Add warning for partial imports (#3608)

go/cmd/dolt/commands/tblcmds/import.go

@@ -500,6 +500,12 @@ func newImportSqlEngineMover(ctx context.Context, dEnv *env.DoltEnv, rdSchema sc
 		return nil, &mvdata.DataMoverCreationError{ErrType: mvdata.SchemaErr, Cause: err}
 	}
 
+	// Leave a warning if the import operation is operating on fewer columns than the relevant table's schema.
+	// This can certainly be intentional, but it is often due to typos in the header of a csv file.
+	if rowOperationSchema.GetAllCols().Size() < tableSchema.GetAllCols().Size() {
+		cli.PrintErrln(color.YellowString("Warning: There are fewer columns in the import file's schema than the table's schema.\nIf unintentional, check for any typos in the import file's header."))
+	}
+
 	mv, err := mvdata.NewSqlEngineTableWriter(ctx, dEnv, tableSchema, rowOperationSchema, moveOps, importStatsCB)
 	if err != nil {
 		return nil, &mvdata.DataMoverCreationError{ErrType: mvdata.CreateWriterErr, Cause: err}

integration-tests/bats/import-create-tables.bats

@@ -112,6 +112,9 @@ teardown() {
     run dolt table import -c --pk=pk test people.csv
     [ "$status" -eq 0 ]
     [[ "$output" =~ "Import completed successfully." ]] || false
+    # Sanity Check
+    ! [[ "$output" =~ "Warning: There are fewer columns in the import file's schema than the table's schema" ]] || false
+
     run dolt sql -q "select * from test"
     [ "$status" -eq 0 ]
     [ "${#lines[@]}" -eq 8 ]
@@ -664,7 +667,7 @@ DELIM
     [ "${lines[1]}" = "0,1,2" ]
 }
 
-@test "import-create-tables: csv files has less columns filled with default value" {
+@test "import-create-tables: csv files has fewer columns filled with default value" {
     cat <<SQL > schema.sql
 CREATE TABLE subset (
     pk INT NOT NULL,
@@ -680,6 +683,7 @@ DELIM
 
     run dolt table import -s schema.sql -c subset data.csv
     [ "$status" -eq 0 ]
+    [[ "$output" =~ "Warning: There are fewer columns in the import file's schema than the table's schema" ]] || false
 
     # schema argument subsets the data and adds empty column
     run dolt sql -r csv -q "select * from subset ORDER BY pk"

integration-tests/bats/import-replace-tables.bats

@@ -336,6 +336,7 @@ DELIM
 
     run dolt table import -r test 1pk5col-ints-updt.csv
     [ "$status" -eq 0 ]
+    [[ "$output" =~ "Warning: There are fewer columns in the import file's schema than the table's schema" ]] || false
     [[ "$output" =~ "Rows Processed: 1, Additions: 1, Modifications: 0, Had No Effect: 0" ]] || false
     [[ "$output" =~ "Import completed successfully." ]] || false
 
@@ -362,6 +363,7 @@ DELIM
     dolt sql -q "insert into subset values (1000, 100, 1000, 10000)"
 
     run dolt table import -r subset data.csv
+    ! [[ "$output" =~ "Warning: There are fewer columns in the import file's schema than the table's schema" ]] || false
     [ "$status" -eq 0 ]
 
     # schema argument subsets the data and adds empty column

integration-tests/bats/import-update-tables.bats

@@ -118,6 +118,8 @@ teardown() {
     [ "$status" -eq 0 ]
     [[ "$output" =~ "Rows Processed: 2, Additions: 2, Modifications: 0, Had No Effect: 0" ]] || false
     [[ "$output" =~ "Import completed successfully." ]] || false
+    # Sanity check
+    ! [[ "$output" =~ "Warning: There are fewer columns in the import file's schema than the table's schema" ]] || false
 
     # Validate that a successful import with no bad rows does not print the following
     ! [[ "$output" =~ "The following rows were skipped:" ]] || false
@@ -553,6 +555,7 @@ DELIM
 
     run dolt table import -u test 1pk5col-ints-updt.csv
     [ "$status" -eq 0 ]
+    [[ "$output" =~ "Warning: There are fewer columns in the import file's schema than the table's schema" ]] || false
     [[ "$output" =~ "Rows Processed: 1, Additions: 1, Modifications: 0, Had No Effect: 0" ]] || false
     [[ "$output" =~ "Import completed successfully." ]] || false
 
@@ -573,6 +576,7 @@ DELIM
 
     run dolt table import -u test 1pk5col-ints-updt.csv
     [ "$status" -eq 0 ]
+    [[ "$output" =~ "Warning: There are fewer columns in the import file's schema than the table's schema" ]] || false
     [[ "$output" =~ "Rows Processed: 1, Additions: 1, Modifications: 0, Had No Effect: 0" ]] || false
     [[ "$output" =~ "Import completed successfully." ]] || false
 
@@ -595,6 +599,7 @@ DELIM
     [ "$status" -eq 0 ]
     [[ "$output" =~ "Rows Processed: 1, Additions: 1, Modifications: 0, Had No Effect: 0" ]] || false
     [[ "$output" =~ "Import completed successfully." ]] || false
+    ! [[ "$output" =~ "Warning: There are fewer columns in the import file's schema than the table's schema" ]] || false
 
     run dolt sql -r csv -q "select * from test"
     [ "${lines[1]}" = "0,1,2,3,4,6" ]
@@ -615,6 +620,7 @@ DELIM
     [ "$status" -eq 0 ]
     [[ "$output" =~ "Rows Processed: 1, Additions: 1, Modifications: 0, Had No Effect: 0" ]] || false
     [[ "$output" =~ "Import completed successfully." ]] || false
+    ! [[ "$output" =~ "Warning: There are fewer columns in the import file's schema than the table's schema" ]] || false
 
     run dolt sql -r csv -q "select * from test"
     [ "${lines[1]}" = "0,1,2,3,4,6" ]
@@ -635,6 +641,7 @@ DELIM
 
     run dolt table import -u test 1pk5col-ints-updt.csv
     [ "$status" -eq 0 ]
+    [[ "$output" =~ "Warning: There are fewer columns in the import file's schema than the table's schema" ]] || false
     [[ "$output" =~ "Rows Processed: 1, Additions: 0, Modifications: 1, Had No Effect: 0" ]] || false
     [[ "$output" =~ "Import completed successfully." ]] || false
 
@@ -674,6 +681,7 @@ DELIM
 
     run dolt table import -u keyless data.csv
     [ "$status" -eq 0 ]
+    [[ "$output" =~ "Warning: There are fewer columns in the import file's schema than the table's schema" ]] || false
     [[ "$output" =~ "Rows Processed: 1, Additions: 1, Modifications: 0, Had No Effect: 0" ]] || false
     [[ "$output" =~ "Import completed successfully." ]] || false