Add markdown support

_automation/grammars.json

@@ -322,5 +322,16 @@
     "reference": "v0.5.0",
     "revision": "6129a83eeec7d6070b1c0567ec7ce3509ead607c",
     "updateBasedOn": "tag"
+  },
+  {
+    "language": "markdown",
+    "url": "https://github.com/tree-sitter-grammars/tree-sitter-markdown",
+    "files": [
+      "parser.c",
+      "scanner.c"
+    ],
+    "reference": "v0.2.3",
+    "revision": "62516e8c78380e3b51d5b55727995d2c511436d8",
+    "updateBasedOn": "tag"
   }
 ]

_automation/main.go

@@ -239,6 +239,8 @@ func (s *UpdateService) downloadGrammar(ctx context.Context, g *Grammar) {
 		s.downloadYaml(ctx, g)
 	case "php":
 		s.downloadPhp(ctx, g)
+	case "markdown":
+		s.downloadMarkdown(ctx, g)
 	default:
 		s.defaultGrammarDownload(ctx, g)
 	}
@@ -434,6 +436,34 @@ func (s *UpdateService) downloadTypescript(ctx context.Context, g *Grammar) {
 	}
 }
 
+// markdown is special as it contains 2 different grammars
+func (s *UpdateService) downloadMarkdown(ctx context.Context, g *Grammar) {
+	url := g.ContentURL()
+
+	langs := []string{"tree-sitter-markdown", "tree-sitter-markdown-inline"}
+	for _, lang := range langs {
+		s.makeDir(ctx, fmt.Sprintf("%s/%s", g.Language, lang))
+
+		s.downloadFile(
+			ctx,
+			fmt.Sprintf("%s/%s/%s/src/tree_sitter/parser.h", url, g.Revision, lang),
+			fmt.Sprintf("%s/%s/parser.h", g.Language, lang),
+			nil,
+		)
+
+		for _, f := range g.Files {
+			s.downloadFile(
+				ctx,
+				fmt.Sprintf("%s/%s/%s/src/%s", url, g.Revision, lang, f),
+				fmt.Sprintf("%s/%s/%s", g.Language, lang, f),
+				map[string]string{
+					`"tree_sitter/parser.h"`: `"parser.h"`,
+				},
+			)
+		}
+	}
+}
+
 // for yaml grammar scanner.cc includes schema.generated.cc file
 // it causes cgo to compile schema.generated.cc twice and throw duplicate symbols error
 func (s *UpdateService) downloadYaml(ctx context.Context, g *Grammar) {

bindings.go

@@ -405,6 +405,10 @@ func (t SymbolType) String() string {
 	return symbolTypeNames[t]
 }
 
+func (n Node) ID() uintptr {
+	return uintptr(n.c.id)
+}
+
 // StartByte returns the node's start byte.
 func (n Node) StartByte() uint32 {
 	return uint32(C.ts_node_start_byte(n.c))
@@ -433,6 +437,15 @@ func (n Node) EndPoint() Point {
 	}
 }
 
+func (n Node) Range() Range {
+	return Range{
+		StartByte:  n.StartByte(),
+		EndByte:    n.EndByte(),
+		StartPoint: n.StartPoint(),
+		EndPoint:   n.EndPoint(),
+	}
+}
+
 // Symbol returns the node's type as a Symbol.
 func (n Node) Symbol() Symbol {
 	return C.ts_node_symbol(n.c)

