Fix #2704 (#3379)

* Fix bind file scan
* Fix parquet and npy reader bindFunc

dataset/copy-test/node/parquet/copy.cypher

@@ -1 +1 @@
-COPY tableOfTypes FROM "dataset/copy-test/node/parquet/types_50k*.parquet" (HEADER=true);
+COPY tableOfTypes FROM "dataset/copy-test/node/parquet/types_50k*.parquet";

src/binder/bind/bind_import_database.cpp

@@ -55,12 +55,19 @@ std::unique_ptr<BoundStatement> Binder::bindImportDatabaseClause(const Statement
             copyFromStatement->getSource())
                              ->filePaths;
         KU_ASSERT(filePaths.size() == 1);
+        auto fileType = bindFileType(filePaths);
         auto copyFilePath = boundFilePath + "/" + filePaths[0];
-        auto csvConfig = CSVReaderConfig::construct(
-            bindParsingOptions(copyFromStatement->getParsingOptionsRef()));
-        auto csvQuery = stringFormat("COPY {} FROM \"{}\" {};", copyFromStatement->getTableName(),
-            copyFilePath, csvConfig.option.toCypher());
-        finalQueryStatements += csvQuery;
+        std::string query;
+        if (fileType == FileType::CSV) {
+            auto csvConfig = CSVReaderConfig::construct(
+                bindParsingOptions(copyFromStatement->getParsingOptionsRef()));
+            query = stringFormat("COPY {} FROM \"{}\" {};", copyFromStatement->getTableName(),
+                copyFilePath, csvConfig.option.toCypher());
+        } else {
+            query = stringFormat("COPY {} FROM \"{}\";", copyFromStatement->getTableName(),
+                copyFilePath);
+        }
+        finalQueryStatements += query;
     }
     finalQueryStatements += getQueryFromFile(fs, boundFilePath, ImportDBConstants::MACRO_NAME);
     return std::make_unique<BoundImportDatabase>(boundFilePath, finalQueryStatements);

src/processor/operator/persistent/reader/npy/npy_reader.cpp

@@ -296,6 +296,7 @@ static void bindColumns(const common::ReaderConfig& readerConfig,
 static std::unique_ptr<function::TableFuncBindData> bindFunc(main::ClientContext* /*context*/,
     function::TableFuncBindInput* input) {
     auto scanInput = reinterpret_cast<function::ScanTableFuncBindInput*>(input);
+    KU_ASSERT(scanInput->config.options.empty());
     std::vector<std::string> detectedColumnNames;
     std::vector<common::LogicalType> detectedColumnTypes;
     bindColumns(scanInput->config, detectedColumnNames, detectedColumnTypes);

src/processor/operator/persistent/reader/parquet/parquet_reader.cpp

@@ -2,6 +2,7 @@
 
 #include <fcntl.h>
 
+#include "common/exception/binder.h"
 #include "common/exception/copy.h"
 #include "common/file_system/virtual_file_system.h"
 #include "common/string_format.h"
@@ -652,6 +653,9 @@ static std::unique_ptr<function::TableFuncBindData> bindFunc(main::ClientContext
     function::TableFuncBindInput* input) {
     auto scanInput =
         ku_dynamic_cast<function::TableFuncBindInput*, function::ScanTableFuncBindInput*>(input);
+    if (!scanInput->config.options.empty()) {
+        throw BinderException{"Copy from Parquet cannot have options."};
+    }
     std::vector<std::string> detectedColumnNames;
     std::vector<common::LogicalType> detectedColumnTypes;
     bindColumns(scanInput, detectedColumnNames, detectedColumnTypes);

test/test_files/copy/copy_node_parquet.test

@@ -53,3 +53,8 @@
 -STATEMENT MATCH (row:tableOfTypes) WHERE 0 <= row.doubleColumn AND row.doubleColumn <= 10 AND 0 <= row.int64Column AND row.int64Column <= 10 RETURN count(*);
 ---- 1
 546
+
+-LOG CopyWithOptionsErrorTest
+-STATEMENT COPY tableOfTypes FROM "${KUZU_ROOT_DIRECTORY}/dataset/copy-test/node/parquet/types_50k*.parquet" (HEADER=true);
+---- error
+Binder exception: Copy from Parquet cannot have options.

test/test_files/copy/copy_partial_column.test

@@ -184,7 +184,7 @@ Copy exception: Found NULL, which violates the non-null constraint of the primar
 ---- ok
 -STATEMENT COPY tableOfTypes12 (id, int64Column, doubleColumn, booleanColumn, dateColumn, stringColumn, listOfInt64,
 listOfString, listOfListOfInt64, structColumn) FROM
-"${KUZU_ROOT_DIRECTORY}/dataset/copy-test/node/parquet/types_50k*.parquet" (HEADER=true);
+"${KUZU_ROOT_DIRECTORY}/dataset/copy-test/node/parquet/types_50k*.parquet";
 ---- ok
 -STATEMENT MATCH (row:tableOfTypes12) WHERE row.id >= 20 AND row.id <= 24 RETURN row.*;
 ---- 5
@@ -205,6 +205,6 @@ listOfString, listOfListOfInt64, structColumn) FROM
 ---- ok
 -STATEMENT COPY tableOfTypes12 (id, dateColumn, doubleColumn, booleanColumn, int64Column, stringColumn, listOfInt64,
 listOfString, listOfListOfInt64, structColumn) FROM
-"${KUZU_ROOT_DIRECTORY}/dataset/copy-test/node/parquet/types_50k*.parquet" (HEADER=true);
+"${KUZU_ROOT_DIRECTORY}/dataset/copy-test/node/parquet/types_50k*.parquet";
 ---- error
 Binder exception: Column `dateColumn` type mismatch. Expected DATE but got INT64.

test/test_files/exceptions/copy/wrong_header.test

@@ -111,7 +111,7 @@ Binder exception: Number of columns mismatch. Expected 4 but got 3.
 -CASE NodeUnmatchedNumColumns
 -STATEMENT create node table person (ID1 SERIAL, ID INT64, fName INT64, age INT64, PRIMARY KEY (ID1))
 ---- ok
--STATEMENT COPY person FROM "${KUZU_ROOT_DIRECTORY}/dataset/copy-test/node/parquet/types_50k_1.parquet" (HEADER=true)
+-STATEMENT COPY person FROM "${KUZU_ROOT_DIRECTORY}/dataset/copy-test/node/parquet/types_50k_1.parquet"
 ---- error
 Binder exception: Number of columns mismatch. Expected 3 but got 10.
 
@@ -120,6 +120,6 @@ Binder exception: Number of columns mismatch. Expected 3 but got 10.
 ---- ok
 -STATEMENT create rel table knows (FROM person TO person, time date, age INT64)
 ---- ok
--STATEMENT COPY knows FROM "${KUZU_ROOT_DIRECTORY}/dataset/demo-db/parquet/follows.parquet" (HEADER=true)
+-STATEMENT COPY knows FROM "${KUZU_ROOT_DIRECTORY}/dataset/demo-db/parquet/follows.parquet"
 ---- error
 Binder exception: Number of columns mismatch. Expected 4 but got 3.