duckdb 0.7.2-dev12.0 → 0.7.2-dev1244.0

package/src/duckdb/third_party/re2/re2/re2.cc

@@ -566,6 +566,15 @@ static int ascii_strcasecmp(const char* a, const char* b, size_t len) {
   return 0;
 }
 
+RE2::Anchor RE2::Anchored() const {
+  if (prog_->anchor_start()) {
+    if (prog_->anchor_end()) {
+      return Anchor::ANCHOR_BOTH;
+    }
+    return Anchor::ANCHOR_START;
+  }
+  return Anchor::UNANCHORED;
+}
 
 /***** Actual matching and rewriting code *****/
 

package/src/duckdb/third_party/re2/re2/re2.h

@@ -475,6 +475,8 @@ class RE2 {
     ANCHOR_BOTH         // Anchor at start and end
   };
 
+  Anchor Anchored() const;
+
   // Return the number of capturing subpatterns, or -1 if the
   // regexp wasn't valid on construction.  The overall match ($0)
   // does not count: if the regexp is "(a)(b)", returns 2.

package/src/duckdb/ub_extension_icu_third_party_icu_i18n.cpp

@@ -350,15 +350,15 @@
 
 #include "extension/icu/third_party/icu/i18n/double-conversion-strtod.cpp"
 
+#include "extension/icu/third_party/icu/i18n/double-conversion-bignum-dtoa.cpp"
+
 #include "extension/icu/third_party/icu/i18n/double-conversion-string-to-double.cpp"
 
 #include "extension/icu/third_party/icu/i18n/double-conversion-bignum.cpp"
 
-#include "extension/icu/third_party/icu/i18n/double-conversion-fast-dtoa.cpp"
+#include "extension/icu/third_party/icu/i18n/double-conversion-cached-powers.cpp"
 
 #include "extension/icu/third_party/icu/i18n/double-conversion-double-to-string.cpp"
 
-#include "extension/icu/third_party/icu/i18n/double-conversion-bignum-dtoa.cpp"
-
-#include "extension/icu/third_party/icu/i18n/double-conversion-cached-powers.cpp"
+#include "extension/icu/third_party/icu/i18n/double-conversion-fast-dtoa.cpp"
 

package/src/duckdb/ub_extension_json_json_functions.cpp

@@ -20,6 +20,8 @@
 
 #include "extension/json/json_functions/json_valid.cpp"
 
+#include "extension/json/json_functions/json_serialize_sql.cpp"
+
 #include "extension/json/json_functions/read_json.cpp"
 
 #include "extension/json/json_functions/read_json_objects.cpp"

package/src/duckdb/ub_src_common_serializer.cpp

@@ -6,3 +6,5 @@
 
 #include "src/common/serializer/buffered_serializer.cpp"
 
+#include "src/common/serializer/enum_serializer.cpp"
+

package/src/duckdb/ub_src_common_types.cpp

@@ -36,6 +36,8 @@
 
 #include "src/common/types/interval.cpp"
 
+#include "src/common/types/list_segment.cpp"
+
 #include "src/common/types/partitioned_column_data.cpp"
 
 #include "src/common/types/row_data_collection.cpp"

package/src/duckdb/ub_src_execution_physical_plan.cpp

@@ -2,6 +2,8 @@
 
 #include "src/execution/physical_plan/plan_any_join.cpp"
 
+#include "src/execution/physical_plan/plan_asof_join.cpp"
+
 #include "src/execution/physical_plan/plan_column_data_get.cpp"
 
 #include "src/execution/physical_plan/plan_comparison_join.cpp"

package/src/duckdb/ub_src_function_aggregate_distributive.cpp

@@ -2,6 +2,8 @@
 
 #include "src/function/aggregate/distributive/bitagg.cpp"
 
+#include "src/function/aggregate/distributive/bitstring_agg.cpp"
+
 #include "src/function/aggregate/distributive/count.cpp"
 