markdown/binding.go

@@ -0,0 +1,156 @@
+package markdown
+
+import (
+	"context"
+
+	sitter "github.com/smacker/go-tree-sitter"
+	tree_sitter_markdown "github.com/smacker/go-tree-sitter/markdown/tree-sitter-markdown"
+	tree_sitter_markdown_inline "github.com/smacker/go-tree-sitter/markdown/tree-sitter-markdown-inline"
+)
+
+type MarkdownTree struct {
+	blockTree     *sitter.Tree
+	inlineTrees   []*sitter.Tree
+	inlineIndices map[uintptr]int
+}
+
+func (t *MarkdownTree) Edit(edit sitter.EditInput) {
+	t.blockTree.Edit(edit)
+	for _, tree := range t.inlineTrees {
+		tree.Edit(edit)
+	}
+}
+
+func (t *MarkdownTree) BlockTree() *sitter.Tree {
+	return t.blockTree
+}
+
+func (t *MarkdownTree) InlineTree(parent *sitter.Node) *sitter.Tree {
+	if parent == nil {
+		return nil
+	}
+
+	index, ok := t.inlineIndices[parent.ID()]
+	if ok {
+		return t.inlineTrees[index]
+	}
+
+	return nil
+}
+
+func (t *MarkdownTree) InlineRootNode(parent *sitter.Node) *sitter.Node {
+	tree := t.InlineTree(parent)
+	if tree == nil {
+		return nil
+	}
+
+	return tree.RootNode()
+}
+
+func (t *MarkdownTree) InlineTrees() []*sitter.Tree {
+	return t.inlineTrees
+}
+
+func (t *MarkdownTree) Iter(f func(node *Node) bool) {
+	root := t.blockTree.RootNode()
+	t.iter(&Node{root, t.InlineRootNode(root)}, f)
+}
+
+func (t *MarkdownTree) iter(node *Node, f func(node *Node) bool) (goNext bool) {
+	goNext = f(node)
+	if !goNext {
+		return goNext
+	}
+
+	childCount := node.NamedChildCount()
+	for i := 0; i < int(childCount); i++ {
+		child := node.NamedChild(i)
+
+		goNext = t.iter(&Node{Node: child, Inline: t.InlineRootNode(child)}, f)
+		if !goNext {
+			return goNext
+		}
+	}
+
+	return true
+}
+
+type Node struct {
+	*sitter.Node
+	Inline *sitter.Node
+}
+
+func ParseCtx(ctx context.Context, oldTree *MarkdownTree, content []byte) (*MarkdownTree, error) {
+	p := sitter.NewParser()
+	p.SetLanguage(tree_sitter_markdown.GetLanguage())
+
+	var old *sitter.Tree
+	if oldTree != nil {
+		old = oldTree.blockTree
+	}
+	tree, err := p.ParseCtx(ctx, old, content)
+	if err != nil {
+		return nil, err
+	}
+
+	res := &MarkdownTree{
+		blockTree:     tree,
+		inlineTrees:   []*sitter.Tree{},
+		inlineIndices: map[uintptr]int{},
+	}
+
+	p.SetLanguage(tree_sitter_markdown_inline.GetLanguage())
+
+	q, err := sitter.NewQuery([]byte(`(inline) @inline`), tree_sitter_markdown.GetLanguage())
+	if err != nil {
+		return nil, err
+	}
+
+	qc := sitter.NewQueryCursor()
+	qc.Exec(q, tree.RootNode())
+
+	idx := int(0)
+	for {
+		match, ok := qc.NextMatch()
+		if !ok {
+			break
+		}
+
+		for _, capture := range match.Captures {
+			r := capture.Node.Range()
+			ranges := []sitter.Range{}
+			for i := 0; i < int(capture.Node.NamedChildCount()); i++ {
+				child := capture.Node.NamedChild(i)
+				childRange := child.Range()
+				ranges = append(ranges, sitter.Range{
+					StartPoint: r.StartPoint,
+					StartByte:  r.StartByte,
+					EndPoint:   childRange.EndPoint,
+					EndByte:    childRange.EndByte,
+				})
+
+				r.StartPoint = childRange.EndPoint
+				r.StartByte = childRange.EndByte
+			}
+
+			ranges = append(ranges, r)
+			p.SetIncludedRanges(ranges)
+			var old *sitter.Tree
+			if oldTree != nil && idx < len(oldTree.inlineTrees) {
+				old = oldTree.inlineTrees[idx]
+			}
+
+			inlineTree, err := p.ParseCtx(ctx, old, content)
+			if err != nil {
+				return nil, err
+			}
+
+			res.inlineTrees = append(res.inlineTrees, inlineTree)
+			res.inlineIndices[capture.Node.ID()] = idx
+			idx++
+		}
+	}
+	qc.Close()
+
+	return res, nil
+}

