Merge pull request #1441 from kuzudb/copy-npy-grammar-change

Change npy copy grammar
kuzudb · Apr 5, 2023 · 2f5b1b4 · 2f5b1b4
2 parents 0fe05bd + 7587212
commit 2f5b1b4
Show file tree

Hide file tree

Showing 12 changed files with 3,294 additions and 3,276 deletions.
diff --git a/dataset/npy-1d/copy.cypher b/dataset/npy-1d/copy.cypher
@@ -1 +1 @@
-npy copy npytable from ("dataset/npy-1d/one_dim_int64.npy",  "dataset/npy-1d/one_dim_int32.npy",  "dataset/npy-1d/one_dim_int16.npy",  "dataset/npy-1d/one_dim_double.npy", "dataset/npy-1d/one_dim_float.npy");
+copy npytable from ("dataset/npy-1d/one_dim_int64.npy",  "dataset/npy-1d/one_dim_int32.npy",  "dataset/npy-1d/one_dim_int16.npy",  "dataset/npy-1d/one_dim_double.npy", "dataset/npy-1d/one_dim_float.npy") by column;
diff --git a/dataset/npy-20k/copy.cypher b/dataset/npy-20k/copy.cypher
@@ -1 +1 @@
-npy copy npytable from ("dataset/npy-20k/id_int64.npy", "dataset/npy-20k/two_dim_float.npy");
+copy npytable from ("dataset/npy-20k/id_int64.npy", "dataset/npy-20k/two_dim_float.npy") by column;
diff --git a/dataset/npy-2d/copy.cypher b/dataset/npy-2d/copy.cypher
@@ -1 +1 @@
-npy copy npytable from ("dataset/npy-2d/id_int64.npy", "dataset/npy-2d/two_dim_int64.npy", "dataset/npy-2d/two_dim_int32.npy",  "dataset/npy-2d/two_dim_int16.npy",  "dataset/npy-2d/two_dim_double.npy", "dataset/npy-2d/two_dim_float.npy");
+copy npytable from ("dataset/npy-2d/id_int64.npy", "dataset/npy-2d/two_dim_int64.npy", "dataset/npy-2d/two_dim_int32.npy",  "dataset/npy-2d/two_dim_int16.npy",  "dataset/npy-2d/two_dim_double.npy", "dataset/npy-2d/two_dim_float.npy") by column;
diff --git a/dataset/npy-3d/copy.cypher b/dataset/npy-3d/copy.cypher
@@ -1 +1 @@
-npy copy npytable FROM ("dataset/npy-3d/id_int64.npy", "dataset/npy-3d/three_dim_int64.npy");
+copy npytable FROM ("dataset/npy-3d/id_int64.npy", "dataset/npy-3d/three_dim_int64.npy") by column;
diff --git a/src/antlr4/Cypher.g4 b/src/antlr4/Cypher.g4
@@ -20,7 +20,7 @@ kU_CopyCSV
     : COPY SP oC_SchemaName SP FROM SP kU_FilePaths ( SP? '(' SP? kU_ParsingOptions SP? ')' )? ;
 
 kU_CopyNPY
-    : NPY SP COPY SP oC_SchemaName SP FROM ( SP '(' SP? StringLiteral ( SP? ',' SP? StringLiteral )* ')' ) ;
+    : COPY SP oC_SchemaName SP FROM SP '(' SP? StringLiteral ( SP? ',' SP? StringLiteral )* ')' SP BY SP COLUMN ;
 
 kU_FilePaths
     : '[' SP? StringLiteral ( SP? ',' SP? StringLiteral )* ']'
@@ -41,6 +41,8 @@ FROM : ( 'F' | 'f' ) ( 'R' | 'r' ) ( 'O' | 'o' ) ( 'M' | 'm' );
 
 NPY : ( 'N' | 'n' ) ( 'P' | 'p' ) ( 'Y' | 'y' ) ;
 
+COLUMN : ( 'C' | 'c' ) ( 'O' | 'o' ) ( 'L' | 'l' ) ( 'U' | 'u' ) ( 'M' | 'm' ) ( 'N' | 'n' ) ;
+
 kU_DDL
     : kU_CreateNode
         | kU_CreateRel

diff --git a/src/binder/bind/bind_copy.cpp b/src/binder/bind/bind_copy.cpp
@@ -22,7 +22,7 @@ std::unique_ptr<BoundStatement> Binder::bindCopy(const Statement& statement) {
     std::unordered_map<common::property_id_t, std::string> propertyToNpyMap;
     if (expectedFileType == common::CopyDescription::FileType::UNKNOWN &&
         actualFileType == common::CopyDescription::FileType::NPY) {
-        throw BinderException("Please use NPY COPY statement for copying npy files.");
+        throw BinderException("Please use COPY FROM BY COLUMN statement for copying npy files.");
     }
     if (expectedFileType == common::CopyDescription::FileType::NPY &&
         actualFileType != expectedFileType) {

diff --git a/test/copy/copy_npy_test.cpp b/test/copy/copy_npy_test.cpp
@@ -246,66 +246,66 @@ TEST_F(CopyLargeNpyTest, CopyLargeNpyTest) {
 