 #include "src/function/aggregate/distributive/first.cpp"

package/src/duckdb/ub_src_function_scalar_bit.cpp

@@ -0,0 +1,2 @@
+#include "src/function/scalar/bit/bitstring.cpp"
+

package/src/duckdb/ub_src_function_scalar_map.cpp

@@ -4,5 +4,9 @@
 
 #include "src/function/scalar/map/map_from_entries.cpp"
 
+#include "src/function/scalar/map/map_entries.cpp"
+
+#include "src/function/scalar/map/map_keys_values.cpp"
+
 #include "src/function/scalar/map/cardinality.cpp"
 

package/src/duckdb/ub_src_function_scalar_string.cpp

@@ -58,3 +58,5 @@
 
 #include "src/function/scalar/string/jaro_winkler.cpp"
 
+#include "src/function/scalar/string/hex.cpp"
+

package/src/duckdb/ub_src_function_scalar_string_regexp.cpp

@@ -0,0 +1,4 @@
+#include "src/function/scalar/string/regexp/regexp_util.cpp"
+
+#include "src/function/scalar/string/regexp/regexp_extract_all.cpp"
+

package/src/duckdb/ub_src_main_capi.cpp

@@ -26,6 +26,8 @@
 
 #include "src/main/capi/result-c.cpp"
 
+#include "src/main/capi/stream-c.cpp"
+
 #include "src/main/capi/table_function-c.cpp"
 
 #include "src/main/capi/threading-c.cpp"

package/src/duckdb/ub_src_optimizer_rule.cpp

@@ -24,5 +24,7 @@
 
 #include "src/optimizer/rule/in_clause_simplification_rule.cpp"
 
+#include "src/optimizer/rule/ordered_aggregate_optimizer.cpp"
+
 #include "src/optimizer/rule/regex_optimizations.cpp"
 

package/src/duckdb/ub_src_parser.cpp

@@ -6,6 +6,8 @@
 
 #include "src/parser/constraint.cpp"
 
+#include "src/parser/common_table_expression_info.cpp"
+
 #include "src/parser/expression_util.cpp"
 
 #include "src/parser/keyword_helper.cpp"

package/src/duckdb/ub_src_parser_parsed_data.cpp

@@ -1,7 +1,9 @@
-#include "src/parser/parsed_data/alter_function_info.cpp"
-
 #include "src/parser/parsed_data/alter_info.cpp"
 
+#include "src/parser/parsed_data/alter_scalar_function_info.cpp"
+
+#include "src/parser/parsed_data/alter_table_function_info.cpp"
+
 #include "src/parser/parsed_data/alter_table_info.cpp"
 
 #include "src/parser/parsed_data/create_info.cpp"

package/src/duckdb/ub_src_parser_statement.cpp

@@ -26,6 +26,8 @@
 
 #include "src/parser/statement/load_statement.cpp"
 
+#include "src/parser/statement/multi_statement.cpp"
+
 #include "src/parser/statement/pragma_statement.cpp"
 
 #include "src/parser/statement/prepare_statement.cpp"

package/src/duckdb/ub_src_parser_tableref.cpp

@@ -6,6 +6,8 @@
 
 #include "src/parser/tableref/joinref.cpp"
 
+#include "src/parser/tableref/pivotref.cpp"
+
 #include "src/parser/tableref/subqueryref.cpp"
 
 #include "src/parser/tableref/table_function.cpp"

package/src/duckdb/ub_src_parser_transform_statement.cpp

@@ -50,6 +50,8 @@
 
 #include "src/parser/transform/statement/transform_set.cpp"
 
+#include "src/parser/transform/statement/transform_pivot_stmt.cpp"
+
 #include "src/parser/transform/statement/transform_prepare.cpp"
 
 #include "src/parser/transform/statement/transform_show.cpp"

package/src/duckdb/ub_src_parser_transform_tableref.cpp

@@ -4,6 +4,8 @@
 
 #include "src/parser/transform/tableref/transform_from.cpp"
 
+#include "src/parser/transform/tableref/transform_pivot.cpp"
+
 #include "src/parser/transform/tableref/transform_subquery.cpp"
 