markdown/binding_test.go

@@ -0,0 +1,101 @@
+package markdown_test
+
+import (
+	"context"
+	"testing"
+
+	"github.com/smacker/go-tree-sitter/markdown"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestMarkdown(t *testing.T) {
+	assert := assert.New(t)
+
+	content := "# Hello\n- This is a image: ![image](https://example.com/image.jpg \"a image\")"
+	tree, err := markdown.ParseCtx(context.Background(), nil, []byte(content))
+	assert.NoError(err)
+
+	assert.Equal(
+		"(document (section (atx_heading (atx_h1_marker) heading_content: (inline)) (list (list_item (list_marker_minus) (paragraph (inline))))))",
+		tree.BlockTree().RootNode().String(),
+	)
+
+	assert.Equal(
+		"(inline)",
+		tree.InlineTrees()[0].RootNode().String(),
+	)
+
+	assert.Equal(
+		"(inline (image (image_description) (link_destination) (link_title)))",
+		tree.InlineTrees()[1].RootNode().String(),
+	)
+}
+
+func TestIter(t *testing.T) {
+	assert := assert.New(t)
+
+	content := "# Hello\n- This two image: ![image](https://example.com/image.jpg \"a image\"), ![apple](https://example.com/apple.jpg \"a apple\")"
+	tree, err := markdown.ParseCtx(context.Background(), nil, []byte(content))
+	assert.NoError(err)
+
+	type BlockWithInline struct {
+		Node       string
+		InlineNode string
+	}
+
+	expected := []BlockWithInline{
+		{"(document (section (atx_heading (atx_h1_marker) heading_content: (inline)) (list (list_item (list_marker_minus) (paragraph (inline))))))", ""},
+		{"(section (atx_heading (atx_h1_marker) heading_content: (inline)) (list (list_item (list_marker_minus) (paragraph (inline)))))", ""},
+		{"(atx_heading (atx_h1_marker) heading_content: (inline))", ""},
+		{"(atx_h1_marker)", ""},
+		{"(inline)", "(inline)"},
+		{"(list (list_item (list_marker_minus) (paragraph (inline))))", ""},
+		{"(list_item (list_marker_minus) (paragraph (inline)))", ""},
+		{"(list_marker_minus)", ""},
+		{"(paragraph (inline))", ""},
+		{"(inline)", "(inline (image (image_description) (link_destination) (link_title)) (image (image_description) (link_destination) (link_title)))"},
+	}
+
+	i := int(0)
+	tree.Iter(func(node *markdown.Node) bool {
+		assert.Equal(expected[i].Node, node.String(), "node mismatch. idx: %d", i)
+		if expected[i].InlineNode != "" || node.Inline != nil {
+			assert.Equal(expected[i].InlineNode, node.Inline.String(), "inline node mismatch. idx: %d", i)
+		}
+
+		i++
+		return true
+	})
+}
+
+func TestIterStop(t *testing.T) {
+	assert := assert.New(t)
+
+	content := "# Hello\n- This two image: ![image](https://example.com/image.jpg \"a image\"), ![apple](https://example.com/apple.jpg \"a apple\")"
+	tree, err := markdown.ParseCtx(context.Background(), nil, []byte(content))
+	assert.NoError(err)
+
+	type BlockWithInline struct {
+		Node       string
+		InlineNode string
+	}
+
+	expected := []string{
+		"(document (section (atx_heading (atx_h1_marker) heading_content: (inline)) (list (list_item (list_marker_minus) (paragraph (inline))))))",
+		"(section (atx_heading (atx_h1_marker) heading_content: (inline)) (list (list_item (list_marker_minus) (paragraph (inline)))))",
+		"(atx_heading (atx_h1_marker) heading_content: (inline))",
+	}
+
+	collected := []string{}
+	tree.Iter(func(node *markdown.Node) bool {
+		collected = append(collected, node.String())
+
+		if node.Type() == "document" || node.Type() == "section" {
+			return true
+		}
+
+		return false
+	})
+
+	assert.Equal(expected, collected)
+}

markdown/tree-sitter-markdown-inline/binding.go

@@ -0,0 +1,15 @@
+package tree_sitter_markdown_inline
+
+//#include "parser.h"
+//TSLanguage *tree_sitter_markdown_inline();
+import "C"
+import (
+	"unsafe"
+
+	sitter "github.com/smacker/go-tree-sitter"
+)
+
+func GetLanguage() *sitter.Language {
+	ptr := unsafe.Pointer(C.tree_sitter_markdown_inline())
+	return sitter.NewLanguage(ptr)
+}

markdown/tree-sitter-markdown-inline/bingding_test.go

@@ -0,0 +1,21 @@
+package tree_sitter_markdown_inline_test
+
+import (
+	"context"
+	"testing"
+
+	sitter "github.com/smacker/go-tree-sitter"
+	tree_sitter_markdown_inline "github.com/smacker/go-tree-sitter/markdown/tree-sitter-markdown-inline"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestGrammar(t *testing.T) {
+	assert := assert.New(t)
+
+	n, err := sitter.ParseCtx(context.Background(), []byte("# Hello world!\n- Here is a picture: ![picture](https://example.com/picture.png)"), tree_sitter_markdown_inline.GetLanguage())
+	assert.NoError(err)
+	assert.Equal(
+		"(inline (image (image_description) (link_destination)))",
+		n.String(),
+	)
+}