update parquet-tool

xitongsys · Mar 21, 2018 · 12272b2 · 12272b2
1 parent 381f931
commit 12272b2
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 9 deletions.
diff --git a/tool/parquet-tools/README.md b/tool/parquet-tools/README.md
@@ -5,12 +5,20 @@ More functions will be added.
 ## Build
 cd parquet-tools && go build parquet-tools
 
+## Description
+### -cmd
+schema: show schema of the parquet file; 
+### -file
+parquet file name;
+### -tag
+print the go struct tags; default is false;
+
 ## Example
 
 ### Output Schema
 
 ```bash
-bash$ ./parquet-tools -cmd=schema -file=a.parquet
+bash$ ./parquet-tools -cmd=schema -file=a.parquet -tag=false
 bash$
 ----- Go struct -----
 parquet_go_root struct{

diff --git a/tool/parquet-tools/SchemaTool/SchemaTool.go b/tool/parquet-tools/SchemaTool/SchemaTool.go
@@ -197,7 +197,7 @@ func (self *Node) OutputJsonSchema() string {
 	return res
 }
 
-func(cNode *Node) getStructTags() string {
+func (cNode *Node) getStructTags() string {
 	rTStr := "REQUIRED"
 	if cNode.SE.GetRepetitionType() == parquet.FieldRepetitionType_OPTIONAL {
 		rTStr = "OPTIONAL"
@@ -227,7 +227,7 @@ func(cNode *Node) getStructTags() string {
 	return tags
 }
 
-func (self *Node) OutputStruct(withName bool) string {
+func (self *Node) OutputStruct(withName bool, withTags bool) string {
 	name := self.SE.GetName()
 
 	res := ""
@@ -249,7 +249,11 @@ func (self *Node) OutputStruct(withName bool) string {
 	if pT == nil && cT == nil {
 		res += rTStr + "struct{\n"
 		for _, cNode := range self.Children {
-			res += cNode.OutputStruct(true) + " " + cNode.getStructTags()  + "\n"
+			if withTags {
+				res += cNode.OutputStruct(true, withTags) + " " + cNode.getStructTags() + "\n"
+			} else {
+				res += cNode.OutputStruct(true, withTags) + "\n"
+			}
 		}
 		res += "}"
 
@@ -258,11 +262,11 @@ func (self *Node) OutputStruct(withName bool) string {
 		keyPT, keyCT := keyNode.SE.Type, keyNode.SE.ConvertedType
 		keyGoTypeStr := ParquetTypeToGoTypeStr(keyPT, keyCT)
 		valNode := self.Children[0].Children[1]
-		res += rTStr + "map[" + keyGoTypeStr + "]" + valNode.OutputStruct(false)
+		res += rTStr + "map[" + keyGoTypeStr + "]" + valNode.OutputStruct(false, withTags)
 
 	} else if cT != nil && *cT == parquet.ConvertedType_LIST {
 		cNode := self.Children[0].Children[0]
-		res += rTStr + "[]" + cNode.OutputStruct(false)
+		res += rTStr + "[]" + cNode.OutputStruct(false, withTags)
 
 	} else {
 		goTypeStr := ParquetTypeToGoTypeStr(pT, cT)
@@ -320,6 +324,6 @@ func (self *SchemaTree) OutputJsonSchema() string {
 
 }
 
-func (self *SchemaTree) OutputStruct() string {
-	return self.Root.OutputStruct(true)
+func (self *SchemaTree) OutputStruct(withTags bool) string {
+	return self.Root.OutputStruct(true, withTags)
 }
diff --git a/tool/parquet-tools/parquet-tools.go b/tool/parquet-tools/parquet-tools.go
@@ -12,6 +12,7 @@ import (
 func main() {
 	cmd := flag.String("cmd", "schema", "command")
 	fileName := flag.String("file", "", "file name")
+	withTags := flag.Bool("tag", false, "show struct tags")
 
 	flag.Parse()
 
@@ -30,7 +31,7 @@ func main() {
 	if *cmd == "schema" {
 		tree := SchemaTool.CreateSchemaTree(pr.SchemaHandler.SchemaElements)
 		fmt.Println("----- Go struct -----")
-		fmt.Printf("%s\n", tree.OutputStruct())
+		fmt.Printf("%s\n", tree.OutputStruct(*withTags))
 		fmt.Println("----- Json schema -----")
 		fmt.Printf("%s\n", tree.OutputJsonSchema())
 	} else {