 #include "src/parser/transform/tableref/transform_table_function.cpp"

package/src/duckdb/ub_src_planner_binder_expression.cpp

@@ -28,6 +28,8 @@
 
 #include "src/planner/binder/expression/bind_positional_reference_expression.cpp"
 
+#include "src/planner/binder/expression/bind_star_expression.cpp"
+
 #include "src/planner/binder/expression/bind_subquery_expression.cpp"
 
 #include "src/planner/binder/expression/bind_unnest_expression.cpp"

package/src/duckdb/ub_src_planner_binder_tableref.cpp

@@ -6,6 +6,8 @@
 
 #include "src/planner/binder/tableref/bind_joinref.cpp"
 
+#include "src/planner/binder/tableref/bind_pivot.cpp"
+
 #include "src/planner/binder/tableref/bind_subqueryref.cpp"
 
 #include "src/planner/binder/tableref/bind_table_function.cpp"

package/src/duckdb/ub_src_planner_expression_binder.cpp

@@ -2,6 +2,8 @@
 
 #include "src/planner/expression_binder/alter_binder.cpp"
 
+#include "src/planner/expression_binder/base_select_binder.cpp"
+
 #include "src/planner/expression_binder/check_binder.cpp"
 
 #include "src/planner/expression_binder/column_alias_binder.cpp"

package/src/duckdb/ub_src_planner_operator.cpp

@@ -2,6 +2,8 @@
 
 #include "src/planner/operator/logical_any_join.cpp"
 
+#include "src/planner/operator/logical_asof_join.cpp"
+
 #include "src/planner/operator/logical_column_data_get.cpp"
 
 #include "src/planner/operator/logical_comparison_join.cpp"

package/src/duckdb/ub_src_storage_statistics.cpp

@@ -4,15 +4,15 @@
 
 #include "src/storage/statistics/distinct_statistics.cpp"
 
-#include "src/storage/statistics/list_statistics.cpp"
+#include "src/storage/statistics/list_stats.cpp"
 
-#include "src/storage/statistics/numeric_statistics.cpp"
+#include "src/storage/statistics/numeric_stats.cpp"
 
-#include "src/storage/statistics/segment_statistics.cpp"
+#include "src/storage/statistics/numeric_stats_union.cpp"
 
-#include "src/storage/statistics/string_statistics.cpp"
+#include "src/storage/statistics/segment_statistics.cpp"
 
-#include "src/storage/statistics/struct_statistics.cpp"
+#include "src/storage/statistics/string_stats.cpp"
 
-#include "src/storage/statistics/validity_statistics.cpp"
+#include "src/storage/statistics/struct_stats.cpp"
 

package/src/duckdb/ub_src_storage_table.cpp

@@ -14,8 +14,6 @@
 
 #include "src/storage/table/persistent_table_data.cpp"
 
-#include "src/storage/table/segment_tree.cpp"
-
 #include "src/storage/table/row_group.cpp"
 
 #include "src/storage/table/row_group_collection.cpp"

package/src/duckdb_node.hpp

@@ -198,7 +198,8 @@ struct TaskHolder {
 
 class Utils {
 public:
-	static Napi::Value CreateError(Napi::Env env, std::string msg);
+	static Napi::Object CreateError(Napi::Env env, duckdb::PreservedError &e);
+	static Napi::Object CreateError(Napi::Env env, std::string msg);
 	static bool OtherIsInt(Napi::Number source);
 
 	template <class T>

package/src/statement.cpp

@@ -78,7 +78,7 @@ struct PrepareTask : public Task {
 
 		auto cb = callback.Value();
 		if (statement.statement->HasError()) {
-			cb.MakeCallback(statement.Value(), {Utils::CreateError(env, statement.statement->error.Message())});
+			cb.MakeCallback(statement.Value(), {Utils::CreateError(env, statement.statement->error)});
 			return;
 		}
 		cb.MakeCallback(statement.Value(), {env.Null(), statement.Value()});
@@ -282,11 +282,11 @@ struct RunPreparedTask : public Task {
 			return;
 		}
 		if (statement.statement->HasError()) {
-			cb.MakeCallback(statement.Value(), {Utils::CreateError(env, statement.statement->GetError())});
+			cb.MakeCallback(statement.Value(), {Utils::CreateError(env, statement.statement->GetErrorObject())});
 			return;
 		}
 		if (result->HasError()) {
-			cb.MakeCallback(statement.Value(), {Utils::CreateError(env, result->GetError())});
+			cb.MakeCallback(statement.Value(), {Utils::CreateError(env, result->GetErrorObject())});
 			return;
 		}
 
