Moved edge index implementation to a separate namespace (#27)

CMakeLists.txt

@@ -50,6 +50,7 @@ endif()
 
 set(SONATA_SRC
     src/common.cpp
+    src/edge_index.cpp
     src/edges.cpp
     src/hdf5_mutex.cpp
     src/nodes.cpp

src/edge_index.cpp

@@ -0,0 +1,111 @@
+/*************************************************************************
+ * Copyright (C) 2018-2019 Blue Brain Project
+ *
+ * This file is part of 'libsonata', distributed under the terms
+ * of the GNU Lesser General Public License.
+ *
+ * See top-level LICENSE.txt file for details.
+ *************************************************************************/
+
+#include "edge_index.h"
+
+#include <bbp/sonata/common.h>
+
+#include <cstdint>
+#include <set>
+#include <vector>
+
+
+namespace bbp {
+namespace sonata {
+namespace edge_index {
+
+namespace {
+
+typedef std::vector<std::vector<uint64_t>> RawIndex;
+
+const char* SOURCE_INDEX_GROUP = "indices/source_to_target";
+const char* TARGET_INDEX_GROUP = "indices/target_to_source";
+const char* NODE_ID_TO_RANGES_DSET = "node_id_to_ranges";
+const char* RANGE_TO_EDGE_ID_DSET = "range_to_edge_id";
+
+}  // unnamed namespace
+
+
+const HighFive::Group sourceIndex(const HighFive::Group& h5Root)
+{
+    if (!h5Root.exist(SOURCE_INDEX_GROUP)) {
+        throw SonataError("No source index group found");
+    }
+    return h5Root.getGroup(SOURCE_INDEX_GROUP);
+}
+
+
+const HighFive::Group targetIndex(const HighFive::Group& h5Root)
+{
+    if (!h5Root.exist(TARGET_INDEX_GROUP)) {
+        throw SonataError("No target index group found");
+    }
+    return h5Root.getGroup(TARGET_INDEX_GROUP);
+}
+
+
+Selection resolve(const HighFive::Group& indexGroup, const NodeID nodeID)
+{
+    if (nodeID >= indexGroup.getDataSet(NODE_ID_TO_RANGES_DSET).getSpace().getDimensions()[0]) {
+        // Returning empty set for out-of-range node IDs, to be aligned with SYN2 reader implementation
+        // TODO: throw a SonataError instead
+        return Selection({});
+    }
+
+    RawIndex primaryRange;
+    indexGroup
+        .getDataSet(NODE_ID_TO_RANGES_DSET)
+        .select({ nodeID, 0 }, { 1, 2 })
+        .read(primaryRange);
+
+    const uint64_t primaryRangeBegin = primaryRange[0][0];
+    const uint64_t primaryRangeEnd = primaryRange[0][1];
+
+    if (primaryRangeBegin >= primaryRangeEnd) {
+        return Selection({});
+    }
+
+    RawIndex secondaryRange;
+    indexGroup
+        .getDataSet(RANGE_TO_EDGE_ID_DSET)
+        .select({ primaryRangeBegin, 0 }, { primaryRangeEnd - primaryRangeBegin, 2 })
+        .read(secondaryRange);
+
+    Selection::Ranges ranges;
+    ranges.reserve(secondaryRange.size());
+
+    for (const auto& row: secondaryRange) {
+        ranges.emplace_back(row[0], row[1]);
+    }
+
+    return Selection(std::move(ranges));
+}
+
+
+Selection resolve(const HighFive::Group& indexGroup, const std::vector<NodeID>& nodeIDs)
+{
+    if (nodeIDs.size() == 1) {
+        return resolve(indexGroup, nodeIDs[0]);
+    }
+    // TODO optimize: bulk read for primary index
+    // TODO optimize: range merging
+    std::set<EdgeID> merged;
+    for (NodeID nodeID : nodeIDs) {
+        const auto ids = resolve(indexGroup, nodeID).flatten();
+        merged.insert(ids.begin(), ids.end());
+    }
+    std::vector<EdgeID> result;
+    result.reserve(merged.size());
+    std::copy(merged.begin(), merged.end(), std::back_inserter(result));
+    return Selection::fromValues(result);
+}
+
+}
+}
+} // namespace bbp::sonata::edge_index

