fix(codegen): handle nullable for date type (#3543)

* fix(codegen): handle nullable for date type Also fix include headers * fix(codegen): handle nulls for date and timestamp always returns allocated structure even it is null, for the cases: 1. construct timestamp from number, even number is < 0 2. construct date from timestamp, even timestamp is null
4paradigm · Oct 23, 2023 · 6fe2a30 · 6fe2a30
1 parent 190992d
commit 6fe2a30
Show file tree

Hide file tree

Showing 18 changed files with 76 additions and 83 deletions.
diff --git a/cases/query/const_query.yaml b/cases/query/const_query.yaml
@@ -126,3 +126,15 @@ cases:
       columns: ['c1 bool', 'c2 int16', 'c3 int', 'c4 double', 'c5 string', 'c6 date', 'c7 timestamp' ]
       rows:
         - [ true, 3, 13, 10.0, 'a string', '2020-05-22', 1590115420000 ]
+  - id: 10
+    mode: procedure-unsupport
+    sql: |
+      select
+        datediff(Date(timestamp(-1)), Date("2021-05-01")) as out1,
+        datediff(Date(timestamp(-2177481600)), Date("2021-05-01")) as out2,
+        datediff(cast(NULL as date), Date("2021-05-01")) as out3
+        ;
+    expect:
+      columns: ["out1 int", "out2 int", "out3 int"]
+      data: |
+        NULL, NULL, NULL
diff --git a/hybridse/src/codegen/array_ir_builder.cc b/hybridse/src/codegen/array_ir_builder.cc
@@ -17,6 +17,7 @@
 #include "codegen/array_ir_builder.h"
 
 #include <string>
+#include "codegen/ir_base_builder.h"
 
 namespace hybridse {
 namespace codegen {

diff --git a/hybridse/src/codegen/cast_expr_ir_builder.cc b/hybridse/src/codegen/cast_expr_ir_builder.cc
@@ -126,6 +126,11 @@ Status CastExprIRBuilder::UnSafeCast(const NativeValue& value,
             StringIRBuilder string_ir_builder(block_->getModule());
             CHECK_STATUS(string_ir_builder.CreateNull(block_, output));
             return base::Status::OK();
+
+        } else if (TypeIRBuilder::IsDatePtr(type)) {
+            DateIRBuilder date_ir(block_->getModule());
+            CHECK_STATUS(date_ir.CreateNull(block_, output));
+            return base::Status::OK();
         } else {
             *output = NativeValue::CreateNull(type);
         }

diff --git a/hybridse/src/codegen/date_ir_builder.cc b/hybridse/src/codegen/date_ir_builder.cc
@@ -19,6 +19,7 @@
 #include <vector>
 #include "codegen/arithmetic_expr_ir_builder.h"
 #include "codegen/ir_base_builder.h"
+#include "codegen/null_ir_builder.h"
 
 namespace hybridse {
 namespace codegen {
@@ -43,6 +44,15 @@ void DateIRBuilder::InitStructType() {
     struct_type_ = stype;
     return;
 }
+
+base::Status DateIRBuilder::CreateNull(::llvm::BasicBlock* block, NativeValue* output) {
+    ::llvm::Value* value = nullptr;
+    CHECK_TRUE(CreateDefault(block, &value), common::kCodegenError, "Fail to construct string")
+    ::llvm::IRBuilder<> builder(block);
+    *output = NativeValue::CreateWithFlag(value, builder.getInt1(true));
+    return base::Status::OK();
+}
+
 bool DateIRBuilder::CreateDefault(::llvm::BasicBlock* block,
                                   ::llvm::Value** output) {
     return NewDate(block, output);
@@ -123,11 +133,10 @@ base::Status DateIRBuilder::CastFrom(::llvm::BasicBlock* block,
         auto cast_func = m_->getOrInsertFunction(
             fn_name,
             ::llvm::FunctionType::get(builder.getVoidTy(),
-                                      {src.GetType(), dist->getType(),
-                                       builder.getInt1Ty()->getPointerTo()},
-                                      false));
-        builder.CreateCall(cast_func,
-                           {src.GetValue(&builder), dist, is_null_ptr});
+                                      {src.GetType(), dist->getType(), builder.getInt1Ty()->getPointerTo()}, false));
+
+        builder.CreateCall(cast_func, {src.GetValue(&builder), dist, is_null_ptr});
+
         ::llvm::Value* should_return_null = builder.CreateLoad(is_null_ptr);
         null_ir_builder.CheckAnyNull(block, src, &should_return_null);
         *output = NativeValue::CreateWithFlag(dist, should_return_null);

