feat: extract title from first H1 heading into memo property (#5726)

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

plugin/markdown/markdown.go

@@ -138,6 +138,26 @@ func (s *service) ExtractTags(content []byte) ([]string, error) {
 	return uniquePreserveCase(tags), nil
 }
 
+// extractHeadingText extracts plain text content from a heading node.
+func extractHeadingText(n gast.Node, source []byte) string {
+	var buf strings.Builder
+	for child := n.FirstChild(); child != nil; child = child.NextSibling() {
+		extractTextFromNode(child, source, &buf)
+	}
+	return buf.String()
+}
+
+// extractTextFromNode recursively extracts plain text from a node and its children.
+func extractTextFromNode(n gast.Node, source []byte, buf *strings.Builder) {
+	if textNode, ok := n.(*gast.Text); ok {
+		buf.Write(textNode.Segment.Value(source))
+		return
+	}
+	for child := n.FirstChild(); child != nil; child = child.NextSibling() {
+		extractTextFromNode(child, source, buf)
+	}
+}
+
 // ExtractProperties computes boolean properties about the content.
 func (s *service) ExtractProperties(content []byte) (*storepb.MemoPayload_Property, error) {
 	root, err := s.parse(content)
@@ -146,12 +166,21 @@ func (s *service) ExtractProperties(content []byte) (*storepb.MemoPayload_Proper
 	}
 
 	prop := &storepb.MemoPayload_Property{}
+	firstBlockChecked := false
 
 	err = gast.Walk(root, func(n gast.Node, entering bool) (gast.WalkStatus, error) {
 		if !entering {
 			return gast.WalkContinue, nil
 		}
 
+		// Check if the first block-level child of the document is an H1 heading.
+		if !firstBlockChecked && n.Parent() != nil && n.Parent().Kind() == gast.KindDocument {
+			firstBlockChecked = true
+			if heading, ok := n.(*gast.Heading); ok && heading.Level == 1 {
+				prop.Title = extractHeadingText(n, content)
+			}
+		}
+
 		switch n.Kind() {
 		case gast.KindLink:
 			prop.HasLink = true
@@ -302,6 +331,8 @@ func (s *service) ExtractAll(content []byte) (*ExtractedData, error) {
 		Property: &storepb.MemoPayload_Property{},
 	}
 
+	firstBlockChecked := false
+
 	// Single walk to collect all data
 	err = gast.Walk(root, func(n gast.Node, entering bool) (gast.WalkStatus, error) {
 		if !entering {
@@ -313,6 +344,14 @@ func (s *service) ExtractAll(content []byte) (*ExtractedData, error) {
 			data.Tags = append(data.Tags, string(tagNode.Tag))
 		}
 
+		// Check if the first block-level child of the document is an H1 heading.
+		if !firstBlockChecked && n.Parent() != nil && n.Parent().Kind() == gast.KindDocument {
+			firstBlockChecked = true
+			if heading, ok := n.(*gast.Heading); ok && heading.Level == 1 {
+				data.Property.Title = extractHeadingText(n, content)
+			}
+		}
+
 		// Extract properties based on node kind
 		switch n.Kind() {
 		case gast.KindLink:

