Merge pull request #3 from jeizenga/eval-crash

Speed up XG construction when IDs don't start at 1

Author: ekg

Diff for: CMakeLists.txt

@@ -77,7 +77,7 @@ set(sdsl-lite-divsufsort_LIB "${INSTALL_DIR}/src/sdsl-lite-build/external/libdiv
 # mmmultimap (memory mapped multimap)
 ExternalProject_Add(mmmultimap
   GIT_REPOSITORY "https://github.com/ekg/mmmultimap.git"
-  GIT_TAG "8a76cec0b819de6e2b855f4edc29993f75b1b26b"
+  GIT_TAG "b92a5c8826141d61413546278719724e0f612c39"
   BUILD_COMMAND ""
   UPDATE_COMMAND ""
   INSTALL_COMMAND "")
@@ -109,8 +109,8 @@ set(gfakluge_tinyFA_INCLUDE "${INSTALL_DIR}/src/gfakluge/src/tinyFA")
 
 # In-place Parallel Super Scalar Samplesort (IPS⁴o), header only
 ExternalProject_Add(ips4o
-  GIT_REPOSITORY "https://github.com/SaschaWitt/ips4o.git"
-  GIT_TAG "bff3ccf0bf349497f2bb10f825d160b792236367"
+  GIT_REPOSITORY "https://github.com/vgteam/ips4o.git"
+  GIT_TAG "22069381cc1bf2df07ee1ff47f6b6073fcfb4508"
   INSTALL_COMMAND ""
   BUILD_COMMAND ""
   CONFIGURE_COMMAND "")
@@ -121,6 +121,7 @@ set(CMAKE_BUILD_TYPE Release)
 
 add_library(xg_objs OBJECT src/xg.cpp)
 
+add_dependencies(xg_objs handlegraph)
 add_dependencies(xg_objs sdsl-lite)
 add_dependencies(xg_objs mmmultimap)
 add_dependencies(xg_objs tayweeargs)

Diff for: src/xg.cpp

@@ -11,7 +11,7 @@
 
 #include "gfakluge.hpp"
 
-#define VERBOSE_DEBUG
+//#define VERBOSE_DEBUG
 //#define debug_algorithms
 //#define debug_component_index
 
@@ -383,31 +383,31 @@ void XG::from_gfa(const std::string& gfa_filename, bool validate, std::string ba
     // set up our enumerators
     auto for_each_sequence = [&](const std::function<void(const std::string& seq, const nid_t& node_id)>& lambda) {
         gfa.for_each_sequence_line_in_file(filename, [&](gfak::sequence_elem s) {
-                nid_t node_id = std::stol(s.name);
-                lambda(s.sequence, node_id);
-            });
+            nid_t node_id = std::stol(s.name);
+            lambda(s.sequence, node_id);
+        });
     };
     auto for_each_edge = [&](const std::function<void(const nid_t& from_id, const bool& from_rev,
                                                       const nid_t& to_id, const bool& to_rev)>& lambda) {
         gfa.for_each_edge_line_in_file(filename, [&](gfak::edge_elem e) {
-                if (e.source_name.empty()) return;
-                nid_t from_id = std::stol(e.source_name);
-                bool from_rev = !e.source_orientation_forward;
-                nid_t to_id = std::stol(e.sink_name);
-                bool to_rev = !e.sink_orientation_forward;
-                lambda(from_id, from_rev, to_id, to_rev);
-            });
+            if (e.source_name.empty()) return;
+            nid_t from_id = std::stol(e.source_name);
+            bool from_rev = !e.source_orientation_forward;
+            nid_t to_id = std::stol(e.sink_name);
+            bool to_rev = !e.sink_orientation_forward;
+            lambda(from_id, from_rev, to_id, to_rev);
+        });
     };
     auto for_each_path_element = [&](const std::function<void(const std::string& path_name,
                                                               const nid_t& node_id, const bool& is_rev,
                                                               const std::string& cigar)>& lambda) {
         gfa.for_each_path_element_in_file(filename, [&](const std::string& path_name_raw, const std::string& node_id_str,
                                                         bool is_rev, const std::string& cigar) {
-                nid_t node_id = std::stol(node_id_str);
-                std::string path_name = path_name_raw;
-                path_name.erase(std::remove_if(path_name.begin(), path_name.end(), [](char c) { return std::isspace(c); }), path_name.end());
-                lambda(path_name, node_id, is_rev, cigar);
-            });
+            nid_t node_id = std::stol(node_id_str);
+            std::string path_name = path_name_raw;
+            path_name.erase(std::remove_if(path_name.begin(), path_name.end(), [](char c) { return std::isspace(c); }), path_name.end());
+            lambda(path_name, node_id, is_rev, cigar);
+        });
     };
     from_enumerators(for_each_sequence, for_each_edge, for_each_path_element, validate, basename);
 }
@@ -587,7 +587,7 @@ void XG::from_enumerators(const std::function<void(const std::function<void(cons
             edge_from_to_mm.append(as_integer(from_handle), as_integer(to_handle));
             edge_to_from_mm.append(as_integer(to_handle), as_integer(from_handle));
         });
-    handle_t max_handle = number_bool_packing::pack(max_id, true);
+    handle_t max_handle = number_bool_packing::pack(r_iv.size(), true);
     edge_from_to_mm.index(as_integer(max_handle));
     edge_to_from_mm.index(as_integer(max_handle));
 

Diff for: src/xg.hpp

@@ -22,17 +22,11 @@
 #include "sdsl/csa_wt.hpp"
 #include "sdsl/suffix_arrays.hpp"
 
-#include <handlegraph/types.hpp>
-#include <handlegraph/iteratee.hpp>
-#include <handlegraph/util.hpp>
-#include <handlegraph/handle_graph.hpp>
-//#include <handlegraph/path_handle_graph.hpp>
-#include <handlegraph/path_position_handle_graph.hpp>
-//#include <handlegraph/mutable_handle_graph.hpp>
-//#include <handlegraph/mutable_path_handle_graph.hpp>
-//#include <handlegraph/mutable_path_mutable_handle_graph.hpp>
-//#include <handlegraph/deletable_handle_graph.hpp>
-//#include <handlegraph/mutable_path_deletable_handle_graph.hpp>
+#include "handlegraph/types.hpp"
+#include "handlegraph/iteratee.hpp"
+#include "handlegraph/util.hpp"
+#include "handlegraph/handle_graph.hpp"
+#include "handlegraph/path_position_handle_graph.hpp"
 
 #include "mmmultimap.hpp"