@@ -427,9 +427,9 @@ struct RunQueryTask : public Task {
 		if (!statement.statement) {
 			deferred.Reject(Utils::CreateError(env, "statement was finalized"));
 		} else if (statement.statement->HasError()) {
-			deferred.Reject(Utils::CreateError(env, statement.statement->GetError()));
+			deferred.Reject(Utils::CreateError(env, statement.statement->GetErrorObject()));
 		} else if (result->HasError()) {
-			deferred.Reject(Utils::CreateError(env, result->GetError()));
+			deferred.Reject(Utils::CreateError(env, result->GetErrorObject()));
 		} else {
 			auto db = statement.connection_ref->database_ref->Value();
 			auto query_result = QueryResult::constructor.New({db});

package/src/utils.cpp

@@ -12,11 +12,36 @@ bool Utils::OtherIsInt(Napi::Number source) {
 	}
 }
 
-Napi::Value Utils::CreateError(Napi::Env env, std::string msg) {
+static void SetString(Napi::Object &obj, const std::string &key, const std::string &value) {
+	obj.Set(Napi::String::New(obj.Env(), key), Napi::String::New(obj.Env(), value));
+}
+
+Napi::Object Utils::CreateError(Napi::Env env, duckdb::PreservedError &error) {
+	auto obj = Utils::CreateError(env, error.Message());
+	if (error.Type() == duckdb::ExceptionType::HTTP) {
+		const auto &e = error.GetError()->AsHTTPException();
+		obj.Set(Napi::String::New(env, "statusCode"), Napi::Number::New(env, e.GetStatusCode()));
+		SetString(obj, "response", e.GetResponseBody());
+		SetString(obj, "reason", e.GetReason());
+
+		auto headers = Napi::Object::New(env);
+		for (const auto &item : e.GetHeaders()) {
+			SetString(headers, item.first, item.second);
+		}
+		obj.Set(Napi::String::New(env, "headers"), headers);
+	}
+
+	SetString(obj, "errorType", duckdb::Exception::ExceptionTypeToString(error.Type()));
+
+	return obj;
+}
+
+Napi::Object Utils::CreateError(Napi::Env env, std::string msg) {
 	auto err = Napi::Error::New(env, Napi::String::New(env, msg).Utf8Value()).Value();
 	Napi::Object obj = err.As<Napi::Object>();
 	obj.Set(Napi::String::New(env, "errno"), Napi::Number::New(env, Database::DUCKDB_NODEJS_ERROR));
-	obj.Set(Napi::String::New(env, "code"), Napi::String::New(env, "DUCKDB_NODEJS_ERROR"));
+	SetString(obj, "code", "DUCKDB_NODEJS_ERROR");
+	SetString(obj, "errorType", "Invalid");
 
 	return obj;
 }

package/test/extension.test.ts

@@ -1,9 +1,8 @@
 import * as duckdb from '..';
-import {Database, TableData} from '..';
+import {Database, DuckDbError, HttpError, TableData} from '..';
 import * as fs from 'fs';
 import * as assert from 'assert';
 import * as path from 'path';
-import {Done} from "mocha";
 
 const extension_base_path = "../../../build/release/extension";
 
@@ -21,46 +20,63 @@ if (fs.existsSync(extension_full_path)) {
     }).filter(a => a) as string[];
 }
 