plugin/markdown/markdown_test.go

@@ -190,6 +190,7 @@ func TestExtractProperties(t *testing.T) {
 		hasCode  bool
 		hasTasks bool
 		hasInc   bool
+		title    string
 	}{
 		{
 			name:     "plain text",
@@ -198,6 +199,7 @@ func TestExtractProperties(t *testing.T) {
 			hasCode:  false,
 			hasTasks: false,
 			hasInc:   false,
+			title:    "",
 		},
 		{
 			name:     "with link",
@@ -206,6 +208,7 @@ func TestExtractProperties(t *testing.T) {
 			hasCode:  false,
 			hasTasks: false,
 			hasInc:   false,
+			title:    "",
 		},
 		{
 			name:     "with inline code",
@@ -214,6 +217,7 @@ func TestExtractProperties(t *testing.T) {
 			hasCode:  true,
 			hasTasks: false,
 			hasInc:   false,
+			title:    "",
 		},
 		{
 			name:     "with code block",
@@ -222,6 +226,7 @@ func TestExtractProperties(t *testing.T) {
 			hasCode:  true,
 			hasTasks: false,
 			hasInc:   false,
+			title:    "",
 		},
 		{
 			name:     "with completed task",
@@ -230,6 +235,7 @@ func TestExtractProperties(t *testing.T) {
 			hasCode:  false,
 			hasTasks: true,
 			hasInc:   false,
+			title:    "",
 		},
 		{
 			name:     "with incomplete task",
@@ -238,6 +244,7 @@ func TestExtractProperties(t *testing.T) {
 			hasCode:  false,
 			hasTasks: true,
 			hasInc:   true,
+			title:    "",
 		},
 		{
 			name:     "mixed tasks",
@@ -246,6 +253,7 @@ func TestExtractProperties(t *testing.T) {
 			hasCode:  false,
 			hasTasks: true,
 			hasInc:   true,
+			title:    "",
 		},
 		{
 			name:     "everything",
@@ -254,6 +262,32 @@ func TestExtractProperties(t *testing.T) {
 			hasCode:  true,
 			hasTasks: true,
 			hasInc:   true,
+			title:    "Title",
+		},
+		{
+			name:    "h1 as first node extracts title",
+			content: "# My Article Title\n\nBody text here.",
+			title:   "My Article Title",
+		},
+		{
+			name:    "h2 as first node does not extract title",
+			content: "## Sub Heading\n\nBody text.",
+			title:   "",
+		},
+		{
+			name:    "h1 not first node does not extract title",
+			content: "Some text\n\n# Heading Later",
+			title:   "",
+		},
+		{
+			name:    "h1 with inline formatting extracts plain text",
+			content: "# Title with **bold** and *italic*\n\nBody.",
+			title:   "Title with bold and italic",
+		},
+		{
+			name:    "empty content has no title",
+			content: "",
+			title:   "",
 		},
 	}
 
@@ -267,6 +301,41 @@ func TestExtractProperties(t *testing.T) {
 			assert.Equal(t, tt.hasCode, props.HasCode, "HasCode")
 			assert.Equal(t, tt.hasTasks, props.HasTaskList, "HasTaskList")
 			assert.Equal(t, tt.hasInc, props.HasIncompleteTasks, "HasIncompleteTasks")
+			assert.Equal(t, tt.title, props.Title, "Title")
+		})
+	}
+}
+
+func TestExtractAllTitle(t *testing.T) {
+	svc := NewService(WithTagExtension())
+
+	tests := []struct {
+		name    string
+		content string
+		title   string
+	}{
+		{
+			name:    "h1 first node",
+			content: "# Article Title\n\nContent with #tag",
+			title:   "Article Title",
+		},
+		{
+			name:    "no h1",
+			content: "Just text with #tag",
+			title:   "",
+		},
+		{
+			name:    "h1 not first",
+			content: "Intro\n\n# Late Heading",
+			title:   "",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			data, err := svc.ExtractAll([]byte(tt.content))
+			require.NoError(t, err)
+			assert.Equal(t, tt.title, data.Property.Title, "Title")
 		})
 	}
 }

proto/api/v1/memo_service.proto

@@ -227,6 +227,8 @@ message Memo {
     bool has_task_list = 2;
     bool has_code = 3;
     bool has_incomplete_tasks = 4;
+    // The title extracted from the first H1 heading, if present.
+    string title = 5;
   }
 }
 

proto/gen/api/v1/memo_service.pb.go

@@ -1679,8 +1679,10 @@ type Memo_Property struct {
 	HasTaskList        bool                   `protobuf:"varint,2,opt,name=has_task_list,json=hasTaskList,proto3" json:"has_task_list,omitempty"`
 	HasCode            bool                   `protobuf:"varint,3,opt,name=has_code,json=hasCode,proto3" json:"has_code,omitempty"`
 	HasIncompleteTasks bool                   `protobuf:"varint,4,opt,name=has_incomplete_tasks,json=hasIncompleteTasks,proto3" json:"has_incomplete_tasks,omitempty"`
-	unknownFields      protoimpl.UnknownFields
-	sizeCache          protoimpl.SizeCache
+	// The title extracted from the first H1 heading, if present.
+	Title         string `protobuf:"bytes,5,opt,name=title,proto3" json:"title,omitempty"`
+	unknownFields protoimpl.UnknownFields
+	sizeCache     protoimpl.SizeCache
 }
 
 func (x *Memo_Property) Reset() {
@@ -1741,6 +1743,13 @@ func (x *Memo_Property) GetHasIncompleteTasks() bool {
 	return false
 }
 
