Implementation feature list: InQueryCall, Unwind, Merge, YieldField, Multi-Match, With, Aggregate, Expression, Exists, cypher::FieldData of type map and List of run-through tests includes TestMultiMatch，Testparameter, TestWith, TestUnwind, TestAdd, TestDelete, TestMerge, TestCreatYago, TestAggregate, TestTopn, TestLdbcSnb, TestOpt, TestCreateLabel, TestEdgeIdQuery #561

jiazhenjiang · 2024-06-19T10:30:43Z

No description provided.

…raph-db into cypher_join_ast

multi match init

wangtao9 · 2024-06-24T08:17:19Z

deps/geax-front-end/include/geax-front-end/ast/AstNode.h

@@ -188,6 +188,8 @@ using StrArray = std::array<const char* const, N>;
    TYPE(MkRecord, kMkRecord, "MkRecord")                                                 \
    TYPE(MkSet, kMkSet, "MkSet")                                                          \
    TYPE(MkTuple, kMkTuple, "MkTuple")                                                    \
+    TYPE(UnwindStatement, kUnwindStatement, "UnwindStatement")                            \


unwind statement不能复用for statement吗

gql for包含 cypher unwind，可以复用

这块的计划是先把其他的都做完，然后再把unwind和for统一起来

wangtao9 · 2024-06-24T08:18:04Z

deps/geax-front-end/include/geax-front-end/ast/AstNode.h

@@ -188,6 +188,8 @@ using StrArray = std::array<const char* const, N>;
    TYPE(MkRecord, kMkRecord, "MkRecord")                                                 \
    TYPE(MkSet, kMkSet, "MkSet")                                                          \
    TYPE(MkTuple, kMkTuple, "MkTuple")                                                    \
+    TYPE(UnwindStatement, kUnwindStatement, "UnwindStatement")                            \
+    TYPE(InQueryProcedureCall, kInQueryProcedureCall, "InQueryProcedureCall")             \


in query call在gql ast中没有吗？在gql语法中呢

gql语法中只有带参数的函数调用和不带参数的函数调用。cypher中in query call本质和procedure call在ast层面是同一个东西，但是在执行计划里因为区分了两种算子，所以需要区分是procedure还是inquery call，暂时通过新增一个ast结构做为区分，后续计划是从算子层面统一。

wangtao9 · 2024-06-24T08:21:11Z

deps/geax-front-end/include/geax-front-end/ast/expr/MkList.h

@@ -18,6 +18,8 @@
 #ifndef GEAXFRONTEND_AST_EXPR_MKLIST_H_
 #define GEAXFRONTEND_AST_EXPR_MKLIST_H_

+#include <vector>


这些新增的头文件符合cpplint吗

这里用到了std::vector,不添加头文件，在后续的编译中会出问题。

wangtao9 · 2024-06-24T08:27:41Z

src/cypher/arithmetic/ast_agg_expr_detector.h

@@ -44,7 +44,8 @@ class AstAggExprDetector : public cypher::AstNodeVisitorImpl {

 private:
    std::any visit(geax::frontend::AggFunc* node) override;
-    std::any visit(geax::frontend::Ref* node) override;
+    std::any visit(geax::frontend::BAggFunc* node) override;


这里删掉了visit Ref，举个例子在处理什么情况

这里是删除了Ref新增了GetField表达式，为了区分 MATCH (n:City) RETURN collect(n.name) + n.name;
MATCH (n:Person) RETURN avg(n.age) + 100;

wangtao9 · 2024-06-24T08:31:02Z

src/cypher/arithmetic/ast_agg_expr_detector.h

@@ -44,7 +44,8 @@ class AstAggExprDetector : public cypher::AstNodeVisitorImpl {

 private:
    std::any visit(geax::frontend::AggFunc* node) override;
-    std::any visit(geax::frontend::Ref* node) override;
+    std::any visit(geax::frontend::BAggFunc* node) override;


这里删掉了visit Ref，举个例子在处理什么情况

这里是删除了Ref新增了GetField表达式，为了区分 MATCH (n:City) RETURN collect(n.name) + n.name;
MATCH (n:Person) RETURN avg(n.age) + 100;

wangtao9 · 2024-06-24T08:40:42Z

src/cypher/cypher_types.h

+            delete array;
+            break;
+        case MAP:
+            delete array;


wangtao9 · 2024-06-24T09:03:12Z

src/cypher/cypher_types.h

        type = rhs.type;
-        if (rhs.type == ARRAY) {
+
+        switch (type) {


两个switch合并一起做

这里要先根据this.type回收资源，然后再根据rhs.type分配资源，合并起来代码可读性很差，不如这样清晰

wangtao9 · 2024-06-24T09:03:20Z

src/cypher/cypher_types.h

-        scalar = std::move(rhs.scalar);
-        array = rhs.array;
-        // rhs.scalar becomes FieldType::NUL after move
+        switch (type) {