src/edge_index.h

@@ -0,0 +1,29 @@
+/*************************************************************************
+ * Copyright (C) 2018-2019 Blue Brain Project
+ *
+ * This file is part of 'libsonata', distributed under the terms
+ * of the GNU Lesser General Public License.
+ *
+ * See top-level LICENSE.txt file for details.
+ *************************************************************************/
+
+#pragma once
+
+#include <bbp/sonata/population.h>
+
+#include <highfive/H5File.hpp>
+#include <highfive/H5Group.hpp>
+
+namespace bbp {
+namespace sonata {
+namespace edge_index {
+
+const HighFive::Group sourceIndex(const HighFive::Group& h5Root);
+const HighFive::Group targetIndex(const HighFive::Group& h5Root);
+
+Selection resolve(const HighFive::Group& indexGroup, NodeID nodeID);
+Selection resolve(const HighFive::Group& indexGroup, const std::vector<NodeID>& nodeIDs);
+
+}
+}
+} // namespace bbp::sonata::edge_index

src/edges.cpp

@@ -9,6 +9,7 @@
 
 #include "population.hpp"
 #include "hdf5_mutex.hpp"
+#include "edge_index.h"
 
 #include <bbp/sonata/common.h>
 #include <bbp/sonata/edges.h>
@@ -19,6 +20,15 @@
 #include <algorithm>
 
 
+namespace {
+
+const char* SOURCE_NODE_ID_DSET = "source_node_id";
+const char* TARGET_NODE_ID_DSET = "target_node_id";
+const char* NODE_POPULATION_ATTR = "node_population";
+
+}  // unnamed namespace
+
+
 namespace bbp {
 namespace sonata {
 
@@ -36,7 +46,7 @@ std::string EdgePopulation::source() const
 {
     HDF5_LOCK_GUARD
     std::string result;
-    impl_->h5Root.getDataSet("source_node_id").getAttribute("node_population").read(result);
+    impl_->h5Root.getDataSet(SOURCE_NODE_ID_DSET).getAttribute(NODE_POPULATION_ATTR).read(result);
     return result;
 }
 
@@ -45,100 +55,45 @@ std::string EdgePopulation::target() const
 {
     HDF5_LOCK_GUARD
     std::string result;
-    impl_->h5Root.getDataSet("target_node_id").getAttribute("node_population").read(result);
+    impl_->h5Root.getDataSet(TARGET_NODE_ID_DSET).getAttribute(NODE_POPULATION_ATTR).read(result);
     return result;
 }
 
 
 std::vector<NodeID> EdgePopulation::sourceNodeIDs(const Selection& selection) const
 {
     HDF5_LOCK_GUARD
-    const auto dset = impl_->h5Root.getDataSet("source_node_id");
+    const auto dset = impl_->h5Root.getDataSet(SOURCE_NODE_ID_DSET);
     return _readSelection<NodeID>(dset, selection);
 }
 
 
 std::vector<NodeID> EdgePopulation::targetNodeIDs(const Selection& selection) const
 {
     HDF5_LOCK_GUARD
-    const auto dset = impl_->h5Root.getDataSet("target_node_id");
+    const auto dset = impl_->h5Root.getDataSet(TARGET_NODE_ID_DSET);
     return _readSelection<NodeID>(dset, selection);
 }
 
