes[idx];
-
-	graph->free_head = node->next_free;
-	node->next_free  = INVALID_NODE_INDEX;
-	node->type       = type;
-	memset(&node->params, 0, sizeof(node->params));
-
-	graph->count++;
-
-	if (idx + 1 > graph->max_count_ever) { graph->max_count_ever = idx + 1; }
-
-	return (PrNodeId){ .index = idx, .generation = node->generation };
-}
 
 wp_intern void _unlinkForward(PrGraph *graph, PrNodeId from, PrEdgeVertex *edge) {
 	if (!prGraphIsActiveNode(graph, from)) { return; }
@@ -211,6 +252,55 @@ wp_intern void _unlinkBackward(PrGraph *graph, PrNodeId to, PrEdgeVertex *edge)
 	}
 }
 
+wp_intern void prGraphAddEdge(PrGraph *graph, PrNodeId from, PrNodeId to) {
+	if (!prGraphIsActiveNode(graph, from) || !prGraphIsActiveNode(graph, to)) { return; }
+	if (from.index == to.index) { return; }
+
+	PrEdgeVertex *edge = (PrEdgeVertex *)prPoolAlloc(&graph->vertex_pool);
+	if (!edge) { return; }
+
+	edge->source = from;
+	edge->target = to;
+
+	/* Link into adjacency chains */
+	PrEdgeVertex *src   = &graph->vertices[from.index];
+	PrEdgeVertex *dst   = &graph->vertices[to.index];
+	edge->next_forward  = src->next_forward;
+	edge->next_backward = dst->next_backward;
+	src->next_forward   = edge;
+	dst->next_backward  = edge;
+
+	/* Check whether the new edge created a cycle */
+	WpAllocator scratch  = wpMemArenaAllocatorInitZero(KiB(16));
+	PrNodeIdArray sorted = prGraphTopologicalSort(graph, &scratch);
+	u64 sorted_count     = sorted ? wpArrayCount(sorted) : 0;
+	if (sorted_count < graph->count) {
+		_unlinkForward(graph, from, edge);
+		_unlinkBackward(graph, to, edge);
+		prPoolFree(&graph->vertex_pool, edge);
+	}
+
+	wpMemArenaAllocatorDestroy(&scratch);
+}
+
+wp_intern PrNodeId prGraphAddNode(PrGraph *graph, PrNodeType type) {
+	u64 idx = graph->free_head;
+	if (idx == INVALID_NODE_INDEX) { return INVALID_NODE_ID; }
+
+	PrNode *node = &graph->nodes[idx];
+
+	graph->free_head = node->next_free;
+	node->next_free  = INVALID_NODE_INDEX;
+	node->type       = type;
+	memset(&node->params, 0, sizeof(node->params));
+
+	graph->count++;
+
+	if (idx + 1 > graph->max_count_ever) { graph->max_count_ever = idx + 1; }
+
+	return (PrNodeId){ .index = idx, .generation = node->generation };
+}
+
 wp_intern void prGraphRemoveNode(PrGraph *graph, PrNodeId id) {
 	if (!prGraphIsActiveNode(graph, id)) { return; }
 
@@ -297,6 +387,10 @@ wp_intern void prGraphDump(const PrGraph *graph) {
 
 i32 main(void) {
 	WpAllocator arena = wpMemArenaAllocatorInitZero(MiB(16));
+	if (wpMemAllocatorInvalid(&arena)) {
+		wpLogFatal(&_log, wpStr8Lit("arena init failed"));
+		return 1;
+	}
 
 	PrGraph graph = {0};
 	prGraphInit(&graph, &arena, 128);
@@ -349,6 +443,32 @@ i32 main(void) {
 	printf("\n");
 	prGraphDump(&graph);
 
+	printf("\n=== Try adding cycle 8→1 (should be rejected) ===\n");
+	prGraphAddEdge(&graph, n8, n1);
+
+	printf("\n=== Try adding 8→1 again (still rejected) ===\n");
+	prGraphAddEdge(&graph, n8, n1);
+
+	printf("\n=== Try adding cycle 7→2 (should be rejected) ===\n\n");
+	prGraphAddEdge(&graph, n7, n2);
+
+	prGraphDump(&graph);
+
+	printf("\n=== Topological sort ===\n");
+	PrNodeId *sorted = prGraphTopologicalSort(&graph, &arena);
+	if (sorted) {
+		u64 n = wpArrayCount(sorted);
+		printf("  count: %llu / %llu active\n",
+		       (unsigned long long)n,
+		       (unsigned long long)graph.count);
+		for (u64 i = 0; i < n; i++) {
+			printf("  [%llu] idx=%llu gen=%llu\n",
+			       (unsigned long long)i,
+			       (unsigned long long)sorted[i].index,
+			       (unsigned long long)sorted[i].generation);
+		}
+	}
+
 	wpMemArenaAllocatorDestroy(&arena);
 
 	return 0;