feat(interactive): Support labels and type Functions in Cypher Qu…

…eries (alibaba#3255)

<!--
Thanks for your contribution! please review
https://github.com/alibaba/GraphScope/blob/main/CONTRIBUTING.md before
opening an issue.
-->

## What do these changes do?
as titled.

<!-- Please give a short brief about these changes. -->

## Related issue number

<!-- Are there any issues opened that will be resolved by merging this
change? -->

Fixes

---------

Co-authored-by: zhanglei1949 <[email protected]>
Co-authored-by: xiaolei.zl <[email protected]>
Co-authored-by: Longbin Lai <[email protected]>

docs/interactive_engine/neo4j/supported_cypher.md

@@ -98,6 +98,8 @@ Note that some Aggregator operators, such as `max()`, we listed here are impleme
 | Branch | Use with `Project` and `Return` | CASE WHEN  | CASE WHEN |  <input type="checkbox" disabled  />| planned |
 | Scalar | Returns the length of a path | length() | length() | <input type="checkbox" disabled checked /> | |
 | List | Fold expressions into a single list | [] | [] | <input type="checkbox" disabled checked /> |   |
+| Labels | Get label name of a vertex type | labels() | labels() | <input type="checkbox" disabled checked /> |   |
+| Type | Get label name of an edge type | type() | type() | <input type="checkbox" disabled checked /> |   |
 
 
 

flex/codegen/src/graph_types.h

@@ -43,6 +43,7 @@ enum class DataType {
   kTime = 11,
   kDate = 12,
   kDateTime = 13,
+  kLabelId = 14
 };
 
 // a parameter const, the real data will be feed at runtime.
@@ -193,6 +194,8 @@ static std::string data_type_2_string(const codegen::DataType& data_type) {
     return EDGE_ID_T;
   case codegen::DataType::kDate:
     return "Date";
+  case codegen::DataType::kLabelId:
+    return "LabelKey";
   default:
     // LOG(FATAL) << "unknown data type" << static_cast<int>(data_type);
     throw std::runtime_error(

flex/codegen/src/hqps/hqps_project_builder.h

@@ -214,6 +214,10 @@ std::string project_variable_mapping_to_string(BuildingContext& ctx,
       } else if (prop.item_case() == common::Property::kLen) {
         prop_names.push_back("length");
         data_types.push_back(codegen::DataType::kLength);
+      } else if (prop.item_case() == common::Property::kLabel) {
+        // return the label id.
+        prop_names.push_back("label");
+        data_types.push_back(codegen::DataType::kLabelId);
       } else {
         LOG(FATAL) << "Unknown property type" << prop.DebugString();
       }

flex/engines/hqps_db/core/operator/project.h

@@ -291,8 +291,31 @@ class ProjectOp {
 
   ///////////////////Project implementation for all data structures.
 
+  /// Special case for project for labelKey
+  template <
+      typename T, typename NODE_T,
+      typename std::enable_if<std::is_same_v<T, LabelKey>>::type* = nullptr>
+  static auto apply_single_project_impl(
+      const GRAPH_INTERFACE& graph, NODE_T& node, const std::string& prop_name,
+      const std::vector<size_t>& repeat_array) {
+    auto size = node.Size();
+    auto label_vec = node.GetLabelVec();
+    std::vector<T> res_prop_vec;
+    CHECK(label_vec.size() == repeat_array.size())
+        << "label size: " << label_vec.size()
+        << " repeat size: " << repeat_array.size();
+    for (auto i = 0; i < repeat_array.size(); ++i) {
+      for (auto j = 0; j < repeat_array[i]; ++j) {
+        res_prop_vec.emplace_back(label_vec[i]);
+      }
+    }
+    return Collection<T>(std::move(res_prop_vec));
+  }
+
   // single label vertex set.
-  template <typename T, typename LabelT, typename VID_T, typename... SET_T>
+  template <
+      typename T, typename LabelT, typename VID_T, typename... SET_T,
+      typename std::enable_if<(!std::is_same_v<T, LabelKey>)>::type* = nullptr>
   static auto apply_single_project_impl(
       const GRAPH_INTERFACE& graph,
       RowVertexSetImpl<LabelT, VID_T, SET_T...>& node,
@@ -315,8 +338,10 @@ class ProjectOp {
   }
 