 
-namespace {
-
-Selection _resolveIndex(const HighFive::Group& indexGroup, const NodeID nodeID)
-{
-    typedef std::vector<std::vector<uint64_t>> RawIndex;
-
-    if (nodeID >= indexGroup.getDataSet("node_id_to_ranges").getSpace().getDimensions()[0]) {
-        // Returning empty set for out-of-range node IDs, to be aligned with SYN2 reader implementation
-        // TODO: throw a SonataError instead
-        return Selection({});
-    }
-
-    RawIndex primaryRange;
-    indexGroup
-        .getDataSet("node_id_to_ranges")
-        .select({ nodeID, 0 }, { 1, 2 })
-        .read(primaryRange);
-
-    const uint64_t primaryRangeBegin = primaryRange[0][0];
-    const uint64_t primaryRangeEnd = primaryRange[0][1];
-
-    if (primaryRangeBegin >= primaryRangeEnd) {
-        return Selection({});
-    }
-
-    RawIndex secondaryRange;
-    indexGroup
-        .getDataSet("range_to_edge_id")
-        .select({ primaryRangeBegin, 0 }, { primaryRangeEnd - primaryRangeBegin, 2 })
-        .read(secondaryRange);
-
-    Selection::Ranges ranges;
-    ranges.reserve(secondaryRange.size());
-
-    for (const auto& row: secondaryRange) {
-        ranges.emplace_back(row[0], row[1]);
-    }
-
-    return Selection(std::move(ranges));
-}
-
-
-Selection _resolveIndex(const HighFive::Group& indexGroup, const std::vector<NodeID>& nodeIDs)
-{
-    if (nodeIDs.size() == 1) {
-        return _resolveIndex(indexGroup, nodeIDs[0]);
-    }
-    // TODO optimize: bulk read for primary index
-    // TODO optimize: range merging
-    std::set<EdgeID> result;
-    for (NodeID nodeID : nodeIDs) {
-        const auto ids = _resolveIndex(indexGroup, nodeID).flatten();
-        result.insert(ids.begin(), ids.end());
-    }
-    return _selectionFromValues(result.begin(), result.end());
-}
-
-} // unnamed namespace
-
 
 Selection EdgePopulation::afferentEdges(const std::vector<NodeID>& target) const
 {
     HDF5_LOCK_GUARD
-    const auto& indexGroup = impl_->h5Root.getGroup("indices/target_to_source");
-    return _resolveIndex(indexGroup, target);
+    return edge_index::resolve(
+        edge_index::targetIndex(impl_->h5Root),
+        target
+    );
 }
 
 
 Selection EdgePopulation::efferentEdges(const std::vector<NodeID>& source) const
 {
     HDF5_LOCK_GUARD
-    const auto& indexGroup = impl_->h5Root.getGroup("indices/source_to_target");
-    return _resolveIndex(indexGroup, source);
+    return edge_index::resolve(
+        edge_index::sourceIndex(impl_->h5Root),
+        source
+    );
 }
 
 

src/population.cpp

@@ -49,7 +49,26 @@ Selection::Selection(const Selection::Ranges& ranges)
 
 Selection Selection::fromValues(const Selection::Values& values)
 {
-    return _selectionFromValues(values.begin(), values.end());
+    Selection::Ranges ranges;
+
+    Selection::Range range{ 0, 0 };
+    for (const auto v: values) {
+        if (v == range.second) {
+            ++range.second;
+        } else {
+            if (range.first < range.second) {
+                ranges.push_back(range);
+            }
+            range.first = v;
+            range.second = v + 1;
+        }
+    }
+
+    if (range.first < range.second) {
+        ranges.push_back(range);
+    }
+
+    return Selection(std::move(ranges));
 }
 
 

src/population.hpp

@@ -39,32 +39,6 @@ std::set<std::string> _listChildren(const HighFive::Group& group, const std::set
 }
 
 
-template<typename Iterator>
-Selection _selectionFromValues(Iterator first, Iterator last)
-{
-    Selection::Ranges ranges;
-
-    Selection::Range range{ 0, 0 };
-    for (Iterator it = first; it != last; ++it) {
-        if (*it == range.second) {
-            ++range.second;
-        } else {
-            if (range.first < range.second) {
-                ranges.push_back(range);
-            }
-            range.first = *it;
-            range.second = *it + 1;
-        }
-    }
-
-    if (range.first < range.second) {
-        ranges.push_back(range);
-    }
-
-    return Selection(std::move(ranges));
-}
-
-
 template<typename T>
 std::vector<T> _readChunk(const HighFive::DataSet& dset, const Selection::Range& range)
 {