Added sort_by operator

pkg/yqlib/doc/Sort.md

@@ -1,19 +0,0 @@
-
-## Sort by string field
-Given a sample.yml file of:
-```yaml
-- a: banana
-- a: cat
-- a: apple
-```
-then
-```bash
-yq eval 'sort_by(.a)' sample.yml
-```
-will output
-```yaml
-- a: apple
-- a: banana
-- a: cat
-```
-

pkg/yqlib/doc/headers/sort.md

@@ -0,0 +1,5 @@
+# Sort
+
+Sorts an array. Use `sort` to sort an array as is, or `sort_by` to sort by a particular subfield.
+
+Note that at this stage, `yq` only sorts scalar fields.

pkg/yqlib/doc/sort.md

@@ -0,0 +1,98 @@
+# Sort
+
+Sorts an array. Use `sort` to sort an array as is, or `sort_by` to sort by a particular subfield.
+
+Note that at this stage, `yq` only sorts scalar fields.
+
+## Sort by string field
+Given a sample.yml file of:
+```yaml
+- a: banana
+- a: cat
+- a: apple
+```
+then
+```bash
+yq eval 'sort_by(.a)' sample.yml
+```
+will output
+```yaml
+- a: apple
+- a: banana
+- a: cat
+```
+
+## Sort is stable
+Note the order of the elements in unchanged when equal in sorting.
+
+Given a sample.yml file of:
+```yaml
+- a: banana
+  b: 1
+- a: banana
+  b: 2
+- a: banana
+  b: 3
+- a: banana
+  b: 4
+```
+then
+```bash
+yq eval 'sort_by(.a)' sample.yml
+```
+will output
+```yaml
+- a: banana
+  b: 1
+- a: banana
+  b: 2
+- a: banana
+  b: 3
+- a: banana
+  b: 4
+```
+
+## Sort by numeric field
+Given a sample.yml file of:
+```yaml
+- a: 10
+- a: 100
+- a: 1
+```
+then
+```bash
+yq eval 'sort_by(.a)' sample.yml
+```
+will output
+```yaml
+- a: 1
+- a: 10
+- a: 100
+```
+
+## Sort, nulls come first
+Given a sample.yml file of:
+```yaml
+- 8
+- 3
+- null
+- 6
+- true
+- false
+- cat
+```
+then
+```bash
+yq eval 'sort' sample.yml
+```
+will output
+```yaml
+- null
+- false
+- true
+- 3
+- 6
+- 8
+- cat
+```
+

pkg/yqlib/expression_tokeniser.go

@@ -352,6 +352,8 @@ func initLexer() (*lex.Lexer, error) {
 	lexer.Add([]byte(`from_json`), opToken(decodeOpType))
 
 	lexer.Add([]byte(`sortKeys`), opToken(sortKeysOpType))
+	lexer.Add([]byte(`sort_keys`), opToken(sortKeysOpType))
+
 	lexer.Add([]byte(`load`), opTokenWithPrefs(loadOpType, nil, loadPrefs{loadAsString: false}))
 	lexer.Add([]byte(`strload`), opTokenWithPrefs(loadOpType, nil, loadPrefs{loadAsString: true}))
 	lexer.Add([]byte(`select`), opToken(selectOpType))
@@ -377,6 +379,7 @@ func initLexer() (*lex.Lexer, error) {
 	lexer.Add([]byte(`capture`), opToken(captureOpType))
 	lexer.Add([]byte(`test`), opToken(testOpType))
 
+	lexer.Add([]byte(`sort`), opToken(sortOpType))
 	lexer.Add([]byte(`sort_by`), opToken(sortByOpType))
 
 	lexer.Add([]byte(`any`), opToken(anyOpType))

pkg/yqlib/lib.go

@@ -99,6 +99,7 @@ var getPathOpType = &operationType{Type: "GET_PATH", NumArgs: 0, Precedence: 50,
 
 var explodeOpType = &operationType{Type: "EXPLODE", NumArgs: 1, Precedence: 50, Handler: explodeOperator}
 var sortByOpType = &operationType{Type: "SORT_BY", NumArgs: 1, Precedence: 50, Handler: sortByOperator}
+var sortOpType = &operationType{Type: "SORT", NumArgs: 0, Precedence: 50, Handler: sortOperator}
 var sortKeysOpType = &operationType{Type: "SORT_KEYS", NumArgs: 1, Precedence: 50, Handler: sortKeysOperator}
 var joinStringOpType = &operationType{Type: "JOIN", NumArgs: 1, Precedence: 50, Handler: joinStringOperator}
 var subStringOpType = &operationType{Type: "SUBSTR", NumArgs: 1, Precedence: 50, Handler: substituteStringOperator}

pkg/yqlib/operator_sort.go

@@ -10,6 +10,12 @@ import (
 	yaml "gopkg.in/yaml.v3"
 )
 
+func sortOperator(d *dataTreeNavigator, context Context, expressionNode *ExpressionNode) (Context, error) {
+	selfExpression := &ExpressionNode{Operation: &Operation{OperationType: selfReferenceOpType}}
+	expressionNode.Rhs = selfExpression
+	return sortByOperator(d, context, expressionNode)
+}
+
 // context represents the current matching nodes in the expression pipeline
 //expressionNode is your current expression (sort_by)
 func sortByOperator(d *dataTreeNavigator, context Context, expressionNode *ExpressionNode) (Context, error) {
@@ -44,9 +50,13 @@ func sortByOperator(d *dataTreeNavigator, context Context, expressionNode *Expre
 
 			sortableArray[i] = sortableNode{Node: originalNode, NodeToCompare: nodeToCompare}
 