   // single keyed label vertex set.
-  template <typename T, typename LabelT, typename KEY_T, typename VID_T,
-            typename... SET_T>
+  template <
+      typename T, typename LabelT, typename KEY_T, typename VID_T,
+      typename... SET_T,
+      typename std::enable_if<(!std::is_same_v<T, LabelKey>)>::type* = nullptr>
   static auto apply_single_project_impl(
       const GRAPH_INTERFACE& graph,
       KeyedRowVertexSetImpl<LabelT, KEY_T, VID_T, SET_T...>& node,
@@ -340,7 +365,9 @@ class ProjectOp {
   }
 
   // project for two label vertex set.
-  template <typename T, typename VID_T, typename LabelT, typename... SET_T>
+  template <
+      typename T, typename VID_T, typename LabelT, typename... SET_T,
+      typename std::enable_if<(!std::is_same_v<T, LabelKey>)>::type* = nullptr>
   static auto apply_single_project_impl(
       const GRAPH_INTERFACE& graph,
       TwoLabelVertexSetImpl<VID_T, LabelT, SET_T...>& node,
@@ -379,7 +406,9 @@ class ProjectOp {
   }
 
   // general vertex set.
-  template <typename T, typename VID_T, typename LabelT>
+  template <
+      typename T, typename VID_T, typename LabelT,
+      typename std::enable_if<(!std::is_same_v<T, LabelKey>)>::type* = nullptr>
   static auto apply_single_project_impl(
       const GRAPH_INTERFACE& graph, GeneralVertexSet<VID_T, LabelT>& node,
       const std::string& prop_name_, const std::vector<size_t>& repeat_array) {
@@ -419,8 +448,10 @@ class ProjectOp {
   }
 
   // single label edge set
-  template <typename T, typename NODE_T,
-            typename std::enable_if<NODE_T::is_edge_set>::type* = nullptr>
+  template <
+      typename T, typename NODE_T,
+      typename std::enable_if<NODE_T::is_edge_set &&
+                              (!std::is_same_v<T, LabelKey>)>::type* = nullptr>
   static auto apply_single_project_impl(
       const GRAPH_INTERFACE& graph, NODE_T& node, const std::string& prop_name,
       const std::vector<size_t>& repeat_array) {
@@ -449,7 +480,9 @@ class ProjectOp {
   }
 
   /// Apply project on untyped edge set.
-  template <typename T, typename VID_T, typename LabelT, typename SUB_GRAPH_T>
+  template <
+      typename T, typename VID_T, typename LabelT, typename SUB_GRAPH_T,
+      typename std::enable_if<(!std::is_same_v<T, LabelKey>)>::type* = nullptr>
   static auto apply_single_project_impl(
       const GRAPH_INTERFACE& graph,
       UnTypedEdgeSet<VID_T, LabelT, SUB_GRAPH_T>& node,

flex/engines/hqps_db/core/operator/sink.h

@@ -378,10 +378,11 @@ class SinkOp {
     }
   }
 
-  // sink collection of pod
+  // sink collection of pod, expect for LabelKey type
   template <size_t Ind, size_t act_tag_id, typename T,
-            typename std::enable_if<(!gs::is_vector<T>::value) &&
-                                    (!gs::is_tuple<T>::value)>::type* = nullptr>
+            typename std::enable_if<
+                (!gs::is_vector<T>::value) && (!gs::is_tuple<T>::value) &&
+                (!std::is_same<T, LabelKey>::value)>::type* = nullptr>
   static void sink_col_impl(results::CollectiveResults& results_vec,
                             const Collection<T>& collection,
                             const std::vector<size_t>& repeat_offsets,
@@ -417,6 +418,43 @@ class SinkOp {
     }
   }
 
+  // sink collection of LabelKey
+  template <size_t Ind, size_t act_tag_id>
+  static void sink_col_impl(results::CollectiveResults& results_vec,
+                            const Collection<LabelKey>& collection,
+                            const std::vector<size_t>& repeat_offsets,
+                            int32_t tag_id) {
+    if (repeat_offsets.empty()) {
+      CHECK(collection.Size() == results_vec.results_size())
+          << "size neq " << collection.Size() << " "
+          << results_vec.results_size();
+      for (auto i = 0; i < collection.Size(); ++i) {
+        auto row = results_vec.mutable_results(i);
+        CHECK(row->record().columns_size() == Ind);
+        auto record = row->mutable_record();
+        auto new_col = record->add_columns();
+        new_col->mutable_name_or_id()->set_id(tag_id);
+        auto obj =
+            new_col->mutable_entry()->mutable_element()->mutable_object();
+        obj->set_i32(collection.Get(i).label_id);
+      }
+    } else {
+      CHECK(repeat_offsets.size() == collection.Size());
+      size_t cur_ind = 0;
+      for (auto i = 0; i < collection.Size(); ++i) {
+        for (auto j = 0; j < repeat_offsets[i]; ++j) {
+          auto row = results_vec.mutable_results(cur_ind++);
+          auto record = row->mutable_record();
+          auto new_col = record->add_columns();
+          new_col->mutable_name_or_id()->set_id(tag_id);
+          auto obj =
+              new_col->mutable_entry()->mutable_element()->mutable_object();
+          obj->set_i32(collection.Get(i).label_id);
+        }
+      }
+    }
+  }
+
   // sinke for tuple with one element
   template <size_t Ind, size_t act_tag_id, typename T,
             typename std::enable_if<

