From 429a9de315509ccfc6cf9fca7530490fc274718b Mon Sep 17 00:00:00 2001
From: Oleksandr Kulkov <oleksandr.kulkov@grlab.org>
Date: Thu, 23 May 2024 15:30:16 +0200
Subject: [PATCH] Add row_diff_traverse, row_diff_successor

---
 .github/workflows/main.yml                    |  2 -
 .../representation/base/sequence_graph.cpp    | 44 +++++++++++++++++++
 .../representation/base/sequence_graph.hpp    | 19 ++++++++
 3 files changed, 63 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index c0e2d4ef28..3be8ef96e3 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -7,8 +7,6 @@ on:
     tags:
       - 'v*'
   pull_request:
-    branches:
-      - master
 
 env:
   REGISTRY: ghcr.io
diff --git a/metagraph/src/graph/representation/base/sequence_graph.cpp b/metagraph/src/graph/representation/base/sequence_graph.cpp
index bc19dede84..23cc34e8fa 100644
--- a/metagraph/src/graph/representation/base/sequence_graph.cpp
+++ b/metagraph/src/graph/representation/base/sequence_graph.cpp
@@ -396,6 +396,50 @@ void DeBruijnGraph::call_unitigs(const CallPath &callback,
     ::mtg::graph::call_sequences(*this, callback, num_threads, true, min_tip_size, kmers_in_single_form);
 }
 
+void DeBruijnGraph::row_diff_traverse(size_t num_threads,
+                                      size_t max_length,
+                                      const bit_vector &rd_succ,
+                                      sdsl::bit_vector *terminal) const {
+    sdsl::bit_vector visited(max_index() + 1, false);
+    auto finalised = visited;
+    std::vector<size_t> distance(max_index() + 1);
+    assert(terminal->size() == visited.size());
+    assert(rd_succ.size() == visited.size());
+    auto set_terminal = [&](int v) {
+        distance[v] = 0;
+        (*terminal)[v] = true;
+    };
+    call_nodes([&](node_index v) {
+        if (visited[v]) {
+            return;
+        }
+        static std::stack<node_index> path;
+        while (!visited[v]) {
+            path.push(v);
+            visited[v] = true;
+            if (!has_no_outgoing(v)) {
+                v = row_diff_successor(v, rd_succ);
+            }
+        }
+        if (!finalised[v]) {
+            set_terminal(v);
+            finalised[v] = true;
+        }
+        node_index u = v;
+        while (!path.empty()) {
+            std::tie(u, v) = std::tie(path.top(), u);
+            if (!finalised[u]) {
+                distance[u] = distance[v] + 1;
+                if (distance[u] == max_length) {
+                    set_terminal(u);
+                }
+                finalised[u] = true;
+            }
+            path.pop();
+        }
+    });
+}
+
 /**
  * Traverse graph and iterate over all nodes
  */
diff --git a/metagraph/src/graph/representation/base/sequence_graph.hpp b/metagraph/src/graph/representation/base/sequence_graph.hpp
index 96d13cad34..c85bfb2b0f 100644
--- a/metagraph/src/graph/representation/base/sequence_graph.hpp
+++ b/metagraph/src/graph/representation/base/sequence_graph.hpp
@@ -1,6 +1,8 @@
 #ifndef __SEQUENCE_GRAPH_HPP__
 #define __SEQUENCE_GRAPH_HPP__
 
+#include "common/vectors/bit_vector.hpp"
+
 #include <vector>
 #include <string>
 #include <functional>
@@ -198,6 +200,7 @@ class DeBruijnGraph : public SequenceGraph {
     virtual void call_kmers(const std::function<void(node_index, const std::string&)> &callback) const;
 
     virtual size_t outdegree(node_index) const = 0;
+    virtual bool has_no_outgoing(node_index node) const { return outdegree(node) == 0; }
     virtual bool has_single_outgoing(node_index node) const { return outdegree(node) == 1; }
     virtual bool has_multiple_outgoing(node_index node) const { return outdegree(node) > 1; }
 
@@ -231,6 +234,22 @@ class DeBruijnGraph : public SequenceGraph {
 
     // Call all nodes that have no incoming edges
     virtual void call_source_nodes(const std::function<void(node_index)> &callback) const;
+
+    virtual void row_diff_traverse(size_t num_threads,
+                                   size_t max_length,
+                                   const bit_vector &rd_succ,
+                                   sdsl::bit_vector *terminal) const;
+
+    virtual node_index row_diff_successor(node_index node, const bit_vector &rd_succ) const {
+        node_index succ = npos;
+        adjacent_outgoing_nodes(node, [&](node_index adjacent_node) {
+            if(rd_succ[adjacent_node]) {
+                succ = adjacent_node;
+            }
+        });
+        assert(succ != npos && "a row diff successor must exist");
+        return succ;
+    }
 };