From 4097440be9d749bf1cbf65322a2a3595c9df48dc Mon Sep 17 00:00:00 2001
From: AndreaGuarracino <andreaguarracino@outlook.com>
Date: Mon, 12 Feb 2024 22:01:55 +0100
Subject: [PATCH] avoid duplicated subpaths

---
 src/position.hpp                |  9 +++++++++
 src/subcommand/extract_main.cpp | 11 +++++++++++
 2 files changed, 20 insertions(+)

diff --git a/src/position.hpp b/src/position.hpp
index 9aa1ac4db..69078f8cf 100644
--- a/src/position.hpp
+++ b/src/position.hpp
@@ -88,6 +88,15 @@ struct path_range_t {
     std::string data;
 };
 
+struct path_range_comparator {
+    bool operator() (const path_range_t& lhs, const path_range_t& rhs) const {
+        if (lhs.begin.path != rhs.begin.path) return lhs.begin.path < rhs.begin.path;
+        if (lhs.end.path != rhs.end.path) return lhs.end.path < rhs.end.path;
+        if (lhs.begin.offset != rhs.begin.offset) return lhs.begin.offset < rhs.begin.offset;
+        return lhs.end.offset < rhs.end.offset;
+    }
+};
+
 inline std::string& get_long_path_name(std::tuple<std::string, uint64_t, uint64_t> path_long_start_end) {
 	return std::get<0>(path_long_start_end);
 }
diff --git a/src/subcommand/extract_main.cpp b/src/subcommand/extract_main.cpp
index ce0ecc022..e900a28d8 100644
--- a/src/subcommand/extract_main.cpp
+++ b/src/subcommand/extract_main.cpp
@@ -556,6 +556,17 @@ namespace odgi {
                 return std::binary_search(source_paths_from_path_ranges.begin(), source_paths_from_path_ranges.end(), x);
             }), source_paths->end());
 
+            // We don't cut nodes for the extraction, so close path intervals can generate identical subpaths.
+            // To avoid duplicated subpaths in the final subgraph, we remove duplicated path ranges.
+            {
+                std::set<odgi::path_range_t, odgi::path_range_comparator> unique_path_ranges;
+
+                for (const auto& path_range : path_ranges) {
+                    unique_path_ranges.insert(path_range);
+                }
+
+                path_ranges.assign(unique_path_ranges.begin(), unique_path_ranges.end());
+            }
 
             if (max_dist_subpaths > 0) {
                 // Iterate multiple times to merge subpaths which became mergeable during the first iteration where new nodes were added