diff --git a/hybridse/src/codegen/date_ir_builder.h b/hybridse/src/codegen/date_ir_builder.h
@@ -16,13 +16,9 @@
 
 #ifndef HYBRIDSE_SRC_CODEGEN_DATE_IR_BUILDER_H_
 #define HYBRIDSE_SRC_CODEGEN_DATE_IR_BUILDER_H_
+
 #include "base/fe_status.h"
-#include "codegen/cast_expr_ir_builder.h"
-#include "codegen/null_ir_builder.h"
-#include "codegen/scope_var.h"
 #include "codegen/struct_ir_builder.h"
-#include "llvm/IR/IRBuilder.h"
-#include "proto/fe_type.pb.h"
 
 namespace hybridse {
 namespace codegen {
@@ -31,17 +27,17 @@ class DateIRBuilder : public StructTypeIRBuilder {
  public:
     explicit DateIRBuilder(::llvm::Module* m);
     ~DateIRBuilder();
-    void InitStructType();
-    bool CreateDefault(::llvm::BasicBlock* block, ::llvm::Value** output);
+    void InitStructType() override;
+
+    base::Status CreateNull(::llvm::BasicBlock* block, NativeValue* output);
+    bool CreateDefault(::llvm::BasicBlock* block, ::llvm::Value** output) override;
+
     bool NewDate(::llvm::BasicBlock* block, ::llvm::Value** output);
     bool NewDate(::llvm::BasicBlock* block, ::llvm::Value* date,
                  ::llvm::Value** output);
-    bool CopyFrom(::llvm::BasicBlock* block, ::llvm::Value* src,
-                  ::llvm::Value* dist);
-    base::Status CastFrom(::llvm::BasicBlock* block, const NativeValue& src,
-                          NativeValue* output);
-    base::Status CastFrom(::llvm::BasicBlock* block, ::llvm::Value* src,
-                          ::llvm::Value** output);
+    bool CopyFrom(::llvm::BasicBlock* block, ::llvm::Value* src, ::llvm::Value* dist);
+    base::Status CastFrom(::llvm::BasicBlock* block, const NativeValue& src, NativeValue* output);
+
     bool GetDate(::llvm::BasicBlock* block, ::llvm::Value* date,
                  ::llvm::Value** output);
     bool SetDate(::llvm::BasicBlock* block, ::llvm::Value* date,

diff --git a/hybridse/src/codegen/ir_base_builder.h b/hybridse/src/codegen/ir_base_builder.h
@@ -19,7 +19,6 @@
 
 #include <string>
 #include <vector>
-#include "glog/logging.h"
 #include "llvm/IR/IRBuilder.h"
 #include "node/sql_node.h"
 #include "node/type_node.h"

diff --git a/hybridse/src/codegen/native_value.cc b/hybridse/src/codegen/native_value.cc
@@ -17,7 +17,6 @@
 #include "codegen/native_value.h"
 #include <dlfcn.h>
 #include <execinfo.h>
-#include <signal.h>
 #include "codegen/context.h"
 #include "codegen/ir_base_builder.h"
 

diff --git a/hybridse/src/codegen/predicate_expr_ir_builder.cc b/hybridse/src/codegen/predicate_expr_ir_builder.cc
@@ -17,6 +17,7 @@
 #include "codegen/predicate_expr_ir_builder.h"
 #include "codegen/date_ir_builder.h"
 #include "codegen/ir_base_builder.h"
+#include "codegen/null_ir_builder.h"
 #include "codegen/string_ir_builder.h"
 #include "codegen/timestamp_ir_builder.h"
 #include "codegen/type_ir_builder.h"

diff --git a/hybridse/src/codegen/struct_ir_builder.h b/hybridse/src/codegen/struct_ir_builder.h
@@ -16,12 +16,10 @@
 