+func (x *Memo_Property) GetTitle() string {
+	if x != nil {
+		return x.Title
+	}
+	return ""
+}
+
 // Memo reference in relations.
 type MemoRelation_Memo struct {
 	state protoimpl.MessageState `protogen:"open.v1"`
@@ -1812,7 +1821,7 @@ const file_api_v1_memo_service_proto_rawDesc = "" +
 	"\rreaction_type\x18\x04 \x01(\tB\x03\xe0A\x02R\freactionType\x12@\n" +
 	"\vcreate_time\x18\x05 \x01(\v2\x1a.google.protobuf.TimestampB\x03\xe0A\x03R\n" +
 	"createTime:X\xeaAU\n" +
-	"\x15memos.api.v1/Reaction\x12!memos/{memo}/reactions/{reaction}\x1a\x04name*\treactions2\breaction\"\xd8\b\n" +
+	"\x15memos.api.v1/Reaction\x12!memos/{memo}/reactions/{reaction}\x1a\x04name*\treactions2\breaction\"\xee\b\n" +
 	"\x04Memo\x12\x17\n" +
 	"\x04name\x18\x01 \x01(\tB\x03\xe0A\bR\x04name\x12.\n" +
 	"\x05state\x18\x02 \x01(\x0e2\x13.memos.api.v1.StateB\x03\xe0A\x02R\x05state\x123\n" +
@@ -1837,12 +1846,13 @@ const file_api_v1_memo_service_proto_rawDesc = "" +
 	"\x06parent\x18\x10 \x01(\tB\x19\xe0A\x03\xfaA\x13\n" +
 	"\x11memos.api.v1/MemoH\x00R\x06parent\x88\x01\x01\x12\x1d\n" +
 	"\asnippet\x18\x11 \x01(\tB\x03\xe0A\x03R\asnippet\x12<\n" +
-	"\blocation\x18\x12 \x01(\v2\x16.memos.api.v1.LocationB\x03\xe0A\x01H\x01R\blocation\x88\x01\x01\x1a\x96\x01\n" +
+	"\blocation\x18\x12 \x01(\v2\x16.memos.api.v1.LocationB\x03\xe0A\x01H\x01R\blocation\x88\x01\x01\x1a\xac\x01\n" +
 	"\bProperty\x12\x19\n" +
 	"\bhas_link\x18\x01 \x01(\bR\ahasLink\x12\"\n" +
 	"\rhas_task_list\x18\x02 \x01(\bR\vhasTaskList\x12\x19\n" +
 	"\bhas_code\x18\x03 \x01(\bR\ahasCode\x120\n" +
-	"\x14has_incomplete_tasks\x18\x04 \x01(\bR\x12hasIncompleteTasks:7\xeaA4\n" +
+	"\x14has_incomplete_tasks\x18\x04 \x01(\bR\x12hasIncompleteTasks\x12\x14\n" +
+	"\x05title\x18\x05 \x01(\tR\x05title:7\xeaA4\n" +
 	"\x11memos.api.v1/Memo\x12\fmemos/{memo}\x1a\x04name*\x05memos2\x04memoB\t\n" +
 	"\a_parentB\v\n" +
 	"\t_location\"u\n" +

proto/gen/openapi.yaml

@@ -2590,6 +2590,9 @@ components:
                     type: boolean
                 hasIncompleteTasks:
                     type: boolean
+                title:
+                    type: string
+                    description: The title extracted from the first H1 heading, if present.
             description: Computed properties of a memo.
         OAuth2Config:
             type: object