这里要先根据this.type回收资源，然后再根据rhs.type分配资源，合并起来代码可读性很差，不如这样清晰

wangtao9 · 2024-06-24T09:03:48Z

src/cypher/cypher_types.h

+            scalar = std::move(rhs.scalar);
+            break;
+        }
+        // TODO(lingsu) : rhs should return to the default state
        rhs.type = SCALAR;
        return *this;
    }

    bool operator==(const FieldData& rhs) const {
        if (type != rhs.type)
            throw std::runtime_error("Unable to compare between SCALAR and ARRAY.");


msg要改，比如 between different field data types

wangtao9 · 2024-06-24T09:04:43Z

src/cypher/cypher_types.h

@@ -118,16 +212,30 @@ struct FieldData {
    bool operator>(const FieldData& rhs) const {
        if (type != rhs.type)
            throw std::runtime_error("Unable to compare between SCALAR and ARRAY.");


wangtao9 · 2024-06-24T09:14:52Z

src/cypher/arithmetic/ast_expr_evaluator.cpp

    if (agg_it != registered_agg_funcs.end()) {
        // Evalute Mode
        if (visit_mode_ == VisitMode::EVALUATE) {
            if (agg_pos_ >= agg_ctxs_.size()) {
-                NOT_SUPPORT_AND_THROW();
+                return Entry(cypher::FieldData(lgraph_api::FieldData(0)));


举个例子，什么情况会走到这个分支

我理解是某个aggfunction在统计时，结果为空，代码会执行到这个分支，之前的case里是返回了0，因此这里也返回0

之前的test case里返回的是0，这里是兼容之前的case

wangtao9 · 2024-06-24T09:24:45Z

src/cypher/execution_plan/execution_plan_v2.cpp

    geax::frontend::GEAXErrorCode ret = geax::frontend::GEAXErrorCode::GEAX_SUCCEED;
    // build pattern graph
    PatternGraphMaker pattern_graph_maker(pattern_graphs_);
-    ret = pattern_graph_maker.Build(astNode);
+    ret = pattern_graph_maker.Build(astNode, ctx);


传入ctx什么作用

为了在PatternGraphMaker中处理YieldField时判断是否是正确的返回值

wangtao9 · 2024-06-25T01:30:44Z

src/cypher/rewriter/MultiPathPatternRewriter.h

+ public:
+    explicit MultiPathPatternRewriter(geax::common::ObjectArenaAllocator& allocator)
+        : allocator_(allocator) {}
+    std::any visit(geax::frontend::GraphPattern* node) override {


这个是把什么重写成什么了？举个例子

MATCH (x)<-[:ACTED_IN]-(p)-[:MARRIED]->(y), (p)-[:HAS_CHILD]->(z) RETURN p,x,y,z; 把多段pattern中相同的点挑出来，通过filter确定相同的点，执行的时候对同一个点的查找只执行一次。例子中就是将p挑出来了

…raph-db into complete_branch

spasserby · 2024-06-26T02:40:43Z

test/test_cypher_v2.cpp

@@ -202,7 +205,7 @@ class TestCypherV2 : public TuGraphTest {
                } catch (std::exception& e) {
                    UT_LOG() << e.what();
                    result = e.what();
-                    return false;
+                    return true;


为什么改动

之前的test case里有一些就是要抛异常，这里如果不改，抛异常就被视为失败了

spasserby · 2024-06-26T03:15:59Z

src/cypher/arithmetic/ast_expr_evaluator.cpp

    if (agg_it != registered_agg_funcs.end()) {
        // Evalute Mode
        if (visit_mode_ == VisitMode::EVALUATE) {
            if (agg_pos_ >= agg_ctxs_.size()) {
-                NOT_SUPPORT_AND_THROW();
+                return Entry(cypher::FieldData(lgraph_api::FieldData(0)));


spasserby · 2024-06-26T03:39:32Z

src/cypher/arithmetic/ast_expr_evaluator.cpp

@@ -318,6 +332,36 @@ std::any cypher::AstExprEvaluator::visit(geax::frontend::AggFunc* node) {
 }