 #ifndef HYBRIDSE_SRC_CODEGEN_STRUCT_IR_BUILDER_H_
 #define HYBRIDSE_SRC_CODEGEN_STRUCT_IR_BUILDER_H_
+
 #include "base/fe_status.h"
-#include "codegen/cast_expr_ir_builder.h"
-#include "codegen/scope_var.h"
+#include "codegen/native_value.h"
 #include "codegen/type_ir_builder.h"
-#include "llvm/IR/IRBuilder.h"
-#include "proto/fe_type.pb.h"
 
 namespace hybridse {
 namespace codegen {

diff --git a/hybridse/src/codegen/timestamp_ir_builder.cc b/hybridse/src/codegen/timestamp_ir_builder.cc
@@ -15,14 +15,15 @@
  */
 
 #include "codegen/timestamp_ir_builder.h"
+
 #include <string>
 #include <vector>
+
 #include "codegen/arithmetic_expr_ir_builder.h"
 #include "codegen/ir_base_builder.h"
 #include "codegen/null_ir_builder.h"
 #include "codegen/predicate_expr_ir_builder.h"
 #include "glog/logging.h"
-#include "node/sql_node.h"
 
 using hybridse::common::kCodegenError;
 
@@ -70,29 +71,30 @@ base::Status TimestampIRBuilder::CastFrom(::llvm::BasicBlock* block,
     CondSelectIRBuilder cond_ir_builder;
     PredicateIRBuilder predicate_ir_builder(block);
     NullIRBuilder null_ir_builder;
+
+    // always allocate for returned timestmap even it is null
+    ::llvm::Value* dist = nullptr;
+    if (!CreateDefault(block, &dist)) {
+        status.code = common::kCodegenError;
+        status.msg = "Fail to cast date: create default date fail";
+        return status;
+    }
+
     if (IsNumber(src.GetType())) {
         CHECK_STATUS(cast_builder.Cast(src, builder.getInt64Ty(), &ts));
         NativeValue cond;
         CHECK_STATUS(predicate_ir_builder.BuildGeExpr(
             ts, NativeValue::Create(builder.getInt64(0)), &cond));
-        ::llvm::Value* timestamp;
-        CHECK_TRUE(NewTimestamp(block, ts.GetValue(&builder), &timestamp),
+        CHECK_TRUE(SetTs(block, dist, ts.GetValue(&builder)),
                    kCodegenError,
                    "Fail to cast timestamp: new timestamp(ts) fail");
-        CHECK_STATUS(
-            cond_ir_builder.Select(block, cond, NativeValue::Create(timestamp),
-                                   NativeValue::CreateNull(GetType()), output));
+        CHECK_STATUS(cond_ir_builder.Select(block, cond, NativeValue::Create(dist),
+                                            NativeValue::CreateWithFlag(dist, builder.getInt1(true)), output));
 
     } else if (IsStringPtr(src.GetType()) || IsDatePtr(src.GetType())) {
         ::llvm::IRBuilder<> builder(block);
-        ::llvm::Value* dist = nullptr;
         ::llvm::Value* is_null_ptr = CreateAllocaAtHead(
             &builder, builder.getInt1Ty(), "timestamp_is_null_alloca");
-        if (!CreateDefault(block, &dist)) {
-            status.code = common::kCodegenError;
-            status.msg = "Fail to cast date: create default date fail";
-            return status;
-        }
         ::std::string fn_name = "timestamp." + TypeName(src.GetType());
 
         auto cast_func = m_->getOrInsertFunction(

diff --git a/hybridse/src/codegen/timestamp_ir_builder.h b/hybridse/src/codegen/timestamp_ir_builder.h
@@ -16,12 +16,9 @@
 