+			if nodeToCompare.Kind != yaml.ScalarNode {
+				return Context{}, fmt.Errorf("sort only works for scalars, got %v", nodeToCompare.Tag)
 			}
 
-		sort.Sort(sortableArray)
+		}
+
+		sort.Stable(sortableArray)
 
 		sortedList := &yaml.Node{Kind: yaml.SequenceNode, Tag: "!!seq", Style: candidateNode.Style}
 		sortedList.Content = make([]*yaml.Node, len(candidateNode.Content))
@@ -73,7 +83,27 @@ func (a sortableNodeArray) Less(i, j int) bool {
 	lhs := a[i].NodeToCompare
 	rhs := a[j].NodeToCompare
 
-	if lhs.Tag != rhs.Tag || lhs.Tag == "!!str" {
+	if lhs.Tag == "!!null" && rhs.Tag != "!!null" {
+		return true
+	} else if lhs.Tag != "!!null" && rhs.Tag == "!!null" {
+		return false
+	} else if lhs.Tag == "!!bool" && rhs.Tag != "!!bool" {
+		return true
+	} else if lhs.Tag != "!!bool" && rhs.Tag == "!!bool" {
+		return false
+	} else if lhs.Tag == "!!bool" && rhs.Tag == "!!bool" {
+		lhsTruthy, err := isTruthyNode(lhs)
+		if err != nil {
+			panic(fmt.Errorf("could not parse %v as boolean: %w", lhs.Value, err))
+		}
+
+		rhsTruthy, err := isTruthyNode(rhs)
+		if err != nil {
+			panic(fmt.Errorf("could not parse %v as boolean: %w", rhs.Value, err))
+		}
+
+		return !lhsTruthy && rhsTruthy
+	} else if lhs.Tag != rhs.Tag || lhs.Tag == "!!str" {
 		return strings.Compare(lhs.Value, rhs.Value) < 0
 	} else if lhs.Tag == "!!int" && rhs.Tag == "!!int" {
 		_, lhsNum, err := parseInt(lhs.Value)

pkg/yqlib/operator_sort_test.go

@@ -8,22 +8,56 @@ var sortByOperatorScenarios = []expressionScenario{
 		document:    "[{a: banana},{a: cat},{a: apple}]",
 		expression:  `sort_by(.a)`,
 		expected: []string{
-			"D0, P[], (!!bool)::true\n",
+			"D0, P[], (!!seq)::[{a: apple}, {a: banana}, {a: cat}]\n",
+		},
+	},
+	{
+		description:    "Sort is stable",
+		subdescription: "Note the order of the elements in unchanged when equal in sorting.",
+		document:       "[{a: banana, b: 1}, {a: banana, b: 2}, {a: banana, b: 3}, {a: banana, b: 4}]",
+		expression:     `sort_by(.a)`,
+		expected: []string{
+			"D0, P[], (!!seq)::[{a: banana, b: 1}, {a: banana, b: 2}, {a: banana, b: 3}, {a: banana, b: 4}]\n",
+		},
+	},
+	{
+		description: "Sort by numeric field",
+		document:    "[{a: 10},{a: 100},{a: 1}]",
+		expression:  `sort_by(.a)`,
+		expected: []string{
+			"D0, P[], (!!seq)::[{a: 1}, {a: 10}, {a: 100}]\n",
+		},
+	},
+	{
+		skipDoc:    true,
+		document:   "[{a: 1.1},{a: 1.001},{a: 1.01}]",
+		expression: `sort_by(.a)`,
+		expected: []string{
+			"D0, P[], (!!seq)::[{a: 1.001}, {a: 1.01}, {a: 1.1}]\n",
+		},
+	},
+	{
+		description: "Sort, nulls come first",
+		document:    "[8,3,null,6, true, false, cat]",
+		expression:  `sort`,
+		expected: []string{
+			"D0, P[], (!!seq)::[null, false, true, 3, 6, 8, cat]\n",
+		},
+	},
+	{
+		skipDoc:     true,
+		description: "false before true",
+		document:    "[{a: false, b: 1}, {a: true, b: 2}, {a: false, b: 3}]",
+		expression:  `sort_by(.a)`,
+		expected: []string{
+			"D0, P[], (!!seq)::[{a: false, b: 1}, {a: false, b: 3}, {a: true, b: 2}]\n",
 		},
 	},
-	// {
-	// 	description: "Sort, nulls come first",
-	// 	document:    "[8,3,null,6]",
-	// 	expression:  `sort`,
-	// 	expected: []string{
-	// 		"D0, P[], (!!bool)::[null,3,6,8]\n",
-	// 	},
-	// },
 }
 
 func TestSortByOperatorScenarios(t *testing.T) {
 	for _, tt := range sortByOperatorScenarios {
 		testScenario(t, &tt)
 	}
-	documentScenarios(t, "Sort", sortByOperatorScenarios)
+	documentScenarios(t, "sort", sortByOperatorScenarios)
 }