proto/gen/store/memo.pb.go

@@ -88,8 +88,10 @@ type MemoPayload_Property struct {
 	HasTaskList        bool                   `protobuf:"varint,2,opt,name=has_task_list,json=hasTaskList,proto3" json:"has_task_list,omitempty"`
 	HasCode            bool                   `protobuf:"varint,3,opt,name=has_code,json=hasCode,proto3" json:"has_code,omitempty"`
 	HasIncompleteTasks bool                   `protobuf:"varint,4,opt,name=has_incomplete_tasks,json=hasIncompleteTasks,proto3" json:"has_incomplete_tasks,omitempty"`
-	unknownFields      protoimpl.UnknownFields
-	sizeCache          protoimpl.SizeCache
+	// The title extracted from the first H1 heading, if present.
+	Title         string `protobuf:"bytes,5,opt,name=title,proto3" json:"title,omitempty"`
+	unknownFields protoimpl.UnknownFields
+	sizeCache     protoimpl.SizeCache
 }
 
 func (x *MemoPayload_Property) Reset() {
@@ -150,6 +152,13 @@ func (x *MemoPayload_Property) GetHasIncompleteTasks() bool {
 	return false
 }
 
+func (x *MemoPayload_Property) GetTitle() string {
+	if x != nil {
+		return x.Title
+	}
+	return ""
+}
+
 type MemoPayload_Location struct {
 	state         protoimpl.MessageState `protogen:"open.v1"`
 	Placeholder   string                 `protobuf:"bytes,1,opt,name=placeholder,proto3" json:"placeholder,omitempty"`
@@ -214,16 +223,17 @@ var File_store_memo_proto protoreflect.FileDescriptor
 
 const file_store_memo_proto_rawDesc = "" +
 	"\n" +
-	"\x10store/memo.proto\x12\vmemos.store\"\xa0\x03\n" +
+	"\x10store/memo.proto\x12\vmemos.store\"\xb6\x03\n" +
 	"\vMemoPayload\x12=\n" +
 	"\bproperty\x18\x01 \x01(\v2!.memos.store.MemoPayload.PropertyR\bproperty\x12=\n" +
 	"\blocation\x18\x02 \x01(\v2!.memos.store.MemoPayload.LocationR\blocation\x12\x12\n" +
-	"\x04tags\x18\x03 \x03(\tR\x04tags\x1a\x96\x01\n" +
+	"\x04tags\x18\x03 \x03(\tR\x04tags\x1a\xac\x01\n" +
 	"\bProperty\x12\x19\n" +
 	"\bhas_link\x18\x01 \x01(\bR\ahasLink\x12\"\n" +
 	"\rhas_task_list\x18\x02 \x01(\bR\vhasTaskList\x12\x19\n" +
 	"\bhas_code\x18\x03 \x01(\bR\ahasCode\x120\n" +
-	"\x14has_incomplete_tasks\x18\x04 \x01(\bR\x12hasIncompleteTasks\x1af\n" +
+	"\x14has_incomplete_tasks\x18\x04 \x01(\bR\x12hasIncompleteTasks\x12\x14\n" +
+	"\x05title\x18\x05 \x01(\tR\x05title\x1af\n" +
 	"\bLocation\x12 \n" +
 	"\vplaceholder\x18\x01 \x01(\tR\vplaceholder\x12\x1a\n" +
 	"\blatitude\x18\x02 \x01(\x01R\blatitude\x12\x1c\n" +

proto/store/memo.proto

@@ -17,6 +17,8 @@ message MemoPayload {
     bool has_task_list = 2;
     bool has_code = 3;
     bool has_incomplete_tasks = 4;
+    // The title extracted from the first H1 heading, if present.
+    string title = 5;
   }
 
   message Location {

server/router/api/v1/memo_service_converter.go

@@ -192,6 +192,7 @@ func convertMemoPropertyFromStore(property *storepb.MemoPayload_Property) *v1pb.
 		HasTaskList:        property.HasTaskList,
 		HasCode:            property.HasCode,
 		HasIncompleteTasks: property.HasIncompleteTasks,
+		Title:              property.Title,
 	}
 }