 #ifndef HYBRIDSE_SRC_CODEGEN_TIMESTAMP_IR_BUILDER_H_
 #define HYBRIDSE_SRC_CODEGEN_TIMESTAMP_IR_BUILDER_H_
+
 #include "base/fe_status.h"
-#include "codegen/cast_expr_ir_builder.h"
-#include "codegen/scope_var.h"
 #include "codegen/struct_ir_builder.h"
-#include "llvm/IR/IRBuilder.h"
-#include "proto/fe_type.pb.h"
 
 namespace hybridse {
 namespace codegen {
@@ -33,8 +30,8 @@ class TimestampIRBuilder : public StructTypeIRBuilder {
     void InitStructType();
     bool CreateDefault(::llvm::BasicBlock* block, ::llvm::Value** output);
     bool NewTimestamp(::llvm::BasicBlock* block, ::llvm::Value** output);
-    bool NewTimestamp(::llvm::BasicBlock* block, ::llvm::Value* ts,
-                      ::llvm::Value** output);
+    bool NewTimestamp(::llvm::BasicBlock* block, ::llvm::Value* ts, ::llvm::Value** output);
+
     bool CopyFrom(::llvm::BasicBlock* block, ::llvm::Value* src,
                   ::llvm::Value* dist);
     base::Status CastFrom(::llvm::BasicBlock* block, const NativeValue& src,

diff --git a/hybridse/src/codegen/type_ir_builder.cc b/hybridse/src/codegen/type_ir_builder.cc
@@ -15,6 +15,7 @@
  */
 
 #include "codegen/type_ir_builder.h"
+
 #include "codegen/ir_base_builder.h"
 #include "glog/logging.h"
 #include "node/node_manager.h"

diff --git a/hybridse/src/codegen/type_ir_builder.h b/hybridse/src/codegen/type_ir_builder.h
@@ -18,11 +18,10 @@
 #define HYBRIDSE_SRC_CODEGEN_TYPE_IR_BUILDER_H_
 
 #include <string>
-#include <vector>
+
 #include "base/fe_status.h"
-#include "codec/fe_row_codec.h"
-#include "codegen/ir_base_builder.h"
-#include "node/node_enum.h"
+#include "llvm/IR/Module.h"
+#include "llvm/IR/Type.h"
 #include "node/sql_node.h"
 #include "node/type_node.h"
 

diff --git a/hybridse/src/codegen/udf_ir_builder.cc b/hybridse/src/codegen/udf_ir_builder.cc
@@ -15,19 +15,17 @@
  */
 
 #include "codegen/udf_ir_builder.h"
-#include <iostream>
-#include <utility>
+
 #include <vector>
+
 #include "codegen/context.h"
-#include "codegen/date_ir_builder.h"
 #include "codegen/fn_ir_builder.h"
+#include "codegen/ir_base_builder.h"
 #include "codegen/list_ir_builder.h"
 #include "codegen/null_ir_builder.h"
-#include "codegen/timestamp_ir_builder.h"
+#include "codegen/type_ir_builder.h"
 #include "llvm/IR/Attributes.h"
-#include "node/node_manager.h"
 #include "node/sql_node.h"
-#include "udf/udf.h"
 #include "udf/udf_registry.h"
 
 using ::hybridse::common::kCodegenError;
@@ -162,7 +160,7 @@ Status UdfIRBuilder::BuildLambdaCall(
 Status UdfIRBuilder::BuildCodeGenUdfCall(
     const node::UdfByCodeGenDefNode* fn,
     const std::vector<NativeValue>& args, NativeValue* output) {
-    auto gen_impl = fn->GetGenImpl();
+    std::shared_ptr<udf::LlvmUdfGenBase> gen_impl = fn->GetGenImpl();
 
     ::llvm::Value* ret_null = nullptr;
     for (size_t i = 0; i < fn->GetArgSize(); ++i) {

diff --git a/hybridse/src/codegen/udf_ir_builder.h b/hybridse/src/codegen/udf_ir_builder.h
@@ -17,13 +17,9 @@
 #ifndef HYBRIDSE_SRC_CODEGEN_UDF_IR_BUILDER_H_
 #define HYBRIDSE_SRC_CODEGEN_UDF_IR_BUILDER_H_
 