 TEST_F(CopyNpyFaultTest, CopyNpyInsufficientNumberOfProperties) {
     conn->query("create node table npytable (id INT64,i64 INT64[12],PRIMARY KEY(id));");
-    validateCopyException("npy copy npytable from (\"" +
+    validateCopyException("copy npytable from (\"" +
                               TestHelper::appendKuzuRootPath("dataset/npy-3d/three_dim_int64.npy") +
-                              "\");",
+                              "\") by column;",
         "Binder exception: Number of npy files is not equal to number of properties in table "
         "npytable.");
 }
 
 TEST_F(CopyNpyFaultTest, CopyNpyRedundantProperties) {
     conn->query("create node table npytable (id INT64,i32 INT32, PRIMARY KEY(id));");
     auto f32Path = TestHelper::appendKuzuRootPath("dataset/npy-20k/two_dim_float.npy");
-    validateCopyException("npy copy npytable from (\"" +
+    validateCopyException("copy npytable from (\"" +
                               TestHelper::appendKuzuRootPath("dataset/npy-20k/id_int64.npy") +
-                              "\", \"" + f32Path + "\");",
+                              "\", \"" + f32Path + "\") BY COLUMN;",
         "The type of npy file " + f32Path + " does not match the type defined in table npytable.");
 }
 
 TEST_F(CopyNpyFaultTest, CopyNpyVectorIntoScaler) {
     conn->query("create node table npytable (id INT64,f32 FLOAT, PRIMARY KEY(id));");
     auto f32Path = TestHelper::appendKuzuRootPath("dataset/npy-20k/two_dim_float.npy");
-    validateCopyException("npy copy npytable from (\"" +
+    validateCopyException("copy npytable from (\"" +
                               TestHelper::appendKuzuRootPath("dataset/npy-20k/id_int64.npy") +
-                              "\", \"" + f32Path + "\");",
+                              "\", \"" + f32Path + "\") by column;",
         "Cannot copy a vector property in npy file " + f32Path +
             " to a scalar property in table npytable.");
 }
 
 TEST_F(CopyNpyFaultTest, CopyNpyWithMismatchedTypeOneDimensionTest) {
     conn->query("create node table npytable (id INT64,i32 INT32, PRIMARY KEY(id));");
     auto f32Path = TestHelper::appendKuzuRootPath("dataset/npy-1d/one_dim_float.npy");
-    validateCopyException("npy copy npytable from ( \"" +
+    validateCopyException("copy npytable from ( \"" +
                               TestHelper::appendKuzuRootPath("dataset/npy-1d/one_dim_int64.npy") +
-                              "\", \"" + f32Path + "\");",
+                              "\", \"" + f32Path + "\") by column;",
         "The type of npy file " + f32Path + " does not match the type defined in table npytable.");
 }
 
 TEST_F(CopyNpyFaultTest, CopyNpyWithMismatchedTypeTwoDimensionTest) {
     conn->query("create node table npytable (id INT64,i32 INT32[10], PRIMARY KEY(id));");
     auto f32Path = TestHelper::appendKuzuRootPath("dataset/npy-20k/two_dim_float.npy");
-    validateCopyException("npy copy npytable from (\"" +
+    validateCopyException("copy npytable from (\"" +
                               TestHelper::appendKuzuRootPath("dataset/npy-20k/id_int64.npy") +
-                              "\", \"" + f32Path + "\");",
+                              "\", \"" + f32Path + "\") by column;",
         "The type of npy file " + f32Path + " does not match the type defined in table npytable.");
 }
 
 TEST_F(CopyNpyFaultTest, CopyNpyWithMismatchedDimensionTest) {
     conn->query("create node table npytable (id INT64,f32 FLOAT[12],PRIMARY KEY(id));");
     auto twoDimFloatNpyPath = TestHelper::appendKuzuRootPath("dataset/npy-20k/two_dim_float.npy");
-    validateCopyException("npy copy npytable from (\"" +
+    validateCopyException("copy npytable from (\"" +
                               TestHelper::appendKuzuRootPath("dataset/npy-20k/id_int64.npy") +
-                              "\", \"" + twoDimFloatNpyPath + "\");",
+                              "\", \"" + twoDimFloatNpyPath + "\") by column;",
         "The shape of " + twoDimFloatNpyPath +
             " does not match the length of the fixed list property in table npytable.");
 }
 
 TEST_F(CopyNpyFaultTest, CopyNpyWithMismatchedLengthTest) {
     conn->query("create node table npytable (id INT64,i64 INT64[12],PRIMARY KEY(id));");
     validateCopyException(
-        "npy copy npytable from (\"" +
-            TestHelper::appendKuzuRootPath("dataset/npy-20k/id_int64.npy") + "\", \"" +
-            TestHelper::appendKuzuRootPath("dataset/npy-3d/three_dim_int64.npy") + "\");",
+        "copy npytable from (\"" + TestHelper::appendKuzuRootPath("dataset/npy-20k/id_int64.npy") +
+            "\", \"" + TestHelper::appendKuzuRootPath("dataset/npy-3d/three_dim_int64.npy") +
+            "\") by column;",
         "Number of rows in npy files is not equal to each other.");
 }
 } // namespace testing