flex/engines/hqps_db/core/params.h

@@ -238,6 +238,13 @@ struct LengthKey {
   using length_data_type = int32_t;
 };
 
+struct LabelKey {
+  using label_data_type = int32_t;
+  int32_t label_id;
+  LabelKey() = default;
+  LabelKey(int32_t id) : label_id(id) {}
+};
+
 template <typename T>
 struct is_label_key_prop : std::false_type {};
 

flex/engines/hqps_db/core/utils/hqps_utils.h

@@ -847,6 +847,13 @@ struct to_string_impl<std::string> {
   }
 };
 
+template <>
+struct to_string_impl<LabelKey> {
+  static inline std::string to_string(const LabelKey& label_key) {
+    return std::to_string(label_key.label_id);
+  }
+};
+
 template <>
 struct to_string_impl<Direction> {
   static inline std::string to_string(const Direction& opt) {

flex/engines/hqps_db/structures/multi_edge_set/adj_edge_set.h

@@ -298,6 +298,12 @@ class AdjEdgeSet {
     return builder_t(src_label_, dst_label_, edge_label_, prop_names_, dir_);
   }
 
+  std::vector<LabelKey> GetLabelVec() const {
+    std::vector<LabelKey> label_vec(Size());
+    std::fill(label_vec.begin(), label_vec.end(), {edge_label_});
+    return label_vec;
+  }
+
   iterator begin() const { return iterator(vids_, adj_lists_, 0); }
 
   iterator end() const { return iterator(vids_, adj_lists_, vids_.size()); }
@@ -414,6 +420,12 @@ class AdjEdgeSet<GI, VID_T, LabelT, grape::EmptyType> {
 
   iterator end() const { return iterator(vids_, adj_lists_, vids_.size()); }
 
+  std::vector<LabelKey> GetLabelVec() const {
+    std::vector<LabelKey> label_vec(Size());
+    std::fill(label_vec.begin(), label_vec.end(), {edge_label_});
+    return label_vec;
+  }
+
   size_t Size() const { return size_; }
 
   template <typename EXPR, size_t num_labels>

flex/engines/hqps_db/structures/multi_edge_set/flat_edge_set.h

@@ -173,6 +173,16 @@ class FlatEdgeSet {
 
   iterator end() const { return iterator(vec_, vec_.size()); }
 
+  std::vector<LabelKey> GetLabelVec() const {
+    std::vector<LabelKey> res;
+    res.reserve(vec_.size());
+    for (auto i = 0; i < vec_.size(); ++i) {
+      auto ind = label_triplet_ind_[i];
+      res.emplace_back(label_triplet_[ind][2]);
+    }
+    return res;
+  }
+
   template <size_t col_ind, typename... index_ele_tuple_t_>
   flat_t Flat(
       std::vector<std::tuple<index_ele_tuple_t_...>>& index_ele_tuple) const {
@@ -498,6 +508,15 @@ class SingleLabelEdgeSet {
 
   iterator end() const { return iterator(vec_, vec_.size()); }
 
+  std::vector<LabelKey> GetLabelVec() const {
+    std::vector<LabelKey> res;
+    res.reserve(vec_.size());
+    for (auto i = 0; i < vec_.size(); ++i) {
+      res.emplace_back(label_triplet_[2]);
+    }
+    return res;
+  }
+
   template <size_t col_ind, typename... index_ele_tuple_t_>
   flat_t Flat(
       std::vector<std::tuple<index_ele_tuple_t_...>>& index_ele_tuple) const {

flex/engines/hqps_db/structures/multi_edge_set/general_edge_set.h

@@ -428,6 +428,15 @@ class GeneralEdgeSet<2, GI, VID_T, LabelT, std::tuple<T...>, std::tuple<T...>> {
 
   iterator end() const { return iterator(vids_, adj_lists_, vids_.size()); }
 
+  std::vector<LabelKey> GetLabelVec() const {
+    std::vector<LabelKey> res;
+    res.reserve(Size());
+    for (auto i = 0; i < Size(); ++i) {
+      res.emplace_back(edge_label_);
+    }
+    return res;
+  }
+
   size_t Size() const {
     if (size_ == 0) {
       for (auto i = 0; i < adj_lists_.size(); ++i) {
@@ -612,6 +621,15 @@ class GeneralEdgeSet<2, GI, VID_T, LabelT, std::tuple<grape::EmptyType>,
     bitsets_.swap(other.bitsets_);
   }
 