+function isHTTPException(err: DuckDbError): err is HttpError {
+    return err.errorType === 'HTTP';
+}
+
 // Note: test will pass on http request failing due to connection issues.
-const test_httpfs = async function (db: duckdb.Database, done: Done) {
-    db.all("SELECT id, first_name, last_name FROM PARQUET_SCAN('https://raw.githubusercontent.com/cwida/duckdb/master/data/parquet-testing/userdata1.parquet') LIMIT 3;", function(err: null | Error, rows: TableData) {
+const test_httpfs = async function (db: duckdb.Database) {
+    await new Promise<void>((resolve, reject) => db.all("SELECT id, first_name, last_name FROM PARQUET_SCAN('https://raw.githubusercontent.com/cwida/duckdb/master/data/parquet-testing/userdata1.parquet') LIMIT 3;", function (err: null | Error, rows: TableData) {
         if (err) {
             if (err.message.startsWith("Unable to connect to URL")) {
                 console.warn("Warning: HTTP request failed in extension.test.js");
-                done();
+                resolve();
             } else {
-                throw err;
+                reject(err);
             }
         } else {
             assert.deepEqual(rows, [
-                { id: 1, first_name: 'Amanda', last_name: 'Jordan'},
-                { id: 2, first_name: 'Albert', last_name: 'Freeman'},
-                { id: 3, first_name: 'Evelyn', last_name: 'Morgan'},
+                {id: 1, first_name: 'Amanda', last_name: 'Jordan'},
+                {id: 2, first_name: 'Albert', last_name: 'Freeman'},
+                {id: 3, first_name: 'Evelyn', last_name: 'Morgan'},
             ]);
-            done();
+            resolve();
         }
-    });
+    }));
+
+    await new Promise<void>((resolve) => {
+        db.exec("select * from read_csv_auto('https://example.com/hello.csv')", (err: DuckDbError | null) => {
+            assert.ok(err);
+            assert.ok(isHTTPException(err));
+            if (isHTTPException(err)) {
+                assert.equal(err.statusCode, 404);
+                assert.equal(err.reason, 'Not Found');
+                assert.equal(err.response, '');
+                assert.ok('Content-Length' in err.headers, JSON.stringify(err.headers));
+            }
+            resolve();
+        });
+    })
 };
 
-const test_tpch = async function (db: Database, done:Done) {
-    db.all("CALL DBGEN(sf=0.01);", function(err: null | Error) {
+const test_tpch = async function (db: Database) {
+    await new Promise<void>((resolve, reject) => db.all("CALL DBGEN(sf=0.01);", function (err: null | Error) {
         if (err) {
-            throw err;
+            reject(err);
         }
-        done();
-    });
+        resolve();
+    }));
 };
 
-const test_extension = function(extension_name: string, db: duckdb.Database, done: Done) {
-    switch(extension_name) {
+const test_extension = async function (extension_name: string, db: duckdb.Database) {
+    switch (extension_name) {
         case 'httpfs.duckdb_extension':
-            test_httpfs(db, done);
+            await test_httpfs(db);
             break;
         case 'tpch.duckdb_extension':
-            test_tpch(db, done);
+            await test_tpch(db);
             break;
         default:
-            done();
             break;
     }
 };
@@ -79,13 +95,15 @@ describe('Extension loading', function() {
             continue;
         }
 
-        it(extension_name, function(done) {
-            db.run(`LOAD '${extension_path}';`, function(err: null | Error) {
+        it(extension_name, async function () {
+            await new Promise<void>((resolve, reject) => db.run(`LOAD '${extension_path}';`, function (err: null | Error) {
                 if (err) {
-                    throw err;
+                    reject(err);
                 }
-                test_extension(extension_name, db, done);
-            });
+                resolve()
+            }));
+
+            await test_extension(extension_name, db);
         });
     }
 });

package/test/syntax_error.test.ts

@@ -8,8 +8,10 @@ describe('exec', function() {
     });
 
     it("doesn't crash on a syntax error", function(done) {
-        db.exec("syntax error", function(err: null | Error) {
+        db.exec("syntax error", function(err: null | duckdb.DuckDbError) {
             assert.notEqual(err, null, "Expected an error")
+            assert.equal(err?.errorType, 'Parser');
+            assert.ok(err?.message.startsWith('Parser Error: syntax error at or near "syntax"'))
             done();
         });
     });