-#include <map>
-#include <string>
 #include <vector>
 #include "base/fe_status.h"
 #include "codegen/expr_ir_builder.h"
-#include "codegen/scope_var.h"
-#include "llvm/IR/Module.h"
 
 #include "node/sql_node.h"
 namespace hybridse {

diff --git a/hybridse/src/udf/default_udf_library.cc b/hybridse/src/udf/default_udf_library.cc
@@ -26,7 +26,6 @@
 #include <queue>
 #include <functional>
 
-#include "absl/strings/str_cat.h"
 #include "codegen/date_ir_builder.h"
 #include "codegen/string_ir_builder.h"
 #include "codegen/timestamp_ir_builder.h"
@@ -2169,11 +2168,7 @@ void DefaultUdfLibrary::InitTypeUdf() {
         )");
 
     RegisterExternal("date")
-        .args<Timestamp>(reinterpret_cast<void*>(
-            static_cast<void (*)(Timestamp*, Date*, bool*)>(
-                v1::timestamp_to_date)))
-        .return_by_arg(true)
-        .returns<Nullable<Date>>()
+        .args<Timestamp>(v1::timestamp_to_date)
         .doc(R"(
             @brief Cast timestamp or string expression to date (date >= 1900-01-01)
 
@@ -2192,11 +2187,7 @@ void DefaultUdfLibrary::InitTypeUdf() {
             @endcode
             @since 0.1.0)");
     RegisterExternal("date")
-        .args<StringRef>(reinterpret_cast<void*>(
-            static_cast<void (*)(StringRef*, Date*, bool*)>(
-                v1::string_to_date)))
-        .return_by_arg(true)
-        .returns<Nullable<Date>>();
+        .args<StringRef>(v1::string_to_date);
 
     RegisterExternal("timestamp")
         .args<Date>(reinterpret_cast<void*>(

diff --git a/hybridse/src/udf/udf.cc b/hybridse/src/udf/udf.cc
@@ -20,29 +20,24 @@
 #include <time.h>
 
 #include <ctime>
-#include <map>
-#include <set>
 #include <utility>
 
 #include "absl/strings/ascii.h"
 #include "absl/strings/str_replace.h"
 #include "absl/time/civil_time.h"
 #include "absl/time/time.h"
 #include "base/iterator.h"
-#include "boost/date_time.hpp"
+#include "boost/date_time/gregorian/conversion.hpp"
 #include "boost/date_time/gregorian/parsers.hpp"
-#include "boost/date_time/posix_time/posix_time.hpp"
 #include "bthread/types.h"
-#include "codec/list_iterator_codec.h"
 #include "codec/row.h"
 #include "codec/type_codec.h"
-#include "codegen/fn_ir_builder.h"
 #include "farmhash.h"
 #include "node/node_manager.h"
 #include "node/sql_node.h"
 #include "re2/re2.h"
-#include "udf/default_udf_library.h"
 #include "udf/literal_traits.h"
+#include "udf/udf_library.h"
 #include "vm/jit_runtime.h"
 
 namespace hybridse {
@@ -394,8 +389,7 @@ void bool_to_string(bool v, StringRef *output) {
     }
 }
 
-void timestamp_to_date(Timestamp *timestamp,
-                       Date *output, bool *is_null) {
+void timestamp_to_date(Timestamp *timestamp, Date *output, bool *is_null) {
     time_t time = (timestamp->ts_ + TZ_OFFSET) / 1000;
     struct tm t;
     memset(&t, 0, sizeof(struct tm));
@@ -771,8 +765,7 @@ void string_to_double(StringRef *str, double *out, bool *is_null_ptr) {
     }
     return;
 }
-void string_to_date(StringRef *str, Date *output,
-                    bool *is_null) {
+void string_to_date(StringRef *str, Date *output, bool *is_null) {
     if (19 == str->size_) {
         struct tm timeinfo;
         memset(&timeinfo, 0, sizeof(struct tm));