+  std::vector<LabelKey> GetLabelVec() const {
+    std::vector<LabelKey> res;
+    res.reserve(Size());
+    for (auto i = 0; i < Size(); ++i) {
+      res.emplace_back(edge_label_);
+    }
+    return res;
+  }
+
   iterator begin() const { return iterator(vids_, adj_lists_, 0); }
 
   iterator end() const { return iterator(vids_, adj_lists_, vids_.size()); }

flex/engines/hqps_db/structures/multi_edge_set/untyped_edge_set.h

@@ -195,6 +195,26 @@ class UnTypedEdgeSet {
                     src_vertices_.size());
   }
 
+  std::vector<LabelKey> GetLabelVec() const {
+    std::vector<LabelKey> res;
+    res.reserve(Size());
+    for (auto i = 0; i < src_vertices_.size(); ++i) {
+      auto label_ind = label_indices_[i];
+      auto label = src_labels_[label_ind];
+      if (adj_lists_.find(label) != adj_lists_.end()) {
+        auto& sub_graphs = adj_lists_.at(label);
+        for (auto& sub_graph : sub_graphs) {
+          auto edge_iters = sub_graph.get_edges(src_vertices_[i]);
+          auto edge_label = sub_graph.GetEdgeLabel();
+          for (auto j = 0; j < edge_iters.Size(); ++j) {
+            res.emplace_back(edge_label);
+          }
+        }
+      }
+    }
+    return res;
+  }
+
   size_t Size() const {
     if (size_ == 0) {
       auto iter_vec = generate_iters();

flex/engines/hqps_db/structures/multi_vertex_set/general_vertex_set.h

@@ -78,7 +78,7 @@ auto general_project_vertices_impl(
             break;
           }
         } else {
-          if (expr(eles)) {
+          if (expr(std::get<0>(eles))) {
             res_bitsets[label_id].set_bit(res_vec.size());
             res_vec.push_back(old_vec[i]);
             break;
@@ -381,6 +381,20 @@ class GeneralVertexSet {
 
   LabelT GetLabel(size_t i) const { return label_names_[i]; }
 
+  const std::vector<LabelKey> GetLabelVec() const {
+    std::vector<LabelKey> res;
+    // fill res with vertex labels
+    for (auto i = 0; i < vec_.size(); ++i) {
+      for (auto j = 0; j < bitsets_.size(); ++j) {
+        if (bitsets_[j].get_bit(i)) {
+          res.emplace_back(label_names_[j]);
+          break;
+        }
+      }
+    }
+    return res;
+  }
+
   // generate label indices.
   std::vector<uint8_t> GenerateLabelIndices() const {
     std::vector<uint8_t> label_indices;

flex/engines/hqps_db/structures/multi_vertex_set/keyed_row_vertex_set.h

@@ -569,6 +569,16 @@ class KeyedRowVertexSetImpl {
 
   LabelT GetLabel() const { return v_label_; }
 
+  const std::vector<LabelKey> GetLabelVec() {
+    std::vector<LabelKey> res;
+    // fill res with vertex labels
+    res.reserve(vids_.size());
+    for (auto i = 0; i < vids_.size(); ++i) {
+      res.emplace_back(v_label_);
+    }
+    return res;
+  }
+
   const std::array<std::string, sizeof...(T)>& GetPropNames() const {
     return prop_names_;
   }
@@ -768,6 +778,16 @@ class KeyedRowVertexSetImpl<LabelT, KEY_T, VID_T, grape::EmptyType> {
 
   LabelT GetLabel() const { return v_label_; }
 
+  const std::vector<LabelKey> GetLabelVec() {
+    std::vector<LabelKey> res;
+    // fill res with vertex labels
+    res.reserve(vids_.size());
+    for (auto i = 0; i < vids_.size(); ++i) {
+      res.emplace_back(v_label_);
+    }
+    return res;
+  }
+
   const std::vector<lid_t>& GetVertices() const { return vids_; }
 
   builder_t CreateBuilder() const { return builder_t(v_label_); }