package/src/duckdb/src/include/duckdb/main/loadable_extension.hpp

@@ -1,59 +0,0 @@
-//===----------------------------------------------------------------------===//
-//                         DuckDB
-//
-// duckdb/main/loadable_extension.hpp
-//
-//
-//===----------------------------------------------------------------------===//
-
-#pragma once
-
-#if defined(DUCKDB_BUILD_LOADABLE_EXTENSION) && defined(DUCKDB_EXTENSION_MAIN)
-#if defined(_WIN32) || defined(_WIN64)
-#ifndef NOMINMAX
-#define NOMINMAX
-#endif
-
-#ifndef _WINSOCKAPI_
-#define _WINSOCKAPI_
-#endif
-#include <windows.h>
-
-#undef CreateDirectory
-#undef MoveFile
-#undef RemoveDirectory
-
-#include <delayimp.h>
-
-extern "C" {
-/*
-This is interesting: Windows would normally require a duckdb.dll being on the DLL search path when we load an extension
-using LoadLibrary(). However, there is likely no such dll, because DuckDB was statically linked, or is running as part
-of an R or Python module with a completely different name (that we don't know) or something of the sorts. Amazingly,
-Windows supports lazy-loading DLLs by linking them with /DELAYLOAD. Then a callback will be triggered whenever we access
-symbols in the extension. Since DuckDB is already running in the host process (hopefully), we can use
-GetModuleHandle(NULL) to return the current process so the symbols are looked for there. See here for another
-explanation of this crazy process:
-
-* https://docs.microsoft.com/en-us/cpp/build/reference/linker-support-for-delay-loaded-dlls?view=msvc-160
-* https://docs.microsoft.com/en-us/cpp/build/reference/understanding-the-helper-function?view=msvc-160
-*/
-FARPROC WINAPI duckdb_dllimport_delay_hook(unsigned dliNotify, PDelayLoadInfo pdli) {
-	switch (dliNotify) {
-	case dliNotePreLoadLibrary:
-		if (strcmp(pdli->szDll, "duckdb.dll") != 0) {
-			return NULL;
-		}
-		return (FARPROC)GetModuleHandle(NULL);
-	default:
-		return NULL;
-	}
-
-	return NULL;
-}
-
-ExternC const PfnDliHook __pfnDliNotifyHook2 = duckdb_dllimport_delay_hook;
-ExternC const PfnDliHook __pfnDliFailureHook2 = duckdb_dllimport_delay_hook;
-}
-#endif
-#endif

package/src/duckdb/src/include/duckdb/storage/statistics/list_statistics.hpp

@@ -1,36 +0,0 @@
-//===----------------------------------------------------------------------===//
-//                         DuckDB
-//
-// duckdb/storage/statistics/list_statistics.hpp
-//
-//
-//===----------------------------------------------------------------------===//
-
-#pragma once
-
-#include "duckdb/storage/statistics/base_statistics.hpp"
-#include "duckdb/common/enums/filter_propagate_result.hpp"
-#include "duckdb/storage/statistics/validity_statistics.hpp"
-
-namespace duckdb {
-class Value;
-
-class ListStatistics : public BaseStatistics {
-public:
-	explicit ListStatistics(LogicalType type);
-
-	unique_ptr<BaseStatistics> child_stats;
-
-public:
-	void Merge(const BaseStatistics &other) override;
-	FilterPropagateResult CheckZonemap(ExpressionType comparison_type, const Value &constant) const;
-
-	unique_ptr<BaseStatistics> Copy() const override;
-	void Serialize(FieldWriter &serializer) const override;
-	static unique_ptr<BaseStatistics> Deserialize(FieldReader &source, LogicalType type);
-	void Verify(Vector &vector, const SelectionVector &sel, idx_t count) const override;
-
-	string ToString() const override;
-};
-
-} // namespace duckdb