 std::any cypher::AstExprEvaluator::visit(geax::frontend::BAggFunc* node) {


这部分的作用是？

gql中aggfunction分为单参数和双参数的，cypher没区分这些，因此只用agg不完整，所以加了这个

spasserby · 2024-06-26T03:40:39Z

test/resource/unit_test/topn/cypher/topn.result

@@ -5,7 +5,7 @@ match (v1:Film) return v1.title order by v1.title limit 3;
 match (v1:Film)<-[:ACTED_IN|DIRECTED]-(v2:Person) return v1.title,v2.name as cnt order by cnt desc limit 3;
 [{"cnt":"Vanessa Redgrave","v1.title":"Camelot"},{"cnt":"Richard Harris","v1.title":"Camelot"},{"cnt":"Richard Harris","v1.title":"Harry Potter and the Sorcerer's Stone"}]
 match (:Person {name:'Vanessa Redgrave'})<-[:HAS_CHILD]-(p)-[:ACTED_IN*0..]->(m) return p.name,m order by p.name limit 3;
-[{"m":{"identity":16,"label":"Film","properties":{"title":"Goodbye, Mr. Chips"}},"p.name":"Michael Redgrave"},{"m":{"identity":1,"label":"Person","properties":{"birthyear":1908,"name":"Michael Redgrave"}},"p.name":"Michael Redgrave"},{"m":{"identity":0,"label":"Person","properties":{"birthyear":1910,"name":"Rachel Kempson"}},"p.name":"Rachel Kempson"}]
+[{"m":{"identity":1,"label":"Person","properties":{"birthyear":1908,"name":"Michael Redgrave"}},"p.name":"Michael Redgrave"},{"m":{"identity":16,"label":"Film","properties":{"title":"Goodbye, Mr. Chips"}},"p.name":"Michael Redgrave"},{"m":{"identity":0,"label":"Person","properties":{"birthyear":1910,"name":"Rachel Kempson"}},"p.name":"Rachel Kempson"}]


结果变化？

理论上查询中是order by p.name，因此结果中p.name相同的其他部分，因为未指明排序规则，按照neo4j语法，是无序的

spasserby · 2024-06-26T16:07:44Z

src/cypher/cypher_types.h

            scalar = rhs.scalar;
+            break;
        }
        return *this;
    }

    FieldData& operator=(::cypher::FieldData&& rhs) {


cypher_types多补一些单测，这里有内存泄露
注意一下map和array，可能原来就有

测例比较简单：
std::unordered_map<std::string, lgraph::FieldData> fields;
fields["a"] = lgraph::FieldData();
cypher::FieldData f1(fields);
cypher::FieldData f2;
f2 = std::move(f1);

jiazhenjiang and others added 30 commits April 1, 2024 15:02

add cypher test compare result

6b4bb7d

Merge branch 'master' into cypher_join_ast

bfa33dd

implement query

cad4c0e

Merge branch 'master' into cypher_join_ast

5d8f2b0

implement match query

50834b1

Merge branch 'master' into cypher_join_ast

8a401fe

Merge branch 'master' into cypher_join_ast

e6dcbeb

implement standalone call and create

3a6f99b

Merge branch 'master' into cypher_join_ast

9b8f1f3

Merge branch 'cypher_join_ast' of https://github.com/jiazhenjiang/tug…

d777633

…raph-db into cypher_join_ast

cpplint

f9e3e4f

cpplint

87e28cf

cpplint

85d95f0

fix double format

271348f

update

37855fb

update

95271e6

implement set

e85b571

implement gql set

df86ed6

Merge branch 'master' into cypher_join_ast

745b791

fix test case

bc09ac1

move dml

15037cc

cpplint

918806e

cpplint

285b8aa

add create delete set

93cd07b

implement unwind

e237903

fix bug

83e342a

implement count(*)

729c1bb

implement expression

f71af20

multi match init

3f27530

Merge pull request #1 from spasserby/multi_match

29a6aed

multi match init

jiazhenjiang and others added 11 commits June 14, 2024 03:36

fix bug

bb5368b

fix bug

aa1b684

implement merge

ecae16b

implement merge

d021e60

fix bug in merge clause

3e3ad93

fix bug

98a8165

merge master

cb5219b

fix bug

c949fe4

cpplint

4e446a4

cpplint

2ecc8a7

Merge branch 'master' into complete_branch

7a71e17

lipanpan03 enabled auto-merge (squash) June 24, 2024 07:31

lipanpan03 approved these changes Jun 24, 2024

View reviewed changes

lipanpan03 disabled auto-merge June 24, 2024 07:32

wangtao9 reviewed Jun 24, 2024

View reviewed changes

jiazhenjiang changed the title ~~Implement cypher query statements in addition to profile， union， remove and list_comprehension~~ Implement Jun 24, 2024

jiazhenjiang changed the title ~~Implement~~ Implementation feature list: InQueryCall, Unwind, Merge, YieldField, Multi-Match, With, Aggregate, Expression, Exists and cypher::FieldData of type map Jun 24, 2024

wangtao9 reviewed Jun 25, 2024

View reviewed changes

jiazhenjiang added 2 commits June 25, 2024 08:17

comments

5e1030d

Merge branch 'complete_branch' of https://github.com/jiazhenjiang/tug…

ce82ac9

…raph-db into complete_branch

ljcui approved these changes Jun 25, 2024

View reviewed changes

spasserby reviewed Jun 26, 2024

View reviewed changes

jiazhenjiang and others added 3 commits June 27, 2024 03:30

comments

8c6023b

cpplint

39edc26

Merge branch 'master' into complete_branch

31c1a32

spasserby approved these changes Jun 27, 2024

View reviewed changes

spasserby merged commit 6e5585d into TuGraph-family:master Jun 27, 2024
5 checks passed

		@@ -318,6 +332,36 @@ std::any cypher::AstExprEvaluator::visit(geax::frontend::AggFunc* node) {
		}

		std::any cypher::AstExprEvaluator::visit(geax::frontend::BAggFunc* node) {

Conversation

jiazhenjiang commented Jun 19, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment